1 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * tree-core.h (tree_base::u): Add an "nelts" field.
6 (tree_vector): Use VECTOR_CST_NELTS as the length.
7 * tree.c (tree_size): Likewise.
8 (make_vector): Initialize VECTOR_CST_NELTS.
9 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13 as the number of units, setting it from VECTOR_CST_NELTS.
14 (const_vector_from_tree): Likewise.
15 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
16 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
17 (fold_negate_expr_1): Likewise.
18 (fold_convert_const): Likewise.
19 (const_binop): Likewise. Differentiate the number of output and
21 (const_unop): Likewise.
22 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
23 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
24 in cases that did the opposite.
26 2017-09-14 Richard Biener <rguenther@suse.de>
28 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
31 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
33 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
35 2017-09-14 Jakub Jelinek <jakub@redhat.com>
38 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
39 if and where to split a bb, except for splitting before debug insn
40 sequences followed by non-label real insn. Delete debug insns
41 in between basic blocks.
43 * combine.c (make_compound_operation_int): Formatting fixes.
45 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
46 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
47 * config/netbsd.h (LINK_EH_SPEC): Likewise.
48 * config/sol2.h (LINK_EH_SPEC): Likewise.
49 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
50 * config/s390/linux.h (LINK_SPEC): Likewise.
51 * config/freebsd.h (LINK_EH_SPEC): Likewise.
52 * config/openbsd.h (LINK_EH_SPEC): Likewise.
53 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
54 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
55 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
56 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
57 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
58 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
60 2017-09-13 Jakub Jelinek <jakub@redhat.com>
62 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
64 (ENDFILE_LINUX_SPEC): Likewise.
65 (LINK_EH_SPEC): Likewise.
66 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
67 (LINK_OS_LINUX_SPEC32): Likewise.
68 (LINK_OS_LINUX_SPEC64): Likewise.
69 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
70 (LINK_OS_LINUX_SPEC): Likewise.
72 2017-09-13 Martin Liska <mliska@suse.cz>
75 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
76 CASE_HIGH is NULL_TREE.
78 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
79 Alan Hayward <alan.hayward@arm.com>
80 David Sherwood <david.sherwood@arm.com>
82 * target.def (secondary_memory_needed): New hook.
83 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
84 instead of SECONDARY_MEMORY_NEEDED.
85 (secondary_memory_needed_mode): Likewise.
86 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
87 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
88 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
89 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
90 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
91 * doc/tm.texi: Regenerate.
92 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
93 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
94 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
95 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
96 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
97 * config/i386/i386.c (inline_secondary_memory_needed): Put the
98 mode argument first and change the reg_class arguments to reg_class_t.
99 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
100 Make static. Update the call to inline_secondary_memory_needed.
101 (ix86_register_move_cost): Update the call to
102 inline_secondary_memory_needed.
103 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
104 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
106 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
107 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
109 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
110 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
111 * config/mips/mips.c (mips_secondary_memory_needed): Make static
112 and match hook interface. Add comment from mips.h.
113 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
114 * config/mmix/mmix.md (truncdfsf2): Refer to
115 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
117 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
118 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
119 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
120 (pa_secondary_memory_needed): New function.
121 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
122 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
123 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
124 (pdp11_secondary_memory_needed): Make static and match hook interface.
125 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
126 * config/powerpcspe/powerpcspe-protos.h
127 (rs6000_secondary_memory_needed_ptr): Delete.
128 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
130 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
131 (rs6000_option_override_internal): Assign to
132 targetm.secondary_memory_needed rather than
133 rs6000_secondary_memory_needed_ptr.
134 (rs6000_secondary_memory_needed): Match hook interface.
135 (rs6000_debug_secondary_memory_needed): Likewise.
136 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
137 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
138 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
139 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
140 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
141 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
143 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
144 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
145 (rs6000_option_override_internal): Assign to
146 targetm.secondary_memory_needed rather than
147 rs6000_secondary_memory_needed_ptr.
148 (rs6000_secondary_memory_needed): Match hook interface.
149 (rs6000_debug_secondary_memory_needed): Likewise.
150 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
151 * config/s390/s390.c (s390_secondary_memory_needed): New function.
152 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
153 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
154 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
155 (sparc_secondary_memory_needed): New function.
156 * lra-constraints.c (check_and_process_move): Refer to
157 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
159 (curr_insn_transform): Likewise.
160 (process_alt_operands): Use targetm.secondary_memory_needed
161 instead of TARGET_SECONDARY_MEMORY_NEEDED.
162 (check_secondary_memory_needed_p): Likewise.
163 (choose_split_class): Likewise.
164 * reload.c: Unconditionally include code that was previously
165 conditional on SECONDARY_MEMORY_NEEDED.
166 (push_secondary_reload): Use targetm.secondary_memory_needed
167 instead of TARGET_SECONDARY_MEMORY_NEEDED.
168 (push_reload): Likewise.
169 * reload1.c: Unconditionally include code that was previously
170 conditional on SECONDARY_MEMORY_NEEDED.
171 (choose_reload_regs): Use targetm.secondary_memory_needed
172 instead of TARGET_SECONDARY_MEMORY_NEEDED.
173 (gen_reload): Likewise.
174 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
176 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
177 Alan Hayward <alan.hayward@arm.com>
178 David Sherwood <david.sherwood@arm.com>
180 * target.def (secondary_memory_needed_mode): New hook:
181 * targhooks.c (default_secondary_memory_needed_mode): Declare.
182 * targhooks.h (default_secondary_memory_needed_mode): New function.
183 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
184 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
185 * doc/tm.texi: Regenerate.
186 * lra-constraints.c (check_and_process_move): Use
187 targetm.secondary_memory_needed_mode instead of
188 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
189 (curr_insn_transform): Likewise.
190 * reload.c (get_secondary_mem): Likewise.
191 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
192 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
194 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
195 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
196 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
197 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
198 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
200 * config/powerpcspe/powerpcspe-protos.h
201 (rs6000_secondary_memory_needed_mode): Delete.
202 * config/powerpcspe/powerpcspe.c
203 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
204 (rs6000_secondary_memory_needed_mode): Make static.
205 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
206 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
208 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
210 (rs6000_secondary_memory_needed_mode): Make static.
211 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
212 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
213 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
214 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
215 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
217 (sparc_secondary_memory_needed_mode): New function.
218 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
220 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
222 * config/aarch64/constraints.md (Umq): New constraint.
223 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
225 (mov<mode>): Update condition.
227 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
229 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
230 when bitposition is the same.
232 2017-09-13 Richard Biener <rguenther@suse.de>
234 * dwarf2out.c (output_die_symbol): Remove.
235 (output_die): Do not output a DIEs symbol.
237 2017-09-13 Richard Biener <rguenther@suse.de>
240 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
241 default-def to avoid breaking iterator update with the weird
242 interaction with cgraph_update_edges_for_call_stmt_node.
244 2017-09-13 Richard Biener <rguenther@suse.de>
246 * tree-cfg.c (verify_gimple_assign_binary): Add verification
247 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
248 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
249 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
251 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
253 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
254 Disable pc relative literal load irrespective of
255 TARGET_FIX_ERR_A53_84341 for default.
257 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
259 * config/sparc/sparc.c (output_return): Output the source location of
260 the insn in the delay slot, if any.
261 (output_sibcall): Likewise.
263 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
266 * common.opt (-static-pie): New alias.
267 (shared): Negate static-pie.
268 (-no-pie): Update help text.
270 (static-pie): New option.
271 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
273 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
274 (LINK_EH_SPEC): Likewise.
275 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
276 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
277 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
278 * gcc.c (LINK_COMMAND_SPEC): Likewise.
279 (init_gcc_specs): Likewise.
280 (init_spec): Likewise.
281 (display_help): Update help message for -pie.
282 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
285 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
287 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
288 (movdi_aarch64): Likewise.
289 (movti_aarch64): Likewise.
291 2017-09-12 Simon Wright <simon@pushface.org>
294 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
295 calculation of the minor version, always output as 0.
297 2017-09-12 Jakub Jelinek <jakub@redhat.com>
300 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
301 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
302 on it early, rather than manual conversion late. For
303 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
304 instead of performing manual conversion.
306 2017-09-12 Carl Love <cel@us.ibm.com>
308 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
309 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
311 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
312 VMULOSW): Add definitions.
313 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
314 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
315 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
316 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
317 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
319 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
321 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
323 (movti_aarch64): Likewise.
324 (movdf_aarch64): Likewise.
325 (movtf_aarch64): Likewise.
326 (load_pairdi): Likewise.
327 (store_pairdi): Likewise.
328 (load_pairdf): Likewise.
329 (store_pairdf): Likewise.
330 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
331 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
332 (ldr_got_small_<mode>): Likewise.
333 (ldr_got_small_28k_<mode>): Likewise.
334 (ldr_got_tiny): Likewise.
335 * config/aarch64/iterators.md (ldst_sz): New.
336 (ldpstp_sz): Likewise.
337 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
339 (thunderx_load): Split load_8 to load_16.
340 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
342 (thunderx2t99_storepair_basic): Split store_8 to store_16.
343 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
344 (xgene1_store_pair): Split store_8 to store_16.
345 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
346 (falkor_st_0_st_sd): Split store_8 to store_16.
348 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
350 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
351 and store1/2/3/4 to store_4/8/12/16.
352 * config/aarch64/aarch64.md: Update for rename.
353 * config/arm/arm.md: Likewise.: Likewise.
354 * config/arm/arm.c: Likewise.
355 * config/arm/thumb1.md: Likewise.
356 * config/arm/thumb2.md: Likewise.
357 * config/arm/vfp.md: Likewise.
358 * config/arm/arm-generic.md: Likewise.
359 * config/arm/arm1020e.md: Likewise.
360 * config/arm/arm1026ejs.md: Likewise.
361 * config/arm/arm1136jfs.md: Likewise.
362 * config/arm/arm926ejs.md: Likewise.
363 * config/arm/cortex-a15.md: Likewise.
364 * config/arm/cortex-a17.md: Likewise.
365 * config/arm/cortex-a5.md: Likewise.
366 * config/arm/cortex-a53.md: Likewise.
367 * config/arm/cortex-a57.md: Likewise.
368 * config/arm/cortex-a7.md: Likewise.
369 * config/arm/cortex-a8.md: Likewise.
370 * config/arm/cortex-a9.md: Likewise.
371 * config/arm/cortex-m4.md: Likewise.
372 * config/arm/cortex-m7.md: Likewise.
373 * config/arm/cortex-r4.md: Likewise.
374 * config/arm/exynos-m1.md: Likewise.
375 * config/arm/fa526.md: Likewise.
376 * config/arm/fa606te.md: Likewise.
377 * config/arm/fa626te.md: Likewise.
378 * config/arm/fa726te.md: Likewise.
379 * config/arm/fmp626.md: Likewise.
380 * config/arm/iwmmxt.md: Likewise.
381 * config/arm/ldmstm.md: Likewise.
382 * config/arm/marvell-pj4.md: Likewise.
383 * config/arm/xgene1.md: Likewise.
384 * config/aarch64/thunderx.md: Likewise.
385 * config/aarch64/thunderx2t99.md: Likewise.
386 * config/aarch64/falkor.md: Likewise.
388 2017-09-12 Martin Liska <mliska@suse.cz>
390 * attribs.c (private_lookup_attribute): New function.
391 * attribs.h (private_lookup_attribute): Declared here.
392 (lookup_attribute): Called from this place.
394 2017-09-12 Richard Biener <rguenther@suse.de>
396 PR tree-optimization/82157
397 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
398 stmts with side-effects.
400 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
401 Alan Hayward <alan.hayward@arm.com>
402 David Sherwood <david.sherwood@arm.com>
404 * target.def (hard_regno_nregs): New hook.
405 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
406 * targhooks.h (default_hard_regno_nregs): Declare.
407 * targhooks.c (default_hard_regno_nregs): New function.
408 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
409 (TARGET_HARD_REGNO_NREGS): ...this hook.
410 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
411 (CLASS_MAX_NREGS): Likewise.
412 * doc/tm.texi: Regenerate.
413 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
414 instead of HARD_REGNO_NREGS.
415 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
416 HARD_REGNO_NREGS in the comment.
417 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
418 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
419 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
420 Return an unsigned int.
421 (TARGET_HARD_REGNO_NREGS): Redefine.
422 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
423 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
424 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
425 (arc_hard_regno_nregs): New function.
426 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
427 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
428 (arm_hard_regno_nregs): New function.
429 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
430 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
431 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
432 (TARGET_HARD_REGNO_NREGS): Redefine.
433 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
434 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
435 (HARD_REGNO_NREGS): Delete.
436 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
437 (cr16_hard_regno_nregs): New function.
438 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
439 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
440 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
441 (cris_hard_regno_nregs): New function.
442 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
443 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
444 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
445 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
446 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
447 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
448 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
449 (frv_hard_regno_nregs): Make static. Take and return an
451 (frv_class_max_nregs): Remove outdated copy of documentation.
452 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
453 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
454 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
455 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
456 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
457 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
458 (TARGET_HARD_REGNO_NREGS): Redefine.
459 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
460 (CLASS_MAX_NREGS): Update comment.
461 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
462 (ia64_hard_regno_nregs): New function.
463 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
464 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
465 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
466 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
467 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
469 (m32c_hard_regno_nregs): Likewise. Make static.
470 (TARGET_HARD_REGNO_NREGS): Redefine.
471 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
472 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
473 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
474 (m68k_hard_regno_nregs): New function.
475 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
476 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
477 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
478 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
479 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
480 Take and return an unsigned int.
481 (TARGET_HARD_REGNO_NREGS): Redefine.
482 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
483 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
484 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
485 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
486 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
487 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
488 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
489 (msp430_hard_regno_nregs): Make static. Take and return an
491 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
492 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
493 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
494 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
495 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
496 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
497 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
498 (TARGET_HARD_REGNO_NREGS): Redefine.
499 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
500 (PA_HARD_REGNO_NREGS): ...this.
501 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
502 (PA_HARD_REGNO_NREGS): ...this.
503 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
504 (pa_hard_regno_nregs): New function.
505 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
506 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
507 (pdp11_hard_regno_nregs): New function.
508 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
509 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
510 (rs6000_hard_regno_nregs_hook): New function.
511 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
512 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
513 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
514 Take and return an unsigned int. Move earlier in file.
515 (TARGET_HARD_REGNO_NREGS): Redefine.
516 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
517 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
518 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
519 (rl78_hard_regno_nregs): Make static. Take and return an
521 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
522 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
523 (rs6000_hard_regno_nregs_hook): New function.
524 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
525 * config/rx/rx.c (rx_hard_regno_nregs): New function.
526 (TARGET_HARD_REGNO_NREGS): Redefine.
527 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
528 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
529 instead of HARD_REGNO_NREGS.
530 (s390_hard_regno_nregs): New function.
531 (s390_hard_regno_mode_ok): Add comment from s390.h.
532 (TARGET_HARD_REGNO_NREGS): Redefine.
533 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
534 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
535 (sh_hard_regno_nregs): New function.
536 (sh_pass_in_reg_p): Use it.
537 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
538 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
539 (sparc_hard_regno_nregs): New function.
540 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
541 * config/spu/spu.c (spu_hard_regno_nregs): New function.
542 (spu_function_arg_advance): Use it, supplying a valid register number.
543 (TARGET_HARD_REGNO_NREGS): Redefine.
544 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
545 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
546 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
547 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
548 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
549 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
550 (CLASS_MAX_NREGS): Remove copy of old documentation.
551 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
552 (visium_hard_regno_nregs): New function.
553 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
554 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
555 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
556 xtensa_hard_regno_nregs): New function.
557 * system.h (HARD_REGNO_NREGS): Poison.
559 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
561 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
562 hard_regno_nregs instead of HARD_REGNO_NREGS.
563 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
564 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
565 (c6x_expand_epilogue): Likewise.
566 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
567 (frv_read_iacc_argument): Likewise.
568 * config/sh/sh.c: Include regs.h.
569 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
570 (regs_used): Likewise.
571 (output_stack_adjust): Likewise.
572 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
573 * expmed.c: Include regs.h.
574 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
575 * ree.c: Include regs.h.
576 (combine_reaching_defs): Use hard_regno_nregs instead of
578 (add_removable_extension): Likewise.
580 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
582 * regs.h (hard_regno_nregs): Turn into a function.
583 (end_hard_regno): Update accordingly.
584 * caller-save.c (setup_save_areas): Likewise.
585 (save_call_clobbered_regs): Likewise.
586 (replace_reg_with_saved_mem): Likewise.
587 (insert_restore): Likewise.
588 (insert_save): Likewise.
589 * combine.c (can_change_dest_mode): Likewise.
590 (move_deaths): Likewise.
591 (distribute_notes): Likewise.
592 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
593 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
594 (rs6000_split_multireg_move): Likewise.
595 (rs6000_register_move_cost): Likewise.
596 (rs6000_memory_move_cost): Likewise.
597 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
598 (rs6000_split_multireg_move): Likewise.
599 (rs6000_register_move_cost): Likewise.
600 (rs6000_memory_move_cost): Likewise.
601 * cselib.c (cselib_reset_table): Likewise.
602 (cselib_lookup_1): Likewise.
603 * emit-rtl.c (set_mode_and_regno): Likewise.
604 * function.c (aggregate_value_p): Likewise.
605 * ira-color.c (setup_profitable_hard_regs): Likewise.
606 (check_hard_reg_p): Likewise.
607 (calculate_saved_nregs): Likewise.
608 (assign_hard_reg): Likewise.
609 (improve_allocation): Likewise.
610 (calculate_spill_cost): Likewise.
611 * ira-emit.c (modify_move_list): Likewise.
612 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
613 (ira_hard_reg_in_set_p): Likewise.
614 * ira.c (setup_reg_mode_hard_regset): Likewise.
615 (clarify_prohibited_class_mode_regs): Likewise.
616 (check_allocation): Likewise.
617 * lra-assigns.c (find_hard_regno_for_1): Likewise.
618 (lra_setup_reg_renumber): Likewise.
619 (setup_try_hard_regno_pseudos): Likewise.
620 (spill_for): Likewise.
621 (assign_hard_regno): Likewise.
622 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
623 * lra-constraints.c (in_class_p): Likewise.
624 (lra_constraint_offset): Likewise.
625 (simplify_operand_subreg): Likewise.
626 (lra_constraints): Likewise.
627 (split_reg): Likewise.
628 (split_if_necessary): Likewise.
629 (invariant_p): Likewise.
630 (inherit_in_ebb): Likewise.
631 * lra-lives.c (process_bb_lives): Likewise.
632 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
633 (get_hard_regs): Likewise.
634 (do_remat): Likewise.
635 * lra-spills.c (assign_spill_hard_regs): Likewise.
636 * mode-switching.c (create_pre_exit): Likewise.
637 * postreload.c (reload_combine_recognize_pattern): Likewise.
638 * recog.c (peep2_find_free_register): Likewise.
639 * regcprop.c (kill_value_regno): Likewise.
640 (set_value_regno): Likewise.
641 (copy_value): Likewise.
642 (maybe_mode_change): Likewise.
643 (find_oldest_value_reg): Likewise.
644 (copyprop_hardreg_forward_1): Likewise.
645 * regrename.c (check_new_reg_p): Likewise.
646 (regrename_do_replace): Likewise.
647 * reload.c (push_reload): Likewise.
648 (combine_reloads): Likewise.
649 (find_dummy_reload): Likewise.
650 (operands_match_p): Likewise.
651 (find_reloads): Likewise.
652 (find_equiv_reg): Likewise.
653 (reload_adjust_reg_for_mode): Likewise.
654 * reload1.c (count_pseudo): Likewise.
655 (count_spilled_pseudo): Likewise.
656 (find_reg): Likewise.
657 (clear_reload_reg_in_use): Likewise.
658 (free_for_value_p): Likewise.
659 (allocate_reload_reg): Likewise.
660 (choose_reload_regs): Likewise.
661 (reload_adjust_reg_for_temp): Likewise.
662 (emit_reload_insns): Likewise.
663 (delete_output_reload): Likewise.
664 * rtlanal.c (subreg_get_info): Likewise.
665 * sched-deps.c (sched_analyze_reg): Likewise.
666 * sel-sched.c (init_regs_for_mode): Likewise.
667 (mark_unavailable_hard_regs): Likewise.
668 (choose_best_reg_1): Likewise.
669 (verify_target_availability): Likewise.
670 * valtrack.c (dead_debug_insert_temp): Likewise.
671 * var-tracking.c (track_loc_p): Likewise.
672 (emit_note_insn_var_location): Likewise.
673 * varasm.c (make_decl_rtl): Likewise.
674 * reginfo.c (choose_hard_reg_mode): Likewise.
675 (init_reg_modes_target): Refer directly to
676 this_target_regs->x_hard_regno_nregs.
678 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
680 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
681 instead of hard_regno_nregs.
683 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
685 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
686 end_hard_regno instead of hard_regno_nregs.
687 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
688 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
689 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
690 * ira-color.c (improve_allocation): Likewise.
691 * lra-assigns.c (find_hard_regno_for_1): Likewise.
692 * lra-lives.c (mark_regno_live): Likewise.
693 (mark_regno_dead): Likewise.
694 * lra-remat.c (operand_to_remat): Likewise.
695 * lra.c (collect_non_operand_hard_regs): Likewise.
696 * postreload.c (reload_combine_note_store): Likewise.
697 (move2add_valid_value_p): Likewise.
698 * reload.c (regno_clobbered_p): Likewise.
700 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
702 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
704 * config/v850/v850.c (v850_reorg): Likewise.
705 * reload.c (refers_to_regno_for_reload_p): Likewise.
706 (find_equiv_reg): Likewise.
707 * reload1.c (reload_reg_reaches_end_p): Likewise.
709 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
711 * caller-save.c (add_used_regs): Use REG_NREGS instead of
713 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
714 * config/arm/arm.c (output_move_neon): Likewise.
715 (arm_attr_length_move_neon): Likewise.
716 (neon_split_vcombine): Likewise.
717 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
718 (c6x_mark_reg_written): Likewise.
719 (c6x_dwarf_register_span): Likewise.
720 * config/i386/i386.c (ix86_save_reg): Likewise.
721 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
722 (rws_access_reg): Likewise.
723 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
724 * mode-switching.c (create_pre_exit): Likewise.
725 * ree.c (combine_reaching_defs): Likewise.
726 (add_removable_extension): Likewise.
727 * regcprop.c (find_oldest_value_reg): Likewise.
728 (copyprop_hardreg_forward_1): Likewise.
729 * reload.c (reload_inner_reg_of_subreg): Likewise.
730 (push_reload): Likewise.
731 (combine_reloads): Likewise.
732 (find_dummy_reload): Likewise.
733 (reload_adjust_reg_for_mode): Likewise.
734 * reload1.c (find_reload_regs): Likewise.
735 (forget_old_reloads_1): Likewise.
736 (reload_reg_free_for_value_p): Likewise.
737 (reload_adjust_reg_for_temp): Likewise.
738 (emit_reload_insns): Likewise.
739 (delete_output_reload): Likewise.
740 * sel-sched.c (choose_best_reg_1): Likewise.
741 (choose_best_pseudo_reg): Likewise.
743 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
744 Alan Hayward <alan.hayward@arm.com>
745 David Sherwood <david.sherwood@arm.com>
747 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
748 * target.def (slow_unaligned_access): New hook.
749 * targhooks.h (default_slow_unaligned_access): Declare.
750 * targhooks.c (default_slow_unaligned_access): New function.
751 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
752 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
753 * doc/tm.texi: Regenerate.
754 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
755 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
756 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
758 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
759 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
761 (rs6000_slow_unaligned_access): New function.
762 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
763 (expand_block_compare): Likewise.
764 (expand_strn_compare): Likewise.
765 (rs6000_rtx_costs): Likewise.
766 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
767 (riscv_slow_unaligned_access): Likewise.
768 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
769 (riscv_slow_unaligned_access_p): ...this and make static.
770 (riscv_option_override): Update accordingly.
771 (riscv_slow_unaligned_access): New function.
772 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
773 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
774 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
775 (rs6000_slow_unaligned_access): New function.
776 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
777 (rs6000_rtx_costs): Likewise.
778 * config/rs6000/rs6000-string.c (expand_block_compare)
779 (expand_strn_compare): Use targetm.slow_unaligned_access instead
780 of SLOW_UNALIGNED_ACCESS.
781 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
782 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
783 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
784 of SLOW_UNALIGNED_ACCESS.
785 * expmed.c (simple_mem_bitfield_p): Likewise.
786 * expr.c (alignment_for_piecewise_move): Likewise.
787 (emit_group_load_1): Likewise.
788 (emit_group_store): Likewise.
789 (copy_blkmode_from_reg): Likewise.
790 (emit_push_insn): Likewise.
791 (expand_assignment): Likewise.
792 (store_field): Likewise.
793 (expand_expr_real_1): Likewise.
794 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
795 * lra-constraints.c (simplify_operand_subreg): Likewise.
796 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
797 * gimple-ssa-store-merging.c: Likewise in block comment at start
799 * tree-ssa-strlen.c: Include target.h.
800 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
801 of SLOW_UNALIGNED_ACCESS.
802 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
804 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
806 PR rtl-optimization/82185
807 * expmed.c (emit_store_flag_int): Only test tem if it has been
810 2017-09-12 Richard Biener <rguenther@suse.de>
813 * match.pd ((FTYPE) N CMP CST): Fix typo.
815 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
817 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
819 (mips_near_type_p): Add 'short_call' attribute as a synonym
821 * doc/extend.texi (short_call): Document new function attribute.
823 2017-09-12 Jakub Jelinek <jakub@redhat.com>
826 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
827 assertion check that in the condition.
828 (get_atomic_generic_size): Likewise. Before testing if parameter
829 has pointer type, if it has array type, call for C++
830 default_conversion to perform array-to-pointer conversion.
832 2017-09-12 Richard Biener <rguenther@suse.de>
834 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
835 for operations we cannot scalarize.
837 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
839 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
840 vectors heap vectors. Clean up comments.
841 Make visited_bbs a reference.
842 (profitable_jump_thread_path): Make GC
843 vectors heap vectors. Clean up comments.
845 (convert_and_register_jump_thread_path): Make GC vectors heap
847 (check_subpath_and_update_thread_path): Same. Clean up comments.
848 Make visited_bbs a reference.
849 (handle_phi): Abstract common code to to
850 register_jump_thread_path_if_profitable.
851 Rename VAR_BB to DEF_BB.
853 Make GC vectors heap vectors.
854 Make visited_bbs a reference.
855 (handle_assignment): Same.
856 (register_jump_thread_path_if_profitable): New.
857 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
859 Make GC vectors heap vectors. Clean up comments.
860 Make visited_bbs a reference.
861 (find_jump_threads_backwards): Make visited_bbs live in the stack.
862 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
865 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
868 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
869 words of E_DImode object are reachable by xtensa_uimm8x4 access.
871 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
873 Revert r251800 and r251799.
875 2017-09-11 Martin Jambor <mjambor@suse.cz>
878 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
879 arguments in advance.
880 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
881 use it to find predecessor edges.
882 (naive_outof_ssa): Collect vector of predecessors.
884 2017-09-08 Jason Merrill <jason@redhat.com>
886 PR c++/70029 - ICE with ref-qualifier and -flto
887 * langhooks.h (struct lang_hooks_for_types): Add
888 copy_lang_qualifiers.
889 * attribs.c (build_type_attribute_qual_variant): Use it.
890 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
892 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
893 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
895 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
898 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
899 (*mulsi3_sp64): New instruction.
900 (mulsi3): New expander.
902 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
904 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
906 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
908 * sancov.c: Include memmodel.h.
910 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
913 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
916 2017-09-07 Carl Love <cel@us.ibm.com>
918 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
919 the sldi instruction.
921 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
923 * sancov.c: Include tm_p.h.
925 2017-09-07 Jakub Jelinek <jakub@redhat.com>
928 * output.h (switch_to_other_text_partition): New declaration.
929 * varasm.c (switch_to_other_text_partition): New function.
930 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
931 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
932 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
933 to the other text partition before emitting LCL label and switch back
934 after emitting the word after it.
936 2017-09-07 Richard Biener <rguenther@suse.de>
938 * passes.def (pass_split_crit_edges): Remove instance before PRE.
939 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
940 critical edges here, after loop init.
941 (pass_data_pre): Remove PROP_no_crit_edges flags.
942 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
943 for valueization of call args to avoid leaking VN_TOP.
944 (visit_use): Assert we do not visit default defs.
945 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
946 Use error_mark_node to more easily detect leaking VN_TOP.
947 All default-defs are varying, not VN_TOP. Mark them visited.
948 (run_scc_vn): Make code match comment.
950 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
952 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
953 OPTION_MASK_FLOAT128_KEYWORD.
954 (POWERPC_MASKS): Likewise.
955 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
956 support for the -mfloat128-type option, and make -mfloat128
957 default on PowerPC Linux systems. Define or undefine
958 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
959 Define __float128 to be __ieee128 if IEEE 128-bit support is
960 enabled, or undefine it.
961 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
962 Delete defining __FLOAT128_TYPE__.
963 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
964 -mfloat128-type option and make -mfloat128 default on PowerPC
966 (TARGET_FLOAT128_TYPE): Likewise.
967 (-mfloat128-type): Likewise.
968 * config/rs6000/rs6000.c (rs6000_option_override_internal):
969 Delete the -mfloat128-type option and make -mfloat128 default on
970 PowerPC Linux systems. Always use __ieee128 to be the keyword for
971 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
972 128-bit floating point is enabled. Change tests from using
973 -mfloat128-type to -mfloat128.
974 (rs6000_mangle_type): Use the correct mangling for the __float128
975 type even if normal long double is restricted to 64-bits.
976 (floatn_mode): Enable the _Float128 type by default on VSX Linux
978 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
979 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
980 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
982 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
983 documentation for -mfloat128.
985 2017-09-06 Olivier Hainque <hainque@adacore.com>
987 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
989 2017-09-06 Wish Wu <wishwu007@gmail.com>
990 Jakub Jelinek <jakub@redhat.com>
992 * asan.c (initialize_sanitizer_builtins): Add
993 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
994 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
995 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
996 BT_FN_VOID_UINT64_PTR variables.
997 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
998 (BT_FN_VOID_UINT16_UINT16): Likewise.
999 (BT_FN_VOID_UINT32_UINT32): Likewise.
1000 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1001 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1002 (BT_FN_VOID_UINT64_PTR): Likewise.
1003 * common.opt (flag_sanitize_coverage): New variable.
1004 (fsanitize-coverage=trace-pc): Remove.
1005 (fsanitize-coverage=): Add.
1006 * flag-types.h (enum sanitize_coverage_code): New enum.
1007 * fold-const.c (fold_range_test): Disable non-short-circuit
1008 optimization if flag_sanitize_coverage.
1009 (fold_truth_andor): Likewise.
1010 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1011 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1012 (coverage_sanitizer_opts): New array.
1013 (get_closest_sanitizer_option): Add OPTS argument, handle also
1014 OPT_fsanitize_coverage_.
1015 (parse_sanitizer_options): Adjusted to also handle
1016 OPT_fsanitize_coverage_.
1017 (common_handle_option): Add OPT_fsanitize_coverage_.
1018 * sancov.c (instrument_comparison, instrument_switch): New function.
1019 (sancov_pass): Add trace-cmp support.
1020 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1021 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1022 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1023 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1024 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1025 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1026 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1027 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1028 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1029 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1031 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1033 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1034 error. Only quit immediately if parsing is complete.
1035 (BEGIN): Initialize fatal_err and parse_done.
1036 (begin fpu, end fpu): Check number of arguments.
1037 (begin arch, end arch): Likewise.
1038 (begin cpu, end cpu): Likewise.
1039 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1040 (optalias): Likewise.
1042 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1044 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1045 * config/arm/arm-isa.h: Delete. Move definitions to ...
1046 * arm-cpus.in: ... here. Use new feature and fgroup values.
1047 * config/arm/arm.c (arm_option_override): Use lower case for feature
1049 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1050 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1051 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1053 (print_isa_bits_for): New function.
1054 (gen_isa): New function.
1055 (gen_comm_data): Use print_isa_bits_for.
1056 (define feature): New keyword.
1057 (define fgroup): New keyword.
1058 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1059 (arm-isa.h): Add rule to generate file.
1060 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1061 case for feature bit names.
1063 2017-09-06 Richard Biener <rguenther@suse.de>
1065 * tree-ssa-pre.c (NECESSARY): Remove.
1066 (create_expression_by_pieces): Do not touch pass-local flags.
1067 (insert_into_preds_of_block): Likewise.
1068 (do_pre_regular_insertion): Likewise.
1069 (eliminate_insert): Likewise.
1070 (eliminate_dom_walker::before_dom_children): Likewise.
1071 (fini_eliminate): Do not look at inserted_exprs.
1072 (mark_operand_necessary): Remove.
1073 (remove_dead_inserted_code): Replace with simple work-list
1074 algorithm based on inserted_exprs and SSA uses.
1075 (pass_pre::execute): Re-order fini_eliminate and
1076 remove_dead_inserted_code.
1078 2017-09-06 Olivier Hainque <hainque@adacore.com>
1080 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1081 for VxWorks 7. Adjust surrounding comments.
1083 2017-09-06 Richard Biener <rguenther@suse.de>
1085 * gimple-ssa-strength-reduction.c
1086 (find_candidates_dom_walker::before_dom_children): Also allow
1089 2017-09-06 Richard Biener <rguenther@suse.de>
1091 PR tree-optimization/82108
1092 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1093 for gap in the non-permutation SLP case.
1095 2017-09-06 Martin Jambor <mjambor@suse.cz>
1097 PR tree-optimization/82078
1098 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1099 add_access_to_work_queue...
1100 (build_accesses_from_assign): ...here.
1101 (propagate_all_subaccesses): Make sure racc is the group
1102 representative, if there is one.
1104 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1107 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1110 2017-09-06 Richard Biener <rguenther@suse.de>
1112 * gimple-ssa-strength-reduction.c
1113 (find_candidates_dom_walker::before_doom_children): Use a
1114 type and not a mode check.
1116 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1119 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1121 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1123 2017-09-05 Jeff Law <law@redhat.com>
1125 PR tree-optimization/64910
1126 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1127 cases where we have 3 or more operands.
1129 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1132 * omp-low.c (lower_omp_for): Recompute tree invariant if
1133 gimple_omp_for_initial/final is ADDR_EXPR.
1136 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1137 into gimple val before gimplification fo the COND_EXPR.
1139 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1141 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1142 REGION_COPY argument.
1143 (thread_through_all_blocks): Remove unused argument to
1144 duplicate_thread_path.
1146 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1147 Alan Hayward <alan.hayward@arm.com>
1148 David Sherwood <david.sherwood@arm.com>
1150 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1151 Take a scalar_mode rather than a machine_mode.
1152 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1153 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1154 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1155 (aarch64_gen_adjusted_ldpstp): Likewise.
1156 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1158 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1163 Take a scalar_int_mode instead of a machine_mode.
1164 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1165 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1166 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1167 (aarch64_simd_attr_length_rglist): Delete.
1168 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1169 a scalar_int_mode instead of a machine_mode.
1170 (aarch64_add_offset): Likewise.
1171 (aarch64_internal_mov_immediate): Likewise
1172 (aarch64_add_constant_internal): Likewise.
1173 (aarch64_add_constant): Likewise.
1174 (aarch64_movw_imm): Likewise.
1175 (aarch64_rtx_arith_op_extract_p): Likewise.
1176 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1177 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1178 Remove assert that the mode isn't a vector.
1179 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1180 (aarch64_expand_mov_immediate): Update calls after above changes.
1181 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1182 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1183 (aarch64_move_imm): Likewise.
1184 (aarch64_can_const_movi_rtx_p): Likewise.
1185 (aarch64_strip_extend): Likewise.
1186 (aarch64_extr_rtx_p): Likewise.
1187 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1188 a CONST_INT when the mode parameter is VOIDmode.
1189 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1191 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1193 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1194 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1195 (bitwise_type_for_mode): Update accordingly.
1197 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1199 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1200 * stor-layout.c (mode_for_size_tree): Likewise.
1201 (mode_for_array): Update accordingly.
1202 (layout_decl): Likewise.
1203 (compute_record_mode): Likewise. Only set the mode once.
1205 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1207 * target.def (get_mask_mode): Change return type to opt_mode.
1209 * doc/tm.texi: Regenerate.
1210 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1211 * targhooks.c (default_get_mask_mode): Likewise.
1212 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1213 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1214 targetm.get_mask_mode.
1215 * tree.c (build_truth_vector_type): Likewise.
1217 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1219 * machmode.h (mode_for_vector): Return an opt_mode.
1220 * stor-layout.c (mode_for_vector): Likewise.
1221 (mode_for_int_vector): Update accordingly.
1222 (layout_type): Likewise.
1223 * config/i386/i386.c (emit_memmov): Likewise.
1224 (ix86_expand_set_or_movmem): Likewise.
1225 (ix86_expand_vector_init): Likewise.
1226 (ix86_get_mask_mode): Likewise.
1227 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1229 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1230 * expmed.c (extract_bit_field_1): Likewise.
1231 * expr.c (expand_expr_real_2): Likewise.
1232 * optabs-query.c (can_vec_perm_p): Likewise.
1233 (can_vec_mask_load_store_p): Likewise.
1234 * optabs.c (expand_vec_perm): Likewise.
1235 * targhooks.c (default_get_mask_mode): Likewise.
1236 * tree-vect-stmts.c (vectorizable_store): Likewise.
1237 (vectorizable_load): Likewise.
1238 (get_vectype_for_scalar_type_and_size): Likewise.
1240 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1242 * machmode.h (mode_for_int_vector): New function.
1243 * stor-layout.c (mode_for_int_vector): Likewise.
1244 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1245 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1246 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1247 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1248 (s390_expand_vcond): Likewise.
1250 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1252 * machmode.h (opt_machine_mode): New type.
1253 (opt_mode<T>): Allow construction from anything that can be
1255 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1256 (mode_for_size): Return an opt_machine_mode.
1257 * stor-layout.c (mode_for_size): Likewise.
1258 (mode_for_size_tree): Update call accordingly.
1259 (bitwise_mode_for_mode): Likewise.
1260 (make_fract_type): Likewise.
1261 (make_accum_type): Likewise.
1262 * caller-save.c (replace_reg_with_saved_mem): Update call
1264 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1265 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1266 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1267 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1268 * expmed.c (extract_bit_field_1): Likewise.
1269 * reload.c (get_secondary_mem): Likewise.
1270 * varasm.c (assemble_integer): Likewise.
1271 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1274 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1276 * machmode.h (decimal_float_mode_for_size): New function.
1277 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1278 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1279 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1280 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1281 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1282 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1284 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1286 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1287 (get_builtin_sync_mode): Likewise.
1288 (expand_ifn_atomic_compare_exchange): Likewise.
1289 (expand_builtin_atomic_clear): Likewise.
1290 (expand_builtin_atomic_test_and_set): Likewise.
1291 (fold_builtin_atomic_always_lock_free): Likewise.
1292 * calls.c (compute_argument_addresses): Likewise.
1293 (emit_library_call_value_1): Likewise.
1294 (store_one_arg): Likewise.
1295 * combine.c (combine_instructions): Likewise.
1296 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1297 * config/arm/arm.c (arm_function_value): Likewise.
1298 (aapcs_allocate_return_reg): Likewise.
1299 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1300 * config/i386/i386.c (construct_container): Likewise.
1301 (ix86_gimplify_va_arg): Likewise.
1302 (ix86_expand_sse_cmp): Likewise.
1303 (emit_memmov): Likewise.
1304 (emit_memset): Likewise.
1305 (expand_small_movmem_or_setmem): Likewise.
1306 (ix86_expand_pextr): Likewise.
1307 (ix86_expand_pinsr): Likewise.
1308 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1309 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1311 * config/mips/mips.c (mips_function_value_1) Likewise.
1312 (mips_block_move_straight): Likewise.
1313 (mips_expand_ins_as_unaligned_store): Likewise.
1314 * config/powerpcspe/powerpcspe.c
1315 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1316 (rs6000_darwin64_record_arg_flush): Likewise.
1317 * config/rs6000/rs6000.c
1318 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1319 (rs6000_darwin64_record_arg_flush): Likewise.
1320 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1321 (sparc_function_value_1): Likewise.
1322 * config/spu/spu.c (adjust_operand): Likewise.
1323 (spu_emit_branch_or_set): Likewise.
1324 (arith_immediate_p): Likewise.
1325 * emit-rtl.c (gen_lowpart_common): Likewise.
1326 * expr.c (expand_expr_real_1): Likewise.
1327 * function.c (assign_parm_setup_block): Likewise.
1328 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1329 * reload1.c (alter_reg): Likewise.
1330 * stor-layout.c (mode_for_vector): Likewise.
1331 (layout_type): Likewise.
1333 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1335 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1336 (spu_convert_move): Likewise.
1337 * lower-subreg.c (resolve_simple_move): Likewise.
1339 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1342 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1343 define_insn to a define_expand.
1344 (altivec_vsum2sws_direct): New define_insn.
1345 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1347 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1349 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1352 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1357 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1358 realign stack if stack alignment needed is less than incoming
1361 2017-09-05 Marek Polacek <polacek@redhat.com>
1364 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1367 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1369 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1371 2017-09-05 Richard Biener <rguenther@suse.de>
1373 PR tree-optimization/82084
1374 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1376 2017-09-05 Richard Biener <rguenther@suse.de>
1378 PR tree-optimization/82102
1379 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1381 2017-09-05 Martin Liska <mliska@suse.cz>
1383 PR tree-optimization/82032
1384 * tree-cfg.c (generate_range_test): New function.
1385 * tree-cfg.h (generate_range_test): Declared here.
1386 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1387 (cleanup_control_expr_graph): Use it.
1388 * tree-switch-conversion.c (try_switch_expansion): Remove
1390 (emit_case_nodes): Use generate_range_test.
1392 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1395 * config/i386/i386.md (*<btsc><mode>_mask): Add
1396 TARGET_USE_BT to insn constraint.
1397 (*btr<mode>_mask): Ditto.
1399 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1401 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1402 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1404 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1407 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1408 TARGET_NEON and TARGET_IWMMXT.
1409 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1410 TARGET_NEON and TARGET_IWMMXT.
1411 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1413 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1415 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1416 (ix86_rewrite_tls_address): Ditto.
1417 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1418 (ix86_rewrite_tls_address_1): Ditto.
1419 (ix86_rewrite_tls_address): Ditto.
1420 * config/i386/predicates.md (tls_address_pattern): New predicate.
1421 * config/i386/i386.md (TLS address splitter): New splitter.
1423 2017-09-04 Richard Biener <rguenther@suse.de>
1425 PR tree-optimization/82084
1426 * fold-const.h (can_native_encode_string_p): Declare.
1427 * fold-const.c (can_native_encode_string_p): Factor out from ...
1428 (native_encode_string): ... here.
1429 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1430 vectorizing stores from constants we later cannot handle.
1432 2017-09-04 Marek Polacek <polacek@redhat.com>
1435 * doc/invoke.texi: Update -Wtautological-compare documentation.
1437 2017-09-04 Jeff Law <law@redhat.com>
1439 PR tree-optimization/64910
1440 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1441 swap the first and last operand if the last is a constant.
1443 2017-09-04 Marek Polacek <polacek@redhat.com>
1446 * convert.c (do_narrow): When sanitizing signed integer overflows,
1447 bail out for signed types.
1448 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1450 2017-09-04 Richard Biener <rguenther@suse.de>
1452 PR tree-optimization/82060
1453 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1454 Move devirtualization after stmt folding and before EH/AB/noreturn
1455 cleanup to get the stmt refs canonicalized. Use a bool instead
1456 of gimple_modified_p since that doesn't work for NOPs. Schedule
1457 NOPs generated by folding for removal.
1459 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1460 Alan Hayward <alan.hayward@arm.com>
1461 David Sherwood <david.sherwood@arm.com>
1463 * coretypes.h (pad_direction): New enum.
1464 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1465 (FUNCTION_ARG_PADDING): Likewise.
1466 * target.def (function_arg_padding): New hook.
1467 * targhooks.h (default_function_arg_padding): Declare.
1468 * targhooks.c (default_function_arg_padding): New function.
1469 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1470 (TARGET_FUNCTION_ARG_PADDING): ...this.
1471 * doc/tm.texi: Regenerate.
1472 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1473 instead of direction.
1474 (compute_argument_addresses): Likewise.
1475 (load_register_parameters): Likewise.
1476 (emit_library_call_value_1): Likewise.
1477 (store_one_arg): Use targetm.calls.function_arg_padding instead
1478 of FUNCTION_ARG_PADDING.
1479 (must_pass_in_stack_var_size_or_pad): Likewise.
1480 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1481 (emit_group_store): Likewise.
1482 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1483 instead of FUNCTION_ARG_PADDING.
1484 (emit_push_insn): Likewise, and propagate enum change throughout
1486 * function.h (direction): Delete.
1487 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1489 * function.c (assign_parm_find_stack_rtl): Likewise.
1490 (assign_parm_setup_block_p): Likewise.
1491 (assign_parm_setup_block): Likewise.
1492 (gimplify_parameters): Likewise.
1493 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1494 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1496 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1497 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1498 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1499 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1500 (aarch64_function_arg_padding): ...this new function.
1501 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1502 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1503 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1504 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1505 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1506 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1507 (arm_pad_arg_upward): Replace with...
1508 (arm_function_arg_padding): ...this new function.
1509 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1511 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1512 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1513 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1514 (ia64_hpux_function_arg_padding): Replace with...
1515 (ia64_function_arg_padding): ...this new function. Use pad_direction
1516 instead of direction. Check for TARGET_HPUX.
1517 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1518 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1519 (iq2000_function_arg_padding): New function.
1520 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1521 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1522 (mips_function_arg_padding): ...this new function.
1523 (mips_pad_reg_upward): Update accordingly.
1524 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1525 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1526 targetm.calls.function_arg_padding.
1527 (FUNCTION_ARG_PADDING): Delete.
1528 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1529 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1530 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1531 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1532 (nios2_block_reg_padding): Return pad_direction instead of direction.
1533 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1534 instead of direction.
1535 (nios2_function_arg_padding): Likewise. Make static.
1536 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1537 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1538 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1539 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1540 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1541 (pa_function_arg_padding): Make static. Return pad_direction instead
1543 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1544 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1545 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1546 instead of direction. Use targetm.calls.function_arg_padding.
1547 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1548 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1549 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1550 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1551 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1553 (function_arg_padding): Rename to...
1554 (rs6000_function_arg_padding): ...this. Make static. Return
1555 pad_direction instead of direction.
1556 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1557 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1558 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1559 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1560 instead of direction. Use targetm.calls.function_arg_padding.
1561 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1562 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1563 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1564 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1565 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1566 (function_arg_padding): Rename to...
1567 (rs6000_function_arg_padding): ...this. Make static. Return
1568 pad_direction instead of direction.
1569 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1570 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1571 * config/s390/s390.c (s390_function_arg_padding): New function.
1572 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1573 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1574 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1575 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1576 (function_arg_padding): Rename to...
1577 (sparc_function_arg_padding): ...this. Make static. Return
1578 pad_direction instead of direction.
1579 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1580 * config/spu/spu.c (spu_function_arg_padding): New function.
1581 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1582 * system.h (FUNCTION_ARG_PADDING): Poison.
1584 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1585 Alan Hayward <alan.hayward@arm.com>
1586 David Sherwood <david.sherwood@arm.com>
1588 * target.def (modes_tieable_p): New hook.
1589 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1590 (TARGET_MODES_TIEABLE_P): ...this.
1591 * doc/tm.texi.in: Regenerate.
1592 * hooks.h (hook_bool_mode_mode_true): Declare.
1593 * hooks.c (hook_bool_mode_mode_true): New function.
1594 * combine.c (subst): Use targetm.modes_tieable_p instead of
1596 * dse.c (find_shift_sequence): Likewise.
1597 * expmed.c (extract_low_bits): Likewise.
1598 * lower-subreg.c: Include target.h.
1599 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1601 * rtlanal.c (rtx_cost): Likewise.
1602 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1603 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1604 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1605 (TARGET_MODES_TIEABLE_P): Redefine.
1606 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1607 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1608 (TARGET_MODES_TIEABLE_P): Redefine.
1609 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1610 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1611 (arc_modes_tieable_p): New function.
1612 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1613 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1614 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1615 (arm_modes_tieable_p): Make static.
1616 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1617 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1618 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1619 (TARGET_MODES_TIEABLE_P): Redefine.
1620 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1621 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1622 (TARGET_MODES_TIEABLE_P): Redefine.
1623 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1624 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1625 (cr16_modes_tieable_p): New function.
1626 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1627 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1628 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1629 (TRULY_NOOP_TRUNCATION): Update comment.
1630 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1631 (TRULY_NOOP_TRUNCATION): Update comment.
1632 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1633 (frv_modes_tieable_p): New function.
1634 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1635 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1636 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1637 (TARGET_MODES_TIEABLE_P): Redefine.
1638 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1639 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1640 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1641 (TARGET_MODES_TIEABLE_P): Redefine.
1642 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1643 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1644 (ia64_modes_tieable_p): New function.
1645 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1646 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1647 (iq2000_modes_tieable_p): New function.
1648 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1649 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1650 (lm32_modes_tieable_p): New function.
1651 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1652 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1653 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1654 (TARGET_MODES_TIEABLE_P): Redefine.
1655 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1656 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1657 (m32r_modes_tieable_p): New function.
1658 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1659 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1660 (m68k_modes_tieable_p): New function.
1661 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1662 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1663 (mcore_modes_tieable_p): New function.
1664 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1665 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1667 (TARGET_MODES_TIEABLE_P): Redefine.
1668 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1669 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1670 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1671 (TARGET_MODES_TIEABLE_P): Redefine.
1672 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1673 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1674 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1675 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1676 (mn10300_modes_tieable_p): ...this and make static.
1677 (TARGET_MODES_TIEABLE_P): Redefine.
1678 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1679 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1680 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1681 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1682 (msp430_modes_tieable_p): Make static.
1683 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1684 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1685 (TARGET_MODES_TIEABLE_P): Redefine.
1686 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1687 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1688 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1689 (TARGET_MODES_TIEABLE_P): Redefine.
1690 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1691 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1692 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1693 (TARGET_MODES_TIEABLE_P): Redefine.
1694 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1695 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1696 (pdp11_modes_tieable_p): New function.
1697 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1698 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1699 (rs6000_modes_tieable_p): New function.
1700 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1701 * config/powerpcspe/powerpcspe.md: Update comment.
1702 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1703 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1704 (TARGET_MODES_TIEABLE_P): Redefine.
1705 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1706 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1707 (rl78_modes_tieable_p): New function.
1708 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1709 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1710 (rs6000_modes_tieable_p): New function.
1711 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1712 * config/rs6000/rs6000.md: Update comment.
1713 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1714 * config/rx/rx.c (rx_modes_tieable_p): New function.
1715 (TARGET_MODES_TIEABLE_P): Redefine.
1716 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1717 * config/s390/s390.c (s390_modes_tieable_p): New function.
1718 (TARGET_MODES_TIEABLE_P): Redefine.
1719 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1720 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1721 (sh_modes_tieable_p): New function.
1722 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1723 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1724 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1725 (sparc_modes_tieable_p): Make static.
1726 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1727 * config/spu/spu.c (spu_modes_tieable_p): New function.
1728 (TARGET_MODES_TIEABLE_P): Redefine.
1729 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1730 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1731 (TARGET_MODES_TIEABLE_P): Redefine.
1732 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1733 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1734 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1735 * config/v850/v850.c (v850_modes_tieable_p): New function.
1736 (TARGET_MODES_TIEABLE_P): Redefine.
1737 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1738 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1739 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1740 (visium_modes_tieable_p): New function.
1741 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1742 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1743 (xtensa_modes_tieable_p): New function.
1744 * system.h (MODES_TIEABLE_P): Poison.
1746 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1747 Alan Hayward <alan.hayward@arm.com>
1748 David Sherwood <david.sherwood@arm.com>
1750 * target.def (hard_regno_mode_ok): New hook.
1751 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1752 (TARGET_HARD_REGNO_MODE_OK): ...this.
1753 * doc/tm.texi.in: Regenerate.
1754 * hooks.h (hook_bool_uint_mode_true): Declare.
1755 * hooks.c (hook_bool_uint_mode_true): New function.
1756 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1758 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1759 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1760 instead of HARD_REGNO_MODE_OK.
1761 * caller-save.c: Include target.h.
1762 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1764 * combine.c (can_combine_p): Likewise.
1765 (combinable_i3pat): Likewise.
1766 (can_change_dest_mode): Likewise.
1767 * expr.c (init_expr_target): Likewise.
1768 (convert_move): Likewise.
1769 (convert_modes): Likewise.
1770 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1771 (setup_prohibited_mode_move_regs): Likewise.
1772 * ira.h (target_ira): Likewise.
1773 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1774 * lra-constraints.c (process_alt_operands): Likewise.
1775 (split_reg): Likewise.
1776 * recog.c (peep2_find_free_register): Likewise.
1777 * ree.c (combine_reaching_defs): Likewise.
1778 * regcprop.c (maybe_mode_change): Likewise.
1779 * reginfo.c (init_reg_sets_1): Likewise.
1780 (choose_hard_reg_mode): Likewise.
1781 (simplifiable_subregs): Likewise.
1782 * regrename.c (check_new_reg_p): Likewise.
1783 * reload.c (find_valid_class): Likewise.
1784 (find_valid_class_1): Likewise.
1785 (reload_inner_reg_of_subreg): Likewise.
1786 (push_reload): Likewise.
1787 (combine_reloads): Likewise.
1788 (find_dummy_reload): Likewise.
1789 (find_reloads): Likewise.
1790 * reload1.c (find_reg): Likewise.
1791 (set_reload_reg): Likewise.
1792 (allocate_reload_reg): Likewise.
1793 (choose_reload_regs): Likewise.
1794 (reload_adjust_reg_for_temp): Likewise.
1795 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1796 (simplify_subreg_regno): Likewise.
1797 * sel-sched.c (init_regs_for_mode): Likewise.
1798 * varasm.c (make_decl_rtl): Likewise.
1799 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1800 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1802 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1803 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1804 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1805 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1806 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1807 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1808 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1809 (arc_mode_class): Delete.
1810 (HARD_REGNO_MODE_OK): Delete.
1811 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1812 (arc_hard_regno_mode_ok): Rename old array to...
1813 (arc_hard_regno_mode_ok_modes): ...this.
1814 (arc_conditional_register_usage): Update accordingly.
1815 (arc_mode_class): Make static.
1816 (arc_hard_regno_mode_ok): New function.
1817 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1818 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1819 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1820 (arm_hard_regno_mode_ok): Make static.
1821 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1823 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1824 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1825 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1827 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1828 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1829 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1830 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1831 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1832 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1833 * config/bfin/predicates.md (valid_reg_operand): Use
1834 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1835 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1836 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1837 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1838 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1839 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1840 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1841 (cr16_hard_regno_mode_ok): Make static and return a bool.
1842 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1843 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1844 (cris_hard_regno_mode_ok): New function.
1845 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1846 (epiphany_mode_class): Delete.
1847 (HARD_REGNO_MODE_OK): Delete.
1848 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1849 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1850 (hard_regno_mode_ok): Rename to...
1851 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1852 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1853 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1855 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1856 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1857 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1858 (frv_hard_regno_mode_ok): Make static and return a bool.
1859 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1861 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1862 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1863 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1864 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1866 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1867 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1868 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1869 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1871 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1872 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1873 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1874 (ia64_hard_regno_mode_ok): New function.
1875 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1876 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1877 (iq2000_hard_regno_mode_ok): New function.
1878 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1879 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1880 (lm32_hard_regno_mode_ok): New function.
1881 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1882 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1883 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1884 instead of HARD_REGNO_MODE_OK.
1885 (m32c_hard_regno_ok): Rename to...
1886 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1887 (m32c_cannot_change_mode_class): Update accordingly.
1888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1889 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1890 (m32r_mode_class): Delete.
1891 (HARD_REGNO_MODE_OK): Delete.
1892 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1893 (m32r_hard_regno_mode_ok): Rename to...
1894 (m32r_hard_regno_modes): ...this.
1895 (m32r_mode_class): Make static.
1896 (m32r_hard_regno_mode_ok): New function.
1897 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1898 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1899 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1900 (m68k_hard_regno_mode_ok): Make static.
1901 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1902 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1903 (mcore_hard_regno_mode_ok): New function.
1904 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1905 (HARD_REGNO_MODE_OK): Delete.
1906 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1908 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1909 (microblaze_hard_regno_mode_ok): New function.
1910 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1911 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1912 (mips_hard_regno_mode_ok): Delete.
1913 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1914 (mips_hard_regno_mode_ok_p): ...this and make static.
1915 (mips_hard_regno_mode_ok_p): Rename to...
1916 (mips_hard_regno_mode_ok_uncached): ...this.
1917 (mips_hard_regno_mode_ok): New function.
1918 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1919 of HARD_REGNO_MODE_OK.
1920 (mips_option_override): Update after above name changes.
1921 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1922 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1923 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1924 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1925 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1926 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1927 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1928 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1929 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1930 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1931 (msp430_hard_regno_mode_ok): Make static and return a bool.
1932 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1933 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1934 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1936 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1937 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1938 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1939 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1940 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1941 (PA_HARD_REGNO_MODE_OK): ...this
1942 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1943 (PA_HARD_REGNO_MODE_OK): ...this.
1944 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1945 (pa_hard_regno_mode_ok): New function.
1946 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1947 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1948 (pdp11_hard_regno_mode_ok): New function.
1949 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1950 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1952 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1954 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1955 (rs6000_hard_regno_mode_ok): Rename to...
1956 (rs6000_hard_regno_mode_ok_uncached): ...this.
1957 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1958 (rs6000_hard_regno_mode_ok): New function.
1959 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1960 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1961 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1962 (riscv_hard_regno_mode_ok): ...this and make static.
1963 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1964 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1965 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1966 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1967 (rl78_hard_regno_mode_ok): Make static and return bool.
1968 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1969 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1971 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1973 (rs6000_hard_regno_mode_ok): Rename to...
1974 (rs6000_hard_regno_mode_ok_uncached): ...this.
1975 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1976 (rs6000_hard_regno_mode_ok): New function.
1977 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1978 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1979 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1980 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1981 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1982 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1983 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1984 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1985 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1986 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1987 (sh_hard_regno_mode_ok): Make static.
1988 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1989 instead of HARD_REGNO_MODE_OK.
1990 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1991 (sparc_mode_class): Delete.
1992 (HARD_REGNO_MODE_OK): Delete.
1993 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1994 (hard_regno_mode_classes): Make static.
1995 (sparc_mode_class): Likewise.
1996 (sparc_hard_regno_mode_ok): New function.
1997 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1998 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1999 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2002 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2003 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2004 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2005 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2006 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2007 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2008 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2009 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2010 (visium_hard_regno_mode_ok): New function.
2011 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2012 instead of HARD_REGNO_MODE_OK.
2013 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2014 (HARD_REGNO_MODE_OK): Delete.
2015 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2016 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2017 (xtensa_option_override): Update accordingly.
2018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2019 (xtensa_hard_regno_mode_ok): New function.
2020 * system.h (HARD_REGNO_MODE_OK): Poison.
2022 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2023 Alan Hayward <alan.hayward@arm.com>
2024 David Sherwood <david.sherwood@arm.com>
2026 * target.def (hard_regno_call_part_clobbered): New hook.
2027 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2028 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2029 * doc/tm.texi: Regenerate.
2030 * hooks.h (hook_bool_uint_mode_false): Declare.
2031 * hooks.c (hook_bool_uint_mode_false): New function.
2032 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2033 * cselib.c (cselib_process_insn): Use
2034 targetm.hard_regno_call_part_clobbered instead of
2035 HARD_REGNO_CALL_PART_CLOBBERED.
2036 * ira-conflicts.c (ira_build_conflicts): Likewise.
2037 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2038 * lra-constraints.c (need_for_call_save_p): Likewise.
2039 * lra-lives.c: Include target.h.
2040 (check_pseudos_live_through_calls): Use
2041 targetm.hard_regno_call_part_clobbered instead of
2042 HARD_REGNO_CALL_PART_CLOBBERED.
2043 * regcprop.c: Include target.h.
2044 (copyprop_hardreg_forward_1): Use
2045 targetm.hard_regno_call_part_clobbered instead of
2046 HARD_REGNO_CALL_PART_CLOBBERED.
2047 * reginfo.c (choose_hard_reg_mode): Likewise.
2048 * regrename.c (check_new_reg_p): Likewise.
2049 * reload.c (find_equiv_reg): Likewise.
2050 * reload1.c (emit_reload_insns): Likewise.
2051 * sched-deps.c (deps_analyze_insn): Likewise.
2052 * sel-sched.c (init_regs_for_mode): Likewise.
2053 (mark_unavailable_hard_regs): Likewise.
2054 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2055 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2056 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2058 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2059 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2060 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2062 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2064 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2065 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2066 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2068 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2069 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2070 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2072 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2073 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2075 * config/powerpcspe/powerpcspe.c
2076 (rs6000_hard_regno_call_part_clobbered): New function.
2077 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2078 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2079 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2081 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2082 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2083 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2085 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2086 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2087 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2089 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2090 Alan Hayward <alan.hayward@arm.com>
2091 David Sherwood <david.sherwood@arm.com>
2093 * rtl.h (subreg_memory_offset): Declare.
2094 * emit-rtl.c (subreg_memory_offset): New function.
2095 * expmed.c (store_bit_field_1): Use it.
2096 * expr.c (undefined_operand_subword_p): Likewise.
2097 * simplify-rtx.c (simplify_subreg): Likewise.
2099 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2101 PR rtl-optimization/57448
2104 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2105 using atomic_load pattern.
2106 (expand_atomic_store): Likewise.
2108 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2111 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2112 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2113 handling. Use replace_call_with_value with NULL instead of
2114 gsi_replace, unlink_stmt_vdef and release_defs.
2116 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2119 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2121 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2124 * rtl.h (emit_library_call_value_1): Declare.
2125 (emit_library_call): Replace declaration with a series of overloads.
2126 Remove the parameter count argument.
2127 (emit_library_call_value): Likewise.
2128 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2129 with an "rtx_mode_t *".
2130 (emit_library_call_value): Delete.
2131 (emit_library_call): Likewise.
2132 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2133 (asan_emit_allocas_unpoison): Likewise.
2134 * builtins.c (expand_builtin_powi): Likewise.
2135 (expand_asan_emit_allocas_unpoison): Likewise.
2136 * cfgexpand.c (expand_main_function): Likewise.
2137 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2138 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2139 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2140 * config/arm/arm.c (arm_trampoline_init): Likewise.
2141 (arm_call_tls_get_addr): Likewise.
2142 (arm_expand_divmod_libfunc): Likewise.
2143 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2144 (smulsi3_highpart): Likewise.
2145 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2146 (c6x_expand_compare): Likewise.
2147 (c6x_expand_movmem): Likewise.
2148 * config/frv/frv.c (frv_trampoline_init): Likewise.
2149 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2150 (ix86_expand_divmod_libfunc): Likewise.
2151 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2152 (ia64_expand_compare): Likewise.
2153 (ia64_profile_hook): Likewise.
2154 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2155 (nonlocal_goto): Likewise.
2156 (restore_stack_nonlocal): Likewise.
2157 * config/m32r/m32r.c (block_move_call): Likewise.
2158 (m32r_trampoline_init): Likewise.
2159 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2160 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2161 (m68k_call_m68k_read_tp): Likewise.
2162 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2163 (microblaze_expand_divide): Likewise.
2164 * config/mips/mips.h (mips_args): Likewise.
2165 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2166 (MIPS_ICACHE_SYNC): Likewise.
2167 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2168 (nios2_trampoline_init): Likewise.
2169 * config/pa/pa.c (hppa_tls_call): Likewise.
2170 (pa_trampoline_init): Likewise.
2171 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2172 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2173 (expand_strn_compare): Likewise.
2174 (rs6000_generate_compare): Likewise.
2175 (rs6000_expand_float128_convert): Likewise.
2176 (output_profile_hook): Likewise.
2177 (rs6000_trampoline_init): Likewise.
2178 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2179 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2180 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2181 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2182 (rs6000_generate_compare): Likewise.
2183 (rs6000_expand_float128_convert): Likewise.
2184 (output_profile_hook): Likewise.
2185 (rs6000_trampoline_init): Likewise.
2186 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2187 * config/sh/sh.c (sh_trampoline_init): Likewise.
2188 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2189 (sparc_emit_float_lib_cmp): Likewise.
2190 (sparc32_initialize_trampoline): Likewise.
2191 (sparc64_initialize_trampoline): Likewise.
2192 (sparc_profile_hook): Likewise.
2193 * config/spu/spu.c (ea_load_store): Likewise.
2194 * config/spu/spu.md (floatunssidf2): Likewise.
2195 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2196 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2197 * config/visium/visium.c (expand_block_move_4): Likewise.
2198 (expand_block_move_2): Likewise.
2199 (expand_block_move_1): Likewise.
2200 (expand_block_set_4): Likewise.
2201 (expand_block_set_2): Likewise.
2202 (expand_block_set_1): Likewise.
2203 (visium_trampoline_init): Likewise.
2204 (visium_profile_hook): Likewise.
2205 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2206 (xtensa_setup_frame_addresses): Likewise.
2207 (xtensa_trampoline_init): Likewise.
2208 * except.c (sjlj_emit_function_enter): Likewise.
2209 (sjlj_emit_function_exit): Likewise.
2210 * explow.c (allocate_dynamic_stack_space): Likewise.
2211 (probe_stack_range): Likewise.
2212 * expr.c (convert_mode_scalar): Likewise.
2213 * optabs.c (expand_binop): Likewise.
2214 (expand_twoval_binop_libfunc): Likewise.
2215 (expand_unop): Likewise.
2216 (prepare_cmp_insn): Likewise.
2217 (prepare_float_lib_cmp): Likewise.
2218 (expand_float): Likewise.
2219 (expand_fix): Likewise.
2220 (expand_fixed_convert): Likewise.
2221 (maybe_emit_sync_lock_test_and_set): Likewise.
2222 (expand_atomic_compare_and_swap): Likewise.
2223 (expand_mem_thread_fence): Likewise.
2224 (expand_atomic_fetch_op): Likewise.
2226 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2228 * doc/generic.texi (OpenACC): Adjust URL.
2229 * doc/invoke.texi (C Dialect Options): Ditto.
2231 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2233 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2234 predicate for operand 1. Add (m,<S>) constraint.
2235 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2236 Prevent memory operand 1 with register operand 2.
2238 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2240 PR rtl-optimization/82024
2241 * combine.c (try_combine): If the combination result is a PARALLEL,
2242 and we only need to retain the SET in there that would be placed
2243 at I2, check that we can place that at I3 instead, before doing so.
2245 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2248 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2250 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2251 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2254 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2255 Jeff Law <law@redhat.com>
2257 * varasm.c (bss_initializer_p): Do not put constants into .bss
2258 (categorize_decl_for_section): Handle bss_initializer_p returning
2259 false when DECL_INITIAL is NULL.
2261 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2264 * config/s390/s390.c (s390_can_inline_p): New function.
2266 2017-09-01 Jeff Law <law@redhat.com>
2268 PR tree-optimization/82052
2269 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2270 Always initialize the returned slot after a hash table miss
2271 when INSERT is true.
2273 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2275 * config/s390/s390.md (mem_signal_fence): Remove.
2276 * doc/md.texi (mem_signal_fence): Remove.
2277 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2279 * target-insns.def (mem_signal_fence): Remove.
2281 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2284 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2287 * asan.c (create_odr_indicator): Strip name encoding from assembler
2288 name before appending it after __odr_asan_.
2290 2017-09-01 Martin Liska <mliska@suse.cz>
2292 PR tree-optimization/82059
2293 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2294 frequency only when an edge is redirected.
2296 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2298 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2299 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2300 (arc_conditional_register_usage): Remove ARC600 lp_count
2302 (arc_file_start): Emit Tag_ARC_CPU_variation.
2303 (arc_can_use_doloop_p): New conditions to use ZOLs.
2304 (hwloop_fail): New function.
2305 (hwloop_optimize): Likewise.
2306 (hwloop_pattern_reg): Likewise.
2307 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2308 (arc_reorg_loops): New function, calls reorg_loops.
2309 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2310 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2312 (arc_loop_hazard): Remove function, functionality moved into
2314 (arc_hazard): Remove arc_loop_hazard call.
2315 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2316 into hwloop_optimize.
2317 (arc_label_align): Remove ZOL handling.
2318 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2319 * config/arc/arc.md (doloop_begin): Remove pattern.
2320 (doloop_begin_i): Likewise.
2321 (doloop_end_i): Likewise.
2322 (doloop_fallback): Likewise.
2323 (doloop_fallback_m): Likewise.
2324 (doloop_end): Reimplement expand.
2325 (arc_lp): New pattern for LP instruction.
2326 (loop_end): New pattern.
2327 (loop_fail): Likewise.
2328 (decrement_and_branch_until_zero): Likewise.
2329 * config/arc/arc.opt (mlpc-width): New option.
2330 * doc/invoke.texi (mlpc-width): Document option.
2332 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2334 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2335 (arc_ccfsm_advance): Fix checking for delay slots.
2336 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2338 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2340 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2342 (movhi_insn): Update store instruction constraint which are saving
2343 6-bit short immediates.
2344 (movsi_insn): Consider also short scaled load operations.
2345 (zero_extendhisi2_i): Use Usd constraint instead of T.
2346 (extendhisi2_i): Add q constraint.
2347 (arc_clzsi2): Add type and length attributes.
2348 (arc_ctzsi2): Likewise.
2349 * config/arc/constraints.md (Usc): Update constraint, the
2350 assembler can parse two relocations for a single instruction.
2352 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2354 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2355 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2357 2017-08-31 Olivier Hainque <hainque@adacore.com>
2359 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2360 match as powerpc-wrs-vxworks*.
2362 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2364 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2365 register constraint for by-element operand.
2366 (aarch64_mls_elt_merge<mode>): Likewise.
2368 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2370 * config/arc/arc.c (arc_can_follow_jump): Check for short
2373 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2375 * config.gcc: Use g.opt for arc.
2376 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2377 functionality moved to ...
2378 (legitimate_scaled_address_p): New function, ...here.
2379 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2380 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2381 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2383 (arc_override_options): Handle G option.
2384 (arc_output_pic_addr_const): Correct function definition.
2385 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2386 (arc_decl_anon_ns_mem_p): Delete.
2387 (arc_in_small_data_p): Overhaul this function to take into
2388 consideration the value given via G option.
2389 (arc_rewrite_small_data_1): Renamed and corrected old
2390 arc_rewrite_small_data function.
2391 (arc_rewrite_small_data): New function.
2392 (small_data_pattern): Don't use pic_offset_table_rtx.
2393 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2394 * config/arc/simdext.md (movmisalignv2hi): Use
2395 prepare_move_operands function.
2397 (movmisalign*): Likewise.
2398 * doc/invoke.texi (ARC options): Document -G option.
2400 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2402 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2404 * config/arc/arc.c (arc_print_operand): Output scalled address for
2405 sdata whenever is possible.
2406 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2407 load/stores are available.
2408 (compact_sda_memory_operand): Check for the alignment required by
2409 code density instructions.
2410 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2412 * config/arc/constraints.md (Usd): Update constraint.
2413 (Us0): New constraint.
2414 (Usc): Update constraint.
2416 2017-08-31 Richard Biener <rguenther@suse.de>
2419 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2422 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2424 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2425 Resize type_signature.
2427 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2428 Alan Hayward <alan.hayward@arm.com>
2429 David Sherwood <david.sherwood@arm.com>
2431 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2432 subregs whose inner modes can be stored in GPRs.
2433 (aarch64_classify_index): Likewise.
2435 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2440 (V_INT_EQUIV): ...this.
2441 (v_cmp_result): Rename to...
2442 (v_int_equiv): ...this.
2443 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2444 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2445 (copysign<mode>3): Likewise.
2446 (aarch64_simd_bsl<mode>_internal): Likewise.
2447 (aarch64_simd_bsl<mode>): Likewise.
2448 (vec_cmp<mode><mode>): Likewise.
2449 (vcond<mode><mode>): Likewise.
2450 (vcond<v_cmp_mixed><mode>): Likewise.
2451 (vcondu<mode><v_cmp_mixed>): Likewise.
2452 (aarch64_cm<optab><mode>): Likewise.
2453 (aarch64_cmtst<mode>): Likewise.
2454 (aarch64_fac<optab><mode>): Likewise.
2455 (vec_perm_const<mode>): Likewise.
2456 (vcond_mask_<mode><v_cmp_result>): Rename to...
2457 (vcond_mask_<mode><v_int_equiv>): ...this.
2458 (vec_cmp<mode><v_cmp_result>): Rename to...
2459 (vec_cmp<mode><v_int_equiv>): ...this.
2461 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2462 Alan Hayward <alan.hayward@arm.com>
2463 David Sherwood <david.sherwood@arm.com>
2465 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2467 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2468 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2469 (UNSPEC_LD4_DREG): New unspecs.
2470 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2471 (aarch64_ld2<mode>_dreg_be): Replace with...
2472 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2474 (aarch64_ld3<mode>_dreg_le)
2475 (aarch64_ld3<mode>_dreg_be): Replace with...
2476 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2478 (aarch64_ld4<mode>_dreg_le)
2479 (aarch64_ld4<mode>_dreg_be): Replace with...
2480 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2483 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2485 PR tree-optimization/81987
2486 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2487 insert an initializer in a location not dominated by the stride
2490 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2492 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2493 on the entire header of the finally block in the fallthru case.
2495 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2497 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2499 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2501 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2502 rs6000_emit_move_from_cr and call renamed function.
2503 (rs6000_emit_prologue): Call renamed functions.
2504 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2505 movesi_from_cr, remove volatile CRs.
2507 2017-08-30 Jon Beniston <jon@beniston.com>
2508 Richard Biener <rguenther@suse.de>
2510 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2511 of VECTOR_MODE_P check.
2512 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2513 element vector types.
2515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2517 * df.h (df_read_modify_subreg_p): Remove in favor of...
2518 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2519 const_rtx instead of an rtx.
2520 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2521 * df-problems.c (df_word_lr_mark_ref): Likewise.
2522 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2523 (mark_pseudo_reg_dead): Likewise.
2524 (mark_ref_dead): Likewise.
2525 * reginfo.c (init_subregs_of_mode): Likewise.
2526 * sched-deps.c (sched_analyze_1): Likewise.
2527 * df-scan.c (df_def_record_1): Likewise.
2528 (df_uses_record): Likewise.
2529 (df_read_modify_subreg_p): Remove in favor of...
2530 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2531 const_rtx instead of an rtx.
2533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2534 Alan Hayward <alan.hayward@arm.com>
2535 David Sherwood <david.sherwood@arm.com>
2537 * rtl.h (partial_subreg_p): New function.
2538 * caller-save.c (save_call_clobbered_regs): Use it.
2539 * calls.c (expand_call): Likewise.
2540 * combine.c (combinable_i3pat): Likewise.
2541 (simplify_set): Likewise.
2542 (make_extraction): Likewise.
2543 (make_compound_operation_int): Likewise.
2544 (gen_lowpart_or_truncate): Likewise.
2545 (force_to_mode): Likewise.
2546 (make_field_assignment): Likewise.
2547 (reg_truncated_to_mode): Likewise.
2548 (record_truncated_value): Likewise.
2549 (move_deaths): Likewise.
2550 * cse.c (record_jump_cond): Likewise.
2551 (cse_insn): Likewise.
2552 * cselib.c (cselib_lookup_1): Likewise.
2553 * expmed.c (extract_bit_field_using_extv): Likewise.
2554 * function.c (assign_parm_setup_reg): Likewise.
2555 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2556 * ira-build.c (create_insn_allocnos): Likewise.
2557 * lra-coalesce.c (merge_pseudos): Likewise.
2558 * lra-constraints.c (match_reload): Likewise.
2559 (simplify_operand_subreg): Likewise.
2560 (curr_insn_transform): Likewise.
2561 * lra-lives.c (process_bb_lives): Likewise.
2562 * lra.c (new_insn_reg): Likewise.
2563 (lra_substitute_pseudo): Likewise.
2564 * regcprop.c (mode_change_ok): Likewise.
2565 (maybe_mode_change): Likewise.
2566 (copyprop_hardreg_forward_1): Likewise.
2567 * reload.c (push_reload): Likewise.
2568 (find_reloads): Likewise.
2569 (find_reloads_subreg_address): Likewise.
2570 * reload1.c (alter_reg): Likewise.
2571 (eliminate_regs_1): Likewise.
2572 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2574 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2576 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2577 back to if statements, including unpack.
2579 2017-08-30 Martin Liska <mliska@suse.cz>
2582 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2584 (func_checker::compare_asm_inputs_outputs): ... this function.
2585 (func_checker::compare_gimple_asm): Use the function to compare
2586 also ASM constrains.
2587 * ipa-icf-gimple.h: Rename the function.
2589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2590 Alan Hayward <alan.hayward@arm.com>
2591 David Sherwood <david.sherwood@arm.com>
2593 * coretypes.h (complex_mode): New type.
2594 * gdbhooks.py (build_pretty_printer): Handle it.
2595 * machmode.h (complex_mode): New class.
2596 (complex_mode::includes_p): New function.
2597 (is_complex_int_mode): Likewise.
2598 (is_complex_float_mode): Likewise.
2599 * genmodes.c (get_mode_class): Handle complex mode classes.
2600 * function.c (expand_function_end): Use is_complex_int_mode.
2602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2603 Alan Hayward <alan.hayward@arm.com>
2604 David Sherwood <david.sherwood@arm.com>
2606 * coretypes.h (scalar_mode_pod): New typedef.
2607 * gdbhooks.py (build_pretty_printer): Handle it.
2608 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2609 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2610 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2611 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2615 Alan Hayward <alan.hayward@arm.com>
2616 David Sherwood <david.sherwood@arm.com>
2618 * machmode.h (mode_for_vector): Take a scalar_mode instead
2620 * stor-layout.c (mode_for_vector): Likewise.
2621 * explow.c (promote_mode): Use as_a <scalar_mode>.
2622 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2625 Alan Hayward <alan.hayward@arm.com>
2626 David Sherwood <david.sherwood@arm.com>
2628 * target.def (preferred_simd_mode): Take a scalar_mode
2629 instead of a machine_mode.
2630 * targhooks.h (default_preferred_simd_mode): Likewise.
2631 * targhooks.c (default_preferred_simd_mode): Likewise.
2632 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2633 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2634 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2635 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2636 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2637 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2638 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2639 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2641 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2642 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2643 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2644 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2645 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2646 * doc/tm.texi: Regenerate.
2647 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2651 Alan Hayward <alan.hayward@arm.com>
2652 David Sherwood <david.sherwood@arm.com>
2654 * target.def (scalar_mode_supported_p): Take a scalar_mode
2655 instead of a machine_mode.
2656 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2657 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2658 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2659 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2660 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2661 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2662 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2663 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2664 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2665 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2666 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2667 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2668 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2669 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2671 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2672 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2673 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2674 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2675 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2677 * doc/tm.texi: Regenerate.
2679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2680 Alan Hayward <alan.hayward@arm.com>
2681 David Sherwood <david.sherwood@arm.com>
2683 * coretypes.h (opt_scalar_mode): New typedef.
2684 * gdbhooks.py (build_pretty_printers): Handle it.
2685 * machmode.h (mode_iterator::get_2xwider): Add overload for
2687 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2689 * expr.c (convert_mode_scalar): Likewise.
2690 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2691 * optabs.c (expand_float): Likewise.
2692 (expand_fix): Likewise.
2693 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2696 Alan Hayward <alan.hayward@arm.com>
2697 David Sherwood <david.sherwood@arm.com>
2699 * optabs.c (expand_float): Explicitly check for scalars before
2700 using a branching expansion.
2701 (expand_fix): Likewise.
2703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2704 Alan Hayward <alan.hayward@arm.com>
2705 David Sherwood <david.sherwood@arm.com>
2707 * expr.c (convert_mode): Split scalar handling out into...
2708 (convert_mode_scalar): ...this new function. Treat the modes
2711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2712 Alan Hayward <alan.hayward@arm.com>
2713 David Sherwood <david.sherwood@arm.com>
2715 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2717 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2720 Alan Hayward <alan.hayward@arm.com>
2721 David Sherwood <david.sherwood@arm.com>
2723 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2724 rather than a machine_mode.
2725 (fixed_from_string): Likewise.
2726 (fixed_convert): Likewise.
2727 (fixed_convert_from_int): Likewise.
2728 (fixed_convert_from_real): Likewise.
2729 (real_convert_from_fixed): Likewise.
2730 * fixed-value.c (fixed_from_double_int): Likewise.
2731 (fixed_from_string): Likewise.
2732 (fixed_convert): Likewise.
2733 (fixed_convert_from_int): Likewise.
2734 (fixed_convert_from_real): Likewise.
2735 (real_convert_from_fixed): Likewise.
2736 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2739 Alan Hayward <alan.hayward@arm.com>
2740 David Sherwood <david.sherwood@arm.com>
2742 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2743 of separate mode class checks. Do not allow vector modes here.
2744 (immed_wide_int_const): Use as_a <scalar_mode>.
2745 * explow.c (trunc_int_for_mode): Likewise.
2746 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2747 (wi::shwi): Likewise.
2748 (wi::min_value): Likewise.
2749 (wi::max_value): Likewise.
2750 * dwarf2out.c (loc_descriptor): Likewise.
2751 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2755 Alan Hayward <alan.hayward@arm.com>
2756 David Sherwood <david.sherwood@arm.com>
2758 * tree.h (SCALAR_TYPE_MODE): New macro.
2759 * expr.c (expand_expr_addr_expr_1): Use it.
2760 (expand_expr_real_2): Likewise.
2761 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2762 (fold_convert_const_fixed_from_int): Likewise.
2763 (fold_convert_const_fixed_from_real): Likewise.
2764 (native_encode_fixed): Likewise
2765 (native_encode_complex): Likewise
2766 (native_encode_vector): Likewise.
2767 (native_interpret_fixed): Likewise.
2768 (native_interpret_real): Likewise.
2769 (native_interpret_complex): Likewise.
2770 (native_interpret_vector): Likewise.
2771 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2772 (simd_clone_adjust_argument_types): Likewise.
2773 (simd_clone_init_simd_arrays): Likewise.
2774 (simd_clone_adjust): Likewise.
2775 * stor-layout.c (layout_type): Likewise.
2776 * tree.c (build_minus_one_cst): Likewise.
2777 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2778 * tree-inline.c (estimate_move_cost): Likewise.
2779 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2780 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2781 (vectorizable_reduction): Likewise.
2782 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2783 (vect_recog_mixed_size_cond_pattern): Likewise.
2784 (check_bool_pattern): Likewise.
2785 (adjust_bool_pattern): Likewise.
2786 (search_type_for_mask_1): Likewise.
2787 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2788 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2789 (vectorizable_load): Likewise.
2790 (vectorizable_store): Likewise.
2791 * ubsan.c (ubsan_encode_value): Likewise.
2792 * varasm.c (output_constant): Likewise.
2794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2795 Alan Hayward <alan.hayward@arm.com>
2796 David Sherwood <david.sherwood@arm.com>
2798 * coretypes.h (scalar_mode): New class.
2799 * machmode.h (scalar_mode): Likewise.
2800 (scalar_mode::includes_p): New function.
2801 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2802 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2803 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2804 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2805 * expmed.c (store_bit_field_1): Likewise.
2806 (extract_bit_field_1): Likewise.
2807 * expr.c (write_complex_part): Likewise.
2808 (read_complex_part): Likewise.
2809 (emit_move_complex_push): Likewise.
2810 (expand_expr_real_2): Likewise.
2811 * function.c (assign_parm_setup_reg): Likewise.
2812 (assign_parms_unsplit_complex): Likewise.
2813 * optabs.c (expand_binop): Likewise.
2814 * rtlanal.c (subreg_get_info): Likewise.
2815 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2816 * varasm.c (output_constant_pool_2): Likewise.
2818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2819 Alan Hayward <alan.hayward@arm.com>
2820 David Sherwood <david.sherwood@arm.com>
2822 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2824 (expmed_mult_highpart_optab): Likewise.
2825 (expmed_mult_highpart): Likewise.
2827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2828 Alan Hayward <alan.hayward@arm.com>
2829 David Sherwood <david.sherwood@arm.com>
2831 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2832 instead of a machine_mode.
2833 (builtin_memset_read_str): Likewise.
2834 * builtins.c (c_readstr): Likewise.
2835 (builtin_memcpy_read_str): Likewise.
2836 (builtin_strncpy_read_str): Likewise.
2837 (builtin_memset_read_str): Likewise.
2838 (builtin_memset_gen_str): Likewise.
2839 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2840 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2841 instead of a machine_mode.
2842 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2844 (make_extraction): Likewise.
2845 (try_widen_shift_mode): Take and return scalar_int_modes instead
2847 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2848 a scalar_int_mode instead of a machine_mode.
2849 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2850 (avr_addr_space_pointer_mode): Likewise.
2851 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2852 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2853 (msp430_unwind_word_mode): Likewise.
2854 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2855 (spu_addr_space_pointer_mode): Likewise.
2856 (spu_addr_space_address_mode): Likewise.
2857 (spu_libgcc_cmp_return_mode): Likewise.
2858 (spu_libgcc_shift_count_mode): Likewise.
2859 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2860 (rl78_addr_space_pointer_mode): Likewise.
2861 (fl78_unwind_word_mode): Likewise.
2862 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2864 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2865 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2866 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2867 (mips_valid_pointer_mode): Likewise.
2868 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2869 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2870 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2872 (ft32_addr_space_address_mode): Likewise.
2873 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2874 scalar_int_mode instead of a machine_mode.
2875 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2877 (m32c_addr_space_address_mode): Likewise.
2878 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2879 (rs6000_eh_return_filter_mode): Likewise.
2880 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2881 (rs6000_eh_return_filter_mode): Likewise.
2882 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2883 (s390_libgcc_shift_count_mode): Likewise.
2884 (s390_unwind_word_mode): Likewise.
2885 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2887 * target.def (mode_rep_extended): Likewise.
2888 (valid_pointer_mode): Likewise.
2889 (addr_space.valid_pointer_mode): Likewise.
2890 (eh_return_filter_mode): Return a scalar_int_mode rather than
2892 (libgcc_cmp_return_mode): Likewise.
2893 (libgcc_shift_count_mode): Likewise.
2894 (unwind_word_mode): Likewise.
2895 (addr_space.pointer_mode): Likewise.
2896 (addr_space.address_mode): Likewise.
2897 * doc/tm.texi: Regenerate.
2898 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2900 (do_jump): Use scalar_int_mode for local variables.
2901 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2902 rather than a machine_mode.
2903 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2904 (scompare_loc_descriptor_wide): Likewise.
2905 (scompare_loc_descriptor_narrow): Likewise.
2906 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2908 * except.c (sjlj_emit_dispatch_table): Likewise.
2909 (expand_builtin_eh_copy_values): Likewise.
2910 * explow.c (convert_memory_address_addr_space_1): Likewise.
2911 Take a scalar_int_mode rather than a machine_mode.
2912 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2913 than a machine_mode.
2914 (memory_address_addr_space): Use scalar_int_mode for local variables.
2915 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2916 rather than a machine_mode.
2917 * expmed.c (mask_rtx): Likewise.
2918 (init_expmed_one_conv): Likewise.
2919 (expand_mult_highpart_adjust): Likewise.
2920 (extract_high_half): Likewise.
2921 (expmed_mult_highpart_optab): Likewise.
2922 (expmed_mult_highpart): Likewise.
2923 (expand_smod_pow2): Likewise.
2924 (expand_sdiv_pow2): Likewise.
2925 (emit_store_flag_int): Likewise.
2926 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2928 (extract_low_bits): Likewise.
2929 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2931 * expr.c (pieces_addr::adjust): Likewise.
2932 (can_store_by_pieces): Likewise.
2933 (store_by_pieces): Likewise.
2934 (clear_by_pieces_1): Likewise.
2935 (expand_expr_addr_expr_1): Likewise.
2936 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2937 (expand_expr_real_1): Likewise.
2938 (try_casesi): Likewise.
2939 * final.c (shorten_branches): Likewise.
2940 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2941 type of "mode" to machine_mode.
2942 * internal-fn.c (expand_arith_overflow_result_store): Take a
2943 scalar_int_mode rather than a machine_mode.
2944 (expand_mul_overflow): Use scalar_int_mode for local variables.
2945 * loop-doloop.c (doloop_modify): Likewise.
2946 (doloop_optimize): Likewise.
2947 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2948 than a machine_mode.
2949 (expand_doubleword_shift_condmove): Likewise.
2950 (expand_doubleword_shift): Likewise.
2951 (expand_doubleword_clz): Likewise.
2952 (expand_doubleword_popcount): Likewise.
2953 (expand_doubleword_parity): Likewise.
2954 (expand_absneg_bit): Use scalar_int_mode for local variables.
2955 (prepare_float_lib_cmp): Likewise.
2956 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2957 rather than a machine_mode.
2958 (convert_memory_address_addr_space): Likewise.
2959 (get_mode_bounds): Likewise.
2960 (get_address_mode): Return a scalar_int_mode rather than a
2962 * rtlanal.c (get_address_mode): Likewise.
2963 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2964 than a machine_mode.
2965 * targhooks.c (default_mode_rep_extended): Likewise.
2966 (default_valid_pointer_mode): Likewise.
2967 (default_addr_space_valid_pointer_mode): Likewise.
2968 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2969 than a machine_mode.
2970 (default_libgcc_cmp_return_mode): Likewise.
2971 (default_libgcc_shift_count_mode): Likewise.
2972 (default_unwind_word_mode): Likewise.
2973 (default_addr_space_pointer_mode): Likewise.
2974 (default_addr_space_address_mode): Likewise.
2975 * targhooks.h (default_eh_return_filter_mode): Likewise.
2976 (default_libgcc_cmp_return_mode): Likewise.
2977 (default_libgcc_shift_count_mode): Likewise.
2978 (default_unwind_word_mode): Likewise.
2979 (default_addr_space_pointer_mode): Likewise.
2980 (default_addr_space_address_mode): Likewise.
2981 (default_mode_rep_extended): Take a scalar_int_mode rather than
2983 (default_valid_pointer_mode): Likewise.
2984 (default_addr_space_valid_pointer_mode): Likewise.
2985 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2987 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2988 rather than a machine_mode.
2989 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2990 for local variables.
2991 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2992 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2993 than a machine_mode.
2995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2996 Alan Hayward <alan.hayward@arm.com>
2997 David Sherwood <david.sherwood@arm.com>
2999 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3000 the mode argument to scalar_int_mode.
3001 (do_jump_by_parts_zero_rtx): Likewise.
3002 (do_jump_by_parts_equality_rtx): Likewise.
3003 (do_jump_by_parts_greater): Take a mode argument.
3004 (do_jump_by_parts_equality): Likewise.
3005 (do_jump_1): Update calls accordingly.
3007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3008 Alan Hayward <alan.hayward@arm.com>
3009 David Sherwood <david.sherwood@arm.com>
3011 * is-a.h (safe_dyn_cast): New function.
3012 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3013 (jump_table_for_label): Likewise.
3014 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3015 instead of an rtx_insn *.
3016 (shorten_branches): Use dyn_cast instead of LABEL_P and
3017 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3018 rtx_jump_table_data::get_data_mode.
3019 (final_scan_insn): Likewise.
3021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3022 Alan Hayward <alan.hayward@arm.com>
3023 David Sherwood <david.sherwood@arm.com>
3025 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3026 trying to combine a full-register integer set with a subreg
3029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3030 Alan Hayward <alan.hayward@arm.com>
3031 David Sherwood <david.sherwood@arm.com>
3033 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3034 that is always either address_mode or pointer_mode.
3036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3037 Alan Hayward <alan.hayward@arm.com>
3038 David Sherwood <david.sherwood@arm.com>
3040 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3041 when the two are known to be equal.
3043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3044 Alan Hayward <alan.hayward@arm.com>
3045 David Sherwood <david.sherwood@arm.com>
3047 * simplify-rtx.c (simplify_const_unary_operation): Use
3048 is_a <scalar_int_mode> instead of checking for a nonzero
3049 precision. Forcibly convert op_mode to a scalar_int_mode
3050 in that case. More clearly differentiate the operand and
3051 result modes and use the former when deciding what the value
3052 of a count-bits operation should be. Use is_int_mode instead
3053 of checking for a MODE_INT. Remove redundant check for whether
3054 this mode has a zero precision.
3056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3057 Alan Hayward <alan.hayward@arm.com>
3058 David Sherwood <david.sherwood@arm.com>
3060 * optabs.c (widen_leading): Change the type of the mode argument
3061 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3062 (widen_bswap): Likewise.
3063 (expand_parity): Likewise.
3064 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3065 (expand_ffs): Likewise.
3066 (epand_unop): Check for scalar integer modes before calling the
3069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3070 Alan Hayward <alan.hayward@arm.com>
3071 David Sherwood <david.sherwood@arm.com>
3073 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3075 (expand_expr_real_1): Update call accordingly.
3077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3078 Alan Hayward <alan.hayward@arm.com>
3079 David Sherwood <david.sherwood@arm.com>
3081 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3082 value_mode arguments. Use scalar_int_mode internally.
3083 (store_bit_field_1): Rename the new integer mode from imode
3084 to op0_mode and use it instead of GET_MODE (op0). Update calls
3085 to store_split_bit_field, store_bit_field_using_insv and
3086 store_fixed_bit_field.
3087 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3088 Use scalar_int_mode internally. Use a bit count rather than a mode
3089 when calculating the largest bit size for get_best_mode.
3090 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3091 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3092 Remove assertion that OP0 has a scalar integer mode.
3093 (store_split_bit_field): Add op0_mode and value_mode arguments.
3094 Update calls to extract_fixed_bit_field.
3095 (extract_bit_field_using_extv): Add an op0_mode argument.
3096 Use scalar_int_mode internally.
3097 (extract_bit_field_1): Rename the new integer mode from imode to
3098 op0_mode and use it instead of GET_MODE (op0). Update calls to
3099 extract_split_bit_field, extract_bit_field_using_extv and
3100 extract_fixed_bit_field.
3101 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3102 to extract_split_bit_field and extract_fixed_bit_field_1.
3103 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3104 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3106 (extract_split_bit_field): Add an op0_mode argument. Update call
3107 to extract_fixed_bit_field.
3109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3110 Alan Hayward <alan.hayward@arm.com>
3111 David Sherwood <david.sherwood@arm.com>
3113 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3114 * explow.c (hard_function_value): Likewise.
3115 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3116 convert_to_mode call outside the loop.
3117 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3118 for the mode iterator. Require the mode specified by max_pieces
3120 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3122 (copy_blkmode_to_reg): Likewise.
3123 (set_storage_via_setmem): Likewise.
3124 * optabs.c (prepare_cmp_insn): Likewise.
3125 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3126 * stor-layout.c (finish_bitfield_representative): Likewise.
3128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3129 Alan Hayward <alan.hayward@arm.com>
3130 David Sherwood <david.sherwood@arm.com>
3132 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3133 * expr.c (convert_move): Use them.
3134 (convert_modes): Likewise.
3135 (store_expr_with_bounds): Likewise.
3137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3138 Alan Hayward <alan.hayward@arm.com>
3139 David Sherwood <david.sherwood@arm.com>
3141 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3142 parameter for the mode of "x". Remove the "known_x", "known_mode"
3143 and "known_ret" arguments. Change the type of the mode argument
3145 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3146 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3147 (reg_num_sign_bit_copies_for_combine): Likewise.
3148 * rtlanal.c (nonzero_bits1): Likewise.
3149 (num_sign_bit_copies1): Likewise.
3150 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3151 (reg_num_sign_bit_copies_general): Likewise.
3152 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3153 (reg_nonzero_bits_general): Likewise.
3155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3156 Alan Hayward <alan.hayward@arm.com>
3157 David Sherwood <david.sherwood@arm.com>
3159 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3160 than in subroutines. Return 1 for non-integer modes.
3161 (cached_num_sign_bit_copies): Change the type of the mode parameter
3163 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3164 classes. Handle CONST_INT_P first and then check whether X also
3165 has a scalar integer mode. Check the same thing for inner registers
3166 of a SUBREG and for values that are being extended or truncated.
3168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3169 Alan Hayward <alan.hayward@arm.com>
3170 David Sherwood <david.sherwood@arm.com>
3172 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3173 in subroutines. Return the mode mask for non-integer modes.
3174 (cached_nonzero_bits): Change the type of the mode parameter
3176 (nonzero_bits1): Likewise. Remove early exit for other mode
3177 classes. Handle CONST_INT_P first and then check whether X
3178 also has a scalar integer mode.
3180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3181 Alan Hayward <alan.hayward@arm.com>
3182 David Sherwood <david.sherwood@arm.com>
3184 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3185 Return a scalar_int_mode and assert that the size is greater than
3187 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3188 (op_by_pieces_d::op_by_pieces_d): Likewise.
3189 (op_by_pieces_d::run): Likewise.
3190 (can_store_by_pieces): Likewise.
3192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3193 Alan Hayward <alan.hayward@arm.com>
3194 David Sherwood <david.sherwood@arm.com>
3196 * combine.c (extract_left_shift): Add a mode argument and update
3198 (make_compound_operation_int): Change the type of the mode parameter
3199 to scalar_int_mode and update the call to extract_left_shift.
3201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3202 Alan Hayward <alan.hayward@arm.com>
3203 David Sherwood <david.sherwood@arm.com>
3205 * combine.c (simplify_and_const_int): Change the type of the mode
3206 parameter to scalar_int_mode.
3207 (simplify_and_const_int_1): Likewise. Update recursive call.
3209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3210 Alan Hayward <alan.hayward@arm.com>
3211 David Sherwood <david.sherwood@arm.com>
3213 * combine.c (simplify_compare_const): Check that the mode is a
3214 scalar_int_mode (rather than VOIDmode) before testing its
3216 (simplify_comparison): Move COMPARISON_P handling out of the
3217 loop and restrict the latter part of the loop to scalar_int_modes.
3218 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3219 and when considering SUBREG_REGs. Use is_int_mode instead of
3220 checking GET_MODE_CLASS against MODE_INT.
3222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3223 Alan Hayward <alan.hayward@arm.com>
3224 David Sherwood <david.sherwood@arm.com>
3226 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3227 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3228 shift_unit_mode and for modes involved in scalar shifts.
3230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3231 Alan Hayward <alan.hayward@arm.com>
3232 David Sherwood <david.sherwood@arm.com>
3234 * combine.c (force_int_to_mode): New function, split out from...
3235 (force_to_mode): ...here. Keep xmode up-to-date and use it
3236 instead of GET_MODE (x).
3238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3239 Alan Hayward <alan.hayward@arm.com>
3240 David Sherwood <david.sherwood@arm.com>
3242 * optabs-query.h (extraction_insn::struct_mode): Change type to
3243 opt_scalar_int_mode and update comment.
3244 (extraction_insn::field_mode): Change type to scalar_int_mode.
3245 (extraction_insn::pos_mode): Likewise.
3246 * combine.c (make_extraction): Update accordingly.
3247 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3248 (get_optab_extraction_insn): Likewise.
3249 * recog.c (simplify_while_replacing): Likewise.
3250 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3251 parameter to opt_scalar_int_mode.
3253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3254 Alan Hayward <alan.hayward@arm.com>
3255 David Sherwood <david.sherwood@arm.com>
3257 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3258 to a scalar_int_mode instead of a machine_mode.
3259 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3260 (get_best_mode): Return a boolean and use a pointer argument to store
3261 the selected mode. Replace the limit mode parameter with a bit limit.
3262 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3263 for the values returned by bit_field_mode_iterator::next_mode.
3264 (store_bit_field): Update call to get_best_mode.
3265 (store_fixed_bit_field): Likewise.
3266 (extract_fixed_bit_field): Likewise.
3267 * expr.c (optimize_bitfield_assignment_op): Likewise.
3268 * fold-const.c (optimize_bit_field_compare): Likewise.
3269 (fold_truth_andor_1): Likewise.
3270 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3271 Update for new type of m_mode.
3272 (get_best_mode): As above.
3274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3275 Alan Hayward <alan.hayward@arm.com>
3276 David Sherwood <david.sherwood@arm.com>
3278 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3279 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3280 (store_bit_field): Check is_a <scalar_int_mode> before calling
3281 strict_volatile_bitfield_p.
3282 (extract_bit_field): Likewise.
3284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3285 Alan Hayward <alan.hayward@arm.com>
3286 David Sherwood <david.sherwood@arm.com>
3288 * target.def (cstore_mode): Return a scalar_int_mode.
3289 * doc/tm.texi: Regenerate.
3290 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3291 * targhooks.h (default_cstore_mode): Likewise.
3292 * targhooks.c (default_cstore_mode): Likewise, using a forced
3294 * expmed.c (emit_cstore): Expect the target of the cstore to be
3297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3298 Alan Hayward <alan.hayward@arm.com>
3299 David Sherwood <david.sherwood@arm.com>
3301 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3303 (niter_desc): Likewise mode.
3304 (iv_analyze): Add a mode parameter.
3306 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3307 and change its type to scalar_int_mode.
3308 * loop-iv.c: Update commentary at head of file.
3309 (iv_constant): Pass the mode paraeter before the rtx it describes
3310 and change its type to scalar_int_mode. Remove VOIDmode handling.
3311 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3312 (iv_extend): Likewise.
3313 (shorten_into_mode): Likewise.
3314 (iv_add): Use scalar_int_mode.
3315 (iv_mult): Likewise.
3316 (iv_shift): Likewise.
3317 (canonicalize_iv_subregs): Likewise.
3318 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3319 it describes and change its mode to scalar_int_mode. Also change
3320 the type of the returned inner_mode to scalar_int_mode.
3321 (get_biv_step): Likewise, turning outer_mode from a pointer
3322 into a direct parameter. Update call to get_biv_step_1.
3323 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3324 iv_constant and get_biv_step.
3325 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3326 and change its type to scalar_int_mode. Don't initialise iv->mode
3327 to VOIDmode and remove later checks for its still being VOIDmode.
3328 Update calls to iv_analyze_op and iv_analyze_expr. Check
3329 is_a <scalar_int_mode> when changing the mode under consideration.
3330 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3331 Update call to iv_analyze_expr.
3332 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3333 inner register is not also a scalar_int_mode. Update call to
3335 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3336 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3337 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3338 separate mode class checks. Update calls to iv_analyze. Remove
3339 fix-up of VOIDmodes after iv_analyze_biv.
3340 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3341 don't have a scalar_int_mode. Update call to biv_p.
3343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3344 Alan Hayward <alan.hayward@arm.com>
3345 David Sherwood <david.sherwood@arm.com>
3347 * cfgexpand.c (convert_debug_memory_address): Use
3348 as_a <scalar_int_mode>.
3349 * combine.c (expand_compound_operation): Likewise.
3350 (make_extraction): Likewise.
3351 (change_zero_ext): Likewise.
3352 (simplify_comparison): Likewise.
3353 * cse.c (cse_insn): Likewise.
3354 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3355 (mem_loc_descriptor): Likewise.
3356 (loc_descriptor): Likewise.
3357 * expmed.c (init_expmed_one_mode): Likewise.
3358 (synth_mult): Likewise.
3359 (emit_store_flag_1): Likewise.
3360 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3361 of a comparison with size.
3362 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3363 (reduce_to_bit_field_precision): Likewise.
3364 * function.c (expand_function_end): Likewise.
3365 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3366 * loop-doloop.c (doloop_modify): Likewise.
3367 * optabs.c (expand_binop): Likewise.
3368 (expand_unop): Likewise.
3369 (expand_copysign_absneg): Likewise.
3370 (prepare_cmp_insn): Likewise.
3371 (maybe_legitimize_operand): Likewise.
3372 * recog.c (const_scalar_int_operand): Likewise.
3373 * rtlanal.c (get_address_mode): Likewise.
3374 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3375 (simplify_cond_clz_ctz): Likewise.
3376 * tree-nested.c (get_nl_goto_field): Likewise.
3377 * tree.c (build_vector_type_for_mode): Likewise.
3378 * var-tracking.c (use_narrower_mode): Likewise.
3380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3381 Alan Hayward <alan.hayward@arm.com>
3382 David Sherwood <david.sherwood@arm.com>
3384 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3385 * builtins.c (expand_builtin_signbit): Use it.
3386 * cfgexpand.c (expand_debug_expr): Likewise.
3387 * dojump.c (do_jump): Likewise.
3388 (do_compare_and_jump): Likewise.
3389 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3390 * expmed.c (make_tree): Likewise.
3391 * expr.c (expand_expr_real_2): Likewise.
3392 (expand_expr_real_1): Likewise.
3393 (try_casesi): Likewise.
3394 * fold-const-call.c (fold_const_call_ss): Likewise.
3395 * fold-const.c (unextend): Likewise.
3396 (extract_muldiv_1): Likewise.
3397 (fold_single_bit_test): Likewise.
3398 (native_encode_int): Likewise.
3399 (native_encode_string): Likewise.
3400 (native_interpret_int): Likewise.
3401 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3402 * internal-fn.c (expand_addsub_overflow): Likewise.
3403 (expand_neg_overflow): Likewise.
3404 (expand_mul_overflow): Likewise.
3405 (expand_arith_overflow): Likewise.
3406 * match.pd: Likewise.
3407 * stor-layout.c (layout_type): Likewise.
3408 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3409 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3410 * tree-ssanames.c (get_range_info): Likewise.
3411 * tree-switch-conversion.c (array_value_type) Likewise.
3412 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3413 (vect_recog_divmod_pattern): Likewise.
3414 (vect_recog_mixed_size_cond_pattern): Likewise.
3415 * tree-vrp.c (extract_range_basic): Likewise.
3416 (simplify_float_conversion_using_ranges): Likewise.
3417 * tree.c (int_fits_type_p): Likewise.
3418 * ubsan.c (instrument_bool_enum_load): Likewise.
3419 * varasm.c (mergeable_string_section): Likewise.
3420 (narrowing_initializer_constant_valid_p): Likewise.
3421 (output_constant): Likewise.
3423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3424 Alan Hayward <alan.hayward@arm.com>
3425 David Sherwood <david.sherwood@arm.com>
3427 * machmode.h (NARROWEST_INT_MODE): New macro.
3428 * expr.c (alignment_for_piecewise_move): Use it instead of
3429 GET_CLASS_NARROWEST_MODE (MODE_INT).
3430 (push_block): Likewise.
3431 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3433 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3436 Alan Hayward <alan.hayward@arm.com>
3437 David Sherwood <david.sherwood@arm.com>
3439 * postreload.c (move2add_valid_value_p): Change the type of the
3440 mode parameter to scalar_int_mode.
3441 (move2add_use_add2_insn): Add a mode parameter and use it instead
3443 (move2add_use_add3_insn): Likewise.
3444 (reload_cse_move2add): Update accordingly.
3446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3447 Alan Hayward <alan.hayward@arm.com>
3448 David Sherwood <david.sherwood@arm.com>
3450 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3452 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3453 * optabs.c (expand_unop): Likewise.
3455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3456 Alan Hayward <alan.hayward@arm.com>
3457 David Sherwood <david.sherwood@arm.com>
3459 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3460 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3461 (popcount_loc_descriptor): Likewise.
3462 (bswap_loc_descriptor): Likewise.
3463 (rotate_loc_descriptor): Likewise.
3464 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3465 calling the functions above.
3467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3468 Alan Hayward <alan.hayward@arm.com>
3469 David Sherwood <david.sherwood@arm.com>
3471 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3473 (try_combine): Likewise.
3474 (simplify_if_then_else): Likewise.
3475 * cse.c (cse_insn): Likewise.
3476 * dwarf2out.c (mem_loc_descriptor): Likewise.
3477 * emit-rtl.c (gen_lowpart_common): Likewise.
3478 * simplify-rtx.c (simplify_truncation): Likewise.
3479 (simplify_binary_operation_1): Likewise.
3480 (simplify_const_relational_operation): Likewise.
3481 (simplify_ternary_operation): Likewise.
3482 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3485 Alan Hayward <alan.hayward@arm.com>
3486 David Sherwood <david.sherwood@arm.com>
3488 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3489 * reload.c (push_reload): Likewise.
3490 (find_reloads): Likewise.
3492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3493 Alan Hayward <alan.hayward@arm.com>
3494 David Sherwood <david.sherwood@arm.com>
3496 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3497 (make_compound_operation_int): Likewise.
3498 (change_zero_ext): Likewise.
3499 * expr.c (convert_move): Likewise.
3500 (convert_modes): Likewise.
3501 * fwprop.c (forward_propagate_subreg): Likewise.
3502 * loop-iv.c (get_biv_step_1): Likewise.
3503 * optabs.c (widen_operand): Likewise.
3504 * postreload.c (move2add_valid_value_p): Likewise.
3505 * recog.c (simplify_while_replacing): Likewise.
3506 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3507 (simplify_binary_operation_1): Likewise. Remove redundant
3508 mode equality check.
3510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3511 Alan Hayward <alan.hayward@arm.com>
3512 David Sherwood <david.sherwood@arm.com>
3514 * combine.c (combine_simplify_rtx): Add checks for
3515 is_a <scalar_int_mode>.
3516 (simplify_if_then_else): Likewise.
3517 (make_field_assignment): Likewise.
3518 (simplify_comparison): Likewise.
3519 * ifcvt.c (noce_try_bitop): Likewise.
3520 * loop-invariant.c (canonicalize_address_mult): Likewise.
3521 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3524 Alan Hayward <alan.hayward@arm.com>
3525 David Sherwood <david.sherwood@arm.com>
3527 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3528 is_a <scalar_int_mode> instead of != BLKmode.
3530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3531 Alan Hayward <alan.hayward@arm.com>
3532 David Sherwood <david.sherwood@arm.com>
3534 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3535 instead of != VOIDmode.
3536 * combine.c (if_then_else_cond): Likewise.
3537 (change_zero_ext): Likewise.
3538 * dwarf2out.c (mem_loc_descriptor): Likewise.
3539 (loc_descriptor): Likewise.
3540 * rtlanal.c (canonicalize_condition): Likewise.
3541 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3544 Alan Hayward <alan.hayward@arm.com>
3545 David Sherwood <david.sherwood@arm.com>
3547 * simplify-rtx.c (simplify_binary_operation_1): Use
3548 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3551 Alan Hayward <alan.hayward@arm.com>
3552 David Sherwood <david.sherwood@arm.com>
3554 * wide-int.h (int_traits<unsigned char>) New class.
3555 (int_traits<unsigned short>) Likewise.
3556 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3557 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3559 * combine.c (set_nonzero_bits_and_sign_copies): Use
3560 is_a <scalar_int_mode>.
3561 (find_split_point): Likewise.
3562 (combine_simplify_rtx): Likewise.
3563 (simplify_logical): Likewise.
3564 (expand_compound_operation): Likewise.
3565 (expand_field_assignment): Likewise.
3566 (make_compound_operation): Likewise.
3567 (extended_count): Likewise.
3568 (change_zero_ext): Likewise.
3569 (simplify_comparison): Likewise.
3570 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3571 (ucompare_loc_descriptor): Likewise.
3572 (minmax_loc_descriptor): Likewise.
3573 (mem_loc_descriptor): Likewise.
3574 (loc_descriptor): Likewise.
3575 * expmed.c (init_expmed_one_mode): Likewise.
3576 * lra-constraints.c (lra_constraint_offset): Likewise.
3577 * optabs.c (prepare_libcall_arg): Likewise.
3578 * postreload.c (move2add_note_store): Likewise.
3579 * reload.c (operands_match_p): Likewise.
3580 * rtl.h (load_extend_op): Likewise.
3581 * rtlhooks.c (gen_lowpart_general): Likewise.
3582 * simplify-rtx.c (simplify_truncation): Likewise.
3583 (simplify_unary_operation_1): Likewise.
3584 (simplify_binary_operation_1): Likewise.
3585 (simplify_const_binary_operation): Likewise.
3586 (simplify_const_relational_operation): Likewise.
3587 (simplify_subreg): Likewise.
3588 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3589 * var-tracking.c (adjust_mems): Likewise.
3590 (prepare_call_arguments): Likewise.
3592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3593 Alan Hayward <alan.hayward@arm.com>
3594 David Sherwood <david.sherwood@arm.com>
3596 * machmode.h (is_int_mode): New fuction.
3597 * combine.c (find_split_point): Use it.
3598 (combine_simplify_rtx): Likewise.
3599 (simplify_if_then_else): Likewise.
3600 (simplify_set): Likewise.
3601 (simplify_shift_const_1): Likewise.
3602 (simplify_comparison): Likewise.
3603 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3604 * cse.c (notreg_cost): Likewise.
3605 (cse_insn): Likewise.
3606 * cselib.c (cselib_lookup_1): Likewise.
3607 * dojump.c (do_jump_1): Likewise.
3608 (do_compare_rtx_and_jump): Likewise.
3609 * dse.c (get_call_args): Likewise.
3610 * dwarf2out.c (rtl_for_decl_init): Likewise.
3611 (native_encode_initializer): Likewise.
3612 * expmed.c (emit_store_flag_1): Likewise.
3613 (emit_store_flag): Likewise.
3614 * expr.c (convert_modes): Likewise.
3615 (store_field): Likewise.
3616 (expand_expr_real_1): Likewise.
3617 * fold-const.c (fold_read_from_constant_string): Likewise.
3618 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3619 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3620 * optabs.c (expand_binop): Likewise.
3621 (expand_unop): Likewise.
3622 (expand_abs_nojump): Likewise.
3623 (expand_one_cmpl_abs_nojump): Likewise.
3624 * simplify-rtx.c (mode_signbit_p): Likewise.
3625 (val_signbit_p): Likewise.
3626 (val_signbit_known_set_p): Likewise.
3627 (val_signbit_known_clear_p): Likewise.
3628 (simplify_relational_operation_1): Likewise.
3629 * tree.c (vector_type_mode): Likewise.
3631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3632 Alan Hayward <alan.hayward@arm.com>
3633 David Sherwood <david.sherwood@arm.com>
3635 * machmode.h (smallest_mode_for_size): Fix formatting.
3636 (smallest_int_mode_for_size): New function.
3637 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3638 instead of smallest_mode_for_size.
3639 * combine.c (make_extraction): Likewise.
3640 * config/arc/arc.c (arc_expand_movmem): Likewise.
3641 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3642 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3643 * config/s390/s390.c (s390_expand_insv): Likewise.
3644 * config/sparc/sparc.c (assign_int_registers): Likewise.
3645 * config/spu/spu.c (spu_function_value): Likewise.
3646 (spu_function_arg): Likewise.
3647 * coverage.c (get_gcov_type): Likewise.
3648 (get_gcov_unsigned_t): Likewise.
3649 * dse.c (find_shift_sequence): Likewise.
3650 * expmed.c (store_bit_field_1): Likewise.
3651 * expr.c (convert_move): Likewise.
3652 (store_field): Likewise.
3653 * internal-fn.c (expand_arith_overflow): Likewise.
3654 * optabs-query.c (get_best_extraction_insn): Likewise.
3655 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3656 * stor-layout.c (layout_type): Likewise.
3657 (initialize_sizetypes): Likewise.
3658 * targhooks.c (default_get_mask_mode): Likewise.
3659 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3662 Alan Hayward <alan.hayward@arm.com>
3663 David Sherwood <david.sherwood@arm.com>
3665 * machmode.h (opt_mode::else_blk): New function.
3666 (int_mode_for_mode): Declare.
3667 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3668 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3670 * cfgexpand.c (expand_debug_expr): Likewise.
3671 * combine.c (gen_lowpart_or_truncate): Likewise.
3672 (gen_lowpart_for_combine): Likewise.
3673 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3674 * config/avr/avr.c (avr_to_int_mode): Likewise.
3675 (avr_out_plus_1): Likewise.
3676 (avr_out_plus): Likewise.
3677 (avr_out_round): Likewise.
3678 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3679 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3680 (s390_expand_vcond): Likewise.
3681 * config/spu/spu.c (spu_split_immediate): Likewise.
3682 (spu_expand_mov): Likewise.
3683 * dse.c (get_stored_val): Likewise.
3684 * expmed.c (store_bit_field_1): Likewise.
3685 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3687 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3688 (extract_low_bits): Likewise.
3689 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3690 handling rather than repeating the check.
3691 (emit_group_store): Likewise.
3692 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3693 * optabs.c (expand_absneg_bit): Likewise.
3694 (expand_copysign_absneg): Likewise.
3695 (expand_copysign_bit): Likewise.
3696 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3697 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3698 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3699 * var-tracking.c (prepare_call_arguments): Likewise.
3700 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3701 int_mode_for_mode instead of mode_for_size.
3702 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3705 Alan Hayward <alan.hayward@arm.com>
3706 David Sherwood <david.sherwood@arm.com>
3708 * machmode.h (int_mode_for_size): New function.
3709 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3710 instead of mode_for_size.
3711 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3713 * combine.c (expand_field_assignment): Use int_mode_for_size
3714 instead of mode_for_size.
3715 (make_extraction): Likewise.
3716 (simplify_shift_const_1): Likewise.
3717 (simplify_comparison): Likewise.
3718 * dojump.c (do_jump): Likewise.
3719 * dwarf2out.c (mem_loc_descriptor): Likewise.
3720 * emit-rtl.c (init_derived_machine_modes): Likewise.
3721 * expmed.c (flip_storage_order): Likewise.
3722 (convert_extracted_bit_field): Likewise.
3723 * expr.c (copy_blkmode_from_reg): Likewise.
3724 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3725 * internal-fn.c (expand_mul_overflow): Likewise.
3726 * lower-subreg.c (simple_move): Likewise.
3727 * optabs-libfuncs.c (init_optabs): Likewise.
3728 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3729 * tree.c (vector_type_mode): Likewise.
3730 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3731 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3732 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3733 * tree-vect-stmts.c (vectorizable_load): Likewise.
3734 (vectorizable_store): Likewise.
3736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3737 Alan Hayward <alan.hayward@arm.com>
3738 David Sherwood <david.sherwood@arm.com>
3740 * coretypes.h (pod_mode): New type.
3741 (scalar_int_mode_pod): New typedef.
3742 * machmode.h (pod_mode): New class.
3743 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3744 * genmodes.c (emit_mode_int_n): Update accordingly.
3745 * lower-subreg.h (target_lower_subreg): Change type to
3746 scalar_int_mode_pod.
3747 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3748 scalar_int_mode_pod.
3750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3751 Alan Hayward <alan.hayward@arm.com>
3752 David Sherwood <david.sherwood@arm.com>
3754 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3755 machine_mode to scalar_int_mode.
3756 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3757 (rs6000_option_override_internal): Remove cast to int.
3758 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3759 machine_mode to scalar_int_mode.
3760 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3761 (rs6000_option_override_internal): Remove cast to int.
3762 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3763 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3765 * config/s390/s390.c (s390_expand_builtin): Likewise.
3766 * coretypes.h (scalar_int_mode): New type.
3767 (opt_scalar_int_mode): New typedef.
3768 * machmode.h (scalar_int_mode): New class.
3769 (scalar_int_mode::includes_p): New function.
3770 (byte_mode): Change type to scalar_int_mode.
3771 (word_mode): Likewise.
3772 (ptr_mode): Likewise.
3773 * emit-rtl.c (byte_mode): Likewise.
3774 (word_mode): Likewise.
3775 (ptr_mode): Likewise.
3776 (init_derived_machine_modes): Update accordingly.
3777 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3778 and MODE_PARTIAL_INT.
3779 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3780 opt_scalar_int_mode.
3782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3783 Alan Hayward <alan.hayward@arm.com>
3784 David Sherwood <david.sherwood@arm.com>
3786 * target.def (libgcc_floating_mode_supported_p): Take a
3788 * doc/tm.texi: Regenerate.
3789 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3791 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3792 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3796 Alan Hayward <alan.hayward@arm.com>
3797 David Sherwood <david.sherwood@arm.com>
3799 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3800 * doc/tm.texi: Regenerate.
3801 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3802 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3803 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3804 * targhooks.h (default_floatn_mode): Likewise.
3805 * targhooks.c (default_floatn_mode): Likewise.
3806 * tree.c (build_common_tree_nodes): Update accordingly.
3808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3809 Alan Hayward <alan.hayward@arm.com>
3810 David Sherwood <david.sherwood@arm.com>
3812 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3813 (mode_iterator::iterate_p): Likewise.
3814 (mode_iterator::get_wider): Likewise.
3815 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3818 Alan Hayward <alan.hayward@arm.com>
3819 David Sherwood <david.sherwood@arm.com>
3821 * coretypes.h (opt_scalar_float_mode): New typedef.
3822 * machmode.h (float_mode_for_size): New function.
3823 * emit-rtl.c (double_mode): Delete.
3824 (init_emit_once): Use float_mode_for_size.
3825 * stor-layout.c (layout_type): Likewise.
3826 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3829 Alan Hayward <alan.hayward@arm.com>
3830 David Sherwood <david.sherwood@arm.com>
3832 * output.h (assemble_real): Take a scalar_float_mode.
3833 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3834 * config/arm/arm.md (consttable_4): Likewise.
3835 (consttable_8): Likewise.
3836 (consttable_16): Likewise.
3837 * config/mips/mips.md (consttable_float): Likewise.
3838 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3839 * varasm.c (assemble_real): Take a scalar_float_mode.
3840 (output_constant_pool_2): Update accordingly.
3841 (output_constant): Likewise.
3843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3844 Alan Hayward <alan.hayward@arm.com>
3845 David Sherwood <david.sherwood@arm.com>
3847 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3848 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3849 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3850 (native_encode_real): Likewise.
3851 (native_interpret_real): Likewise.
3852 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3853 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3856 Alan Hayward <alan.hayward@arm.com>
3857 David Sherwood <david.sherwood@arm.com>
3859 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3860 <scalar_float_mode>. Simplify.
3861 (gen_extend_conv_libfunc): Likewise.
3863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3864 Alan Hayward <alan.hayward@arm.com>
3865 David Sherwood <david.sherwood@arm.com>
3867 * coretypes.h (scalar_float_mode): New type.
3868 * machmode.h (mode_traits::from_int): Use machine_mode if
3869 USE_ENUM_MODES is defined.
3870 (is_a): New function.
3872 (dyn_cast): Likewise.
3873 (scalar_float_mode): New class.
3874 (scalar_float_mode::includes_p): New function.
3875 (is_float_mode): Likewise.
3876 * gdbhooks.py (MachineModePrinter): New class.
3877 (build_pretty_printer): Use it for scalar_float_mode.
3878 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3879 (format_helper::format_helper): Turn into a template.
3880 * genmodes.c (get_mode_class): New function.
3881 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3882 or machine_mode if none.
3883 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3884 as_a <scalar_float_mode>.
3885 * dwarf2out.c (mem_loc_descriptor): Likewise.
3886 (insert_float): Likewise.
3887 (add_const_value_attribute): Likewise.
3888 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3889 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3890 (expand_unop): Update accordingly.
3891 (expand_abs_nojump): Likewise.
3892 (expand_copysign_absneg): Take a scalar_float_mode.
3893 (expand_copysign_bit): Likewise.
3894 (expand_copysign): Update accordingly.
3896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3897 Alan Hayward <alan.hayward@arm.com>
3898 David Sherwood <david.sherwood@arm.com>
3900 * coretypes.h (opt_mode): New class.
3901 * machmode.h (opt_mode): Likewise.
3902 (opt_mode::else_void): New function.
3903 (opt_mode::require): Likewise.
3904 (opt_mode::exists): Likewise.
3905 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3906 (GET_MODE_2XWIDER_MODE): Likewise.
3907 (mode_iterator::get_wider): Update accordingly.
3908 (mode_iterator::get_2xwider): Likewise.
3909 (mode_iterator::get_known_wider): Likewise, turning into a template.
3910 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3911 forcing a wider mode to exist.
3912 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3913 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3914 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3915 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3916 * lower-subreg.c (init_lower_subreg): Likewise.
3917 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3918 on the final iteration.
3919 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3920 a wider mode exists before asking for a move pattern.
3921 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3922 forcing a wider mode to exist.
3923 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3924 returning false if no such mode exists.
3925 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3926 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3927 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3928 Avoid checking for a MODE_INT if we already know the mode is not a
3930 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3931 forcing a wider mode to exist.
3932 (expmed_mult_highpart_optab): Likewise.
3933 (expmed_mult_highpart): Likewise.
3934 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3936 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3937 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3938 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3939 * internal-fn.c (expand_mul_overflow): Update use of
3940 GET_MODE_2XWIDER_MODE.
3941 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3942 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3943 GET_MODE_WIDER_MODE.
3944 (convert_plusminus_to_widen): Likewise.
3945 * tree-switch-conversion.c (array_value_type): Likewise.
3946 * var-tracking.c (emit_note_insn_var_location): Likewise.
3947 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3948 Return false inside rather than outside the loop if no wider mode
3950 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3951 and GET_MODE_2XWIDER_MODE
3952 (can_compare_p): Use else_void.
3953 * gdbhooks.py (OptMachineModePrinter): New class.
3954 (build_pretty_printer): Use it for opt_mode.
3956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3957 Alan Hayward <alan.hayward@arm.com>
3958 David Sherwood <david.sherwood@arm.com>
3960 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3961 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3964 Alan Hayward <alan.hayward@arm.com>
3965 David Sherwood <david.sherwood@arm.com>
3967 * machmode.h (mode_traits): New structure.
3968 (get_narrowest_mode): New function.
3969 (mode_iterator::start): Likewise.
3970 (mode_iterator::iterate_p): Likewise.
3971 (mode_iterator::get_wider): Likewise.
3972 (mode_iterator::get_known_wider): Likewise.
3973 (mode_iterator::get_2xwider): Likewise.
3974 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3975 (FOR_EACH_MODE): Likewise.
3976 (FOR_EACH_MODE_FROM): Likewise.
3977 (FOR_EACH_MODE_UNTIL): Likewise.
3978 (FOR_EACH_WIDER_MODE): Likewise.
3979 (FOR_EACH_2XWIDER_MODE): Likewise.
3980 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3981 * combine.c (simplify_comparison): Likewise
3982 * config/i386/i386.c (type_natural_mode): Likewise.
3983 * cse.c (cse_insn): Likewise.
3984 * dse.c (find_shift_sequence): Likewise.
3985 * emit-rtl.c (init_derived_machine_modes): Likewise.
3986 (init_emit_once): Likewise.
3987 * explow.c (hard_function_value): Likewise.
3988 * expmed.c (extract_fixed_bit_field_1): Likewise.
3989 (extract_bit_field_1): Likewise.
3990 (expand_divmod): Likewise.
3991 (emit_store_flag_1): Likewise.
3992 * expr.c (init_expr_target): Likewise.
3993 (convert_move): Likewise.
3994 (alignment_for_piecewise_move): Likewise.
3995 (widest_int_mode_for_size): Likewise.
3996 (emit_block_move_via_movmem): Likewise.
3997 (copy_blkmode_to_reg): Likewise.
3998 (set_storage_via_setmem): Likewise.
3999 (compress_float_constant): Likewise.
4000 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4001 * optabs-query.c (get_best_extraction_insn): Likewise.
4002 * optabs.c (expand_binop): Likewise.
4003 (expand_twoval_unop): Likewise.
4004 (expand_twoval_binop): Likewise.
4005 (widen_leading): Likewise.
4006 (widen_bswap): Likewise.
4007 (expand_parity): Likewise.
4008 (expand_unop): Likewise.
4009 (prepare_cmp_insn): Likewise.
4010 (prepare_float_lib_cmp): Likewise.
4011 (expand_float): Likewise.
4012 (expand_fix): Likewise.
4013 (expand_sfix_optab): Likewise.
4014 * postreload.c (move2add_use_add2_insn): Likewise.
4015 * reg-stack.c (reg_to_stack): Likewise.
4016 * reginfo.c (choose_hard_reg_mode): Likewise.
4017 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4018 * stor-layout.c (mode_for_size): Likewise.
4019 (smallest_mode_for_size): Likewise.
4020 (mode_for_vector): Likewise.
4021 (finish_bitfield_representative): Likewise.
4022 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4023 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4024 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4025 * var-tracking.c (prepare_call_arguments): Likewise.
4027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4028 Alan Hayward <alan.hayward@arm.com>
4029 David Sherwood <david.sherwood@arm.com>
4031 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4032 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4033 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4034 * machmode.h (mode_size): Move earlier in file.
4035 (mode_precision): Likewise.
4036 (mode_inner): Likewise.
4037 (mode_nunits): Likewise.
4038 (mode_unit_size): Likewise.
4039 (unit_unit_precision): Likewise.
4040 (mode_wider): Likewise.
4041 (mode_2xwider): Likewise.
4042 (machine_mode): New class.
4043 (mode_to_bytes): New function.
4044 (mode_to_bits): Likewise.
4045 (mode_to_precision): Likewise.
4046 (mode_to_inner): Likewise.
4047 (mode_to_unit_size): Likewise.
4048 (mode_to_unit_precision): Likewise.
4049 (mode_to_nunits): Likewise.
4050 (GET_MODE_SIZE): Use mode_to_bytes.
4051 (GET_MODE_BITSIZE): Use mode_to_bits.
4052 (GET_MODE_PRECISION): Use mode_to_precision.
4053 (GET_MODE_INNER): Use mode_to_inner.
4054 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4055 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4056 (GET_MODE_NUNITS): Use mode_to_nunits.
4057 * system.h (ALWAYS_INLINE): New macro.
4058 * config/powerpcspe/powerpcspe-c.c
4059 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4060 int for arg1_mode and arg2_mode.
4062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4063 Alan Hayward <alan.hayward@arm.com>
4064 David Sherwood <david.sherwood@arm.com>
4066 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4067 Prefix mode names with E_ in case statements.
4068 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4069 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4070 (aarch64_split_simd_move): Likewise.
4071 (aarch64_gen_storewb_pair): Likewise.
4072 (aarch64_gen_loadwb_pair): Likewise.
4073 (aarch64_gen_store_pair): Likewise.
4074 (aarch64_gen_load_pair): Likewise.
4075 (aarch64_get_condition_code_1): Likewise.
4076 (aarch64_constant_pool_reload_icode): Likewise.
4077 (get_rsqrte_type): Likewise.
4078 (get_rsqrts_type): Likewise.
4079 (get_recpe_type): Likewise.
4080 (get_recps_type): Likewise.
4081 (aarch64_gimplify_va_arg_expr): Likewise.
4082 (aarch64_simd_container_mode): Likewise.
4083 (aarch64_emit_load_exclusive): Likewise.
4084 (aarch64_emit_store_exclusive): Likewise.
4085 (aarch64_expand_compare_and_swap): Likewise.
4086 (aarch64_gen_atomic_cas): Likewise.
4087 (aarch64_emit_bic): Likewise.
4088 (aarch64_emit_atomic_swap): Likewise.
4089 (aarch64_emit_atomic_load_op): Likewise.
4090 (aarch64_evpc_trn): Likewise.
4091 (aarch64_evpc_uzp): Likewise.
4092 (aarch64_evpc_zip): Likewise.
4093 (aarch64_evpc_ext): Likewise.
4094 (aarch64_evpc_rev): Likewise.
4095 (aarch64_evpc_dup): Likewise.
4096 (aarch64_gen_ccmp_first): Likewise.
4097 (aarch64_gen_ccmp_next): Likewise.
4098 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4099 (alpha_emit_xfloating_libcall): Likewise.
4100 (emit_insxl): Likewise.
4101 (alpha_arg_type): Likewise.
4102 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4103 (arc_preferred_simd_mode): Likewise.
4104 (arc_secondary_reload): Likewise.
4105 (get_arc_condition_code): Likewise.
4106 (arc_print_operand): Likewise.
4107 (arc_legitimate_constant_p): Likewise.
4108 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4109 * config/arc/arc.md (casesi_load): Likewise.
4110 (casesi_compact_jump): Likewise.
4111 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4112 (cc_use_register): Likewise.
4113 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4114 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4115 (arm_init_iwmmxt_builtins): Likewise.
4116 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4117 (neon_expand_vector_init): Likewise.
4118 (arm_attr_length_move_neon): Likewise.
4119 (maybe_get_arm_condition_code): Likewise.
4120 (arm_emit_vector_const): Likewise.
4121 (arm_preferred_simd_mode): Likewise.
4122 (arm_output_iwmmxt_tinsr): Likewise.
4123 (thumb1_output_casesi): Likewise.
4124 (thumb2_output_casesi): Likewise.
4125 (arm_emit_load_exclusive): Likewise.
4126 (arm_emit_store_exclusive): Likewise.
4127 (arm_expand_compare_and_swap): Likewise.
4128 (arm_evpc_neon_vuzp): Likewise.
4129 (arm_evpc_neon_vzip): Likewise.
4130 (arm_evpc_neon_vrev): Likewise.
4131 (arm_evpc_neon_vtrn): Likewise.
4132 (arm_evpc_neon_vext): Likewise.
4133 (arm_validize_comparison): Likewise.
4134 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4135 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4136 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4137 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4138 (c6x_preferred_simd_mode): Likewise.
4139 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4140 (epiphany_rtx_costs): Likewise.
4141 * config/epiphany/predicates.md (proper_comparison_operator):
4143 * config/frv/frv.c (condexec_memory_operand): Likewise.
4144 (frv_emit_move): Likewise.
4145 (output_move_single): Likewise.
4146 (output_condmove_single): Likewise.
4147 (frv_hard_regno_mode_ok): Likewise.
4148 (frv_matching_accg_mode): Likewise.
4149 * config/h8300/h8300.c (split_adds_subs): Likewise.
4150 (h8300_rtx_costs): Likewise.
4151 (h8300_print_operand): Likewise.
4152 (compute_mov_length): Likewise.
4153 (output_logical_op): Likewise.
4154 (compute_logical_op_length): Likewise.
4155 (compute_logical_op_cc): Likewise.
4156 (h8300_shift_needs_scratch_p): Likewise.
4157 (output_a_shift): Likewise.
4158 (compute_a_shift_length): Likewise.
4159 (compute_a_shift_cc): Likewise.
4160 (expand_a_rotate): Likewise.
4161 (output_a_rotate): Likewise.
4162 * config/i386/i386.c (classify_argument): Likewise.
4163 (function_arg_advance_32): Likewise.
4164 (function_arg_32): Likewise.
4165 (function_arg_64): Likewise.
4166 (function_value_64): Likewise.
4167 (ix86_gimplify_va_arg): Likewise.
4168 (ix86_legitimate_constant_p): Likewise.
4169 (put_condition_code): Likewise.
4170 (split_double_mode): Likewise.
4171 (ix86_avx256_split_vector_move_misalign): Likewise.
4172 (ix86_expand_vector_logical_operator): Likewise.
4173 (ix86_split_idivmod): Likewise.
4174 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4175 (ix86_build_const_vector): Likewise.
4176 (ix86_build_signbit_mask): Likewise.
4177 (ix86_match_ccmode): Likewise.
4178 (ix86_cc_modes_compatible): Likewise.
4179 (ix86_expand_branch): Likewise.
4180 (ix86_expand_sse_cmp): Likewise.
4181 (ix86_expand_sse_movcc): Likewise.
4182 (ix86_expand_int_sse_cmp): Likewise.
4183 (ix86_expand_vec_perm_vpermi2): Likewise.
4184 (ix86_expand_vec_perm): Likewise.
4185 (ix86_expand_sse_unpack): Likewise.
4186 (ix86_expand_int_addcc): Likewise.
4187 (ix86_split_to_parts): Likewise.
4188 (ix86_vectorize_builtin_gather): Likewise.
4189 (ix86_vectorize_builtin_scatter): Likewise.
4190 (avx_vpermilp_parallel): Likewise.
4191 (inline_memory_move_cost): Likewise.
4192 (ix86_tieable_integer_mode_p): Likewise.
4193 (x86_maybe_negate_const_int): Likewise.
4194 (ix86_expand_vector_init_duplicate): Likewise.
4195 (ix86_expand_vector_init_one_nonzero): Likewise.
4196 (ix86_expand_vector_init_one_var): Likewise.
4197 (ix86_expand_vector_init_concat): Likewise.
4198 (ix86_expand_vector_init_interleave): Likewise.
4199 (ix86_expand_vector_init_general): Likewise.
4200 (ix86_expand_vector_set): Likewise.
4201 (ix86_expand_vector_extract): Likewise.
4202 (emit_reduc_half): Likewise.
4203 (ix86_emit_i387_round): Likewise.
4204 (ix86_mangle_type): Likewise.
4205 (ix86_expand_round_sse4): Likewise.
4206 (expand_vec_perm_blend): Likewise.
4207 (canonicalize_vector_int_perm): Likewise.
4208 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4209 (expand_vec_perm_1): Likewise.
4210 (expand_vec_perm_interleave3): Likewise.
4211 (expand_vec_perm_even_odd_pack): Likewise.
4212 (expand_vec_perm_even_odd_1): Likewise.
4213 (expand_vec_perm_broadcast_1): Likewise.
4214 (ix86_vectorize_vec_perm_const_ok): Likewise.
4215 (ix86_expand_vecop_qihi): Likewise.
4216 (ix86_expand_mul_widen_hilo): Likewise.
4217 (ix86_expand_sse2_abs): Likewise.
4218 (ix86_expand_pextr): Likewise.
4219 (ix86_expand_pinsr): Likewise.
4220 (ix86_preferred_simd_mode): Likewise.
4221 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4222 * config/i386/sse.md (*andnot<mode>3): Likewise.
4223 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4224 (*<code><mode>3): Likewise.
4225 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4226 (ia64_expand_atomic_op): Likewise.
4227 (ia64_arg_type): Likewise.
4228 (ia64_mode_to_int): Likewise.
4229 (ia64_scalar_mode_supported_p): Likewise.
4230 (ia64_vector_mode_supported_p): Likewise.
4231 (expand_vec_perm_broadcast): Likewise.
4232 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4233 (iq2000_function_arg_advance): Likewise.
4234 (iq2000_function_arg): Likewise.
4235 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4236 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4237 (m68k_libcall_value): Likewise.
4238 (m68k_function_value): Likewise.
4239 (sched_attr_op_type): Likewise.
4240 * config/mcore/mcore.c (mcore_output_move): Likewise.
4241 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4243 (microblaze_function_arg): Likewise.
4244 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4245 (mips_print_operand): Likewise.
4246 (mips_mode_ok_for_mov_fmt_p): Likewise.
4247 (mips_vector_mode_supported_p): Likewise.
4248 (mips_preferred_simd_mode): Likewise.
4249 (mips_expand_vpc_loongson_even_odd): Likewise.
4250 (mips_expand_vec_unpack): Likewise.
4251 (mips_expand_vi_broadcast): Likewise.
4252 (mips_expand_vector_init): Likewise.
4253 (mips_expand_vec_reduc): Likewise.
4254 (mips_expand_msa_cmp): Likewise.
4255 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4256 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4257 (cc_flags_for_mode): Likewise.
4258 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4259 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4260 (nds32_output_casesi_pc_relative): Likewise.
4261 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4262 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4263 (nvptx_gen_unpack): Likewise.
4264 (nvptx_gen_pack): Likewise.
4265 (nvptx_gen_shuffle): Likewise.
4266 (nvptx_gen_wcast): Likewise.
4267 (nvptx_preferred_simd_mode): Likewise.
4268 * config/pa/pa.c (pa_secondary_reload): Likewise.
4269 * config/pa/predicates.md (base14_operand): Likewise.
4270 * config/powerpcspe/powerpcspe-c.c
4271 (altivec_resolve_overloaded_builtin): Likewise.
4272 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4274 (rs6000_preferred_simd_mode): Likewise.
4275 (output_vec_const_move): Likewise.
4276 (rs6000_expand_vector_extract): Likewise.
4277 (rs6000_split_vec_extract_var): Likewise.
4278 (reg_offset_addressing_ok_p): Likewise.
4279 (rs6000_legitimate_offset_address_p): Likewise.
4280 (rs6000_legitimize_address): Likewise.
4281 (rs6000_emit_set_const): Likewise.
4282 (rs6000_const_vec): Likewise.
4283 (rs6000_emit_move): Likewise.
4284 (spe_build_register_parallel): Likewise.
4285 (rs6000_darwin64_record_arg_recurse): Likewise.
4286 (swap_selector_for_mode): Likewise.
4287 (spe_init_builtins): Likewise.
4288 (paired_init_builtins): Likewise.
4289 (altivec_init_builtins): Likewise.
4290 (do_load_for_compare): Likewise.
4291 (rs6000_generate_compare): Likewise.
4292 (rs6000_expand_float128_convert): Likewise.
4293 (emit_load_locked): Likewise.
4294 (emit_store_conditional): Likewise.
4295 (rs6000_output_function_epilogue): Likewise.
4296 (rs6000_handle_altivec_attribute): Likewise.
4297 (rs6000_function_value): Likewise.
4298 (emit_fusion_gpr_load): Likewise.
4299 (emit_fusion_p9_load): Likewise.
4300 (emit_fusion_p9_store): Likewise.
4301 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4302 (fusion_gpr_mem_load): Likewise.
4303 (fusion_addis_mem_combo_load): Likewise.
4304 (fusion_addis_mem_combo_store): Likewise.
4305 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4306 (fusion_gpr_mem_load): Likewise.
4307 (fusion_addis_mem_combo_load): Likewise.
4308 (fusion_addis_mem_combo_store): Likewise.
4309 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4311 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4312 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4313 (rs6000_preferred_simd_mode): Likewise.
4314 (output_vec_const_move): Likewise.
4315 (rs6000_expand_vector_extract): Likewise.
4316 (rs6000_split_vec_extract_var): Likewise.
4317 (reg_offset_addressing_ok_p): Likewise.
4318 (rs6000_legitimate_offset_address_p): Likewise.
4319 (rs6000_legitimize_address): Likewise.
4320 (rs6000_emit_set_const): Likewise.
4321 (rs6000_const_vec): Likewise.
4322 (rs6000_emit_move): Likewise.
4323 (rs6000_darwin64_record_arg_recurse): Likewise.
4324 (swap_selector_for_mode): Likewise.
4325 (paired_init_builtins): Likewise.
4326 (altivec_init_builtins): Likewise.
4327 (rs6000_expand_float128_convert): Likewise.
4328 (emit_load_locked): Likewise.
4329 (emit_store_conditional): Likewise.
4330 (rs6000_output_function_epilogue): Likewise.
4331 (rs6000_handle_altivec_attribute): Likewise.
4332 (rs6000_function_value): Likewise.
4333 (emit_fusion_gpr_load): Likewise.
4334 (emit_fusion_p9_load): Likewise.
4335 (emit_fusion_p9_store): Likewise.
4336 * config/rx/rx.c (rx_gen_move_template): Likewise.
4337 (flags_from_mode): Likewise.
4338 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4339 (s390_slb_comparison): Likewise.
4340 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4341 (s390_vector_mode_supported_p): Likewise.
4342 (s390_cc_modes_compatible): Likewise.
4343 (s390_match_ccmode_set): Likewise.
4344 (s390_canonicalize_comparison): Likewise.
4345 (s390_emit_compare_and_swap): Likewise.
4346 (s390_branch_condition_mask): Likewise.
4347 (s390_rtx_costs): Likewise.
4348 (s390_secondary_reload): Likewise.
4349 (__SECONDARY_RELOAD_CASE): Likewise.
4350 (s390_expand_cs): Likewise.
4351 (s390_preferred_simd_mode): Likewise.
4352 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4353 * config/sh/sh.c (sh_print_operand): Likewise.
4354 (dump_table): Likewise.
4355 (sh_secondary_reload): Likewise.
4356 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4357 * config/sh/sh.md (casesi_worker_1): Likewise.
4358 (casesi_worker_2): Likewise.
4359 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4360 (fcc_comparison_operator): Likewise.
4361 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4362 (emit_soft_tfmode_cvt): Likewise.
4363 (sparc_preferred_simd_mode): Likewise.
4364 (output_cbranch): Likewise.
4365 (sparc_print_operand): Likewise.
4366 (sparc_expand_vec_perm_bmask): Likewise.
4367 (vector_init_bshuffle): Likewise.
4368 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4369 (spu_vector_mode_supported_p): Likewise.
4370 (spu_expand_insv): Likewise.
4371 (spu_emit_branch_or_set): Likewise.
4372 (spu_handle_vector_attribute): Likewise.
4373 (spu_builtin_splats): Likewise.
4374 (spu_builtin_extract): Likewise.
4375 (spu_builtin_promote): Likewise.
4376 (spu_expand_sign_extend): Likewise.
4377 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4378 (tilegx_simd_int): Likewise.
4379 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4380 (tilepro_simd_int): Likewise.
4381 * config/v850/v850.c (const_double_split): Likewise.
4382 (v850_print_operand): Likewise.
4383 (ep_memory_offset): Likewise.
4384 * config/vax/vax.c (vax_rtx_costs): Likewise.
4385 (vax_output_int_move): Likewise.
4386 (vax_output_int_add): Likewise.
4387 (vax_output_int_subtract): Likewise.
4388 * config/visium/predicates.md (visium_branch_operator): Likewise.
4389 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4390 (visium_print_operand_address): Likewise.
4391 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4392 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4393 (xtensa_expand_conditional_branch): Likewise.
4394 (xtensa_copy_incoming_a7): Likewise.
4395 (xtensa_output_literal): Likewise.
4396 * dfp.c (decimal_real_maxval): Likewise.
4397 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4400 Alan Hayward <alan.hayward@arm.com>
4401 David Sherwood <david.sherwood@arm.com>
4403 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4404 (mode_nunits_inline): Likewise.
4405 (mode_inner_inline): Likewise.
4406 (mode_unit_size_inline): Likewise.
4407 (mode_unit_precision_inline): Likewise.
4408 (emit_insn_modes_h): Likewise. Also emit a #define of the
4410 (emit_mode_wider): Add an E_ prefix to mode names.
4411 (emit_mode_complex): Likewise.
4412 (emit_mode_inner): Likewise.
4413 (emit_mode_adjustments): Likewise.
4414 (emit_mode_int_n): Likewise.
4415 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4416 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4417 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4418 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4419 (CRC32_BUILTIN, ENTRY): Likewise.
4420 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4421 (aarch64_pop_regs): Likewise.
4422 (aarch64_process_components): Likewise.
4423 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4424 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4425 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4426 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4427 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4428 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4430 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4431 (pcmpistr): Likewise.
4432 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4433 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4434 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4436 * config/rl78/rl78.c (mduc_regs): Likewise.
4437 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4438 (htm_expand_builtin): Likewise.
4439 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4440 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4441 * config/xtensa/xtensa.c (print_operand): Likewise.
4442 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4443 (NUM_MODE_VECTOR_INT): Likewise.
4444 * genoutput.c (null_operand): Likewise.
4445 (output_operand_data): Likewise.
4446 * genrecog.c (print_parameter_value): Likewise.
4447 * lra.c (debug_operand_data): Likewise.
4449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4451 * dwarf2out.c (scompare_loc_descriptor_wide)
4452 (scompare_loc_descriptor_narrow): New functions, split out from...
4453 (scompare_loc_descriptor): ...here.
4454 * expmed.c (emit_store_flag_int): New function, split out from...
4455 (emit_store_flag): ...here.
4457 2017-08-30 Richard Biener <rguenther@suse.de>
4459 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4460 (dwarf2out_early_finish): Move setting of AT_pubnames from
4461 early debug output to early finish.
4463 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4465 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4466 and -mdata-region to the assembler.
4468 2017-08-30 Richard Biener <rguenther@suse.de>
4470 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4472 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4473 (dwarf2out_early_global_decl): Only generate a DIE for the
4474 abstract origin if it doesn't already exist or is a declaration DIE.
4475 (resolve_addr): Do not add the linkage name twice when
4476 generating a stub DIE for the DW_TAG_GNU_call_site target.
4478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4480 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4481 Use machine_mode rather than int for arg1_mode.
4483 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4486 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4487 that the second argument of the built-in functions to unpack
4488 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4489 switch statement instead a lot of if statements.
4490 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4491 Allow 64-bit values to be in Altivec registers as well as
4492 traditional floating point registers.
4493 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4495 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4497 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4498 MAX_REGS_PER_ADDRESS == 1.
4500 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4502 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4503 * config/i386/i386.c (ix86_option_override_internal): Simplify
4504 setting of opts->x_flag_entry.
4506 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4507 Jakub Jelinek <jakub@redhat.com>
4508 Richard Biener <rguenther@suse.de>
4510 PR tree-optimization/81503
4511 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4512 folded constant fits in the target type; reorder tests for clarity.
4514 2017-08-29 Martin Liska <mliska@suse.cz>
4516 * passes.def: Include pass_lower_switch.
4517 * stmt.c (dump_case_nodes): Remove and move to
4518 tree-switch-conversion.
4519 (case_values_threshold): Likewise.
4520 (expand_switch_as_decision_tree_p): Likewise.
4521 (emit_case_decision_tree): Likewise.
4522 (expand_case): Likewise.
4523 (balance_case_nodes): Likewise.
4524 (node_has_low_bound): Likewise.
4525 (node_has_high_bound): Likewise.
4526 (node_is_bounded): Likewise.
4527 (emit_case_nodes): Likewise.
4528 (struct simple_case_node): New struct.
4529 (add_case_node): Remove.
4530 (emit_case_dispatch_table): Use vector instead of case_list.
4531 (reset_out_edges_aux): Remove.
4532 (compute_cases_per_edge): Likewise.
4533 (expand_case): Build list of simple_case_node.
4534 (expand_sjlj_dispatch_table): Use it.
4535 * tree-switch-conversion.c (struct case_node): Moved from
4536 stmt.c and adjusted.
4537 (emit_case_nodes): Likewise.
4538 (node_has_low_bound): Likewise.
4539 (node_has_high_bound): Likewise.
4540 (node_is_bounded): Likewise.
4541 (case_values_threshold): Likewise.
4542 (reset_out_edges_aux): Likewise.
4543 (compute_cases_per_edge): Likewise.
4544 (add_case_node): Likewise.
4545 (dump_case_nodes): Likewise.
4546 (balance_case_nodes): Likewise.
4547 (expand_switch_as_decision_tree_p): Likewise.
4548 (emit_jump): Likewise.
4549 (emit_case_decision_tree): Likewise.
4550 (try_switch_expansion): Likewise.
4551 (do_jump_if_equal): Likewise.
4552 (emit_cmp_and_jump_insns): Likewise.
4553 (fix_phi_operands_for_edge): New function.
4554 (record_phi_operand_mapping): Likewise.
4555 (class pass_lower_switch): New pass.
4556 (pass_lower_switch::execute): New function.
4557 (make_pass_lower_switch): Likewise.
4558 (conditional_probability):
4559 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4560 * tree-pass.h: Add make_pass_lower_switch.
4562 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4565 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4568 2017-08-29 Richard Biener <rguenther@suse.de>
4570 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4571 we do not add a DW_AT_inline attribute twice.
4572 (gen_subprogram_die): Remove code setting DW_AT_inline on
4573 DECL_ABSTRACT_P nodes.
4575 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4577 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4578 calls to internal functions.
4579 (gimplify_modify_expr): Likewise.
4580 * tree-call-cdce.c (use_internal_fn): Likewise.
4581 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4582 (convert_to_divmod): Set the nothrow flag.
4583 * tree-if-conv.c (predicate_mem_writes): Likewise.
4584 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4585 (vectorizable_call): Likewise.
4586 (vectorizable_store): Likewise.
4587 (vectorizable_load): Likewise.
4588 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4589 (vect_recog_mask_conversion_pattern): Likewise.
4591 2017-08-29 Martin Liska <mliska@suse.cz>
4594 * gcc.c (driver_handle_option): Add new argument.
4595 * opts-common.c (handle_option): Pass
4596 target_option_override_hook.
4597 * opts-global.c (lang_handle_option): Add new option.
4598 (set_default_handlers): Add new argument.
4599 (decode_options): Likewise.
4600 * opts.c (target_handle_option): Likewise.
4601 (common_handle_option): Call target_option_override_hook.
4602 * opts.h (struct cl_option_handler_func): Add hook for
4603 target option override.
4604 (struct cl_option_handlers): Likewise.
4605 (set_default_handlers): Add new argument.
4606 (decode_options): Likewise.
4607 (common_handle_option): Likewise.
4608 (target_handle_option): Likewise.
4609 * toplev.c (toplev::main): Pass targetm.target_option.override
4612 2017-08-29 Richard Biener <rguenther@suse.de>
4613 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4615 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4616 life to the active subtree.
4618 2017-08-28 Jeff Law <law@redhat.com>
4620 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4621 derive_equivalences.
4622 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4624 (edge_info::derive_equivalences): New private member function
4626 * tree-ssa-dom.c (class edge_info): Changed from a struct
4627 to a class. Add ctor/dtor, methods and data members.
4628 (edge_info::edge_info): Renamed from allocate_edge_info.
4629 Initialize additional members.
4630 (edge_info::~edge_info): New.
4631 (free_dom_edge_info): Delete the edge info.
4632 (record_edge_info): Use new class & associated member functions.
4633 Tighten forms for testing for edge equivalences.
4634 (record_temporary_equivalences): Iterate over the simple
4635 equivalences rather than assuming there's only one per edge.
4636 (cprop_into_successor_phis): Iterate over the simple
4637 equivalences rather than assuming there's only one per edge.
4638 (optimize_stmt): Use operand_equal_p rather than pointer
4639 equality for mini-DSE code.
4641 2017-08-28 Nathan Sidwell <nathan@acm.org>
4643 * gcc.c (execute): Fold SIGPIPE handling into switch
4644 statement. Adjust internal error message.
4646 2017-08-28 Richard Biener <rguenther@suse.de>
4649 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4650 Do nothing for removed DIEs.
4652 2017-08-28 Richard Biener <rguenther@suse.de>
4654 PR tree-optimization/81977
4655 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4658 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4661 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4662 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4663 using targetm.gen_mem_thread_fence.
4665 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4668 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4669 predicate to register_operand. Reorder operands.
4670 (*btr<mode>): Ditto.
4671 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4672 (*btr<mode>_mask): Ditto.
4674 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4676 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4677 * config/rs6000/xmmintrin.h: New file.
4678 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4680 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4683 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4684 parameter and_insn and return it.
4685 (recombine_lvx_pattern): Insert a copy to ensure availability of
4686 the base register of the copied masking operation at the point of
4687 the instruction replacement.
4688 (recombine_stvx_pattern): Likewise.
4690 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4692 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4693 undocumented switches.
4694 (-mpower9-dform-vector): Likewise.
4695 (-mpower9-dform): Likewise.
4696 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4697 comments to delete references to -mpower9-dform* switches.
4698 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4699 Delete reference to -mpower9-dform* switches, test for
4700 -mpower9-vector instead.
4701 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4702 (OTHER_P9_VECTOR_MASKS): Likewise.
4703 (POWERPC_MASKS): Likewise.
4704 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4705 tests against -mpower9-dform* to -mpower9-vector. Delete code
4706 that checked for -mpower9-dform* consistancy with other options.
4707 Add test for -mpower9-misc to enable other power9 switches.
4708 (rs6000_init_hard_regno_mode_ok): Likewise.
4709 (rs6000_option_override_internal): Likewise.
4710 (rs6000_emit_prologue): Likewise.
4711 (rs6000_emit_epilogue): Likewise.
4712 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4713 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4714 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4716 (emit_fusion_p9_store): Likewise.
4717 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4718 resetting these macros if the assembler does not support ISA 3.0
4720 (TARGET_P9_DFORM_VECTOR): Likewise.
4721 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4722 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4724 2017-08-25 Alan Modra <amodra@gmail.com>
4726 PR rtl-optimization/81747
4727 * cse.c (cse_extended_basic_block): Don't attempt to record
4728 equivalences for degenerate conditional branches that branch
4729 to their fall-through.
4731 2017-08-24 Martin Sebor <msebor@redhat.com>
4734 * gimple-fold.c (size_must_be_zero_p): New function.
4735 (gimple_fold_builtin_memory_op): Call it.
4737 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4739 * config/rs6000/mm_malloc.h: New file.
4741 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4743 PR tree-optimization/81913
4744 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4745 analysis when either IVs in condition can wrap.
4747 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4749 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4750 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4752 2017-08-24 Richard Biener <rguenther@suse.de>
4755 * targhooks.c (default_target_can_inline_p): Properly
4756 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4757 is present and always compare.
4758 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4759 infer -mfpmath=sse from TARGET_SSE_P.
4760 (ix86_can_inline_p): Properly use target_option_default_node when
4761 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4763 2017-08-24 Richard Biener <rguenther@suse.de>
4766 * dwarf2out.c (output_die): Handle flag_generate_offload like
4768 (output_comp_unit): Likewise.
4769 (gen_array_type_die): Likewise.
4770 (dwarf2out_early_finish): Likewise.
4771 (note_variable_value_in_expr): Likewise.
4772 (dwarf2out_finish): Likewise. Adjust assert.
4773 * cgraphunit.c (symbol_table::compile): Move setting of
4774 flag_generate_offload earlier ...
4775 (symbol_table::finalize_compilation_unit): ... here, before
4776 early debug finalization.
4778 2017-08-24 Richard Biener <rguenther@suse.de>
4780 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4781 and ipa-fnsummary.h.
4782 (ix86_can_inline_p): When ix86_fpmath flags do not match
4783 check whether the callee uses FP math at all.
4785 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4788 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4789 instead of TYPE_PRECISION.
4791 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4792 Alan Hayward <alan.hayward@arm.com>
4793 David Sherwood <david.sherwood@arm.com>
4795 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4797 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4799 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4801 (ix86_frame::stack_realign_allocate): New field.
4802 (struct machine_frame_state): Modify comments.
4803 (machine_frame_state::sp_realigned_fp_end): New field.
4804 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4806 (sp_valid_at): Add assertion to assure no attempt to access invalid
4807 offset of a realigned stack.
4808 (fp_valid_at): Likewise.
4809 (choose_baseaddr): Modify comments.
4810 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4811 ix86_expand_prologue.
4812 (ix86_expand_prologue): Modify stack realignment and allocation.
4813 (ix86_expand_epilogue): Modify comments.
4814 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4815 avx2_runtime, avx512f, and avx512f_runtime.
4817 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4819 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4820 (mstackrealign): Do not init to -1.
4821 * config/i386/i386.c (ix86_option_override_internal):
4822 Check opts_set, not opts when setting default value of
4823 opts->x_ix86_force_align_arg_pointer.
4825 2017-08-23 Richard Biener <rguenther@suse.de>
4827 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4828 lang_hooks.decl_printable_name.
4829 * print-rtl-function.c (print_rtx_function): Likewise.
4830 * tree-pretty-print.c (dump_function_header): Likewise.
4832 2017-08-23 Richard Biener <rguenther@suse.de>
4835 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4836 -g0 at compile-time.
4838 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4841 * doc/sourcebuild.texi (Other hardware attributes):
4844 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4847 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4848 Fix single-use check.
4850 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4852 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4854 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4856 * doc/install.texi: Modify to add more details on running selected
4859 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4861 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4862 is combined with -mabi=ms.
4863 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4866 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4868 PR tree-optimization/81488
4869 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4870 and cached_basis fields.
4871 (MAX_SPREAD): New constant.
4872 (alloc_cand_and_find_basis): Initialize new fields.
4873 (clear_visited): New function.
4874 (create_phi_basis_1): Rename from create_phi_basis, set visited
4875 and cached_basis fields.
4876 (create_phi_basis): New wrapper function.
4877 (phi_add_costs_1): Rename from phi_add_costs, add spread
4878 parameter, set visited field, short-circuit when limits reached.
4879 (phi_add_costs): New wrapper function.
4880 (record_phi_increments_1): Rename from record_phi_increments, set
4882 (record_phi_increments): New wrapper function.
4883 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4884 (phi_incr_cost): New wrapper function.
4885 (all_phi_incrs_profitable_1): Rename from
4886 all_phi_incrs_profitable, set visited field.
4887 (all_phi_incrs_profitable): New wrapper function.
4889 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4890 Alan Hayward <alan.hayward@arm.com>
4891 David Sherwood <david.sherwood@arm.com>
4893 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4894 that takes the outer and inner modes.
4895 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4896 comparison as the canonical test for a paradoxical subreg.
4897 * combine.c (simplify_set): Use paradoxical_subreg_p.
4898 (make_extraction): Likewise.
4899 (force_to_mode): Likewise.
4900 (rtx_equal_for_field_assignment_p): Likewise.
4901 (gen_lowpart_for_combine): Likewise.
4902 (simplify_comparison): Likewise.
4903 * cse.c (equiv_constant): Likewise.
4904 * expmed.c (store_bit_field_1): Likewise.
4905 * final.c (alter_subreg): Likewise.
4906 * fwprop.c (propagate_rtx): Likewise.
4907 (forward_propagate_subreg): Likewise.
4908 * ira-conflicts.c (ira_build_conflicts): Likewise.
4909 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4910 * lra-constraints.c (curr_insn_transform): Likewise.
4911 (split_reg): Likewise.
4912 * lra-eliminations.c (move_plus_up): Likewise.
4913 (lra_eliminate_regs_1): Likewise.
4914 * recog.c (general_operand): Likewise.
4915 * ree.c (combine_reaching_defs): Likewise.
4916 * reload.c (push_reload): Likewise.
4917 (find_reloads): Likewise.
4918 * reload1.c (elimination_effects): Likewise.
4919 (compute_reload_subreg_offset): Likewise.
4920 (choose_reload_regs): Likewise.
4921 * rtlanal.c (subreg_lsb_1): Likewise.
4922 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4923 (simplify_subreg): Likewise.
4924 * var-tracking.c (track_loc_p): Likewise.
4925 * emit-rtl.c (byte_lowpart_offset): Likewise.
4926 (paradoxical_subreg_p): Delete out-of-line definition.
4928 2017-08-22 Jeff Law <law@redhat.com>
4930 PR tree-optimization/81741
4931 PR tree-optimization/71947
4932 * tree-ssa-dom.c: Include tree-inline.h.
4933 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4934 equivalences if one is more expensive to compute than the other.
4935 * tree-ssa-scopedtables.h (class const_or_copies): Make
4936 record_const_or_copy_raw method private.
4937 (class avail_exprs_stack): New method simplify_binary_operation.
4938 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4939 avail_exprs_stack::simplify_binary_operation as needed.
4940 (avail_exprs_stack::simplify_binary_operation): New function.
4942 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4944 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4945 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4946 (DOT_SYMBOLS): Likewise.
4947 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4948 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4949 (RS6000_ABI_NAME): Likewise.
4950 (TARGET_CMODEL): Likewise.
4951 (TOC_SECTION_ASM_OP): Likewise.
4952 (SET_CMODEL): New macro.
4953 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4955 2017-08-22 Richard Biener <rguenther@suse.de>
4957 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4958 to allow for free-lang-data replacements similar to verify_type_variant.
4960 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4961 Alan Hayward <alan.hayward@arm.com>
4962 David Sherwood <david.sherwood@arm.com>
4964 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4965 VOIDmode for the LABEL_REF.
4967 2017-08-22 Richard Biener <rguenther@suse.de>
4969 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4971 2017-08-22 Richard Biener <rguenther@suse.de>
4973 * common.opt (feliminate-dwarf2-dups): Ignore.
4974 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4975 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4976 same_die_p_wrap, compute_section_prefix,
4977 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4978 (comdat_symbol_id, comdat_symbol_number): Likewise.
4979 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4981 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4982 (output_die): Mark unreachable path unreachable.
4983 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4984 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4985 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4986 (dwarf2out_early_finish): Likewise.
4988 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4990 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4992 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4995 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4996 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4997 (avr_attribute_table) <io, io_low, address>: Initialize
4998 .decl_required with true.
5000 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5002 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5003 undocumented debugging options.
5004 (-mvsx-scalar-double): Likewise.
5005 (-mallow-df-permute): Likewise.
5006 (-mvectorize-builtins): Likewise.
5007 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5008 (rs6000_builtin_vectorized_function): Likewise.
5009 (rs6000_builtin_md_vectorized_function): Likewise.
5010 (rs6000_opt_vars): Likewise.
5012 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5015 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5016 (*btrq_imm): Rename from *btrq.
5017 (*btcq_imm): Rename from *btcq.
5018 (btsc): New code attribute.
5019 (*<btsc><mode>): New insn pattern.
5020 (*btr<mode>): Ditto.
5021 (*<btsc><mode>_mask): New insn_and_split pattern.
5022 (*btr<mode>_mask): Ditto.
5024 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5025 Alan Hayward <alan.hayward@arm.com>
5026 David Sherwood <david.sherwood@arm.com>
5028 * function.c (pad_below): Simplify padding calculation.
5030 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5031 Alan Hayward <alan.hayward@arm.com>
5032 David Sherwood <david.sherwood@arm.com>
5034 * target.def (function_prologue): Remove frame size argument.
5035 (function_epilogue): Likewise.
5036 * doc/tm.texi: Regenerate.
5037 * final.c (final_start_function): Update call to function_prologue.
5038 (final_end_function): Update call to function_epilogue.
5039 (default_function_pro_epilogue): Remove frame size argument.
5040 * output.h (default_function_pro_epilogue): Likewise.
5041 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5042 (arm_output_function_prologue): Likewise.
5043 * config/frv/frv.c (frv_function_prologue): Likewise.
5044 (frv_function_epilogue): Likewise.
5045 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5046 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5047 (ia64_output_function_epilogue): Likewise.
5048 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5049 (m32r_output_function_epilogue): Likewise.
5050 * config/microblaze/microblaze.c (microblaze_function_prologue)
5051 (microblaze_function_epilogue): Likewise.
5052 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5053 (mips_output_function_epilogue): Likewise.
5054 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5055 (mmix_target_asm_function_epilogue): Likewise.
5056 * config/msp430/msp430.c (msp430_start_function): Likewise.
5057 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5058 (nds32_asm_function_epilogue): Likewise.
5059 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5060 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5061 (pa_output_function_epilogue): Likewise.
5062 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5063 (rs6000_output_function_epilogue): Likewise.
5064 * config/rl78/rl78.c (rl78_start_function): Likewise.
5065 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5066 (rs6000_output_function_epilogue): Likewise.
5067 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5068 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5069 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5070 (sparc_asm_function_epilogue): Likewise.
5072 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5074 * tree.h (type_has_mode_precision_p): New function.
5075 * convert.c (convert_to_integer_1): Use it.
5076 * expr.c (expand_expr_real_2): Likewise.
5077 (expand_expr_real_1): Likewise.
5078 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5079 * match.pd: Likewise.
5080 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5081 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5082 * tree-tailcall.c (process_assignment): Likewise.
5083 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5084 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5085 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5086 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5087 (vectorizable_assignment): Likewise.
5088 (vectorizable_shift): Likewise.
5089 (vectorizable_operation): Likewise.
5090 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5092 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5094 * match.pd: Add pow (C, x) simplification.
5096 2017-08-21 Richard Biener <rguenther@suse.de>
5098 PR tree-optimization/81900
5099 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5100 for blocks with abnormal predecessors.
5101 (compute_antic): Do not set visited flag prematurely.
5103 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5106 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5108 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5110 * stor-layout.h (vector_type_mode): Move to...
5111 * tree.h (vector_type_mode): ...here.
5112 * stor-layout.c (vector_type_mode): Move to...
5113 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5115 2017-08-21 Richard Biener <rguenther@suse.de>
5117 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5118 register_external_die hooks.
5119 (debug_false_tree_charstarstar_uhwistar): Declare.
5120 (debug_nothing_tree_charstar_uhwi): Likewise.
5121 * debug.c (do_nothing_debug_hooks): Adjust.
5122 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5123 (debug_nothing_tree_charstar_uhwi): Likewise.
5124 * dbxout.c (dbx_debug_hooks): Adjust.
5125 (xcoff_debug_hooks): Likewise.
5126 * sdbout.c (sdb_debug_hooks): Likewise.
5127 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5128 * dwarf2out.c (macinfo_label_base): New global.
5129 (dwarf2out_register_external_die): New function for the
5130 register_external_die hook.
5131 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5132 (dwarf2_debug_hooks): Use them.
5133 (dwarf2_lineno_debug_hooks): Adjust.
5134 (struct die_struct): Add with_offset flag.
5135 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5136 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5137 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5138 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5139 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5140 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5141 defining section names for the early LTO debug variants.
5142 (reset_indirect_string): New helper.
5143 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5144 (print_dw_val): Add support for offsetted symbol references.
5145 (get_ultimate_context): Split out from is_cxx.
5146 (is_cxx): Use get_ultimate_context.
5147 (is_fortran): Add decl overload.
5148 (compute_comp_unit_symbol): Split out worker from
5149 compute_section_prefix.
5150 (compute_section_prefix): Call compute_comp_unit_symbol and
5151 set comdat_type_p here.
5152 (output_die): Skip DIE symbol output for the LTO added one.
5153 Handle DIE symbol references with offset.
5154 (output_comp_unit): Guard section name mangling properly.
5155 For LTO debug sections emit a symbol at the section beginning
5156 which we use to refer to its DIEs.
5157 (add_abstract_origin_attribute): For DIEs registered via
5158 dwarf2out_register_external_die directly refer to the early
5159 DIE rather than indirectly through the shadow one we created.
5160 Remove obsolete call to dwarf2out_abstract_function for
5161 non-function/block origins.
5162 (gen_array_type_die): When generating early LTO debug do
5163 not emit DW_AT_string_length.
5164 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5165 late when in LTO. As suggested place a gcc_unreachable for
5166 the DECL_ABSTRACT_P case.
5167 (gen_subprogram_die): Avoid another specification DIE
5168 for early built declarations/definitions for the late LTO case.
5169 (gen_variable_die): Add type references for late duplicated VLA dies
5171 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5172 we have the abstract instance already.
5173 (process_scope_var): Adjust decl DIE contexts in LTO which
5174 first puts them in limbo.
5175 (gen_decl_die): Do not generate type DIEs late apart from
5176 types for VLAs or for decls we do not yet have a DIE. Do not
5177 call dwarf2out_abstract_function late.
5178 (dwarf2out_early_global_decl): Make sure to create DIEs
5179 for abstract instances of a decl first.
5180 (dwarf2out_late_global_decl): Adjust comment.
5181 (output_macinfo_op): With multiple macro sections use
5182 macinfo_label_base to distinguish labels.
5183 (output_macinfo): Likewise. Update macinfo_label_base.
5184 Pass in the line info label.
5185 (note_variable_value_in_expr): When generating LTO resolve
5186 all variable values here by generating DIEs as needed.
5187 (init_sections_and_labels): Add early LTO debug flag parameter
5188 and generate different sections and names if set. Add generation
5189 counter for the labels so we can have multiple of them.
5190 (reset_dies): Helper to allow DIEs to be output multiple times.
5191 (dwarf2out_finish): When outputting DIEs to the fat part of an
5192 LTO object first reset DIEs.
5193 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5194 (modified_type_die): Check for decl_ultimate_origin being self
5196 (gen_type_die_with_usage): Likewise.
5197 (gen_typedef_die): Allow decl_ultimate_origin being self.
5198 (set_decl_abstract_flags): Remove.
5199 (set_block_abstract_flags): Likewise.
5200 (dwarf2out_abstract_function): Treat the early generated DIEs
5201 as the abstract copy and only add DW_AT_inline and
5202 DW_AT_artificial here and call set_decl_origin_self.
5203 If the DIE has an abstract origin don't do anything.
5204 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5205 if we have none yet (Go fails to build one, PR78628).
5206 (variably_modified_type_p): Prevent endless recursion for Ada
5207 cyclic pointer types.
5208 * lto-streamer-in.c: Include debug.h.
5209 (dref_queue): New global.
5210 (lto_read_tree_1): Stream in DIE references.
5211 (lto_input_tree): Register DIE references.
5212 (input_function): Stream DECL_DEBUG_ARGS.
5213 * lto-streamer-out.c: Include debug.h.
5214 (lto_write_tree_1): Output DIE references.
5215 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5216 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5217 (output_function): Stream DECL_DEBUG_ARGS.
5218 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5219 Stream DECL_ABSTRACT_ORIGIN.
5220 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5221 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5222 DECL_CONTEXT for file-scope decls.
5223 * lto-streamer.h (struct dref_entry): Declare.
5224 (dref_queue): Likewise.
5225 * cfgexpand.c (pass_expand::execute): Do not call the
5226 outlining_inline_function hook here.
5227 * lto-wrapper.c (debug_obj): New global.
5228 (tool_cleanup): Unlink it if required.
5229 (debug_objcopy): New function.
5230 (run_gcc): Handle early debug sections in the IL files by
5231 extracting them to separate files, partially linkin them and
5232 feeding the result back as result to the linker.
5233 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5234 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5235 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5236 sections into a separate segment.
5237 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5239 (darwin_asm_dwarf_section): Likewise.
5240 (darwin_asm_output_dwarf_offset): Likewise.
5241 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5243 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5244 Alan Hayward <alan.hayward@arm.com>
5245 David Sherwood <david.sherwood@arm.com>
5247 * read-md.h (md_reader::record_potential_iterator_use): Replace
5248 pointer argument with an rtx and an index.
5249 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5250 (apply_mode_iterator): Likewise.
5251 (apply_code_iterator): Likewise.
5252 (apply_int_iterator): Likewise.
5253 (apply_subst_iterator): Likewise.
5254 (record_iterator_use): Likewise.
5255 (record_attribute_use): Likewise.
5256 (md_reader::record_potential_iterator_use): Likewise. Update calls
5257 to record_iterator_use and apply_iterator.
5258 (iterator_use): Replace ptr with x and index.
5259 (attribute_use): Likewise.
5260 (apply_attribute_uses): Update calls to apply_iterator.
5261 (apply_iterators): Likewise. Update initialization of iterator_use.
5262 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5263 and record_potential_iterator_use.
5264 (rtx_reader::read_rtx_operand): Likewise.
5266 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5267 Alan Hayward <alan.hayward@arm.com>
5268 David Sherwood <david.sherwood@arm.com>
5270 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5273 2017-08-21 Richard Biener <rguenther@suse.de>
5276 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5277 at struct end conservatively when comparing common bases.
5279 2017-08-21 Richard Biener <rguenther@suse.de>
5281 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5282 (mem_ref_in_stmt): Remove.
5283 (determine_max_movement): Use ref index to get at the reference.
5284 (invariantness_dom_walker::before_dom_children): Deal with
5285 lim data already initialized.
5286 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5288 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5290 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5291 (TARGET_ISA_ROUND): Ditto.
5292 (TARGET_ROUND): Ditto.
5293 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5294 * config/i386/i386.md: Ditto.
5295 * config/i386/sse.md: Ditto.
5296 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5297 with OPTION_MASK_ISA_SSE4_1.
5299 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5302 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5303 __builtin_ia32_lzcnt_u16.
5305 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5308 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5309 (rs6000_set_current_function): Rewrite function to use it.
5311 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5314 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5315 and TYPE_WARN_IF_NOT_ALIGN.
5316 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5317 (handle_warn_if_not_align): New.
5318 (place_union_field): Call handle_warn_if_not_align.
5319 (place_field): Call handle_warn_if_not_align.
5320 Copy TYPE_WARN_IF_NOT_ALIGN.
5321 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5322 (layout_type): Likewise.
5323 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5325 (tree_decl_common): Add warn_if_not_align.
5326 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5327 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5328 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5329 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5330 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5331 * doc/extend.texi: Document warn_if_not_aligned attribute.
5332 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5334 2017-08-17 Martin Liska <mliska@suse.cz>
5337 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5338 (get_data_dependence): Use it as pointer type.
5339 (distribute_loop): Likewise.
5341 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5343 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5344 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5345 (p8_vmrgow_<mode>_direct): New define_insn.
5346 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5347 handle endianness for vmrgew and vmrgow permute patterns.
5349 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5351 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5352 * config/rs6000/rs6000-cpus.def: Remove comment.
5353 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5354 (POWERPC_MASKS): Likewise.
5355 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5356 use of TARGET_VSX_TIMODE.
5357 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5358 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5359 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5360 (rs6000_option_override_internal): Remove dead code.
5361 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5362 (rs6000_legitimize_reload_address): Likewise.
5363 (rs6000_legitimate_address_p): Likewise.
5364 (rs6000_opt_masks): Delete "vsx-timode".
5365 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5366 from function comment.
5367 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5368 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5369 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5371 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5372 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5373 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5376 (define_peephole2): Likewise.
5378 2017-08-17 Martin Sebor <msebor@redhat.com>
5381 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5382 past the end of an array.
5383 (test_pp_format): Add test cases.
5385 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5387 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5388 missing ECF_NOTHROW flags.
5390 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5393 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5394 operands residing in integer registers.
5395 (*vsx_le_perm_load_<mode>): Likewise.
5396 (*vsx_le_perm_store_<mode>): Likewise.
5397 (define_peephole2): Add peepholes to optimize the above.
5399 2017-08-17 Marek Polacek <polacek@redhat.com>
5402 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5403 to mimic what shorten_compare did. Change the return type to bool.
5404 (fold_cond_expr_with_comparison): Update call to
5405 operand_equal_for_comparison_p.
5406 (fold_ternary_loc): Likewise.
5408 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5410 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5412 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5414 2017-08-17 Richard Biener <rguenther@suse.de>
5416 * tree-ssa-structalias.c (solve_graph): When propagating
5417 to successors update the graphs succ edges and avoid duplicate work.
5419 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5422 * config/i386/i386.c (ix86_option_override_internal): Save target
5423 specific options after ix86_stack_protector_guard_reg was changed.
5425 2017-08-17 Richard Biener <rguenther@suse.de>
5427 PR tree-optimization/81827
5428 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5429 (new_var_info): Initialize it conservatively.
5430 (get_call_vi): Mark register vars.
5431 (new_scalar_tmp_constraint_exp): Likewise.
5432 (handle_rhs_call): Likewise.
5433 (handle_const_call): Likewise.
5434 (create_function_info_for): Likewise.
5435 (solve_constraints): Sort varinfos to separate register from
5436 non-register vars to pack points-to solution bitmaps during
5439 2017-08-17 Marek Polacek <polacek@redhat.com>
5441 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5443 2017-08-17 Richard Biener <rguenther@suse.de>
5445 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5446 to true when overflow is undefined and we saturated the result.
5448 2017-08-17 Alan Modra <amodra@gmail.com>
5451 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5452 Don't use store multiple if only one reg needs saving.
5453 (interesting_frame_related_regno): New function.
5454 (rs6000_frame_related): Don't emit frame info for regs that
5456 (rs6000_emit_epilogue): Likewise.
5458 2017-08-16 Nathan Sidwell <nathan@acm.org>
5460 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5461 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5462 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5463 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5464 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5465 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5466 (verify_type): Adjust for TYPE_BINFO move.
5467 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5468 process TYPE_BINFO directly.
5469 (hash_tree): Likewise.
5470 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5472 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5475 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5477 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5479 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5482 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5483 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5484 Add ix86_binary_operator_ok to insn constraint.
5485 (*iordi_1_bts): Ditto.
5486 (*xordi_1_btc): Ditto.
5487 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5488 Update corresponding peephole2 pattern.
5492 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5494 PR tree-optimization/81832
5495 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5496 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5498 2017-08-16 Marek Polacek <polacek@redhat.com>
5501 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5502 regarding size_zero_node.
5504 2017-08-16 Martin Liska <mliska@suse.cz>
5507 * config.gcc: Respect previously set extra_objs in case
5510 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5512 PR tree-optimization/81835
5513 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5514 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5515 not depend on the phi.
5517 2017-08-16 Alan Modra <amodra@gmail.com>
5519 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5522 2017-08-16 Alan Modra <amodra@gmail.com>
5524 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5525 (save_reg_p): ..into this. Update all callers.
5526 (first_reg_to_save): Simplify.
5528 2017-08-16 Alan Modra <amodra@gmail.com>
5530 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5533 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5537 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5538 constant count if that count is less than 32.
5540 2017-08-15 Nathan Sidwell <nathan@acm.org>
5542 * gcc.c (execute): Emit friendlier message if inferior is killed
5543 by an external cause.
5545 2017-08-15 Richard Biener <rguenther@suse.de>
5547 PR tree-optimization/81790
5548 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5549 CONSTRUCTORs from simplifying and VN.
5551 2017-08-14 Martin Sebor <msebor@redhat.com>
5553 * builtin-attrs.def: Add comments.
5555 2017-08-14 Martin Sebor <msebor@redhat.com>
5558 * doc/extend.texi (attribute nonstring): Document new attribute.
5560 2017-08-14 Martin Sebor <msebor@redhat.com>
5563 * tree-diagnostic.c (default_tree_printer): Handle %G.
5564 * gimple-pretty-print.h (percent_G_format): Declare new function.
5565 * gimple-pretty-print.c (percent_G_format): Define.
5566 * tree-pretty-print.c (percent_K_format): Add argument.
5568 2017-08-14 Martin Sebor <msebor@redhat.com>
5570 PR translation/79998
5571 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5572 Remove a stray space.
5574 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5577 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5578 (*iordi_1_bts): Ditto.
5579 (*xordi_1_btc): Ditto.
5581 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5584 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5585 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5587 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5588 quoted strings, and make more translator-friendly.
5589 (darwin_rs6000_override_options): Likewise.
5590 (rs6000_option_override_internal): Likewise.
5591 (rs6000_return_in_memory): Fix overlong line.
5592 (init_cmulative_args): Use quoted strings, and make more
5593 translator-friendly.
5594 (rs6000_pass_by_reference): Fix overlong line.
5595 (def_builtin): Use quoted strings.
5596 (altivec_expand_predicate_builtin): Use quoted strings, and make
5597 more translator-friendly.
5598 (htm_expand_builtin): Use quoted strings.
5599 (cpu_expand_builtin): Use quoted strings, and make more
5600 translator-friendly.
5601 (altivec_expand_builtin): Likewise.
5602 (paired_expand_predicate_builtin): Likewise.
5603 (rs6000_invalid_builtin): Likewise.
5604 (builtin_function_type): Use quoted strings.
5605 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5606 more translator-friendly.
5607 (rs6000_trampoline_init): Likewise.
5608 (rs6000_handle_altivec_attribute): Likewise.
5609 (rs6000_inner_target_options): Use quoted strings.
5610 (rs6000_disable_incompatible_switches): Likewise.
5611 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5612 strings, and make more translator-friendly.
5613 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5615 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5617 PR tree-optimization/81799
5618 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5619 cond_expr to simple gimple operand.
5621 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5624 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5626 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5630 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5631 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5632 TARGET_GASISR_PROLOGUES.
5633 * config/avr/avr.c (avr_option_override): Same.
5634 (avr_pass_pre_proep::execute): Same.
5636 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5639 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5640 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5642 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5644 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5645 define_insn_and_split. Split to a memory load from 0 in
5646 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5647 using PTR mode iterator.
5648 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5649 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5650 (*add_tp_<mode>): Redefine as define_insn_and_split.
5651 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5652 address space. Merge with *add_tp_x32 using PTR mode iterator.
5653 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5654 Split to an add with a memory load from 0 in
5655 DEFAULT_TLS_SEG_REG address space.
5657 2017-08-12 Andrew Pinski <apinski@cavium.com>
5659 * config/aarch64/aarch64-option-extensions.def (rdma):
5660 Fix feature string to what Linux prints out in /proc/cpuinfo.
5662 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5665 * dwarf2out.c (modified_type_die): For C typedef types that have
5666 an ultimate origin, process the ultimate origin instead of the
5668 (gen_typedef_die): Assert that input DECLs have no ultimate
5670 (gen_type_die_with_usage): For typedef variants that have an
5671 ultimate origin, just call gen_decl_die on the original DECL.
5672 (process_scope_var): Avoid creating DIEs for local typedefs and
5673 concrete static variables.
5675 2017-08-12 Alan Modra <amodra@gmail.com>
5679 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5680 match gnu-user.h startfile.
5681 (ENDFILE_LINUX_SPEC): Similarly.
5683 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5686 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5688 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5690 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5691 * config/aarch64/aarch64.md (mov<mode>): Change.
5692 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5693 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5694 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5696 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5698 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5699 for storage order barriers.
5701 2017-08-11 Martin Liska <mliska@suse.cz>
5704 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5705 variables of void type.
5707 2017-08-11 Martin Liska <mliska@suse.cz>
5709 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5710 TARGET_SUPPORTS_ALIASES.
5711 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5712 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5713 (optimize_weakref): Likewise.
5714 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5715 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5716 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5718 2017-08-11 Martin Liska <mliska@suse.cz>
5721 * config/i386/i386.c (make_resolver_func): Do complete
5722 refactoring of the function.
5724 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5727 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5728 * config/i386/i386.c (ix86_stack_protect_guard): Use
5729 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5730 * doc/invoke.texi (x86 Options): Document
5731 -mstack-protector-guard-symbol= option.
5733 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5735 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5736 * config/i386/i386.c (ix86_split_stack_guard): New function.
5737 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5738 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5739 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5740 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5741 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5742 (split_stack_space_check): Call ix86_split_stack_guard.
5744 2017-08-10 Martin Sebor <msebor@redhat.com>
5746 * print-tree.c (print_node): Print location using the established
5748 Replace spaces with colons.
5749 (debug_raw, debug): Ditto.
5751 2017-08-10 Martin Sebor <msebor@redhat.com>
5754 * pretty-print.c (pp_format): Correct the handling of %s precision.
5756 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5759 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5761 (ix86_finalize_stack_frame_flags): This. Also clear
5762 frame_pointer_needed if -fno-omit-frame-pointer is used without
5764 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5765 with ix86_finalize_stack_frame_flags.
5766 (ix86_expand_epilogue): Likewise.
5767 (ix86_expand_split_stack_prologue): Likewise.
5768 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5770 2017-08-10 Martin Liska <mliska@suse.cz>
5773 * c-attribs.c (handle_target_attribute):
5774 Report warning for an empty string argument of target attribute.
5776 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5779 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5781 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5782 or DECL_NONLOCAL labels.
5783 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5784 or DECL_NONLOCAL labels here.
5786 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5788 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5789 to indicate when early gimple folding has been disabled.
5790 (rs6000_gimple_fold_builtin): Add debug content.
5791 (rs6000_invalid_builtin): Fix whitespace.
5792 (rs6000_expand_builtin): Fix whitespace.
5793 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5795 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5798 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5799 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5801 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5803 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5804 (qdf24xx): Likewise.
5805 * config/aarch64/aarch64.md: Include falkor.md.
5806 * config/aarch64/falkor.md: New.
5808 2017-08-09 Marek Polacek <polacek@redhat.com>
5811 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5812 * diagnostic.c (emit_diagnostic): Add a comment.
5813 (emit_diagnostic_valist): New function.
5815 2017-08-09 Marek Polacek <polacek@redhat.com>
5818 * input.c (make_location): New overload.
5819 * input.h (make_location): Declare.
5821 2017-08-08 Alan Modra <amodra@gmail.com>
5822 H.J. Lu <hongjiu.lu@intel.com>
5825 * gcc.c (NO_PIE_SPEC): Delete.
5826 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5828 (LINK_PIE_SPEC): ..to here.
5829 (LINK_COMMAND_SPEC): Support -no-pie.
5830 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5831 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5833 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5834 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5835 (ENDFILE_CRTEND_SPEC): Similarly.
5837 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5840 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5841 (mstack-protector-guard-offset=): Ditto.
5842 * config/i386/i386.c (ix86_option_override): Handle
5843 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5845 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5846 ix86_stack_protect_guard_offset variables.
5847 (TARGET_STACK_PROTECT_GUARD): Always define.
5848 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5849 and -mstack-protector-guard-offset= options.
5851 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5853 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5854 boundary case for the last candidate.
5856 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5858 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5859 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5861 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5864 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5865 * config/aarch64/aarch64-builtins.c
5866 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5867 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5868 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5870 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5871 Andrew Pinski <pinskia@gmail.com>
5874 * internal-fn.def (XORSIGN): New.
5875 * optabs.def (xorsign_optab): New.
5876 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5877 (convert_expand_mult_copysign): New.
5878 (pass_optimize_widening_mul::execute): Call
5879 convert_expand_mult_copysign.
5881 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5883 PR tree-optimization/81354
5884 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5885 Insert on edges rather than explicitly creating landing pads.
5886 (analyze_candidates_and_replace): Commit edge inserts.
5888 2017-08-08 Richard Biener <rguenther@suse.de>
5891 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5892 (expand_simple_operations): Also look through ADDR_EXPRs with
5893 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5895 2017-08-08 Richard Biener <rguenther@suse.de>
5897 PR tree-optimization/81723
5898 * tree-vect-slp.c (struct bst_traits): New hash traits.
5899 (bst_fail): New global.
5900 (vect_build_slp_tree_2): New worker, split out from ...
5901 (vect_build_slp_tree): ... this now wrapping it with using
5902 bst_fail set to cache SLP tree build fails. Properly handle
5904 (vect_analyze_slp_instance): Allocate and free bst_fail.
5906 2017-08-08 Martin Liska <mliska@suse.cz>
5909 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5910 LABEL_DECLs that can be from a different function.
5912 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5914 PR tree-optimization/81744
5915 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5916 loop's number of iterations.
5918 2017-08-08 Martin Liska <mliska@suse.cz>
5920 * asan.c: Include header files.
5921 * attribs.c (build_decl_attribute_variant): New function moved
5923 (build_type_attribute_qual_variant): Likewise.
5924 (cmp_attrib_identifiers): Likewise.
5925 (simple_cst_list_equal): Likewise.
5926 (omp_declare_simd_clauses_equal): Likewise.
5927 (attribute_value_equal): Likewise.
5928 (comp_type_attributes): Likewise.
5929 (build_type_attribute_variant): Likewise.
5930 (lookup_ident_attribute): Likewise.
5931 (remove_attribute): Likewise.
5932 (merge_attributes): Likewise.
5933 (merge_type_attributes): Likewise.
5934 (merge_decl_attributes): Likewise.
5935 (merge_dllimport_decl_attributes): Likewise.
5936 (handle_dll_attribute): Likewise.
5937 (attribute_list_equal): Likewise.
5938 (attribute_list_contained): Likewise.
5939 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5940 (lookup_attribute_by_prefix): Likewise.
5941 * bb-reorder.c: Include header files.
5942 * builtins.c: Likewise.
5943 * calls.c: Likewise.
5944 * cfgexpand.c: Likewise.
5945 * cgraph.c: Likewise.
5946 * cgraphunit.c: Likewise.
5947 * convert.c: Likewise.
5948 * dwarf2out.c: Likewise.
5949 * final.c: Likewise.
5950 * fold-const.c: Likewise.
5951 * function.c: Likewise.
5952 * gimple-expr.c: Likewise.
5953 * gimple-fold.c: Likewise.
5954 * gimple-pretty-print.c: Likewise.
5955 * gimple.c: Likewise.
5956 * gimplify.c: Likewise.
5957 * hsa-common.c: Likewise.
5958 * hsa-gen.c: Likewise.
5959 * internal-fn.c: Likewise.
5960 * ipa-chkp.c: Likewise.
5961 * ipa-cp.c: Likewise.
5962 * ipa-devirt.c: Likewise.
5963 * ipa-fnsummary.c: Likewise.
5964 * ipa-inline.c: Likewise.
5965 * ipa-visibility.c: Likewise.
5967 * lto-cgraph.c: Likewise.
5968 * omp-expand.c: Likewise.
5969 * omp-general.c: Likewise.
5970 * omp-low.c: Likewise.
5971 * omp-offload.c: Likewise.
5972 * omp-simd-clone.c: Likewise.
5973 * opts-global.c: Likewise.
5974 * passes.c: Likewise.
5975 * predict.c: Likewise.
5976 * sancov.c: Likewise.
5977 * sanopt.c: Likewise.
5978 * symtab.c: Likewise.
5979 * toplev.c: Likewise.
5980 * trans-mem.c: Likewise.
5981 * tree-chkp.c: Likewise.
5982 * tree-eh.c: Likewise.
5983 * tree-into-ssa.c: Likewise.
5984 * tree-object-size.c: Likewise.
5985 * tree-parloops.c: Likewise.
5986 * tree-profile.c: Likewise.
5987 * tree-ssa-ccp.c: Likewise.
5988 * tree-ssa-live.c: Likewise.
5989 * tree-ssa-loop.c: Likewise.
5990 * tree-ssa-sccvn.c: Likewise.
5991 * tree-ssa-structalias.c: Likewise.
5992 * tree-ssa.c: Likewise.
5993 * tree-streamer-in.c: Likewise.
5994 * tree-vectorizer.c: Likewise.
5995 * tree-vrp.c: Likewise.
5997 * ubsan.c: Likewise.
5998 * varasm.c: Likewise.
5999 * varpool.c: Likewise.
6000 * tree.c: Remove functions moved to attribs.[ch].
6002 * config/aarch64/aarch64.c: Add attrs.h header file.
6003 * config/alpha/alpha.c: Likewise.
6004 * config/arc/arc.c: Likewise.
6005 * config/arm/arm.c: Likewise.
6006 * config/avr/avr.c: Likewise.
6007 * config/bfin/bfin.c: Likewise.
6008 * config/c6x/c6x.c: Likewise.
6009 * config/cr16/cr16.c: Likewise.
6010 * config/cris/cris.c: Likewise.
6011 * config/darwin.c: Likewise.
6012 * config/epiphany/epiphany.c: Likewise.
6013 * config/fr30/fr30.c: Likewise.
6014 * config/frv/frv.c: Likewise.
6015 * config/ft32/ft32.c: Likewise.
6016 * config/h8300/h8300.c: Likewise.
6017 * config/i386/winnt.c: Likewise.
6018 * config/ia64/ia64.c: Likewise.
6019 * config/iq2000/iq2000.c: Likewise.
6020 * config/lm32/lm32.c: Likewise.
6021 * config/m32c/m32c.c: Likewise.
6022 * config/m32r/m32r.c: Likewise.
6023 * config/m68k/m68k.c: Likewise.
6024 * config/mcore/mcore.c: Likewise.
6025 * config/microblaze/microblaze.c: Likewise.
6026 * config/mips/mips.c: Likewise.
6027 * config/mmix/mmix.c: Likewise.
6028 * config/mn10300/mn10300.c: Likewise.
6029 * config/moxie/moxie.c: Likewise.
6030 * config/msp430/msp430.c: Likewise.
6031 * config/nds32/nds32-isr.c: Likewise.
6032 * config/nds32/nds32.c: Likewise.
6033 * config/nios2/nios2.c: Likewise.
6034 * config/nvptx/nvptx.c: Likewise.
6035 * config/pa/pa.c: Likewise.
6036 * config/pdp11/pdp11.c: Likewise.
6037 * config/powerpcspe/powerpcspe.c: Likewise.
6038 * config/riscv/riscv.c: Likewise.
6039 * config/rl78/rl78.c: Likewise.
6040 * config/rx/rx.c: Likewise.
6041 * config/s390/s390.c: Likewise.
6042 * config/sh/sh.c: Likewise.
6043 * config/sol2.c: Likewise.
6044 * config/sparc/sparc.c: Likewise.
6045 * config/spu/spu.c: Likewise.
6046 * config/stormy16/stormy16.c: Likewise.
6047 * config/tilegx/tilegx.c: Likewise.
6048 * config/tilepro/tilepro.c: Likewise.
6049 * config/v850/v850.c: Likewise.
6050 * config/vax/vax.c: Likewise.
6051 * config/visium/visium.c: Likewise.
6052 * config/xtensa/xtensa.c: Likewise.
6054 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6057 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6058 constraints since the -mupper-regs-* switches have been
6060 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6061 into a vector from a double word element that was extracted from
6062 another vector, and eliminate extra XXPERMDI instructions.
6063 (vsx_concat_<mode>_2): Likewise.
6064 (vsx_concat_<mode>_3): Likewise.
6065 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6066 concat to allow optimizing inserts from previous extracts.
6068 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6070 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6071 memory reference to a SSP offset in TLS address space.
6072 (ix86_print_operand) <case '@'>: Remove.
6073 (ix86_print_operand_punct_valid_p): Remove '@' code.
6074 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6076 (stack_tls_protect_set_<mode>): Remove.
6077 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6078 (stack_tls_protect_test_<mode>): Remove.
6079 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6081 2017-08-07 Olivier Hainque <hainque@adacore.com>
6084 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6086 2017-08-07 Douglas Rupp <rupp@adacore.com>
6088 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6089 variable was referenced as multidir in command.
6091 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6094 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6097 2017-08-07 Martin Liska <mliska@suse.cz>
6099 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6100 * config/rl78/rl78.c: Add include of attribs.h.
6101 * config/sh/sh.c: Likewise.
6102 * config/v850/v850.c: Likewise.
6104 2017-08-07 Tom de Vries <tom@codesourcery.com>
6107 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6109 2017-08-07 Martin Liska <mliska@suse.cz>
6111 * config/mips/mips.c: Include attribs.h.
6113 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6116 * doc/invoke.texi: Document change in behvaior for -Ofast for
6119 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6121 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6123 (aarch64_pop_regs): Likewise.
6124 (aarch64_gen_load_pair): Likewise.
6125 (aarch64_save_callee_saves): Likewise.
6126 (aarch64_restore_callee_saves): Likewise.
6128 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6130 * config/i386/i386.c: Revert the last change.
6132 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6135 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6137 (ix86_finalize_stack_frame_flags): This. Also clear
6138 frame_pointer_needed if -fno-omit-frame-pointer is used without
6140 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6141 with ix86_finalize_stack_frame_flags.
6142 (ix86_expand_epilogue): Likewise.
6143 (ix86_expand_split_stack_prologue): Likewise.
6145 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6148 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6149 to P_AES for Westmere.
6151 2017-08-07 Jonathan Yong <10walls@gmail.com>
6153 * config/i386/mingw.opt (fset-stack-executable): Removed.
6154 * config/i386/cygming.opt (fset-stack-executable): Moved
6156 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6158 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6160 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6162 2017-08-07 Marek Polacek <polacek@redhat.com>
6165 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6167 2017-08-07 Martin Liska <mliska@suse.cz>
6169 * attribs.h (canonicalize_attr_name): New function.
6170 (cmp_attribs): Move from c-format.c and adjusted.
6171 (is_attribute_p): Moved from tree.h.
6172 * tree-inline.c: Add new includes.
6173 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6174 (private_is_attribute_p): Remove.
6175 (private_lookup_attribute): Likewise.
6176 (private_lookup_attribute_by_prefix): Simplify.
6177 (remove_attribute): Use is_attribute_p.
6178 * tree.h: Remove removed declarations.
6180 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6183 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6184 instead of computing it in the function. Formatting fix.
6185 (expand_case): Don't rely on default_edge being the first edge,
6186 clear it if removing it, pass default_edge to
6187 emit_case_dispatch_table.
6188 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6191 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6193 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6194 insn in the function, emit NOP after the insn.
6196 2017-08-06 Tom de Vries <tom@codesourcery.com>
6198 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6201 2017-08-06 Tom de Vries <tom@codesourcery.com>
6203 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6206 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6208 PR tree-optimization/57371
6209 * match.pd: New pattern.
6211 2017-08-04 Marek Polacek <polacek@redhat.com>
6214 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6215 perform the computation in offset_int.
6217 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6219 PR tree-optimization/81136
6220 * tree-vectorizer.h: Include tree-hash-traits.h.
6221 (vec_base_alignments): New typedef.
6222 (vec_info): Add a base_alignments field.
6223 (vect_record_base_alignments): Declare.
6224 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6226 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6227 (create_data_ref): Add an is_conditional_in_stmt argument.
6228 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6229 the is_conditional_in_stmt field.
6230 (data_ref_loc): Add an is_conditional_in_stmt field.
6231 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6232 (find_data_references_in_stmt): Update call to create_data_ref.
6233 (graphite_find_data_references_in_stmt): Likewise.
6234 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6235 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6236 (vect_record_base_alignment): New function.
6237 (vect_record_base_alignments): Likewise.
6238 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6239 for nested statements even if we fail to compute a misalignment.
6240 Use pooled base alignments for unconditional references.
6241 (vect_find_same_alignment_drs): Compare base addresses instead
6243 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6244 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6246 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6248 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6249 Add an explicit name for the enum. Use auto_vec for slp_instances
6251 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6253 (_bb_vec_info): Add a constructor and destructor.
6254 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6255 (destroy_loop_vec_info): Delete.
6256 (vect_destroy_datarefs): Likewise.
6257 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6258 (vec_info::vec_info): New function.
6259 (vec_info::~vec_info): Likewise.
6260 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6261 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6262 destroy_loop_vec_info.
6263 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6264 (_loop_vec_info::_loop_vec_info): ...this.
6265 (destroy_loop_vec_info): Replace with...
6266 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6267 the stmt_vec_infos. Leave handling of vec_info information to its
6268 destructor. Remove explicit vector releases.
6269 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6270 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6271 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6272 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6273 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6274 (destroy_bb_vec_info): Replace with...
6275 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6276 information to its destructor.
6277 (vect_slp_analyze_bb_1): Use new and delete instead of
6278 new_bb_vec_info and destroy_bb_vec_info.
6279 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6282 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6284 * tree-data-ref.h (subscript): Add access_fn field.
6285 (data_dependence_relation): Add could_be_independent_p.
6286 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6287 (same_access_functions): Move to tree-data-ref.c.
6288 * tree-data-ref.c (ref_contains_union_access_p): New function.
6289 (access_fn_component_p): Likewise.
6290 (access_fn_components_comparable_p): Likewise.
6291 (dr_analyze_indices): Add a reference to access_fn_component_p.
6292 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6294 (constant_access_functions): Likewise.
6295 (add_other_self_distances): Likewise.
6296 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6297 (initialize_data_dependence_relation): Use XCNEW and remove
6298 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6299 of access functions that have the same type. Allow the
6300 subsequence to end with different bases in some circumstances.
6301 Record the chosen access functions in SUB_ACCESS_FN.
6302 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6303 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6304 (subscript_dependence_tester_1): Likewise dra and drb.
6305 (build_classic_dist_vector): Update calls accordingly.
6306 (subscript_dependence_tester): Likewise.
6307 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6308 DDR_COULD_BE_INDEPENDENT_P.
6309 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6310 comp_alias_ddrs instead of may_alias_ddrs.
6311 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6313 (vect_analyze_data_ref_dependence): Use it if
6314 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6315 distance vectors if that fails.
6316 (dependence_distance_ge_vf): New function.
6317 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6318 LOOP_VINFO_MAY_ALIAS_DDRS.
6320 2017-08-04 Richard Biener <rguenther@suse.de>
6323 * fold-const.c (fold_binary_loc): Properly restrict
6324 minus_var0 && minus_var1 case when associating undefined overflow
6327 2017-08-04 Tom de Vries <tom@codesourcery.com>
6329 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6331 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6333 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6334 Don't start diagnostic messages with a capital letter.
6335 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6337 (rs6000_invalid_builtin): Likewise.
6338 (rs6000_trampoline_init): Likewise.
6340 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6343 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6344 after setting changeable df flags.
6346 2017-08-03 Richard Biener <rguenther@suse.de>
6348 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6349 up if the use is in USE - X.
6351 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6353 * toplev.c (dumpfile.h): New include.
6354 (internal_error_reentered): New static function. Use it...
6355 (internal_error_function): ...here to handle reentered internal_error.
6357 2017-08-03 Richard Biener <rguenther@suse.de>
6360 * fold-const.c (split_tree): Add minus_var and minus_con
6361 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6362 here but always use minus_*.
6363 (associate_trees): Assert we never associate with MINUS_EXPR
6364 and NULL first operand. Do not recurse for PLUS_EXPR operands
6365 when associating as MINUS_EXPR either.
6366 (fold_binary_loc): Track minus_var and minus_con.
6368 2017-08-03 Tom de Vries <tom@codesourcery.com>
6371 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6372 ACCEL_COMPILER, apply finish_options on
6373 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6375 2017-08-03 Tom de Vries <tom@codesourcery.com>
6378 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6379 function_entry_patch_area_size > 0.
6381 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6384 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6385 instead of 10??LU, perform unit multiplication in wide_int,
6386 don't change alloc_object_size_limit if the limit is larger
6389 PR tree-optimization/81655
6390 PR tree-optimization/81588
6391 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6392 the case when ranges[i].low and high are 1 for unsigned type with
6396 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6397 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6399 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6401 * tree-vrp.h: Add include guard.
6403 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6406 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6407 (ud2): New insn pattern.
6408 * config/i386/i386.c (ix86_expand_epilogue):
6409 For naked functions, generate ud2 instead of trap insn.
6411 2017-08-02 Marek Polacek <polacek@redhat.com>
6414 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6416 2017-08-02 Tom de Vries <tom@codesourcery.com>
6417 Cesar Philippidis <cesar@codesourcery.com>
6419 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6420 Add missing edge probabilities.
6422 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6424 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6427 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6430 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6431 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6432 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6433 prologue_seq sequences - if any.
6435 2017-08-02 Richard Biener <rguenther@suse.de>
6437 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6438 via vectors if supported, integer extracts via punning if supported
6439 or otherwise vector extracts.
6441 2017-08-02 Richard Biener <rguenther@suse.de>
6443 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6445 (bitmap_insert_into_set): ... this.
6447 2017-08-02 Richard Biener <rguenther@suse.de>
6449 PR tree-optimization/81633
6451 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6453 PR tree-optimization/71752
6454 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6456 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6458 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6459 (machine_function::call_ms2sysv_pad_out): Remove field.
6460 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6461 (ix86_compute_frame_layout): Likewise.
6463 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6466 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6467 attribute with interrupt attribute.
6469 2017-08-01 Andrew Pinski <apinski@cavium.com>
6471 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6472 BIT_INSERT_EXPR's operand 1
6473 to see if the types precision matches.
6475 2017-08-01 Martin Liska <mliska@suse.cz>
6478 * builtins.c (expand_builtin_memcpy_args): Remove.
6479 (expand_builtin_memcpy): Call newly added function
6480 expand_builtin_memory_copy_args.
6481 (expand_builtin_memcpy_with_bounds): Likewise.
6482 (expand_builtin_mempcpy): Remove last argument.
6483 (expand_builtin_mempcpy_with_bounds): Likewise.
6484 (expand_builtin_memory_copy_args): New function created from
6485 expand_builtin_mempcpy_args with small modifications.
6486 (expand_builtin_mempcpy_args): Remove.
6487 (expand_builtin_stpcpy): Remove unused argument.
6488 (expand_builtin): Likewise.
6489 (expand_builtin_with_bounds): Likewise.
6491 2017-08-01 Martin Liska <mliska@suse.cz>
6494 Make mempcpy more optimal (PR middle-end/70140).
6496 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6499 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6500 __builtin_vec_cmpne verify both arguments are compatible vectors
6501 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6502 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6503 move computation of aligned to after checking the argument types.
6507 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6508 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6511 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6512 Doug Rupp <rupp@adacore.com>
6513 Olivier Hainque <hainque@adacore.com>
6515 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6516 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6518 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6519 for TARGET_OS_CPP_BUILTIN.
6520 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6521 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6523 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6524 passing required abi options to the assembler for EABI configurations.
6525 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6526 of .text.hot and .text.unlikely sections for kernel modules when
6527 using ARM style exceptions.
6528 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6529 options. Add EXTRA_CC1_SPEC.
6530 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6532 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6534 (ARM_TARGET2_DWARF_FORMAT): Define.
6535 * config/arm/t-vxworks: Adjust multilib control to removal of the
6536 Diab command line options.
6538 2017-08-01 Martin Liska <mliska@suse.cz>
6540 PR gcov-profile/81561
6541 * gcov.c (unblock): Make unblocking safe as we need to preserve
6542 index correspondence of blocks and block_lists.
6544 2017-08-01 Richard Biener <rguenther@suse.de>
6546 PR tree-optimization/81181
6547 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6548 (compute_antic): ... end of iteration here.
6550 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6552 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6553 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6554 (ftree-slp-vectorize): Likewise.
6555 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6556 can no longer be set independent of flag_tree_loop_vectorize.
6557 * omp-general.c (emp_max_vf): Likewise.
6558 * opts.c (enable_fdo_optimizations): Remove references to
6559 flag_tree_vectorize, these are now implicit.
6560 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6561 and leave it for the options machinery.
6563 2017-08-01 Martin Liska <mliska@suse.cz>
6566 * builtins.c (expand_builtin_memcpy_args): Remove.
6567 (expand_builtin_memcpy): Call newly added function
6568 expand_builtin_memory_copy_args.
6569 (expand_builtin_memcpy_with_bounds): Likewise.
6570 (expand_builtin_mempcpy): Remove last argument.
6571 (expand_builtin_mempcpy_with_bounds): Likewise.
6572 (expand_builtin_memory_copy_args): New function created from
6573 expand_builtin_mempcpy_args with small modifications.
6574 (expand_builtin_mempcpy_args): Remove.
6575 (expand_builtin_stpcpy): Remove unused argument.
6576 (expand_builtin): Likewise.
6577 (expand_builtin_with_bounds): Likewise.
6579 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6582 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6583 print "ds:" only for immediates in generic address space.
6585 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6588 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6589 (ix86_function_ok_for_sibcall): Return false for naked functions.
6591 2017-08-01 Richard Biener <rguenther@suse.de>
6593 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6594 (compute_antic): Seed worklist with exit block predecessors.
6595 * cfganal.c (dfs_find_deadend): For a cycle return the source
6596 of the edge closing it.
6598 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6600 * config/aarch64/aarch64.c
6601 (aarch64_can_const_movi_rtx_p): Move 0 check.
6603 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6605 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6606 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6608 * match.pd: Ditto in address comparison pattern.
6610 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6612 PR tree-optimization/81627
6613 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6614 closed ssa form for store-store chain.
6616 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6618 PR tree-optimization/81620
6619 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6620 for store-store chain.
6622 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6624 PR tree-optimization/81588
6625 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6626 ranges[i].in_p, invert comparison code ccode. For >/>=,
6627 swap rhs1 and rhs2 and comparison code unconditionally,
6628 for </<= don't do that. Don't swap rhs1/rhs2 again if
6629 ranges[i].in_p, instead invert comparison code ccode if
6630 opcode or oe->rank is BIT_IOR_EXPR.
6633 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6634 a direct optab to conversion optab.
6635 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6636 with GET_MODE_INNER as last argument instead of optab_handler.
6637 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6638 vector extraction if possible and optab is available.
6639 * expr.c (store_constructor): Use convert_optab_handler instead
6640 of optab_handler. Use vector initialization from smaller
6641 vectors if possible and optab is available.
6642 * tree-vect-stmts.c (vectorizable_load): Likewise.
6643 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6645 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6646 of vec_init from half-sized vectors with the same element mode.
6647 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6648 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6649 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6650 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6651 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6652 after mode in gen_vec_extract* calls.
6653 (vec_extract<mode>): Renamed to ...
6654 (vec_extract<mode><ssescalarmodelower>): ... this.
6655 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6656 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6657 element mode after mode in gen_vec_init* calls.
6658 (VEC_INIT_HALF_MODE): New mode iterator.
6659 (vec_init<mode>): Renamed to ...
6660 (vec_init<mode><ssescalarmodelower>): ... this.
6661 (vec_init<mode><ssehalfvecmodelower>): New expander.
6662 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6663 (vec_extractv2sfsf): ... this.
6664 (vec_initv2sf): Renamed to ...
6665 (vec_initv2sfsf): ... this.
6666 (vec_extractv2si): Renamed to ...
6667 (vec_extractv2sisi): ... this.
6668 (vec_initv2si): Renamed to ...
6669 (vec_initv2sisi): ... this.
6670 (vec_extractv4hi): Renamed to ...
6671 (vec_extractv4hihi): ... this.
6672 (vec_initv4hi): Renamed to ...
6673 (vec_initv4hihi): ... this.
6674 (vec_extractv8qi): Renamed to ...
6675 (vec_extractv8qiqi): ... this.
6676 (vec_initv8qi): Renamed to ...
6677 (vec_initv8qiqi): ... this.
6678 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6679 (vec_init<mode>): Renamed to ...
6680 (vec_init<mode><VEC_base_l>): ... this.
6681 (vec_extract<mode>): Renamed to ...
6682 (vec_extract<mode><VEC_base_l>): ... this.
6683 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6684 (vec_initv2sfsf): ... this.
6685 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6686 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6687 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6688 element mode after mode in gen_vec_init* calls.
6689 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6690 (vec_init<mode><Vel>): ... this.
6691 (vec_extract<mode>): Renamed to ...
6692 (vec_extract<mode><Vel>): ... this.
6693 * config/aarch64/iterators.md (Vel): New mode attribute.
6694 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6695 Add element mode after mode in gen_vec_extract* calls.
6696 * config/s390/vector.md (non_vec_l): New mode attribute.
6697 (vec_extract<mode>): Renamed to ...
6698 (vec_extract<mode><non_vec_l>): ... this.
6699 (vec_init<mode>): Renamed to ...
6700 (vec_init<mode><non_vec_l>): ... this.
6701 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6702 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6704 * config/arm/iterators.md (V_elem_l): New mode attribute.
6705 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6706 (vec_extract<mode><V_elem_l>): ... this.
6707 (vec_extractv2di): Renamed to ...
6708 (vec_extractv2didi): ... this.
6709 (vec_init<mode>): Renamed to ...
6710 (vec_init<mode><V_elem_l>): ... this.
6711 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6712 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6713 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6714 Add element mode after gen_vec_extract* calls.
6715 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6716 (vec_init<mode><unitmode>): ... this.
6717 (vec_extract<mode>): Renamed to ...
6718 (vec_extract<mode><unitmode>): ... this.
6719 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6720 (vec_init<mode><unitmode>): ... this.
6721 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6722 (vec_initv2sfsf): ... this.
6723 (vec_extractv2sf): Renamed to ...
6724 (vec_extractv2sfsf): ... this.
6725 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6726 Add element mode after gen_vec_extract* calls.
6727 * config/mips/mips.md (unitmode): New mode iterator.
6728 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6729 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6730 * config/spu/spu.md (inner_l): New mode attribute.
6731 (vec_init<mode>): Renamed to ...
6732 (vec_init<mode><inner_l>): ... this.
6733 (vec_extract<mode>): Renamed to ...
6734 (vec_extract<mode><inner_l>): ... this.
6735 * config/sparc/sparc.md (veltmode): New mode iterator.
6736 (vec_init<VMALL:mode>): Renamed to ...
6737 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6738 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6739 (vec_initv2sisi): ... this.
6740 (vec_initv2sf): Renamed to ...
6741 (vec_initv2sfsf): ... this.
6742 (vec_extractv2sf): Renamed to ...
6743 (vec_extractv2sfsf): ... this.
6744 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6745 (vec_init<mode>): Renamed to ...
6746 (vec_init<mode><VEC_base_l>): ... this.
6747 (vec_extract<mode>): Renamed to ...
6748 (vec_extract<mode><VEC_base_l>): ... this.
6749 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6750 (vec_initv2sfsf): ... this.
6751 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6752 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6753 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6754 gen_vec_init* calls.
6756 2017-08-01 Richard Biener <rguenther@suse.de>
6758 PR tree-optimization/81297
6759 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6760 TREE_OVERFLOW from INTEGER_CSTs.
6762 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6764 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6766 2017-07-31 Carl Love <cel@us.ibm.com>
6768 * config/rs6000/rs6000-c: Add support for built-in functions
6769 vector signed char vec_xl_be (signed long long, signed char *);
6770 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6771 vector signed int vec_xl_be (signed long long, signed int *);
6772 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6773 vector signed long long vec_xl_be (signed long long, signed long long *);
6774 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6775 vector signed short vec_xl_be (signed long long, signed short *);
6776 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6777 vector double vec_xl_be (signed long long, double *);
6778 vector float vec_xl_be (signed long long, float *);
6779 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6780 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6781 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6783 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6784 (altivec_expand_builtin): Add switch statement to call
6785 altivec_expand_xl_be for each builtin.
6786 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6787 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6788 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6789 __builtin_vsx_le_be_v16qi.
6790 * doc/extend.texi: Update the built-in documentation file for the
6791 new built-in functions.
6793 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6796 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6798 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6800 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6802 * config.gcc: Add z14.
6803 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6804 CPU model numbers for z13s and z14.
6805 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6807 * config/s390/s390-opts.h (enum processor_type): Rename
6808 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6809 * config/s390/s390.c (processor_table): Add field for CPU name to
6810 be passed to Binutils.
6811 (s390_asm_output_machine_for_arch): Use the new field in
6812 processor_table for Binutils.
6813 (s390_expand_builtin): Replace arch12 with z14.
6814 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6815 (s390_get_sched_attrmask): Likewise.
6816 (s390_get_unit_mask): Likewise.
6817 * config/s390/s390.opt: Add z14 to processor_type enum.
6819 2017-07-31 Martin Jambor <mjambor@suse.cz>
6822 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6823 regardless of optimization level.
6825 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6826 Martin Liska <mliska@suse.cz>
6828 * predict.def: Remove old comment and adjust probability.
6829 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6832 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6835 * config/i386/i386.c (ix86_function_naked): New function.
6836 (ix86_can_use_return_insn_p): Return false for naked functions.
6837 (ix86_expand_prologue): Skip prologue for naked functions.
6838 (ix86_expand_epilogue): Skip epilogue for naked functions
6839 and emit trap instruction.
6840 (ix86_warn_func_return): New function.
6841 (ix86_attribute_table): Add "naked" attribute specification.
6842 (TARGET_WARN_FUNC_RETURN): Define.
6843 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6845 2017-07-31 Martin Liska <mliska@suse.cz>
6847 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6848 (dump_gimple_bb_header): Always dump BB info.
6849 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6851 2017-07-31 Martin Liska <mliska@suse.cz>
6854 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6855 also with current_function_decl non-null equality.
6857 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6860 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6861 change type to the element type, instead add eltype variable and
6862 use it where we are interested in the element type.
6864 PR tree-optimization/81603
6865 * ipa-polymorphic-call.c
6866 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6867 offset arithmetic in offset_int, bail out if the resulting bit offset
6868 doesn't fit into shwi.
6870 2017-07-31 Martin Liska <mliska@suse.cz>
6872 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6873 (gimplify_save_expr): Fix comment.
6875 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6878 * config/i386/i386.c (ix86_function_arg): Update arguments for
6880 (ix86_compute_frame_layout): Set the initial stack offset to
6881 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6882 INCOMING_FRAME_SP_OFFSET.
6883 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6884 stack before exception handler returns.
6885 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6886 the 'ERROR_CODE' for exception handler.
6888 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6890 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6891 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6892 (ASM_OUTPUT_REG_POP): Ditto.
6893 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6894 instead of asm_fprintf to output pure string.
6896 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6898 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6899 to imported_module_or_decl hook.
6900 (debug_nothing_tree_tree_tree_bool): Remove.
6901 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6902 * debug.c (do_nothing_debug_hooks): Use
6903 debug_nothing_tree_tree_tree_bool_bool instead of
6904 debug_nothing_tree_tree_tree_bool.
6905 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6906 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6907 * sdbout.c (sdb_debug_hooks): Likewise.
6908 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6909 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6911 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6912 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6913 attribute, don't add anything.
6915 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6917 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6918 (fold_build2_stat_loc): Likewise.
6919 (fold_build3_stat_loc): Likewise.
6920 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6921 (fold_build1_loc): Remove macro.
6922 (fold_build2_loc): Likewise.
6923 (fold_build3_loc): Likewise.
6925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6927 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6928 (gimple_build_debug_bind_source_stat): Likewise.
6929 * gimple.h (gimple_build_debug_bind): Remove macro.
6930 (gimple_build_debug_bind_source): Likewise.
6932 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6934 * bitmap.c (bitmap_alloc): Adjust.
6935 (bitmap_gc_alloc): Likewise.
6936 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6938 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6940 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6941 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6942 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6943 (bitmap_gc_alloc_stat): Likewise.
6944 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6946 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6948 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6949 * rtl.h (shallow_copy_rtx): Remove macro.
6951 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6953 * emit-rtl.c (gen_raw_REG): Adjust.
6954 * gengenrtl.c (gendef): Likewise.
6955 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6956 * rtl.h (rtx_alloc): Remove macro.
6958 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6960 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6961 (build_tree_list_stat): Likewise.
6962 * tree.h (build_tree_list): Remove macro.
6963 (build_tree_list_vec): Likewise.
6965 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6967 * tree.c (make_vector_stat): Remove _stat from name.
6968 (build_vector_stat): Likewise.
6969 * tree.h (make_vector_stat): Remove macro.
6970 (build_vector_stat): Likewise.
6972 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6974 * tree.h (build_var_debug_value): Remove prototype.
6976 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6978 * tree.c (tree_cons_stat): Remove _stat from name.
6979 * tree.h (tree_cons): Remove macro.
6981 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6983 * tree.c (build_vl_exp_stat): Remove _stat from name.
6984 * tree.h (build_vl_exp): Remove macro.
6986 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6988 * tree.c (build_decl_stat): Remove _stat from name.
6989 * tree.h (build_decl): Remove macro.
6991 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6993 * gimple.c (gimple_build_with_ops_stat): Adjust.
6994 (gimple_alloc_stat): Remove _stat from name.
6995 * gimple.h (gimple_alloc): Remove macro.
6997 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6999 * tree.c (make_tree_vec_stat): Remove _stat from name.
7000 (grow_tree_vec_stat): Likewise.
7001 * tree.h (make_tree_vec_stat): Adjust prototype.
7002 (grow_tree_vec_stat): Likewise.
7003 (make_tree_vec): Remove macro.
7004 (grow_tree_vec): Likewise.
7006 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7008 * fold-const.c (fold_build1_stat_loc): Adjust.
7009 (fold_build2_stat_loc): Likewise.
7010 (fold_build3_stat_loc): Likewise.
7011 * tree.c (build0_stat): Remove _stat from name.
7012 (build1_stat): Likewise.
7013 (build2_stat): Likewise.
7014 (build3_stat): Likewise.
7015 (build4_stat): Likewise.
7016 (build5_stat): Likewise.
7017 * tree.h (build1_loc): Remove macro, and rename _stat function
7019 (build2_loc): Likewise.
7020 (build3_loc): Likewise.
7021 (build4_loc): Likewise.
7022 (build5_loc): Likewise.
7024 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7026 * tree.c (make_int_cst_stat): Remove _stat from name.
7027 * tree.h (make_int_cst_stat): Adjust prototype.
7028 (make_int_cst): Remove macro.
7030 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7032 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7033 * tree.h (make_tree_binfo_stat): Adjust prototype.
7034 (make_tree_binfo): Remove.
7036 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7038 * tree.c (copy_node_stat): Rename to copy_node.
7039 (build_distinct_type_copy): Adjust.
7040 * tree.h (copy_node_stat): Adjust prototype.
7041 (copy_node): Remove macro.
7043 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7045 * tree.c (make_node_stat): rename to make_node.
7046 (build_tree_list_stat): Adjust.
7047 (build0_stat): Likewise.
7048 (build2_stat): Likewise.
7049 (build3_stat): Likewise.
7050 (build4_stat): Likewise.
7051 (build5_stat): Likewise.
7052 (build_decl_stat): Likewise.
7053 * tree.h (make_node_stat): Adjust prototype.
7054 (make_node): remove macro.
7056 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7058 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7059 (PPC_FEATURE2_SCV): Likewise.
7060 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7062 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7064 * config/aarch64/aarch64.c
7065 (aarch64_internal_mov_immediate): Add new special pattern.
7066 * config/aarch64/aarch64.md (*movdi_aarch64):
7067 Add reg/32bit const mov case.
7069 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7070 Richard Sandiford <richard.sandiford@linaro.org>
7072 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7073 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7074 Add integer and movi cases.
7075 (movi-split-hf-df-sf split, fp16): New.
7076 (enabled): Added TARGET_FP_F16INST.
7077 * config/aarch64/iterators.md (GPF_HF): New.
7078 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7080 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7082 * config/aarch64/aarch64.c
7083 (aarch64_simd_container_mode): Add prototype.
7084 (aarch64_expand_mov_immediate): Add HI support.
7085 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7086 (aarch64_can_const_movi_rtx_p): New.
7087 (aarch64_preferred_reload_class):
7088 Remove restrictions of using FP registers for certain SIMD operations.
7089 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7090 (aarch64_valid_floating_const): Add integer move validation.
7091 (aarch64_simd_imm_scalar_p): Remove.
7092 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7093 (aarch64_legitimate_constant_p): Expand list of supported cases.
7094 * config/aarch64/aarch64-protos.h
7095 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7096 (aarch64_reinterpret_float_as_int): New.
7097 (aarch64_simd_imm_scalar_p): Remove.
7098 * config/aarch64/constraints.md (Uvi): New.
7099 (Dd): Split into Ds and new Dd.
7100 * config/aarch64/aarch64.md (*movsi_aarch64):
7102 (*movdi_aarch64): Add SIMD mov case.
7104 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7106 * tree-predcom.c: (struct chain): Handle store-store chain in which
7107 stores for elimination only store loop invariant values.
7108 (execute_pred_commoning_chain): Ditto.
7109 (prepare_initializers_chain_store_elim): Ditto.
7110 (prepare_finalizers): Ditto.
7111 (is_inv_store_elimination_chain): New function.
7112 (initialize_root_vars_store_elim_1): New function.
7114 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7116 * tree-predcom.c: Revise general description of the pass.
7117 (enum chain_type): New enum type for store elimination.
7118 (struct chain): New field supporting store elimination.
7119 (struct component): Ditto.
7120 (dump_chain): Dump store-stores chain.
7121 (release_chain): Release resources.
7122 (split_data_refs_to_components): Compute and create component
7123 contains only stores for elimination.
7124 (get_chain_last_ref_at): New function.
7125 (make_invariant_chain): Initialization.
7126 (make_rooted_chain): Specify chain type in parameter and record it.
7127 (add_looparound_copies): Skip for store-stores chain.
7128 (determine_roots_comp): Compute type of chain and pass it to
7130 (initialize_root_vars_store_elim_2): New function.
7131 (finalize_eliminated_stores): New function.
7132 (remove_stmt): Handle store for elimination.
7133 (execute_pred_commoning_chain): Execute predictive commoning on
7135 (determine_unroll_factor): Skip unroll for store-stores chain.
7136 (prepare_initializers_chain_store_elim): New function.
7137 (prepare_initializers_chain): Hanlde store-store chain.
7138 (prepare_finalizers_chain, prepare_finalizers): New function.
7139 (tree_predictive_commoning_loop): Return integer value indicating
7140 if loop is unrolled or lcssa form is corrupted.
7141 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7143 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7145 * tree-predcom.c (initialize_root): Delete.
7146 (execute_pred_commoning_chain): Initialize root vars and replace
7147 reference of non-combined chain directly, rather than call above
7150 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7152 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7153 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7155 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7157 * tree-predcom.c (struct chain): New field init_seq.
7158 (release_chain): Release init_seq.
7159 (prepare_initializers_chain): Record intialization stmts in above
7161 (insert_init_seqs): New function.
7162 (tree_predictive_commoning_loop): Call insert_init_seqs.
7164 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7166 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7168 2017-07-28 Richard Biener <rguenther@suse.de>
7170 * match.pd: Remove superfluous :c.
7171 * genmatch.c (simplify::id): Add member.
7172 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7174 (current_id): New global.
7175 (dt_node::parent): Move from ...
7176 (dt_operand::parent): ... here. Add for_id member.
7177 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7178 (decision_tree::find_node): Relax order requirement when
7179 merging DT_TRUE nodes to ones inbetween the current simplify
7180 and the one we try to merge with. Add diagnostic whenever
7181 we need to enforce pattern order by not merging.
7182 (decision_tree::insert): Set current_id.
7183 (decision_tree::print_node): Dump parent node and for_id.
7184 (parser::last_id): Add member.
7185 (parser::push_simplify): Assign unique id.
7186 (parser::parser): Initialize last_id.
7188 2017-07-28 Martin Liska <mliska@suse.cz>
7191 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7192 gimple_build_debug_bind.
7194 2017-07-28 Richard Biener <rguenther@suse.de>
7196 PR tree-optimization/81502
7197 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7199 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7201 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7202 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7203 for BIT_FIELD_REF args.
7204 * fold-const.c (make_bit_field_ref): Likewise.
7205 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7207 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7210 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7211 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7212 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7213 Or it into SANITIZER_UNDEFINED.
7214 * ubsan.c: Include gimple-fold.h and varasm.h.
7215 (ubsan_expand_ptr_ifn): New function.
7216 (instrument_pointer_overflow): New function.
7217 (maybe_instrument_pointer_overflow): New function.
7218 (instrument_object_size): Formatting fix.
7219 (pass_ubsan::execute): Call instrument_pointer_overflow
7220 and maybe_instrument_pointer_overflow.
7221 * internal-fn.c (expand_UBSAN_PTR): New function.
7222 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7223 * sanitizer.def (__ubsan_handle_pointer_overflow,
7224 __ubsan_handle_pointer_overflow_abort): New builtins.
7225 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7226 * internal-fn.def (UBSAN_PTR): New internal function.
7227 * opts.c (sanitizer_opts): Add pointer-overflow.
7228 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7229 * fold-const.c (build_range_check): Compute pointer range check in
7230 integral type if pointer arithmetics would be needed. Formatting
7233 2017-07-28 Martin Liska <mliska@suse.cz>
7236 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7237 parameters that are of a variable-length.
7239 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7241 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7243 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7244 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7245 (CRT_CALL_STATIC_FUNCTION): Likewise.
7246 (ASM_DEFAULT_SPEC): New define.
7247 (ASM_SPEC32): Likewise.
7248 (ASM_SPEC64): Likewise.
7249 (ASM_SPEC_COMMON): Likewise.
7250 (ASM_SPEC): Likewise.
7251 (INVALID_64BIT): Likewise.
7252 (LINK_OS_DEFAULT_SPEC): Likewise.
7253 (LINK_OS_SPEC32): Likewise.
7254 (LINK_OS_SPEC64): Likewise.
7255 (POWERPC_LINUX): Likewise.
7256 (PTRDIFF_TYPE): Likewise.
7257 (RESTORE_FP_PREFIX): Likewise.
7258 (RESTORE_FP_SUFFIX): Likewise.
7259 (SAVE_FP_PREFIX): Likewise.
7260 (SAVE_FP_SUFFIX): Likewise.
7261 (SIZE_TYPE): Likewise.
7262 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7263 (TARGET_64BIT): Likewise.
7264 (TARGET_64BIT): Likewise.
7265 (TARGET_AIX): Likewise.
7266 (WCHAR_TYPE_SIZE): Likewise.
7267 (WCHAR_TYPE): Undefine.
7268 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7269 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7270 (CPP_OS_RTEMS_SPEC): Delete.
7271 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7272 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7274 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7276 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7278 PR tree-optimization/81578
7279 * tree-parloops.c (build_new_reduction): Bail out if
7280 reduction_code isn't one of the standard OpenMP reductions.
7281 Move the details printing after that decision.
7283 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7285 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7286 related to reload_in_progress.
7287 (splat_input_operand): Likewise.
7288 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7290 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7292 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7293 (TARGET_INSTANTIATE_DECLS): Likewise.
7294 (legitimate_indexed_address_p): Delete reload_in_progress code.
7295 (rs6000_debug_legitimate_address_p): Likewise.
7296 (rs6000_eliminate_indexed_memrefs): Likewise.
7297 (rs6000_emit_le_vsx_store): Likewise.
7298 (rs6000_emit_move_si_sf_subreg): Likewise.
7299 (rs6000_emit_move): Likewise.
7300 (register_to_reg_type): Likewise.
7301 (rs6000_pre_atomic_barrier): Likewise.
7302 (rs6000_machopic_legitimize_pic_address): Likewise.
7303 (rs6000_allocate_stack_temp): Likewise.
7304 (rs6000_address_for_fpconvert): Likewise.
7305 (rs6000_address_for_altivec): Likewise.
7306 (rs6000_secondary_memory_needed_rtx): Delete function.
7307 (rs6000_check_sdmode): Likewise.
7308 (rs6000_alloc_sdmode_stack_slot): Likewise.
7309 (rs6000_instantiate_decls): Likewise.
7310 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7311 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7312 Delete reload_in_progress.
7313 (*vec_reload_and_plus_<mptrsize>): Likewise.
7314 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7315 (vsx_div_v2di): Likewise.
7316 (vsx_udiv_v2di): Likewise.
7318 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7320 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7321 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7322 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7323 (rs6000_debug_reg_global): Delete print of LRA status.
7324 (rs6000_option_override_internal): Delete dead LRA related code.
7325 (rs6000_lra_p): Delete function.
7326 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7328 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7330 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7331 * config/riscv/rtems.h: New file.
7333 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7334 Sudakshina Das <sudi.das@arm.com>
7336 * config/aarch64/aarch64.md
7337 (define_split for and<mode>3nr_compare): Move
7338 non aarch64_logical_operand to a register.
7339 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7340 register immediate operand to a register.
7341 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7343 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7346 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7348 2017-07-27 Richard Biener <rguenther@suse.de>
7350 PR tree-optimization/81573
7351 PR tree-optimization/81494
7352 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7353 multi defuse cycle case.
7355 2017-07-27 Richard Biener <rguenther@suse.de>
7357 PR tree-optimization/81571
7358 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7361 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7363 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7364 earlier and only if MASK_FPU is set. Adjust formatting.
7366 2017-07-27 Martin Liska <mliska@suse.cz>
7368 * opt-functions.awk: Add validation of value of Init.
7369 * optc-gen.awk: Pass new argument.
7371 2017-07-27 Martin Liska <mliska@suse.cz>
7373 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7374 Fix wrong condition.
7376 2017-07-27 Martin Liska <mliska@suse.cz>
7378 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7379 BBs and edges seen by autoFDO.
7381 2017-07-27 Richard Biener <rguenther@suse.de>
7383 PR tree-optimization/81502
7384 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7385 with incompatible but same sized type.
7386 (execute_update_addresses_taken): Likewise.
7388 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7390 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7391 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7393 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7396 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7397 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7398 Change s_operand to memory_operand.
7400 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7402 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7403 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7404 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7405 Emit instructions rather than returning an expression. Handle TFmode
7406 and KFmode by casting to TImode.
7407 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7408 (rs6000_emit_le_vsx_store): Likewise.
7409 * config/rs6000/vsx.md (VSX_TI): New iterator.
7410 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7411 (*vsx_le_undo_permute_<mode>): Likewise.
7412 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7413 emit the split sequence.
7414 (*vsx_le_perm_store_<mode>): Likewise.
7416 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7418 PR tree-optimization/81555
7419 PR tree-optimization/81556
7420 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7421 if true, force CHANGED for the recursive invocation.
7422 (reassociate_bb): Remember original length of ops array, pass
7423 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7425 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7426 attributes for noipa attribute. For naked attribute use
7427 lookup_attribute first before lookup_attribute_spec.
7428 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7430 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7431 for functions with noipa attribute.
7432 (cgraph_externally_visible_p): Return true for functions with noipa
7434 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7435 for functions with noipa attribute.
7436 * doc/extend.texi: Document noipa function attribute.
7437 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7438 also for functions with noipa attribute.
7439 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7441 2017-07-26 Andrew Pinski <apinski@cavium.com>
7443 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7444 vec_unalign_load_cost and vec_unalign_store_cost.
7446 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7448 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7449 -mvsx-small-integer option.
7450 (ISA_3_0_MASKS_IEEE): Likewise.
7451 (OTHER_VSX_VECTOR_MASKS): Likewise.
7452 (POWERPC_MASKS): Likewise.
7453 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7454 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7455 code, only testing for DImode being allowed in non-VSX floating
7457 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7458 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7460 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7461 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7462 TARGET_P8_VECTOR test.
7463 (rs6000_secondary_reload_simple_move): Likewise.
7464 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7465 since TARGET_P9_VECTOR was already tested.
7466 (rs6000_opt_masks): Remove -mvsx-small-integer.
7467 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7468 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7470 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7471 test for TARGET_VEXTRACTUB was used, and that uses
7473 (p9 extract splitter): Likewise.
7474 (vsx_extract_<mode>_di_p9): Likewise.
7475 (vsx_extract_<mode>_store_p9): Likewise.
7476 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7477 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7478 the elimination of TARGET_VSX_SMALL_INTEGER.
7479 (vsx_extract_<mode>_p8): Likewise.
7480 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7481 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7482 (vsx_set_<mode>_p9): Likewise.
7483 (vsx_set_v4sf_p9): Likewise.
7484 (vsx_set_v4sf_p9_zero): Likewise.
7485 (vsx_insert_extract_v4sf_p9): Likewise.
7486 (vsx_insert_extract_v4sf_p9_2): Likewise.
7487 * config/rs6000/rs6000.md (sign extend splitter): Change
7488 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7489 (floatsi<mode>2_lfiwax_mem): Likewise.
7490 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7491 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7492 since a test for TARGET_P9_VECTOR was used.
7493 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7494 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7495 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7496 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7497 TARGET_P8_VECTOR test.
7498 (fix_trunc<mode>si2_stfiwx): Likewise.
7499 (fix_trunc<mode>si2_internal): Likewise.
7500 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7501 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7503 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7504 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7505 TARGET_P8_VECTOR test.
7506 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7507 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7508 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7510 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7511 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7512 since a test for TARGET_P9_VECTOR was used.
7513 (splitter for loading small constants): Likewise.
7515 2017-07-26 Andrew Pinski <apinski@cavium.com>
7517 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7520 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7523 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7524 (fp_valid_at): Likewise.
7526 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7528 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7529 (qdf24xx_addrcost_table): Likewise.
7530 (cortexa57_tunings): Update to use generic_branch_cost.
7531 (cortexa72_tunings): Likewise.
7532 (cortexa73_tunings): Likewise.
7533 (qdf24xx_tunings): Likewise.
7535 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7537 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7538 (thunderx2t99_branch_cost): Likewise.
7539 (cortexa35_tunings): Update to use generic_branch_cost.
7540 (cortexa53_tunings): Likewise.
7541 (cortexa57_tunings): Likewise.
7542 (cortexa72_tunings): Likewise.
7543 (cortexa73_tunings): Likewise.
7544 (thunderx2t99_tunings): Likewise.
7546 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7548 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7549 (sparc_option_override): Honour MASK_FSMULD.
7550 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7551 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7552 * config/sparc/sparc.opt (mfsmuld): New option.
7553 * doc/invoke.texi (mfsmuld): Document option.
7555 2017-07-26 Marek Polacek <polacek@redhat.com>
7558 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7560 2017-07-26 Richard Biener <rguenther@suse.de>
7562 * gimple-match-head.c (do_valueize): Return OP if valueize
7564 (get_def): New helper to get at the def stmt of a SSA name
7566 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7567 do_valueize to get at the def stmt.
7568 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7570 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7573 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7575 2017-07-26 Martin Liska <mliska@suse.cz>
7578 * function.c (expand_function_start): Make expansion of
7579 nonlocal_goto_save_area after parm_birth_insn.
7581 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7583 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7584 from all CPU target flags enable members.
7586 2017-07-26 Richard Biener <rguenther@suse.de>
7588 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7589 (decision_tree::gen): Make 'type' const.
7590 (write_predicate): Likewise.
7592 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7594 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7595 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7596 (rs6000_option_override_internal): Likewise.
7597 (rs6000_expand_vector_set): Likewise.
7598 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7599 (TARGET_UPPER_REGS_SF): Likewise.
7600 (TARGET_UPPER_REGS_DI): Likewise.
7601 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7602 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7603 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7604 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7605 (Splitters for DI constants in Altivec registers): Likewise.
7606 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7607 (vsx_set_v4sf_p9): Likewise.
7608 (vsx_set_v4sf_p9_zero): Likewise.
7609 (vsx_insert_extract_v4sf_p9): Likewise.
7610 (vsx_insert_extract_v4sf_p9_2): Likewise.
7612 2017-07-25 Carl Love <cel@us.ibm.com>
7614 * doc/extend.texi: Update the built-in documentation file for the
7615 existing built-in functions
7616 vector signed char vec_cnttz (vector signed char);
7617 vector unsigned char vec_cnttz (vector unsigned char);
7618 vector signed short vec_cnttz (vector signed short);
7619 vector unsigned short vec_cnttz (vector unsigned short);
7620 vector signed int vec_cnttz (vector signed int);
7621 vector unsigned int vec_cnttz (vector unsigned int);
7622 vector signed long long vec_cnttz (vector signed long long);
7623 vector unsigned long long vec_cnttz (vector unsigned long long);
7625 2017-07-25 Andrew Pinski <apinski@cavium.com>
7627 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7628 accesses where the use is for the first operand of a BIT_INSERT.
7630 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7633 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7634 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7636 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7638 * config/i386/gstabs.h: Delete.
7639 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7641 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7643 * config/i386/i386.c (ix86_decompose_address): Do not check for
7644 register RTX when looking at index_reg or base_reg.
7645 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7647 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7649 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7650 to update EH info here.
7652 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7654 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7656 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7658 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7660 2017-07-25 Torsten Duwe <duwe@suse.de>
7662 * common.opt: Introduce -fpatchable-function-entry
7663 command line option, and its variables function_entry_patch_area_size
7664 and function_entry_patch_area_start.
7665 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7666 including a two-value parser.
7667 * target.def (print_patchable_function_entry): New target hook.
7668 * targhooks.h (default_print_patchable_function_entry): New function.
7669 * targhooks.c (default_print_patchable_function_entry): Likewise.
7670 * toplev.c (process_options): Switch off IPA-RA if
7671 patchable function entries are being generated.
7672 * varasm.c (assemble_start_function): Look at the
7673 patchable-function-entry command line switch and current
7674 function attributes and maybe generate NOP instructions by
7675 calling the print_patchable_function_entry hook.
7676 * doc/extend.texi: Document patchable_function_entry attribute.
7677 * doc/invoke.texi: Document -fpatchable_function_entry
7678 command line option.
7679 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7681 * doc/tm.texi: Re-generate.
7683 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7686 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7687 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7689 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7691 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7693 2017-07-25 Richard Biener <rguenther@suse.de>
7695 PR tree-optimization/81455
7696 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7697 not walk in cycles when looking for guards.
7699 2017-07-25 Richard Biener <rguenther@suse.de>
7701 PR tree-optimization/81529
7702 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7703 when optimizing backedge uses.
7705 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7707 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7709 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7710 to dl_section_ref. On AIX, append an expression to subtract
7711 the size of the section length to dl_section_ref.
7713 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7715 * configure.ac: If any of the config.* scripts fail, exit 1.
7716 * configure: Regenerate.
7718 2017-07-25 Richard Biener <rguenther@suse.de>
7721 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7722 of immediate uses, be more verbose on errors.
7724 2017-07-25 Richard Biener <rguenther@suse.de>
7726 PR tree-optimization/81510
7727 * tree-vect-loop.c (vect_is_simple_reduction): When the
7728 reduction stmt is not inside the loop bail out.
7730 2017-07-25 Richard Biener <rguenther@suse.de>
7732 PR tree-optimization/81303
7733 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7734 profitability check against LOOP_VINFO_NITERSM1.
7736 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7738 * domwalk.c (cmp_bb_postorder): Simplify.
7739 (sort_bbs_postorder): New function. Use it...
7740 (dom_walker::walk): ...here to optimize common cases.
7742 2017-07-25 Martin Liska <mliska@suse.cz>
7745 * ipa-visibility.c (function_and_variable_visibility): Make the
7746 redirection just on target that supports aliasing.
7747 Fix GNU coding style.
7749 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7752 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7754 * config/arm/rtems.h: Likewise.
7755 * config/bfin/rtems.h: Likewise.
7756 * config/i386/rtemself.h: Likewise.
7757 * config/lm32/rtems.h: Likewise.
7758 * config/m32c/rtems.h: Likewise.
7759 * config/m68k/rtemself.h: Likewise.
7760 * config/microblaze/rtems.h: Likewise.
7761 * config/mips/rtems.h: Likewise.
7762 * config/moxie/rtems.h: Likewise.
7763 * config/nios2/rtems.h: Likewise.
7764 * config/powerpcspe/rtems.h: Likewise.
7765 * config/rs6000/rtems.h: Likewise.
7766 * config/rtems.h: Likewise.
7767 * config/sh/rtems.h: Likewise.
7768 * config/sh/rtemself.h: Likewise.
7769 * config/sparc/rtemself.h: Likewise.
7771 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7774 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7775 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7776 * tree-ssa-structalias.c (alias_get_name): Same.
7778 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7781 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7782 instructions if no du chain is found.
7784 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7786 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7788 2017-07-25 Richard Biener <rguenther@suse.de>
7791 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7794 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7796 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7798 (ISA_2_7_MASKS_SERVER): Likewise.
7799 (ISA_3_0_MASKS_IEEE): Likewise.
7800 (OTHER_P8_VECTOR_MASKS): Likewise.
7801 (OTHER_VSX_VECTOR_MASKS): Likewise.
7802 (POWERPC_MASKS): Likewise.
7803 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7804 duplicate list of options.
7805 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7806 explicit -mupper-regs options.
7807 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7808 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7809 alias for -mupper-regs-df.
7810 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7811 (rs6000_init_hard_regno_mode_ok): Likewise.
7812 (rs6000_option_override_internal): Likewise.
7813 (rs6000_opt_masks): Likewise.
7814 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7815 options in terms of whether -mvsx or -mpower8-vector was used.
7816 (TARGET_UPPER_REGS_DI): Likewise.
7817 (TARGET_UPPER_REGS_SF): Likewise.
7818 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7819 -mupper-regs-* options.
7821 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7823 * passes.c (emergency_dump_function): Print some empty lines and a
7824 header before the RTL dump.
7826 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7828 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7830 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7833 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7834 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7836 2017-07-24 Carl Love <cel@us.ibm.com>
7838 * config/rs6000/rs6000-c.c: Add support for built-in functions
7839 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7840 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7841 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7842 vec_extract_fp_from_shortl): Add defines for the two builtins.
7843 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7844 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7846 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7847 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7848 * doc/extend.texi: Update the built-in documentation file for the
7849 new built-in function.
7851 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7854 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7856 * doc/generic.texi: Likewise.
7857 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7858 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7860 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7862 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7863 (aarch64_mls_elt_merge<mode>): Likewise.
7865 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7867 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7868 having __cxa_atexit.
7870 2017-07-23 Michael Collison <michael.collison@arm.com>
7872 * config/arm/arm.c (arm_option_override): Deprecate
7873 use of -mstructure-size-boundary.
7874 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7875 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7877 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7880 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7881 Reduce cost estimate for direct moves.
7883 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7886 * config/i386/i386.c (ix86_option_override_internal): Disable
7887 BMI, BMI2 and TBM instructions for -m16.
7889 2017-07-21 Carl Love <cel@us.ibm.com>
7891 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7892 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7893 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7894 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7895 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7896 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7897 VMULOSW): New enum "unspec" values.
7898 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7899 altivec_vmulosw): New patterns.
7900 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7901 VMULOSW): Add definitions.
7903 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7905 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7906 (qdf24xx): Likewise.
7907 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7908 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7909 (AARCH64_FL_V8_1): Renumber.
7910 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7911 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7912 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7913 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7914 rdma to feature modifiers list.
7916 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7919 * ipa-visibility (function_and_variable_visibility): Convert
7920 recursive PLT call to direct call if appropriate.
7922 2017-07-21 Andrew Pinski <apinski@cavium.com>
7924 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7925 operand 1 to see if the types precision matches.
7926 * fold-const.c (operand_equal_p): Likewise.
7928 2017-07-21 Richard Biener <rguenther@suse.de>
7930 PR tree-optimization/81303
7931 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7932 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7933 (vect_peeling_hash_get_lowest_cost): Adjust.
7934 (vect_enhance_data_refs_alignment): Likewise. Use
7935 vect_get_peeling_costs_all_drs to compute the penalty for no
7936 peeling to match up costs.
7938 2017-07-21 Richard Biener <rguenther@suse.de>
7940 PR tree-optimization/81500
7941 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7942 we didn't identify a reduction path.
7944 2017-07-21 Tom de Vries <tom@codesourcery.com>
7945 Cesar Philippidis <cesar@codesourcery.com>
7947 PR gcov-profile/81442
7948 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7951 2017-07-21 Tom de Vries <tom@codesourcery.com>
7954 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7956 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7957 nvptx_override_options_after_change.
7959 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7961 * dwarf2out.c (output_file_names): Avoid double testing for
7964 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7966 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7968 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7970 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7972 (try_crossjump_to_edge): Do not punt on partitioned functions.
7974 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7976 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7977 Put all BBs reachable only via paths crossing cold region to cold
7979 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7981 2016-07-21 Richard Biener <rguenther@suse.de>
7983 PR tree-optimization/81303
7984 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7985 into account prologue and epilogue iterations when raising
7986 min_profitable_iters to sth at least covering one vector iteration.
7988 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7990 * config/arm/arm.c (arm_test_cpu_arch_dat):
7993 2017-07-20 Nathan Sidwell <nathan@acm.org>
7995 Remove TYPE_METHODS.
7996 * tree.h (TYPE_METHODS): Delete.
7997 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7998 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7999 (dbxout_type_methods): Scan TYPE_FIELDS.
8000 (dbxout_type): Don't check TYPE_METHODS here.
8001 * function.c (use_register_for_decl): Always ignore register for
8002 class types when not optimizing.
8003 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8004 * tree.c (free_lang_data_in_type): Stitch out member functions and
8005 templates from TYPE_FIELDS.
8006 (build_distinct_type_copy, verify_type_variant,
8007 verify_type): Member fns are on TYPE_FIELDS.
8008 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8009 * tree-pretty-print.c (dump_generic_node): Likewise.
8011 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8014 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8015 V2TImode and V4TImode.
8016 (ix86_expand_vector_extract): Likewise.
8017 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8018 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8019 (ssescalarmode): Handle V4TImode and V2TImode.
8020 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8021 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8022 (VEXTRACTI128_MODE): New mode iterator.
8023 (splitter for *vec_extractv?ti first element): New.
8024 (VEC_INIT_MODE): New mode iterator.
8025 (vec_init<mode>): Consolidate 3 expanders into one using
8026 VEC_INIT_MODE mode iterator.
8028 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8030 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8031 non_spilled_static_chain_regno_p.
8033 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8035 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8037 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8039 * bb-reorder.c (connect_traces): Allow copying of blocks within
8042 2017-07-20 Richard Biener <rguenther@suse.de>
8044 * gimple.h (gimple_phi_result): Add gphi * overload.
8045 (gimple_phi_result_ptr): Likewise.
8046 (gimple_phi_arg): Likewise. Adjust index assert to only
8047 allow actual argument accesses rather than all slots available
8049 (gimple_phi_arg_def): Add gphi * overload.
8050 * tree-phinodes.c (make_phi_node): Initialize only actual
8052 (resize_phi_node): Clear memory not covered by old node,
8053 do not initialize excess argument slots.
8054 (reserve_phi_args_for_new_edge): Initialize new argument slot
8057 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8059 PR tree-optimization/81388
8061 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8063 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8064 by removing computation of may_be_zero.
8066 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8067 Tom de Vries <tom@codesourcery.com>
8070 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8071 when gimple level profile disagrees with what RTL expander did.
8073 2017-07-20 Richard Biener <rguenther@suse.de>
8075 PR tree-optimization/61171
8076 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8077 (vect_analyze_stmt): Add slp instance parameter.
8078 (vectorizable_reduction): Likewise.
8079 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8080 (vect_is_simple_reduction): Deal with chains not detected
8081 as SLP reduction chain, specifically not properly associated
8082 chains containing a mix of plus/minus.
8083 (get_reduction_op): Remove.
8084 (get_initial_defs_for_reduction): Simplify, pass in whether
8085 this is a reduction chain, pass in the SLP node for the PHIs.
8086 (vect_create_epilog_for_reduction): Get the SLP instance as
8088 (vectorizable_reduction): Get the SLP instance as arg.
8089 During analysis remember the SLP node with the PHIs in the
8090 instance. Simplify getting at the vectorized reduction PHIs.
8091 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8092 through SLP instance.
8093 (vect_slp_analyze_operations): Likewise.
8094 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8095 (vect_transform_stmt): Likewise.
8097 2017-07-20 Tom de Vries <tom@codesourcery.com>
8099 PR tree-optimization/81489
8100 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8101 read of phi arg location to before loop that modifies phi.
8103 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8105 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8108 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8111 * except.c (execute): Fix ordering issue.
8113 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8115 PR rtl-optimization/81423
8116 * combine.c (make_compound_operation_int): Don't try to optimize
8117 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8119 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8121 PR rtl-optimization/81423
8122 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8123 with a constant that is -1 in the truncated to mode.
8125 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8127 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8128 (determine_unlikely_bbs): ... here.
8129 * predict.h (propagate_unlikely_bbs_forward): Declare.
8130 * cfgexpand.c (pass_expand::execute): Use it.
8131 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8133 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8134 propagate_unlikely_bbs_forward.
8136 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8139 * except.c (maybe_add_nop_after_section_switch): New function.
8142 2017-07-19 Tom de Vries <tom@codesourcery.com>
8144 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8146 2017-07-19 Tom de Vries <tom@codesourcery.com>
8148 * gimple.h (gimple_phi_arg): Make assert more strict.
8150 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8152 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8153 * config/rs6000/mmintrin.h: New file.
8154 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8156 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8158 PR tree-optimization/81346
8159 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8161 2017-07-19 Tom de Vries <tom@codesourcery.com>
8163 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8165 2017-07-19 Tom de Vries <tom@codesourcery.com>
8167 * config/nvptx/nvptx-modes.def: Add V2DImode.
8168 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8169 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8170 (nvptx_output_mov_insn): Handle lack of mov.b128.
8171 (nvptx_print_operand): Handle 'H' and 'L' codes.
8172 (nvptx_vector_mode_supported): Allow V2DImode.
8173 (nvptx_preferred_simd_mode): New function.
8174 (nvptx_data_alignment): New function.
8175 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8176 nvptx_preferred_simd_mode.
8177 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8179 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8181 2017-07-19 Tom de Vries <tom@codesourcery.com>
8183 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8184 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8185 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8186 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8187 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8188 (mov<VECIM>_insn): New define_insn.
8189 (define_expand "mov<VECIM>): New define_expand.
8191 2017-07-19 Tom de Vries <tom@codesourcery.com>
8193 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8195 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8197 PR tree-optimization/81346
8198 * fold-const.h (fold_div_compare, range_check_type): Declare.
8199 * fold-const.c (range_check_type): New function.
8200 (build_range_check): Use range_check_type.
8201 (fold_div_compare): No longer static, rewritten into
8202 a match.pd helper function.
8203 (fold_comparison): Don't call fold_div_compare here.
8204 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8207 2017-07-19 Nathan Sidwell <nathan@acm.org>
8209 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8210 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8211 * tree.c (find_decls_types_r, verify_type): Use
8212 TYPE_{MIN,MAX}_VALUE_RAW.
8213 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8214 (hash_tree): Likewise.
8215 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8217 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8220 2017-07-18 Tom de Vries <tom@codesourcery.com>
8223 * omp-expand.c (expand_omp_for_static_chunk): Handle
8224 equal-argument loop exit phi.
8226 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8229 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8230 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8231 operand 2 predicate.
8232 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8233 operand 2 predicate.
8234 (ror,rol -> rorx splitters): Use const_int_operand as
8235 operand 2 predicate.
8237 2017-06-18 Richard Biener <rguenther@suse.de>
8239 PR tree-optimization/81410
8240 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8241 the gap in the ! slp_perm SLP case after each group.
8243 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8246 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8249 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8252 * predict.c (set_even_probabilities): Cleanup; do not affect
8253 probabilities that are already known.
8254 (combine_predictions_for_bb): Call even when count is set.
8256 2017-07-18 Nathan Sidwell <nathan@acm.org>
8258 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8261 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8264 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8265 optimization for loop niter analysis.
8267 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8270 * config/avr/avr.c (avr_optimize_casesi): Don't use
8271 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8273 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8275 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8276 body_cost_vec from _vect_peel_extended_info.
8277 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8278 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8281 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8283 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8285 2017-07-18 Richard Biener <rguenther@suse.de>
8287 PR tree-optimization/80620
8288 PR tree-optimization/81403
8289 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8290 info when re-using a VN table entry.
8292 2017-07-18 Richard Biener <rguenther@suse.de>
8294 PR tree-optimization/81418
8295 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8296 vectype_in. Verify that with lane-reducing reduction operations
8297 we have a single def-use cycle.
8299 2017-07-17 Carl Love <cel@us.ibm.com>
8301 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8303 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8304 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8305 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8306 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8307 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8308 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8309 VMULOSW): New enum "unspec" values.
8310 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8311 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8312 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8313 altivec_vmulosw): New patterns.
8314 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8315 VMULOSW): Add definitions.
8316 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8318 * config/alpha/alpha.c: Include predict.h.
8320 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8322 * tree-vrp.c (compare_assert_loc): Fix comparison function
8323 to return predictable results.
8325 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8327 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8330 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8331 * doc/invoke.texi (mexpand-adddi): Update text.
8333 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8335 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8336 that also clobbers the CC register. The old expand code is moved
8338 (*arc_clzsi2): ... here.
8339 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8340 the CC register. The old expand code is moved to ...
8341 (arc_ctzsi2): ... here.
8343 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8345 * config/arc/arc.opt (mindexed-loads): Use initial value
8346 TARGET_INDEXED_LOADS_DEFAULT.
8347 (mauto-modify-reg): Use initial value
8348 TARGET_AUTO_MODIFY_REG_DEFAULT.
8349 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8350 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8351 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8352 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8354 2017-07-17 Martin Liska <mliska@suse.cz>
8357 * opts.c (finish_options): Do not allow -fgnu-tm
8358 w/ -fsanitize={kernel-,}address. Say sorry.
8360 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8363 * tree-loop-distribution.c (classify_partition): Only assert on
8364 numer of iterations.
8365 (merge_dep_scc_partitions): Delete prameter. Update function call.
8366 (distribute_loop): Remove code handling loop with unknown niters.
8367 (pass_loop_distribution::execute): Skip loop with unknown niters.
8369 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8372 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8373 function sort_partitions_by_post_order.
8375 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8377 PR tree-optimization/81374
8378 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8379 the max index of basic blocks, rather than number of basic blocks.
8381 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8383 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8385 (arc_legitimate_pic_operand_p): Likewise.
8386 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8388 (arc_needs_pcl_p): Likewise.
8389 (arc_legitimate_pc_offset_p): Likewise.
8390 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8391 function is also used in constrains.md.
8392 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8393 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8394 PLUS. Only return true/false in known cases, otherwise assert.
8395 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8396 is already called in arc_legitimate_constant_p.
8397 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8399 (LEGITIMATE_PIC_OPERAND_P): Use
8400 arc_raw_symbolic_reference_mentioned_p function.
8401 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8406 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8407 Andrew Burgess <andrew.burgess@embecosm.com>
8409 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8410 (arc_return_address_register): New function.
8411 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8412 (arc_handle_fndecl_attribute): Add naked attribute.
8413 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8414 (TARGET_WARN_FUNC_RETURN): Likewise.
8415 (arc_allocate_stack_slots_for_args): New function.
8416 (arc_warn_func_return): Likewise.
8417 (machine_function): Change type fn_type.
8418 (arc_compute_function_type): Consider new naked function type,
8419 change function return type.
8420 (arc_must_save_register): Adapt to handle new
8421 arc_compute_function_type's return type.
8422 (arc_expand_prologue): Likewise.
8423 (arc_expand_epilogue): Likewise.
8424 (arc_return_address_regs): Delete.
8425 (arc_return_address_register): New function.
8426 (arc_epilogue_uses): Use above function.
8427 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8428 (arc_function_type): Change encoding, add naked type.
8429 (ARC_INTERRUPT_P): Change to handle the new encoding.
8430 (ARC_FAST_INTERRUPT_P): Likewise.
8431 (ARC_NORMAL_P): Define.
8432 (ARC_NAKED_P): Likewise.
8433 (arc_compute_function_type): Delete prototype.
8434 * config/arc/arc.md (in_ret_delay_slot): Use
8435 arc_return_address_register function.
8436 (simple_return): Likewise.
8437 (p_return_i): Likewise.
8439 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8441 PR tree-optimization/81428
8442 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8443 can't be built for those types.
8445 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8447 Remove stuff dead since r239246.
8449 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8450 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8451 (avr_inform_devices): Remove dead stuff.
8453 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8455 * config/arm/arm_neon.h: Fix softp typo.
8457 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8459 PR tree-optimization/81365
8460 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8461 aggregate moves onto bb predecessor edges, make sure there are no
8462 loads that could alias the lhs in between the start of bb and the
8465 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8468 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8469 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8470 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8472 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8474 PR tree-optimization/81396
8475 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8476 (init_symbolic_number): Initialize it to 1.
8477 (perform_symbolic_merge): Add n_ops from both operands into the new
8479 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8480 without base_addr as useless if they need more than one operation.
8481 (bswap_replace): Handle !bswap case for NULL base_addr.
8483 2017-07-17 Tom de Vries <tom@codesourcery.com>
8486 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8489 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8491 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8492 conditional builtin define __FIX_LEON3FT_B2BST.
8494 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8496 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8497 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8500 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8502 PR rtl-optimization/81424
8503 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8504 to remove potential trapping from operands if -fnon-call-exceptions.
8506 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8508 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8509 profile_proability for scalling.
8510 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8512 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8514 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8516 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8518 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8519 fixpoint arithmetics.
8521 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8523 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8524 fixpoint arithmetics.
8526 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8528 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8529 fixpoint arithmetics.
8531 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8533 * profile-count.h (profile_probability::from_reg_br_prob_note,
8534 profile_probability::to_reg_br_prob_note): New functions.
8535 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8536 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8537 * predict.c (probability_reliable_p): Update.
8538 (edge_probability_reliable_p): Update.
8539 (br_prob_note_reliable_p): Update.
8540 (invert_br_probabilities): Update.
8541 (add_reg_br_prob_note): New function.
8542 (combine_predictions_for_insn): Update.
8543 * asan.c (asan_clear_shadow): Update.
8544 * cfgbuild.c (compute_outgoing_frequencies): Update.
8545 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8546 (update_br_prob_note): Update.
8547 (rtl_verify_edges): Update.
8548 (purge_dead_edges): Update.
8549 (fixup_reorder_chain): Update.
8550 * emit-rtl.c (try_split): Update.
8551 * ifcvt.c (cond_exec_process_insns): Update.
8552 (cond_exec_process_if_block): Update.
8553 (dead_or_predicable): Update.
8554 * internal-fn.c (expand_addsub_overflow): Update.
8555 (expand_neg_overflow): Update.
8556 (expand_mul_overflow): Update.
8557 * loop-doloop.c (doloop_modify): Update.
8558 * loop-unroll.c (compare_and_jump_seq): Update.
8559 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8560 * predict.h: Update.
8561 * reorg.c (mostly_true_jump): Update.
8563 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8564 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8565 * config/arc/arc.c: (emit_unlikely_jump): Update.
8566 * config/arm/arm.c: (emit_unlikely_jump): Update.
8567 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8568 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8569 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8570 (ix86_print_operand): Update.
8571 (ix86_split_fp_branch): Update.
8572 (predict_jump): Update.
8573 * config/ia64/ia64.c (ia64_print_operand): Update.
8574 * config/mmix/mmix.c (mmix_print_operand): Update.
8575 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8576 (rs6000_expand_split_stack_prologue): Update.
8577 * config/rs6000/rs6000.c: Update.
8578 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8579 (s390_expand_vec_movstr): Update.
8580 (s390_expand_cs_tdsi): Update.
8581 (s390_expand_split_stack_prologue): Update.
8582 * config/sh/sh.c (sh_print_operand): Update.
8583 (expand_cbranchsi4): Update.
8584 (expand_cbranchdi4): Update.
8585 * config/sparc/sparc.c (output_v9branch): Update.
8586 * config/spu/spu.c (get_branch_target): Update.
8587 (ea_load_store_inline): Update.
8588 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8589 * config/tilepro/tilepro.c: Update.
8591 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8593 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8594 (gimplify_save_expr): Likewise.
8596 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8598 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8600 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8602 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8604 * timevar.def (TV_IPA_FNSUMMARY): Define.
8606 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8608 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8609 to back store errata sensitive sequence from being generated.
8610 (sqrtdf2_fix): Likewise.
8612 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8614 * tree-ssa-threadupdate.c (compute_path_counts,
8615 update_joiner_offpath_counts): Use profile_probability.
8617 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8620 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8622 * config/arm/arm-c.c (arm_cpu_builtins): Define
8623 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8625 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8627 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8628 array entries to represent __ieee128 versions of the
8629 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8630 scalar_extract_sig, and scalar_insert_exp built-in functions.
8631 (altivec_resolve_overloaded_builtin): Add special case handling
8632 for the __builtin_scalar_insert_exp function, as represented by
8633 the P9V_BUILTIN_VEC_VSIEDP constant.
8634 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8635 exponent support for __ieee128 argument.
8636 (VSESQP): Add scalar extract signature support for __ieee128
8638 (VSTDCNQP): Add scalar test negative support for __ieee128
8640 (VSIEQP): Add scalar insert exponent support for __int128 argument
8641 with __ieee128 result.
8642 (VSIEQPF): Add scalar insert exponent support for __ieee128
8643 argument with __ieee128 result.
8644 (VSTDCQP): Add scalar test data class support for __ieee128
8646 (VSTDCNQP): Add overload support for scalar test negative with
8648 (VSTDCQP): Add overload support for scalar test data class
8650 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8652 (UNSPEC_VSX_SIEXPQP): New constant.
8653 (xsxexpqp): New insn for VSX scalar extract exponent quad
8655 (xsxsigqp): New insn for VSX scalar extract significand quad
8657 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8658 precision with floating point argument.
8659 (xststdcqp): New expand for VSX scalar test data class quad
8661 (xststdcnegqp): New expand for VSX scalar test negative quad
8663 (xststdcqp): New insn to match expansions for VSX scalar test data
8664 class quad precision and VSX scalar test negative quad precision.
8665 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8666 special case operand checking to enforce that second operand of
8667 VSX scalar test data class with quad precision argument is a 7-bit
8669 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8670 prototypes and descriptions of __ieee128 versions of
8671 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8672 scalar_test_data_class, and scalar_test_neg built-in functions.
8674 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8676 PR tree-optimization/81162
8677 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8678 replace a negate with an add.
8680 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8682 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8684 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8686 * config/arm/arm-c.c (arm_cpu_builtins): Define
8687 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8689 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8691 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8692 (armv8-r): Set ARM Cortex-R52 as default CPU.
8693 * config/arm/arm-tables.opt: Regenerate.
8694 * config/arm/arm-tune.md: Regenerate.
8695 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8697 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8698 extension for -mcpu=cortex-r52.
8700 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8702 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8703 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8704 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8705 (fp-armv8): Define it as FP_ARMv8 only.
8706 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8707 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8709 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8710 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8711 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8712 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8713 than TARGET_FPU_ARMV8.
8714 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8715 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8716 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8718 * config/arm/neon.md (neon_vrint): Likewise.
8719 (neon_vcvt): Likewise.
8720 (neon_<fmaxmin_op><mode>): Likewise.
8721 (<fmaxmin><mode>3): Likewise.
8722 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8723 * config/arm/predicates.md (arm_cond_move_operator): Check against
8724 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8726 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8728 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8731 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8733 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8734 loop in comment with memset.
8736 2017-07-14 Martin Liska <mliska@suse.cz>
8738 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8739 * dwarf2out.c (is_java): Remove the function.
8740 (output_pubname): Remove usage of the function.
8741 (lower_bound_default): Remove usage of DW_LANG_Java.
8742 (gen_compile_unit_die): Likewise.
8743 * gcc.c: Remove compiler defaults for .java and .zip files.
8744 * gimple-expr.c (remove_suffix): Change as there's no longer
8745 extension than 4-letter one.
8746 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8747 (gimplify_save_expr): Likewise.
8748 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8749 as it's possible even for other languages than Java.
8750 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8751 * lto-opts.c (lto_write_options): Remove reference to Java.
8752 * opts.c (strip_off_ending): Update file extension handling.
8753 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8754 * tree-eh.c (lower_resx): Likewise.
8755 * tree.c (free_lang_data_in_type): Remove dead code.
8756 (find_decls_types_r): Likewise.
8757 (build_common_builtin_nodes): Remove Java from a comment.
8758 (verify_type): Remove dead code.
8759 * varasm.c (assemble_external): Remove Java from a comment.
8761 2017-07-14 Martin Liska <mliska@suse.cz>
8763 * opts.c (finish_options): Add quotes.
8764 (common_handle_option): Likewise.
8766 2017-07-14 Martin Liska <mliska@suse.cz>
8768 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8769 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8771 * dwarf2out.c (lower_bound_default): Do not handle
8773 (gen_compile_unit_die): Likewise.
8774 * gcc.c: Remove default extension binding for GNU Pascal.
8775 * stmt.c: Remove Pascal language from a comment.
8776 * xcoffout.c: Likewise.
8778 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8781 * diagnostic-show-locus.c (fixit_cmp): New function.
8782 (layout::layout): Sort m_fixit_hints.
8783 (column_range::column_range): Assert that the values are valid.
8784 (struct char_span): New struct.
8785 (correction::overwrite): New method.
8786 (struct source_line): New struct.
8787 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8788 calls in terms of classes source_line and char_span, and
8789 correction::overwrite.
8790 (selftest::test_overlapped_fixit_printing_2): New function.
8791 (selftest::diagnostic_show_locus_c_tests): Call it.
8793 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8795 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8796 early if there is no lhs.
8798 2017-07-13 Martin Liska <mliska@suse.cz>
8800 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8801 (gen_reference_type_die): Likewise.
8802 * stor-layout.c: Remove Pascal-related comment.
8804 2017-07-13 Martin Liska <mliska@suse.cz>
8806 * opts.c (finish_options): Add quotes to error messages.
8807 (parse_sanitizer_options): Likewise.
8809 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8811 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8813 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8815 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8817 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8819 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8821 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8823 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8826 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8827 provides the hardware capability bits, define the macro
8828 __BUILTIN_CPU_SUPPORTS__.
8829 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8830 if GLIBC does not provide the hardware capability bits. Add a
8831 gcc_unreachable call if the built-in cpu function is neither
8832 __builtin_cpu_is nor __builtin_cpu_supports.
8833 (rs6000_get_function_versions_dispatcher): Change the warning
8834 that an old GLIBC is used which does not export the capability
8835 bits to be an error.
8836 * doc/extend.texi (target_clones attribute): Document the
8837 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8838 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8839 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8840 the macros defined by GCC if the newer GLIBC is available.
8842 2017-07-12 Jeff Law <law@redhat.com>
8844 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8845 remaining includes slightly.
8846 * config/riscv/riscv-builtins.c: Include profile-count.h.
8848 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8851 * config/avr/avr.c (avr_set_current_function): In diagnostic
8852 messages: Quote keywords and (parts of) identifiers.
8853 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8856 2017-07-12 Carl Love <cel@us.ibm.com>
8858 * config/rs6000/rs6000-c.c: Add support for built-in functions
8859 vector bool char vec_revb (vector bool char);
8860 vector bool short vec_revb (vector short char);
8861 vector bool int vec_revb (vector bool int);
8862 vector bool long long vec_revb (vector bool long long);
8863 * doc/extend.texi: Update the built-in documentation file for the
8864 new built-in functions.
8866 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8868 * config/s390/s390.md: Remove movcc splitter.
8870 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8872 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8873 load/store on condition.
8875 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8878 * config/avr/avr.c (avr_encode_section_info)
8879 [progmem && !TREE_READONLY]: Error if progmem object needs
8882 2017-07-11 Michael Collison <michael.collison@arm.com>
8884 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8887 2017-07-11 Carl Love <cel@us.ibm.com>
8889 * config/rs6000/rs6000-c.c: Add support for builtins
8890 vector unsigned int vec_parity_lsbb (vector signed int);
8891 vector unsigned int vec_parity_lsbb (vector unsigned int);
8892 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8893 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8894 vector unsigned long long vec_parity_lsbb (vector signed long long);
8895 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8896 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8897 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8898 * doc/extend.texi: Update the built-in documentation file for the
8899 new built-in functions.
8901 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8903 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8904 (layout::m_primary_loc): New field.
8905 (layout::layout): Initialize new field. Move location filtering
8906 logic from here to...
8907 (layout::maybe_add_location_range): ...this new method. Add
8908 support for filtering to just the lines already specified by other
8910 (layout::will_show_line_p): New method.
8911 (gcc_rich_location::add_location_if_nearby): New method.
8912 (selftest::test_add_location_if_nearby): New test function.
8913 (selftest::diagnostic_show_locus_c_tests): Call it.
8914 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8917 2017-07-11 Tom de Vries <tom@codesourcery.com>
8919 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8920 (bb_first_real_insn): New function.
8921 (nvptx_single): Add extra initialization of broadcasted condition
8924 2017-07-11 Nathan Sidwell <nathan@acm.org>
8926 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8928 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8930 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8931 Binutils doc as TEXI will mess them up.
8932 * doc/invoke.texi (AVR Options): Same here.
8934 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8936 * config/sparc/sparc.opt (mfix-ut700): New option.
8937 (mfix-gr712rc): Likewise.
8938 (sparc_fix_b2bst): New variable.
8939 * doc/invoke.texi (SPARC options): Document them.
8940 (ARM options): Fix warnings.
8941 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8942 instructions to prevent sequences that can trigger the store-store
8943 errata for certain LEON3FT processors.
8944 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8945 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8946 * config/sparc/sparc.md (fix_b2bst): New attribute.
8947 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8949 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8952 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8954 (*rsqrtsf2_sse): Ditto.
8956 (div<mode>3): Macroize insn from divdf3 and divsf3
8957 using MODEF mode iterator.
8959 2017-07-10 Martin Sebor <msebor@redhat.com>
8961 PR tree-optimization/80397
8962 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8963 instead of testing for equality to INTEGER_TYPE.
8965 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8967 * config.gcc: Remove uclibc from arc target spec.
8969 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8971 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8973 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8976 * lto-wrapper.c (remove_option): New function.
8977 (merge_and_complain): Merge PIC/PIE options more realistically.
8979 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8981 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8985 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8986 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8987 * config.in: Regenerate.
8988 * configure: Regenerate.
8989 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8990 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8991 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8992 (TARGET_GASISR_PROLOGUES): ...target mask.
8993 * common/config/avr/avr-common.c
8994 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8995 Set -mgas-isr-prologues.
8996 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8997 INSERT_PASS_BEFORE for it.
8998 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8999 * config/avr/avr.c (avr_option_override)
9000 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9001 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9002 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9003 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9004 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9006 (make_avr_pass_pre_proep): New function.
9007 (emit_push_sfr) <treg>: Add argument to function and use it
9009 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9010 and set machine->gasisr.yes.
9011 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9012 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9013 __gcc_isr.n_pushed to .L__stack_usage.
9014 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9015 (avr_asm_final_postscan_insn): ...this new static function.
9016 * config/avr/avr.h (machine_function)
9017 <is_no_gccisr, use_L__stack_usage>: New fields.
9018 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9019 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9020 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9021 (gasisr, *gasisr): New expander and insn.
9022 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9023 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9024 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9026 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9028 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9031 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9033 Move jump-tables out of .text again.
9036 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9037 (ASM_OUTPUT_ADDR_VEC): New function.
9038 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9039 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9040 INSN_ADDRESSes as asm comment.
9041 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9042 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9043 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9044 * config/avr/avr.md (*tablejump): Adjust comment.
9045 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9046 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9048 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9049 (avr_output_addr_vec): New proto.
9050 (avr_log_t) <insn_addresses>: New field.
9052 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9055 * config/i386/i386.c (ix86_function_arg_advance): Set
9056 outgoing_args_on_stack to true if there are outgoing arguments
9058 (ix86_function_arg): Likewise.
9059 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9060 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9061 * config/i386/i386.h (machine_function): Add
9062 outgoing_args_on_stack.
9064 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9066 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9068 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9070 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9072 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9073 (REAL_H): Remove $(MACHMODE_H).
9074 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9076 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9077 $(MACHMODE_H) and double-int.h.
9078 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9080 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9083 2017-07-07 Andrew Pinski <apinski@cavium.com>
9085 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9086 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9088 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9090 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9091 Add warning if GCC was not configured to link against a GLIBC that
9092 exports the hardware capability bits.
9093 (make_resolver_func): Make resolver function private and not a
9094 COMDAT function. Create the name with clone_function_name instead
9095 of make_unique_name.
9098 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9099 correct operand in doing the split.
9101 2017-07-07 Carl Love <cel@us.ibm.com>
9103 * config/rs6000/rs6000-c: Add support for built-in function
9104 vector unsigned short vec_pack_to_short_fp32 (vector float,
9106 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9107 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9108 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9109 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9110 (convert_4f32_8i16): Add define_expand.
9111 * doc/extend.texi: Update the built-in documentation file for the
9112 new built-in function.
9114 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9116 * config/sparc/m8.md: New file.
9117 * config/sparc/sparc.md: Include m8.md.
9119 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9121 * config/sparc/sparc.opt: New option -mvis4b.
9122 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9123 (sparc_option_override): Handle VIS4B.
9124 (enum sparc_builtins): Define
9125 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9126 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9127 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9128 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9129 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9130 (check_constant_argument): New function.
9131 (sparc_vis_init_builtins): Define builtins
9132 __builtin_vis_dictunpack{8,16,32},
9133 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9134 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9135 __builtin_vis_fpcmpde{8,16,32}shl and
9136 __builtin_vis_fpcmpur{8,16,32}shl.
9137 (sparc_expand_builtin): Check that the constant operands to
9138 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9139 constant and in range.
9140 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9142 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9143 (SPARC_IMM5_P): Likewise.
9144 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9145 (enabled): Handle vis4b.
9146 (UNSPEC_DICTUNPACK): New unspec.
9147 (UNSPEC_FPCMPSHL): Likewise.
9148 (UNSPEC_FPUCMPSHL): Likewise.
9149 (UNSPEC_FPCMPDESHL): Likewise.
9150 (UNSPEC_FPCMPURSHL): Likewise.
9151 (cpu_feature): New CPU feature `vis4b'.
9152 (dictunpack{8,16,32}): New insns.
9153 (FPCSMODE): New mode iterator.
9154 (fpcscond): New code iterator.
9155 (fpcsucond): Likewise.
9156 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9157 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9158 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9159 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9160 * config/sparc/constraints.md: Define constraints `q' for unsigned
9161 2-bit integer constants and `t' for unsigned 5-bit integer
9163 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9165 (imm5_operand_dictunpack16): Likewise.
9166 (imm5_operand_dictunpack32): Likewise.
9167 (imm2_operand): Likewise.
9168 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9169 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9170 ditunpack* and fpcmp*shl builtins.
9172 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9174 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9175 * config.in: Add HAVE_AS_SPARC6 define.
9176 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9178 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9180 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9181 (CPP_CPU_SPEC): Handle m8.
9182 (ASM_CPU_SPEC): Likewise.
9183 * config/sparc/sparc-opts.h (enum processor_type): Add
9185 * config/sparc/sparc.c (m8_costs): New struct.
9186 (sparc_option_override): Handle TARGET_CPU_m8.
9187 (sparc32_initialize_trampoline): Likewise.
9188 (sparc64_initialize_trampoline): Likewise.
9189 (sparc_issue_rate): Likewise.
9190 (sparc_register_move_cost): Likewise.
9191 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9192 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9193 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9194 (CPP_CPU_SPEC): Handle M8.
9195 (ASM_CPU_SPEC): Likewise.
9196 (AS_M8_FLAG): Define.
9197 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9198 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9199 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9201 * configure: Regenerate.
9202 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9205 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9207 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9209 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9210 ("*movdi_insn_sp32"): Do not set v3pipe.
9211 ("*movsi_insn"): Likewise.
9212 ("*movdi_insn_sp64"): Likewise.
9213 ("*movsf_insn"): Likewise.
9214 ("*movdf_insn_sp32"): Likewise.
9215 ("*movdf_insn_sp64"): Likewise.
9216 ("*zero_extendsidi2_insn_sp64"): Likewise.
9217 ("*sign_extendsidi2_insn"): Likewise.
9218 ("*mov<VM32:mode>_insn"): Likewise.
9219 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9220 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9221 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9222 ("<vlop:code><VL:mode>3"): Likewise.
9223 ("*not_<vlop:code><VL:mode>3"): Likewise.
9224 ("*nand<VL:mode>_vis"): Likewise.
9225 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9226 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9227 ("one_cmpl<VL:mode>2"): Likewise.
9228 ("faligndata<VM64:mode>_vis"): Likewise.
9229 ("alignaddrsi_vis"): Likewise.
9230 ("alignaddrdi_vis"): Likweise.
9231 ("alignaddrlsi_vis"): Likewise.
9232 ("alignaddrldi_vis"): Likewise.
9233 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9234 ("bmaskdi_vis"): Likewise.
9235 ("bmasksi_vis"): Likewise.
9236 ("bshuffle<VM64:mode>_vis"): Likewise.
9237 ("cmask8<P:mode>_vis"): Likewise.
9238 ("cmask16<P:mode>_vis"): Likewise.
9239 ("cmask32<P:mode>_vis"): Likewise.
9240 ("pdistn<P:mode>_vis"): Likewise.
9241 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9243 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9245 * config/sparc/sparc.md ("subtype"): New insn attribute.
9246 ("*wrgsr_sp64"): Set insn subtype.
9247 ("*rdgsr_sp64"): Likewise.
9248 ("alignaddrsi_vis"): Likewise.
9249 ("alignaddrdi_vis"): Likewise.
9250 ("alignaddrlsi_vis"): Likewise.
9251 ("alignaddrldi_vis"): Likewise.
9252 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9253 ("fexpand_vis"): Likewise.
9254 ("fpmerge_vis"): Likewise.
9255 ("faligndata<VM64:mode>_vis"): Likewise.
9256 ("bshuffle<VM64:mode>_vis"): Likewise.
9257 ("cmask8<P:mode>_vis"): Likewise.
9258 ("cmask16<P:mode>_vis"): Likewise.
9259 ("cmask32<P:mode>_vis"): Likewise.
9260 ("fchksm16_vis"): Likewise.
9261 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9262 ("fmean16_vis"): Likewise.
9263 ("fp<plusminus_insn>64_vis"): Likewise.
9264 ("<plusminus_insn>v8qi3"): Likewise.
9265 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9266 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9267 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9268 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9269 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9270 ("*movqi_insn"): Likewise.
9271 ("*movhi_insn"): Likewise.
9272 ("*movsi_insn"): Likewise.
9273 ("movsi_pic_gotdata_op"): Likewise.
9274 ("*movdi_insn_sp32"): Likewise.
9275 ("*movdi_insn_sp64"): Likewise.
9276 ("movdi_pic_gotdata_op"): Likewise.
9277 ("*movsf_insn"): Likewise.
9278 ("*movdf_insn_sp32"): Likewise.
9279 ("*movdf_insn_sp64"): Likewise.
9280 ("*zero_extendhisi2_insn"): Likewise.
9281 ("*zero_extendqihi2_insn"): Likewise.
9282 ("*zero_extendqisi2_insn"): Likewise.
9283 ("*zero_extendqidi2_insn"): Likewise.
9284 ("*zero_extendhidi2_insn"): Likewise.
9285 ("*zero_extendsidi2_insn_sp64"): Likewise.
9286 ("ldfsr"): Likewise.
9287 ("prefetch_64"): Likewise.
9288 ("prefetch_32"): Likewise.
9289 ("tie_ld32"): Likewise.
9290 ("tie_ld64"): Likewise.
9291 ("*tldo_ldub_sp32"): Likewise.
9292 ("*tldo_ldub1_sp32"): Likewise.
9293 ("*tldo_ldub2_sp32"): Likewise.
9294 ("*tldo_ldub_sp64"): Likewise.
9295 ("*tldo_ldub1_sp64"): Likewise.
9296 ("*tldo_ldub2_sp64"): Likewise.
9297 ("*tldo_ldub3_sp64"): Likewise.
9298 ("*tldo_lduh_sp32"): Likewise.
9299 ("*tldo_lduh1_sp32"): Likewise.
9300 ("*tldo_lduh_sp64"): Likewise.
9301 ("*tldo_lduh1_sp64"): Likewise.
9302 ("*tldo_lduh2_sp64"): Likewise.
9303 ("*tldo_lduw_sp32"): Likewise.
9304 ("*tldo_lduw_sp64"): Likewise.
9305 ("*tldo_lduw1_sp64"): Likewise.
9306 ("*tldo_ldx_sp64"): Likewise.
9307 ("*mov<VM32:mode>_insn"): Likewise.
9308 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9309 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9311 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9313 * config/sparc/sparc.md ("type"): New insn type viscmp.
9314 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9316 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9317 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9318 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9319 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9321 ("n7_vis_logical_11cycle"): Likewise.
9322 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9323 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9324 * config/sparc/niagara.md ("niag_vis"): Likewise.
9325 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9326 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9328 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9330 * config/sparc/sparc.md: New instruction type `bmask'.
9331 (bmaskdi_vis): Use the `bmask' type.
9332 (bmasksi_vis): Likewise.
9333 * config/sparc/ultra3.md (us3_array): Likewise.
9334 * config/sparc/niagara7.md (n7_array): Likewise.
9335 * config/sparc/niagara4.md (n4_array): Likewise.
9336 * config/sparc/niagara2.md (niag2_vis): Likewise.
9337 (niag3_vis): Likewise.
9338 * config/sparc/niagara.md (niag_vis): Likewise.
9340 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9342 * ipa-comdats.c: Remove optimize check from gate.
9343 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9344 for functions not optimized.
9345 (ipa_fn_summary_read): Skip optimize check.
9346 (ipa_fn_summary_write): Likewise.
9347 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9349 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9351 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9352 (check_callers): Give up on caller which is not optimized.
9353 (inline_small_functions): Likewise.
9354 (ipa_inline): Do not give up when not optimizing.
9355 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9356 away unoptimizes cdtors.
9357 (whole_program_function_and_variable_visibility): Do
9358 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9359 * ipa.c (process_references): Do not check optimize.
9360 (symbol_table::remove_unreachable_nodes): Update optimize check.
9361 (set_writeonly_bit): Update optimize check.
9362 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9363 (pass_ipa_single_use::gate): Remove.
9365 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9367 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9368 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9369 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9370 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9371 permute_load, permute_store, adjust_extract, adjust_splat,
9372 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9373 replace_swap_with_copy, dump_swap_insn_table,
9374 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9375 recombine_lvx_pattern, recombine_stvx_pattern,
9376 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9377 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9378 to file rs6000-p8swap.c.
9379 * config/rs6000/rs6000-p8swap.c: New file.
9380 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9381 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9382 and rs6000*-*-* targets.
9384 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9386 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9388 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9390 * lto-wrapper.c (merge_and_complain): Do not merge
9391 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9392 fsigned_zeros, ftrapping_math, fwrapv.
9393 (append_compiler_options): Do not track these options.
9394 (append_linker_options): Likewie
9396 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9398 * cgraphunit.c (cgraph_node::finalize_function): When
9399 !flag_toplevel_reorde set no_reorder flag.
9400 (varpool_node::finalize_decl): Likewise.
9401 (symbol_table::compile): Drop no toplevel reorder path.
9403 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9405 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9406 edges; zero probability is not better than uninitialized.
9408 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9410 * asan.h (asan_sanitize_allocas_p): Declare.
9411 * asan.c (asan_sanitize_allocas_p): New function.
9412 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9413 (handle_builtin_alloca): Likewise.
9414 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9415 if !asan_sanitize_allocas_p.
9416 * params.def (asan-instrument-allocas): Add new option.
9417 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9418 * opts.c (common_handle_option): Disable allocas sanitization for
9421 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9423 * asan.c: Include gimple-fold.h.
9424 (get_last_alloca_addr): New function.
9425 (handle_builtin_stackrestore): Likewise.
9426 (handle_builtin_alloca): Likewise.
9427 (asan_emit_allocas_unpoison): Likewise.
9428 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9429 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9430 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9431 (instrument_builtin_call): Pass gimple iterator to
9432 get_mem_refs_of_builtin_call.
9433 (last_alloca_addr): New global.
9434 * asan.h (asan_emit_allocas_unpoison): Declare.
9435 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9436 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9437 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9438 if function calls alloca.
9439 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9440 * gimple-fold.h (replace_call_with_value): Declare.
9441 * internal-fn.c: Include asan.h.
9442 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9443 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9445 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9447 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9448 (C_SELFTEST_FLAGS): New.
9449 (CPP_SELFTEST_FLAGS): New.
9450 (SELFTEST_DEPS): New, from deps of s-selftest.
9451 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9452 (CPP_SELFTEST_DEPS): New.
9453 (selftest): Add dependency on s-selftest-c++.
9454 (s-selftest): Rename to...
9455 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9456 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9457 than SELFTEST_FLAGS.
9458 (selftest-gdb): Rename to...
9459 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9461 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9462 (selftest-valgrind): Rename to...
9463 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9465 (selftest-valgrind): Reintroduce as an alias for
9466 selftest-c-valgrind.
9467 (s-selftest-c++): New.
9468 (selftest-c++-gdb): New.
9469 (selftest-c++-valgrind): New.
9471 2017-07-06 Olivier Hainque <hainque@adacore.com>
9473 * gcc.c (process_command): When deciding if undefined variables
9474 should be ignored when processing specs, accept "gcc -v" as well.
9476 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9478 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9479 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9481 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9483 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9484 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9485 * config/arm/arm-tables.opt: Regenerate.
9486 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9488 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9490 2017-07-06 Carl Love <cel@us.ibm.com>
9492 * ChangeLog: Clean up from mid air collision
9494 2017-07-06 Carl Love <cel@us.ibm.com>
9496 * config/rs6000/rs6000-c.c: Add support for built-in functions
9497 vector signed int vec_subc (vector signed int, vector signed int);
9498 vector signed __int128 vec_subc (vector signed __int128,
9499 vector signed __int128);
9500 vector unsigned __int128 vec_subc (vector unsigned __int128,
9501 vector unsigned __int128);
9502 vector signed int vec_sube (vector signed int, vector signed int,
9504 vector unsigned int vec_sube (vector unsigned int,
9505 vector unsigned int,
9506 vector unsigned int);
9507 vector signed __int128 vec_sube (vector signed __int128,
9508 vector signed __int128,
9509 vector signed__int128);
9510 vector unsigned __int128 vec_sube (vector unsigned __int128,
9511 vector unsigned __int128,
9512 vector unsigned __int128);
9513 vector signed int vec_subec (vector signed int, vector signed int,
9515 vector unsigned int vec_subec (vector unsigned int,
9516 vector unsigned int,
9517 vector unsigned int);
9518 vector signed __int128 vec_subec (vector signed __int128,
9519 vector signed __int128,
9520 vector signed__int128);
9521 vector unsigned __int128 vec_subec (vector unsigned __int128,
9522 vector unsigned __int128,
9523 vector unsigned __int128);
9524 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9525 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9526 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9527 BU_ALTIVEC_OVERLOAD_X definitions.
9528 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9529 * doc/extend.texi: Update the built-in documentation file for the new
9532 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9535 * diagnostic-show-locus.c (layout::layout): Use start and finish
9536 spelling location for the start and finish of each range.
9537 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9538 Add unused aspect param.
9539 * input.c (expand_location_1): Add "aspect" param, and use it
9540 to access the correct part of the location.
9541 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9543 (expand_location_to_spelling_point): Likewise.
9544 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9545 param, and pass it to expand_location_1.
9547 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9549 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9550 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9551 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9552 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9553 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9554 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9555 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9556 _mm_maskz_getmant_ss): New intrinsics.
9557 (__builtin_ia32_getexpss128_mask): Changed to ...
9558 __builtin_ia32_getexpss128_round ... this.
9559 (__builtin_ia32_getexpsd128_mask): Changed to ...
9560 __builtin_ia32_getexpsd128_round ... this.
9561 * config/i386/i386-builtin-types.def
9562 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9563 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9564 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9565 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9566 __builtin_ia32_getmantss_mask_round): New builtins.
9567 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9568 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9569 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9570 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9571 * config/i386/sse.md
9572 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9573 avx512f_sgetexp<mode><mask_scalar_name>
9574 <round_saeonly_scalar_name> ... this.
9575 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9576 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9577 vgetexp<ssescalarmodesuffix>
9578 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9579 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9580 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9581 avx512f_vgetmant<mode><mask_scalar_name>
9582 <round_saeonly_scalar_name> ... this.
9583 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9584 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9585 vgetmant<ssescalarmodesuffix>
9586 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9587 %0<mask_scalar_operand4>, %1, %2
9588 <round_saeonly_scalar_mask_op4>, %3} ... this.
9589 * config/i386/subst.md (mask_scalar_operand4,
9590 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9591 round_saeonly_scalar_nimm_predicate): New subst attributes.
9593 2017-07-06 Julia Koval <julia.koval@intel.com>
9595 * config/i386/i386.c (ix86_erase_embedded_rounding):
9596 Remove code for old rounding pattern.
9598 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9600 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9602 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9604 * doc/sourcebuild.texi (Test Directives, Variants of
9605 dg-require-support): Add documentation for dg-require-stack-check.
9607 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9609 * config/i386/subst.md (mask_scalar, round_scalar,
9610 round_saeonly_scalar): New meta-templates.
9611 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9612 round_scalar_mask_operand3, round_scalar_mask_op3,
9613 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9614 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9615 round_saeonly_scalar_constraint,
9616 round_saeonly_scalar_prefix): New subst attribute.
9617 * config/i386/sse.md
9618 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9619 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9620 <round_scalar_name> ... this.
9621 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9622 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9623 <round_scalar_name> ... this.
9624 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9625 <sse>_vm<code><mode>3<mask_scalar_name>
9626 <round_saeonly_scalar_name> ... this.
9627 (v<plusminus_mnemonic><ssescalarmodesuffix>
9628 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9629 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9630 v<plusminus_mnemonic><ssescalarmodesuffix>
9631 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9632 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9633 (v<multdiv_mnemonic><ssescalarmodesuffix>
9634 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9635 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9636 v<multdiv_mnemonic><ssescalarmodesuffix>
9637 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9638 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9639 (v<maxmin_float><ssescalarmodesuffix>
9640 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9641 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9642 v<maxmin_float><ssescalarmodesuffix>
9643 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9644 %0<mask_scalar_operand3>, %1, %<iptr>2
9645 <round_saeonly_scalar_mask_op3>} ... this.
9647 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9649 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9650 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9652 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9653 Alan Hayward <alan.hayward@arm.com>
9654 David Sherwood <david.sherwood@arm.com>
9656 * combine.c (simplify_if_then_else): Remove "enum" before
9658 * compare-elim.c (can_eliminate_compare): Likewise.
9659 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9661 (aarch64_lookup_simd_builtin_type): Likewise.
9662 (aarch64_simd_builtin_type): Likewise.
9663 (aarch64_init_simd_builtin_types): Likewise.
9664 (aarch64_simd_expand_args): Likewise.
9665 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9667 (aarch64_reverse_mask): Likewise.
9668 (aarch64_simd_emit_reg_reg_move): Likewise.
9669 (aarch64_gen_adjusted_ldpstp): Likewise.
9670 (aarch64_ccmp_mode_to_code): Likewise.
9671 (aarch64_operands_ok_for_ldpstp): Likewise.
9672 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9673 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9675 (aarch64_min_divisions_for_recip_mul): Likewise.
9676 (aarch64_reassociation_width): Likewise.
9677 (aarch64_get_condition_code_1): Likewise.
9678 (aarch64_simd_emit_reg_reg_move): Likewise.
9679 (aarch64_simd_attr_length_rglist): Likewise.
9680 (aarch64_reverse_mask): Likewise.
9681 (aarch64_operands_ok_for_ldpstp): Likewise.
9682 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9683 (aarch64_gen_adjusted_ldpstp): Likewise.
9684 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9686 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9687 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9688 (arm_lookup_simd_builtin_type): Likewise.
9689 (arm_simd_builtin_type): Likewise.
9690 (arm_init_simd_builtin_types): Likewise.
9691 (arm_expand_builtin_args): Likewise.
9692 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9693 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9694 (ft32_setup_incoming_varargs): Likewise.
9695 (ft32_function_arg): Likewise.
9696 (ft32_function_arg_advance): Likewise.
9697 (ft32_pass_by_reference): Likewise.
9698 (ft32_arg_partial_bytes): Likewise.
9699 (ft32_valid_pointer_mode): Likewise.
9700 (ft32_addr_space_pointer_mode): Likewise.
9701 (ft32_addr_space_legitimate_address_p): Likewise.
9702 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9704 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9705 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9706 (iamcu_alignment): Likewise.
9707 (canonicalize_vector_int_perm): Likewise.
9708 (ix86_noce_conversion_profitable_p): Likewise.
9709 (ix86_mpx_bound_mode): Likewise.
9710 (ix86_operands_ok_for_move_multiple): Likewise.
9711 * config/microblaze/microblaze-protos.h
9712 (microblaze_expand_conditional_branch_reg): Likewise.
9713 * config/microblaze/microblaze.c
9714 (microblaze_expand_conditional_branch_reg): Likewise.
9715 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9717 (rs6000_reassociation_width): Likewise.
9718 (rs6000_invalid_binary_op): Likewise.
9719 (fusion_p9_p): Likewise.
9720 (emit_fusion_p9_load): Likewise.
9721 (emit_fusion_p9_store): Likewise.
9722 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9724 (riscv_hard_regno_mode_ok_p): Likewise.
9725 (riscv_address_insns): Likewise.
9726 (riscv_split_symbol): Likewise.
9727 (riscv_legitimize_move): Likewise.
9728 (riscv_function_value): Likewise.
9729 (riscv_hard_regno_nregs): Likewise.
9730 (riscv_expand_builtin): Likewise.
9731 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9732 (riscv_build_integer): Likewise.
9733 (riscv_split_integer): Likewise.
9734 (riscv_legitimate_constant_p): Likewise.
9735 (riscv_cannot_force_const_mem): Likewise.
9736 (riscv_regno_mode_ok_for_base_p): Likewise.
9737 (riscv_valid_base_register_p): Likewise.
9738 (riscv_valid_offset_p): Likewise.
9739 (riscv_valid_lo_sum_p): Likewise.
9740 (riscv_classify_address): Likewise.
9741 (riscv_legitimate_address_p): Likewise.
9742 (riscv_address_insns): Likewise.
9743 (riscv_load_store_insns): Likewise.
9744 (riscv_force_binary): Likewise.
9745 (riscv_split_symbol): Likewise.
9746 (riscv_force_address): Likewise.
9747 (riscv_legitimize_address): Likewise.
9748 (riscv_move_integer): Likewise.
9749 (riscv_legitimize_const_move): Likewise.
9750 (riscv_legitimize_move): Likewise.
9751 (riscv_address_cost): Likewise.
9752 (riscv_subword): Likewise.
9753 (riscv_output_move): Likewise.
9754 (riscv_canonicalize_int_order_test): Likewise.
9755 (riscv_emit_int_order_test): Likewise.
9756 (riscv_function_arg_boundary): Likewise.
9757 (riscv_pass_mode_in_fpr_p): Likewise.
9758 (riscv_pass_fpr_single): Likewise.
9759 (riscv_pass_fpr_pair): Likewise.
9760 (riscv_get_arg_info): Likewise.
9761 (riscv_function_arg): Likewise.
9762 (riscv_function_arg_advance): Likewise.
9763 (riscv_arg_partial_bytes): Likewise.
9764 (riscv_function_value): Likewise.
9765 (riscv_pass_by_reference): Likewise.
9766 (riscv_setup_incoming_varargs): Likewise.
9767 (riscv_print_operand): Likewise.
9768 (riscv_elf_select_rtx_section): Likewise.
9769 (riscv_save_restore_reg): Likewise.
9770 (riscv_for_each_saved_reg): Likewise.
9771 (riscv_register_move_cost): Likewise.
9772 (riscv_hard_regno_mode_ok_p): Likewise.
9773 (riscv_hard_regno_nregs): Likewise.
9774 (riscv_class_max_nregs): Likewise.
9775 (riscv_memory_move_cost): Likewise.
9776 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9777 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9778 (rl78_addr_space_address_mode): Likewise.
9779 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9781 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9782 (rs6000_reassociation_width): Likewise.
9783 (rs6000_invalid_binary_op): Likewise.
9784 (fusion_p9_p): Likewise.
9785 (emit_fusion_p9_load): Likewise.
9786 (emit_fusion_p9_store): Likewise.
9787 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9788 (ok_for_simple_move_operands): Likewise.
9789 (ok_for_simple_move_strict_operands): Likewise.
9790 (ok_for_simple_arith_logic_operands): Likewise.
9791 (visium_legitimize_reload_address): Likewise.
9792 (visium_select_cc_mode): Likewise.
9793 (output_cbranch): Likewise.
9794 (visium_split_double_move): Likewise.
9795 (visium_expand_copysign): Likewise.
9796 (visium_expand_int_cstore): Likewise.
9797 (visium_expand_fp_cstore): Likewise.
9798 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9799 (visium_function_arg): Likewise.
9800 (visium_function_arg_advance): Likewise.
9801 (visium_libcall_value): Likewise.
9802 (visium_setup_incoming_varargs): Likewise.
9803 (visium_legitimate_constant_p): Likewise.
9804 (visium_legitimate_address_p): Likewise.
9805 (visium_legitimize_address): Likewise.
9806 (visium_secondary_reload): Likewise.
9807 (visium_register_move_cost): Likewise.
9808 (visium_memory_move_cost): Likewise.
9809 (prepare_move_operands): Likewise.
9810 (ok_for_simple_move_operands): Likewise.
9811 (ok_for_simple_move_strict_operands): Likewise.
9812 (ok_for_simple_arith_logic_operands): Likewise.
9813 (visium_function_value_1): Likewise.
9814 (rtx_ok_for_offset_p): Likewise.
9815 (visium_legitimize_reload_address): Likewise.
9816 (visium_split_double_move): Likewise.
9817 (visium_expand_copysign): Likewise.
9818 (visium_expand_int_cstore): Likewise.
9819 (visium_expand_fp_cstore): Likewise.
9820 (visium_split_cstore): Likewise.
9821 (visium_select_cc_mode): Likewise.
9822 (visium_split_cbranch): Likewise.
9823 (output_cbranch): Likewise.
9824 (visium_print_operand_address): Likewise.
9825 * expmed.c (flip_storage_order): Likewise.
9826 * expmed.h (emit_cstore): Likewise.
9827 (flip_storage_order): Likewise.
9828 * genrecog.c (validate_pattern): Likewise.
9829 * hsa-gen.c (gen_hsa_addr): Likewise.
9830 * internal-fn.c (expand_arith_overflow): Likewise.
9831 * ira-color.c (allocno_copy_cost_saving): Likewise.
9832 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9833 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9834 (process_invariant_for_inheritance): Likewise.
9835 * lra-eliminations.c (move_plus_up): Likewise.
9836 * omp-low.c (lower_oacc_reductions): Likewise.
9837 * simplify-rtx.c (simplify_subreg): Likewise.
9838 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9839 (TARGET_CHKP_BOUND_MODE): Likewise..
9840 * targhooks.c (default_chkp_bound_mode): Likewise.
9841 (default_setup_incoming_vararg_bounds): Likewise.
9842 * targhooks.h (default_chkp_bound_mode): Likewise.
9843 (default_setup_incoming_vararg_bounds): Likewise.
9844 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9845 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9846 (have_whole_vector_shift): Likewise.
9847 * tree-vect-stmts.c (vectorizable_load): Likewise.
9848 * doc/tm.texi: Regenerate.
9850 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9852 Graceful degrade if Binutils PR21472 is not available.
9855 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9856 .rodata in flash test fails.
9857 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9858 * confgure: Regenerate.
9859 * config.in: Regenerate.
9860 * config/avr/avr.c (avr_asm_named_section)
9861 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9862 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9863 (avr_asm_init_sections): Same.
9865 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9867 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9868 (fma<VH:mode>4_intrinsic): Likewise.
9869 (*fmsub<VCVTF:mode>4): Likewise.
9870 (*fmsub<VH:mode>4_intrinsic): Likewise.
9872 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9875 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9876 Don't depend on "optimize > 0".
9877 (out_movhi_r_mr, out_movqi_mr_r): Same.
9878 (out_movhi_mr_r, out_movqi_r_mr): Same.
9879 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9880 io_address_operand on "optimize > 0".
9882 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9884 * tree-loop-distribution.c: Add general explanantion on the pass.
9885 (generate_loops_for_partition): Mark distributed loop.
9886 (pg_add_dependence_edges): New parameter. Handle alias data
9887 dependence specially and record it in the parameter if asked.
9888 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9889 (init_partition_graph_vertices, add_partition_graph_edge): New.
9890 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9891 (free_partition_graph_vdata, build_partition_graph): New.
9892 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9893 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9894 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9895 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9896 (version_for_distribution_p, finalize_partitions): New.
9897 (distribute_loop): Handle alias data dependence specially. Factor
9898 out loop fusion code as functions and call these functions.
9900 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9902 * tree-loop-distribution.c (classify_partition): New parameter and
9903 better handle reduction statement.
9904 (rdg_build_partitions): Revise comment.
9905 (distribute_loop): Compute statements in all partitions and pass it
9906 to classify_partition.
9908 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9910 * tree-loop-distribution.c (enum partition_type): New.
9911 (struct partition): New field type.
9912 (partition_merge_into): Add parameter. Update partition type.
9913 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9914 (build_rdg_partition_for_vertex): Compute partition type.
9915 (rdg_build_partitions): Dump partition type.
9916 (distribute_loop): Update calls to partition_merge_into.
9918 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9920 * tree-loop-distribution.c (struct ddr_hasher): New.
9921 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9923 (classify_partition): Call get_data_dependence.
9924 (pg_add_dependence_edges): Ditto.
9925 (distribute_loop): Release data dependence hash table.
9927 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9929 * tree-loop-distribution.c (ref_base_address): Delete.
9930 (similar_memory_accesses): Rename ...
9931 (share_memory_accesses): ... to this. Check if partitions access
9932 the same memory reference.
9933 (distribute_loop): Call share_memory_accesses.
9935 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9937 * tree-loop-distribution.c (struct partition): New field recording
9939 (partition_alloc, partition_free): Init and release data refs.
9940 (partition_merge_into): Merge data refs.
9941 (build_rdg_partition_for_vertex): Collect data refs for partition.
9942 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9944 (distribute_loop): Remve data refs from vertice data of partition
9947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9949 * tree-loop-distribution.c (params.h): Include header file.
9950 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9951 (datarefs_vec): New global var.
9952 (create_rdg_vertices): Use datarefs_vec directly.
9953 (free_rdg): Don't free data references.
9954 (build_rdg): Update use. Don't free data references.
9955 (distribute_loop): Compute global variable for data references.
9956 Bail out if there are too many data references.
9958 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9960 * tree-loop-distribution.c (loop_nest): New global var.
9961 (build_rdg): Use loop directly, rather than loop nest.
9962 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9964 (distribute_loop): Compute global variable loop nest. Update use.
9966 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9968 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9969 (partition_merge_into): New parameter. Dump reason for fusion.
9970 (distribute_loop): Update use of partition_merge_into.
9972 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9974 * tree-loop-distribution.c (bb_top_order_index): New.
9975 (bb_top_order_index_size, bb_top_order_cmp): New.
9976 (stmts_from_loop): Use topological order.
9977 (pass_loop_distribution::execute): Compute and release topological
9978 order for basic blocks.
9980 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9982 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9985 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9987 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9988 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9989 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9990 * internal-fn.def (LOOP_DIST_ALIAS): New.
9991 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9992 (fold_loop_internal_call): ... this.
9993 (vect_loop_dist_alias_call): New function.
9994 (set_uid_loop_bbs): Call fold_loop_internal_call.
9995 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9998 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10001 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10002 Require dead FLAGS_REG at the beginning of a peephole.
10004 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10007 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10008 arguments in the call to __builtin_ia32_sbb_u32.
10009 (_subborrow_u64): Swap _X and _Y arguments in the call to
10010 __builtin_ia32_sbb_u64.
10012 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10015 * tree-vrp.c (compare_assert_loc): Turn into a function template
10016 with stable template parameter. Only test if a->e is NULL,
10017 !a->e == !b->e has been verified already. Use e == NULL or
10018 e != NULL instead of e or ! e tests. If stable is true, don't use
10019 iterative_hash_expr, on the other side allow a or b or both NULL
10020 and sort the NULLs last.
10021 (process_assert_insertions): Sort using compare_assert_loc<false>
10022 instead of compare_assert_loc, later sort using
10023 compare_assert_loc<true> before calling process_assert_insertions_for
10024 in a loop. Use break instead of continue once seen NULL pointer.
10026 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10028 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10029 Cortex-R7 and Cortex-R8 processors.
10031 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10033 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10034 uninitialized while src is not.
10036 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10038 * common/config/arm/arm-common.c: Adjust include path for
10040 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10041 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10042 (arm-cpu-data.h): Likewise.
10043 (arm-cpu-cdata.h): Likewise.
10044 * config/arm/arm-cpu.h: Delete.
10045 * config/arm/arm-cpu-cdata.h: Delete.
10046 * config/arm/arm-cpu-data.h: Delete.
10048 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10050 * config/arm/arm-cpus.in (cortex-a55): New.
10051 (cortex-a75): Likewise.
10052 (cortex-a75.cortex-a55): Likewise.
10053 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10055 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10056 * config/arm/arm-cpu-cdata.h: Regenerate.
10057 * config/arm/arm-cpu-data.h: Regenerate.
10058 * config/arm/arm-cpu.h: Regenerate.
10059 * config/arm/arm-tables.opt: Regenerate.
10060 * config/arm/arm-tune.md: Regenerate.
10062 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10064 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10066 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10068 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10071 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10073 PR tree-optimization/81292
10074 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10075 full_string_p, also call adjust_related_strinfos if the adjustment
10076 is simple, otherwise invalidate related strinfos.
10078 2017-07-04 Martin Liska <mliska@suse.cz>
10081 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10082 newly created variable as DECL_IGNORED_P.
10084 2017-07-04 Martin Liska <mliska@suse.cz>
10087 * ipa-inline.c (inline_small_functions):
10088 Use xstrdup_for_dump.
10090 2017-07-04 Tom de Vries <tom@codesourcery.com>
10092 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10094 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10097 * config/darwin.c (darwin_function_switched_text_sections):
10100 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10102 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10104 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10106 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10108 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10110 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10111 min_profitable_iters, and th as inclusive lower bounds.
10112 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10113 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10114 for min_profitable_iters and min_profitable_estimate.
10115 (vect_transform_loop): Treat th as an inclusive lower bound.
10116 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10118 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10121 * config/darwin.c (darwin_function_switched_text_sections):
10122 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10123 in two pieces, and suppress the use of buf.
10125 2017-07-03 Nathan Sidwell <nathan@acm.org>
10127 * hash-table.h (hash_table_mod1): Fix indentation.
10129 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10131 PR middle-end/81290
10132 * predict.c (force_edge_cold): Be more careful about propagation
10134 * profile-count.h (profile_probability::guessed,
10135 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10137 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10139 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10141 * doc/invoke.texi (rcpc architecture extension): Document it.
10143 2017-07-03 Richard Biener <rguenther@suse.de>
10145 PR tree-optimization/60510
10146 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10147 the scalar reduction PHI and use it.
10148 (vectorizable_reduction): Properly guard the single_defuse_cycle
10149 path for non-SLP reduction chains where we cannot use it.
10150 Rework reduc_def/index and vector type deduction. Rework
10151 vector operand gathering during reduction op code-gen.
10152 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10153 chains dissolve the chain and leave it to non-SLP reduction
10156 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10158 * tree-data-ref.h (dr_alignment): Declare.
10159 * tree-data-ref.c (dr_alignment): New function.
10160 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10161 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10163 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10165 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10167 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10168 and base_misalignment fields.
10169 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10170 * tree-data-ref.c: Include builtins.h.
10171 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10172 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10173 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10174 * tree-vect-data-refs.c: Include tree-cfg.h.
10175 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10176 fields instead of calculating an alignment here.
10177 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10178 innermost_loop_behavior fields.
10180 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10182 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10184 (DR_STEP_ALIGNMENT): New macro.
10185 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10186 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10187 (create_data_ref): Print it.
10188 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10189 to tell whether the step preserves vector (mis)alignment.
10190 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10191 Move the check for an integer step and generalise to all INTEGER_CST.
10192 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10193 Print the outer step alignment.
10195 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10197 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10198 with offset_alignment.
10199 (DR_ALIGNED_TO): Delete.
10200 (DR_OFFSET_ALIGNMENT): New macro.
10201 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10202 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10203 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10204 (create_data_ref): Likewise.
10205 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10206 (vect_analyze_data_refs): Likewise.
10207 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10208 creating dummy innermost behavior.
10210 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10212 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10213 with a "innermost_loop_behavior *" and refeence tree.
10214 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10215 (create_data_ref): Update call accordingly.
10216 * tree-predcom.c (find_looparound_phi): Likewise.
10218 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10220 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10221 fields with dr_wrt_vec_loop.
10222 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10223 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10224 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10225 (vect_dr_behavior): New function.
10226 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10227 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10228 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10229 track whether the step preserves the misalignment.
10230 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10231 Use vect_dr_behavior.
10232 (vect_setup_realignment): Update call accordingly.
10233 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10234 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10235 call to vect_create_addr_base_for_vector_ref.
10236 (vect_create_cond_for_align_checks): Likewise.
10237 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10238 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10239 (vect_recog_mask_conversion_pattern): Likewise.
10240 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10241 (new_stmt_vec_info): Remove redundant zeroing.
10243 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10245 * common/config/arm/arm-common.c (arm_be8_option): New function.
10246 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10247 (ISA_ARMv6): Add isa_bit_be8.
10248 * config/arm/arm.h (arm_be8_option): Add prototype.
10249 (BE8_SPEC_FUNCTION): New define.
10250 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10251 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10252 (mlittle-endian): Similarly.
10253 (mbe8, mbe32): New options.
10254 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10255 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10257 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10259 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10261 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10263 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10264 (cleanup_tree_cfg_bb): Use it.
10265 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10267 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10269 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10272 * loop-doloop.c (add_test): Update profile.
10274 2017-07-03 Martin Liska <mliska@suse.cz>
10277 * sanopt.c (rewrite_usage_of_param): New function.
10278 (sanitize_rewrite_addressable_params): Likewise.
10279 (pass_sanopt::execute): Call rewrite_usage_of_param.
10281 2017-07-03 Richard Biener <rguenther@suse.de>
10283 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10284 back to using VIEW_CONVERT_EXPR.
10286 2017-07-03 Martin Liska <mliska@suse.cz>
10289 * doc/extend.texi: Document when a resolver function is
10290 generated for target_clones.
10292 2017-07-03 Martin Liska <mliska@suse.cz>
10294 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10295 and shadow memory of auto variables which are subject of
10296 use-after-scope sanitization.
10297 (asan_expand_mark_ifn): Add do set only when is_poison.
10299 2016-07-03 Richard Biener <rguenther@suse.de>
10301 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10303 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10304 (vectorizable_reduction): Perform reduction PHI creation when
10305 visiting a reduction PHI and adjust and simplify code generation
10306 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10307 (vect_transform_loop): Visit reduction PHIs.
10308 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10309 defs into the SLP tree.
10310 (vect_build_slp_tree): Reduction defs terminate the recursion.
10311 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10313 (vect_get_vec_defs_for_stmt_copy): Export.
10314 (vect_get_vec_defs): Likewise.
10315 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10317 (vect_get_vec_defs_for_stmt_copy): Declare.
10318 (vect_get_vec_defs): Likewise.
10320 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10322 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10323 parameter with a "loop" parameter and use it instead of the
10324 loop containing DR_STMT. Don't check simple_iv when doing
10325 BB analysis. Describe the two analysis modes in the comment.
10327 2017-07-03 Tom de Vries <tom@codesourcery.com>
10329 PR tree-optimization/69468
10330 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10331 (find_same_succ_bb): Handle ignore_edge_flags.
10333 2017-07-03 Tom de Vries <tom@codesourcery.com>
10335 PR tree-optimization/81192
10336 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10338 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10340 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10342 2017-07-03 Tom de Vries <tom@codesourcery.com>
10344 PR tree-optimization/81192
10345 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10346 BB_SAME_SUCC (bb) == NULL.
10348 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10350 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10353 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10355 * dumpfile.c: Include profile-count.h
10356 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10358 (insert_cond_bb): Update profile.
10359 * tree-cfg.h (insert_cond_bb): Update prototype.
10360 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10361 * tree-dump.c: Do not include tree-cfg.
10363 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10365 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10367 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10369 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10372 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10374 * tree-complex.c (expand_complex_div_wide): update profile.
10376 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10377 Alan Hayward <alan.hayward@arm.com>
10378 David Sherwood <david.sherwood@arm.com>
10380 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10381 (CORETYPES_H): New define.
10382 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10383 (insn-modes-inline.h, s-modes-inline-h): New rules.
10384 (generated_files): Add insn-modes-inline.h.
10385 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10386 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10387 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10388 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10389 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10390 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10391 (build/gencodes.o, build/genconditions.o): Likewise.
10392 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10393 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10394 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10395 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10396 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10397 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10398 * coretypes.h: Include everything up to real.h for generators.
10399 Include insn-modes.h first. Include wide-int-print.h after
10400 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10401 * machmode.h: Don't include insn-modes.h here.
10402 * function-tests.c: Remove includes of signop.h, machmode.h,
10403 double-int.h and wide-int.h.
10405 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10407 * optc-save-gen.awk: Likewise.
10408 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10409 * godump.c: Remove include of wide-int-print.h.
10410 * pretty-print.h: Likewise.
10411 * wide-int-print.cc: Likewise.
10412 * wide-int.cc: Likewise.
10413 * hash-map-tests.c: Remove include of signop.h.
10414 * hash-set-tests.c: Likewise.
10415 * rtl-tests.c: Likewise.
10416 * mkconfig.sh: Remove include of machmode.h.
10417 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10419 (emit_insn_modes_inline_h): ...this new function. Emit the code
10420 into an insn-modes-inline.h header file, adding appropriate
10421 include guards and end comments.
10422 (emit_insn_modes_c_header): Remove include of machmode.h.
10423 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10425 (main): Handle -i flag and call emit_insn_modes_inline_h when
10428 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10430 * tree-ssa-strlen.c (strinfo): Rename the length field to
10431 nonzero_chars. Add a full_string_p field.
10432 (compare_nonzero_chars, zero_length_string_p): New functions.
10433 (get_addr_stridx): Add an offset_out parameter.
10434 Use compare_nonzero_chars.
10435 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10436 (new_strinfo): Update after above changes to strinfo.
10437 (set_endptr_and_length): Set full_string_p.
10438 (get_string_length): Update after above changes to strinfo.
10439 (unshare_strinfo): Update call to new_strinfo.
10440 (maybe_invalidate): Likewise.
10441 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10442 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10443 as a uhwi instead of an shwi. Update after above changes to
10444 strinfo and new_strinfo.
10445 (zero_length_string): Assert that chainsi contains full strings.
10446 Use zero_length_string_p. Update call to new_strinfo.
10447 (adjust_related_strinfos): Update after above changes to strinfo.
10448 Copy full_string_p from origsi.
10449 (adjust_last_stmt): Use zero_length_string_p.
10450 (handle_builtin_strlen): Update after above changes to strinfo and
10451 new_strinfo. Install the lhs as the string length if the previous
10452 entry didn't describe a full string.
10453 (handle_builtin_strchr): Update after above changes to strinfo
10455 (handle_builtin_strcpy): Likewise.
10456 (handle_builtin_strcat): Likewise.
10457 (handle_builtin_malloc): Likewise.
10458 (handle_pointer_plus): Likewise.
10459 (handle_builtin_memcpy): Likewise. Track nonzero characters
10460 that aren't necessarily followed by a nul terminator.
10461 (handle_char_store): Likewise.
10463 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10465 PR tree-optimization/80769
10466 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10467 for malloc and calloc. Document the new invariant that all related
10468 strinfos have delayed lengths or none do.
10469 (verify_related_strinfos): Move earlier in file.
10470 (set_endptr_and_length): New function, split out from...
10471 (get_string_length): ...here. Also set the lengths of related
10473 (zero_length_string): Assert that chainsi has known (rather than
10475 (adjust_related_strinfos): Likewise.
10477 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10479 PR tree-optimization/81136
10480 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10481 assert that two references with the same misalignment have the same
10482 compile-time misalignment if those compile-time misalignments
10485 2017-07-01 Andi Kleen <ak@linux.intel.com>
10487 * print-tree.c (print_node): Print all attributes.
10489 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10491 * cfg.c (scale_bbs_frequencies): New function.
10492 * cfg.h (scale_bbs_frequencies): Declare it.
10493 * cfgloopanal.c (single_likely_exit): Cleanup.
10494 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10496 (scale_loop_profile): Likewise.
10497 (loop_version): Likewise.
10498 (create_empty_loop_on_edge): Update.
10499 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10500 scale_loop_frequencies, scale_loop_profile, loopify,
10501 loop_version): Update prototypes.
10502 * modulo-sched.c (sms_schedule): Update.
10503 * predict.c (unlikely_executed_edge_p): Also check probability.
10504 (probably_never_executed_edge_p): Fix typo.
10505 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10506 * tree-parloops.c (gen_parallel_loop): Update.
10507 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10508 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10509 * tree-ssa-loop-split.c (split_loop): Update.
10510 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10511 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10512 (vect_loop_versioning): Update.
10513 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10515 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10517 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10519 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10521 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10522 to keep profile consistent.
10524 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10526 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10527 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10528 * profile-count.h (max_safe_multiplier): Make unsigned.
10529 (profile_count::guessed_zero): New.
10531 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10533 * bb-reorder.c (fix_up_crossing_landing_pad,
10534 fix_crossing_conditional_branches): Use make_single_succ_edge
10535 to keep profile consistent.
10537 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10539 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10542 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10545 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10546 the right scopes, make sure cond_jump isn't preserved between multiple
10547 iterations. Search for fallthru edge whenever there are 3+ edges and
10548 use find_fallthru_edge for it.
10550 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10552 Patch by Alexander Monakov <amonakov@ispras.ru>
10553 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10554 probabilities consistently.
10556 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10558 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10559 emit_cmp_and_jump_insns.
10561 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10564 * tree-inline.c (expand_call_inline): Combine profile statuses.
10566 2017-06-30 Andrew Pinski <apinski@cavium.com>
10568 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10569 fold_stmt returned true.
10571 2017-06-30 Nathan Sidwell <nathan@acm.org>
10573 * ggc.h (empty_string): Delete.
10574 * cfgexpand.c (expand_asm_stmt): Use plain "".
10575 * optabs.c (expand_asm_memory_barrier): Likewise.
10576 * stringpool.c (empty_string): Delete.
10577 (digit_vector, digit_string): Delete.
10578 (ggc_alloc_string): Use plain "", don't optimize single digit
10579 strings. Use ggc_alloc_atomic.
10581 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10583 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10584 comparison set and one other set, use the cost of the non-comparison
10587 2017-06-30 Nathan Sidwell <nathan@acm.org>
10589 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10592 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10594 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10595 loops. Remove now unneeded calls to gimple_switch_set_label() that
10596 just set removed labels to NULL_TREE.
10598 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10600 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10601 (set_range_info): ...here. Only call set_range_info_raw if domain
10603 (set_nonzero_bits): Call set_range_info_raw.
10604 * tree-ssanames.h (set_range_info_raw): New.
10606 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10609 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10610 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10611 of nonimmediate_operand and <store_mask_constraint> instead of m
10612 for the input operand. For V8FI iterator, always split if input
10613 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10614 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10615 <store_mask_predicate> instead of register_operand and
10616 <store_mask_constraint> instead of v for the input operand. Make
10617 sure both operands aren't MEMs for if not <mask_applied>.
10619 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10621 * lto-wrapper.c (copy_file) Close both file descriptors before
10624 2017-06-30 Martin Liska <mliska@suse.cz>
10627 * multiple_target.c (create_dispatcher_calls): Make ifunc
10628 also for function that don't have calls or are not referenced.
10630 2017-06-30 Richard Biener <rguenther@suse.de>
10632 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10633 analyze the first scalar stmt. Move vector type computation
10634 for the BB case here from ...
10635 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10636 live operation processing in the SLP case properly.
10638 2017-06-30 Richard Biener <rguenther@suse.de>
10640 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10642 2017-06-30 Martin Liska <mliska@suse.cz>
10645 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10646 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10648 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10650 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10653 2017-06-30 Richard Biener <rguenther@suse.de>
10655 PR tree-optimization/81249
10656 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10657 condition reduction result to original scalar type.
10659 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10661 * profile-count.h (enum profile_quality): Fix typos and whitespace
10664 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10666 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10667 type for branch probabilities.
10669 2017-06-29 Julian Brown <julian@codesourcery.com>
10670 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10672 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10673 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10674 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10675 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10677 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10679 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10680 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10681 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10682 CC usage from generic code to here.
10683 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10684 CC usage into the target macros.
10686 2017-06-29 Maya Rashish <coypu@sdf.org>
10688 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10691 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10693 * arm/arm-builtins.c: Include profile-count.h
10694 * except.c (sjlj_emit_function_enter): Use
10695 profile_probability::unlikely.
10697 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10699 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10700 and tocrel_offset be pointer args rather than implicitly using
10702 (legitimate_constant_pool_address_p, rs6000_emit_move,
10703 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10704 tocrel_offset and use in toc_relative_expr_p call.
10705 (print_operand, print_operand_address): Use static tocrel_base_oac
10706 and tocrel_offset_oac.
10707 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10710 2017-06-29 Maya Rashish <coypu@sdf.org>
10712 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10714 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10716 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10717 objects, take into account only the alignment of 'op0' and 'mode1' if
10720 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10722 * ccmp.c (ccmp_tree_comparison_p): New function.
10723 (ccmp_candidate_p): Update to use above function.
10724 (get_compare_parts): New function.
10725 (expand_ccmp_next): Update to use new functions.
10726 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10728 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10729 take mode as argument.
10730 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10731 * expr.c (expand_expr_real_1): Pass mode as argument.
10733 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10735 * combine.c (combine_instructions): Print insns to dump_file, together
10738 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10740 * asan.c (asan_emit_stack_protection): Update.
10741 (create_cond_insert_point): Update.
10742 * auto-profile.c (afdo_propagate_circuit): Update.
10743 * basic-block.h (struct edge_def): Turn probability to
10744 profile_probability.
10745 (EDGE_FREQUENCY): Update.
10746 * bb-reorder.c (find_traces_1_round): Update.
10747 (better_edge_p): Update.
10748 (sanitize_hot_paths): Update.
10749 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10750 (make_single_succ_edge): Update.
10751 (check_bb_profile): Update.
10752 (dump_edge_info): Update.
10753 (update_bb_profile_for_threading): Update.
10754 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10755 probabilitycount to 0.
10756 * cfgbuild.c (compute_outgoing_frequencies): Update.
10757 * cfgcleanup.c (try_forward_edges): Update.
10758 (outgoing_edges_match): Update.
10759 (try_crossjump_to_edge): Update.
10760 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10761 (expand_gimple_tailcall): Update.
10762 (construct_init_block): Use make_single_succ_edge.
10763 (construct_exit_block): Use make_single_succ_edge.
10764 * cfghooks.c (verify_flow_info): Update.
10765 (redirect_edge_succ_nodup): Update.
10766 (split_edge): Update.
10767 (account_profile_record): Update.
10768 * cfgloopanal.c (single_likely_exit): Update.
10769 * cfgloopmanip.c (scale_loop_profile): Update.
10770 (set_zero_probability): Remove.
10771 (duplicate_loop_to_header_edge): Update.
10772 * cfgloopmanip.h (loop_version): Update prototype.
10773 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10774 (force_nonfallthru_and_redirect): Update.
10775 (update_br_prob_note): Update.
10776 (rtl_verify_edges): Update.
10777 (purge_dead_edges): Update.
10778 (rtl_lv_add_condition_to_bb): Update.
10779 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10780 * cgraphunit.c (init_lowered_empty_function): Update.
10781 (cgraph_node::expand_thunk): Update.
10782 * cilk-common.c: Include profile-count.h
10783 * dojump.c (inv): Remove.
10784 (jumpifnot): Update.
10785 (jumpifnot_1): Update.
10786 (do_jump_1): Update.
10788 (do_jump_by_parts_greater_rtx): Update.
10789 (do_compare_rtx_and_jump): Update.
10790 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10791 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10792 * dwarf2cfi.c: Include profile-count.h
10793 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10794 (sjlj_emit_dispatch_table): Likewise.
10795 * explow.c: Include profile-count.h
10796 * expmed.c (emit_store_flag_force): Update.
10797 (do_cmp_and_jump): Update.
10798 * expr.c (compare_by_pieces_d::generate): Update.
10799 (compare_by_pieces_d::finish_mode): Update.
10800 (emit_block_move_via_loop): Update.
10801 (store_expr_with_bounds): Update.
10802 (store_constructor): Update.
10803 (expand_expr_real_2): Update.
10804 (expand_expr_real_1): Update.
10805 * expr.h (try_casesi, try_tablejump): Update prototypes.
10806 * gimple-pretty-print.c (dump_probability): Update.
10807 (dump_profile): New.
10808 (dump_gimple_label): Update.
10809 (dump_gimple_bb_header): Update.
10810 * graph.c (draw_cfg_node_succ_edges): Update.
10811 * hsa-gen.c (convert_switch_statements): Update.
10812 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10813 (find_if_case_1): Update.
10814 (find_if_case_2): Update.
10815 * internal-fn.c (expand_arith_overflow_result_store): Update.
10816 (expand_addsub_overflow): Update.
10817 (expand_neg_overflow): Update.
10818 (expand_mul_overflow): Update.
10819 (expand_vector_ubsan_overflow): Update.
10820 * ipa-cp.c (good_cloning_opportunity_p): Update.
10821 * ipa-split.c (split_function): Use make_single_succ_edge.
10822 * ipa-utils.c (ipa_merge_profiles): Update.
10823 * loop-doloop.c (add_test): Update.
10824 (doloop_modify): Update.
10825 * loop-unroll.c (compare_and_jump_seq): Update.
10826 (unroll_loop_runtime_iterations): Update.
10827 * lra-constraints.c (lra_inheritance): Update.
10828 * lto-streamer-in.c (input_cfg): Update.
10829 * lto-streamer-out.c (output_cfg): Update.
10830 * mcf.c (adjust_cfg_counts): Update.
10831 * modulo-sched.c (sms_schedule): Update.
10832 * omp-expand.c (expand_omp_for_init_counts): Update.
10833 (extract_omp_for_update_vars): Update.
10834 (expand_omp_ordered_sink): Update.
10835 (expand_omp_for_ordered_loops): Update.
10836 (expand_omp_for_generic): Update.
10837 (expand_omp_for_static_nochunk): Update.
10838 (expand_omp_for_static_chunk): Update.
10839 (expand_cilk_for): Update.
10840 (expand_omp_simd): Update.
10841 (expand_omp_taskloop_for_outer): Update.
10842 (expand_omp_taskloop_for_inner): Update.
10843 * omp-simd-clone.c (simd_clone_adjust): Update.
10844 * optabs.c (expand_doubleword_shift): Update.
10845 (expand_abs): Update.
10846 (emit_cmp_and_jump_insn_1): Update.
10847 (expand_compare_and_swap_loop): Update.
10848 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10849 * predict.c (predictable_edge_p): Update.
10850 (edge_probability_reliable_p): Update.
10851 (set_even_probabilities): Update.
10852 (combine_predictions_for_insn): Update.
10853 (combine_predictions_for_bb): Update.
10854 (propagate_freq): Update.
10855 (estimate_bb_frequencies): Update.
10856 (force_edge_cold): Update.
10857 * profile-count.c (profile_count::dump): Add missing space into dump.
10858 (profile_count::debug): Add newline.
10859 (profile_count::differs_from_p): Explicitly convert to unsigned.
10860 (profile_count::stream_in): Update.
10861 (profile_probability::dump): New member function.
10862 (profile_probability::debug): New member function.
10863 (profile_probability::differs_from_p): New member function.
10864 (profile_probability::differs_lot_from_p): New member function.
10865 (profile_probability::stream_in): New member function.
10866 (profile_probability::stream_out): New member function.
10867 * profile-count.h (profile_count_quality): Rename to ...
10868 (profile_quality): ... this one.
10869 (profile_probability): New.
10870 (profile_count): Update.
10871 * profile.c (compute_branch_probabilities): Update.
10872 * recog.c (peep2_attempt): Update.
10873 * sched-ebb.c (schedule_ebbs): Update.
10874 * sched-rgn.c (find_single_block_region): Update.
10875 (compute_dom_prob_ps): Update.
10876 (schedule_region): Update.
10877 * sel-sched-ir.c (compute_succs_info): Update.
10878 * stmt.c (struct case_node): Update.
10879 (do_jump_if_equal): Update.
10880 (get_outgoing_edge_probs): Update.
10881 (conditional_probability): Update.
10882 (emit_case_dispatch_table): Update.
10883 (expand_case): Update.
10884 (expand_sjlj_dispatch_table): Update.
10885 (emit_case_nodes): Update.
10886 * targhooks.c: Update.
10887 * tracer.c (better_p): Update.
10888 (find_best_successor): Update.
10889 * trans-mem.c (expand_transaction): Update.
10890 * tree-call-cdce.c: Update.
10891 * tree-cfg.c (gimple_split_edge): Upate.
10892 (move_sese_region_to_fn): Upate.
10893 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10894 * tree-eh.c (lower_resx): Upate.
10895 (cleanup_empty_eh_move_lp): Upate.
10896 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10897 * tree-inline.c (copy_edges_for_bb): Update.
10898 (copy_cfg_body): Update.
10899 * tree-parloops.c (gen_parallel_loop): Update.
10900 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10901 (gimple_gen_time_profiler): Update.
10902 * tree-ssa-dce.c (remove_dead_stmt): Update.
10903 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10904 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10905 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10906 (unloop_loops): Update.
10907 (try_peel_loop): Update.
10908 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10909 * tree-ssa-loop-split.c (connect_loops): Update.
10910 (split_loop): Update.
10911 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10912 (hoist_guard): Update.
10913 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10914 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10915 (value_replacement): Update.
10916 * tree-ssa-reassoc.c (branch_fixup): Update.
10917 * tree-ssa-tail-merge.c (replace_block_by): Update.
10918 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10919 (create_edge_and_update_destination_phis): Update.
10920 (compute_path_counts): Update.
10921 (recompute_probabilities): Update.
10922 (update_joiner_offpath_counts): Update.
10923 (freqs_to_counts_path): Update.
10924 (duplicate_thread_path): Update.
10925 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10926 (struct switch_conv_info): Update.
10927 (gen_inbound_check): Update.
10928 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10929 (vect_do_peeling): Update.
10930 (vect_loop_versioning): Update.
10931 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10932 (optimize_mask_stores): Update.
10933 * ubsan.c (ubsan_expand_null_ifn): Update.
10934 * value-prof.c (gimple_divmod_fixed_value): Update.
10935 (gimple_divmod_fixed_value_transform): Update.
10936 (gimple_mod_pow2): Update.
10937 (gimple_mod_pow2_value_transform): Update.
10938 (gimple_mod_subtract): Update.
10939 (gimple_mod_subtract_transform): Update.
10940 (gimple_ic): Update.
10941 (gimple_stringop_fixed_value): Update.
10942 (gimple_stringops_transform): Update.
10943 * value-prof.h: Update.
10945 2017-06-29 Carl Love <cel@us.ibm.com>
10947 * config/rs6000/rs6000-c.c: Add support for built-in functions
10948 vector signed int vec_signed (vector float);
10949 vector signed long long vec_signed (vector double);
10950 vector signed int vec_signed2 (vector double, vector double);
10951 vector signed int vec_signede (vector double);
10952 vector signed int vec_signedo (vector double);
10953 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10954 instruction generator.
10955 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10956 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10957 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10959 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10960 vunsignede_v2df): Add define_expands.
10961 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10962 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10963 VEC_UNSIGNEDO): Add definitions.
10964 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10965 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10966 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10967 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10968 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10969 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10970 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10971 vec_unsignede, vec_unsignedo): Add builtin defines.
10972 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10974 * doc/extend.texi: Update the built-in documentation file for the
10975 new built-in functions.
10977 2017-06-29 Richard Biener <rguenther@suse.de>
10979 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10980 reduction chains to LOOP_VINFO_REDUCTIONS.
10981 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10982 SLP reductions after processing reduction chains.
10984 2017-06-29 Nathan Sidwell <nathan@acm.org>
10986 * builtins.c (fold_builtin_FUNCTION): Use
10987 lang_hooks.decl_printable_name.
10989 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10991 PR middle-end/81194
10992 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10993 with only one label.
10994 * stmt.c (expand_case): Assert NCASES is greater than one.
10996 2017-06-29 Richard Biener <rguenther@suse.de>
10998 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11000 (group_case_labels): Likewise.
11001 (find_taken_edge): Push sanity checking on val to workers...
11002 (find_taken_edge_cond_expr): ... here
11003 (find_taken_edge_switch_expr): ... and here, handle cases
11004 with just a default label.
11005 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11006 (group_case_labels): Likewise.
11007 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11008 group_case_labels does anything cleanup the CFG again.
11010 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11012 PR tree-optimization/81196
11013 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11014 exit condition comparing two IVs.
11016 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11018 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11019 profile to the dummy entry at the end of the list of architectures.
11020 * config/arm/arm-cpu-cdata.h: Regenerated.
11022 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11023 Michael Collison <michael.collison@arm.com>
11026 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11028 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11029 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11030 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11031 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11032 of shift when the shift amount is masked with constant equal to
11033 the size of the mode.
11034 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11037 2017-06-29 Martin Liska <mliska@suse.cz>
11039 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11041 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11044 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11047 (edge_growth_cache_entry::edge_growth_cache_entry): New
11049 (reset_edge_growth_cache): Update to use constructor.
11051 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11053 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11054 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11055 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11057 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11059 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11060 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11062 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11064 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11065 (*-linux-uclibc*): Add t-uclibc tmake_file.
11066 * config/t-musl: New.
11067 * config/t-uclibc: New.
11069 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11071 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11073 (gen_comm_data): Emit architectural setting of arch_prof.
11074 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11076 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11077 (armv8-m.base, armv8-m.main): Likewise.
11078 * arm-protos.h (arm_build_target): Add profile field.
11079 (arch_option): Likewise.
11080 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11082 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11083 arm_active_target.profile.
11085 2017-06-28 Richard Biener <rguenther@suse.de>
11087 PR middle-end/81227
11088 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11089 TYPE_OVERFLOW_WRAPS.
11090 * match.pd (negate_expr_p): Likewise.
11091 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11092 fold_build2, not fold_binary.
11094 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11096 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11097 Convert memory address to Pmode.
11098 (aarch64_print_operand): Assert MEM operands are always Pmode.
11100 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11103 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11104 Remove redundant if.
11105 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11106 * config/arm/aarch-common-protos.h
11107 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11108 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11110 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11113 * multiple_target.c (create_dispatcher_calls): Set the default
11114 clone to be static, not public.
11116 2017-06-28 Richard Biener <rguenther@suse.de>
11118 * tree-vect-loop.c (vectorizable_reduction): Move special
11119 cond reduction IV var creation ...
11120 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11121 parameter. Use STMT_VINFO_VECTYPE.
11122 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11125 2017-06-28 Martin Liska <mliska@suse.cz>
11128 * ipa-visibility.c (non_local_p): Handle visibility.
11130 2017-06-28 Martin Liska <mliska@suse.cz>
11133 * common.opt: Add IntegerRange to various options.
11134 * opt-functions.awk (integer_range_info): New function.
11135 * optc-gen.awk: Add integer_range_info to cl_options struct.
11136 * opts-common.c (decode_cmdline_option): Handle
11137 CL_ERR_INT_RANGE_ARG.
11138 (cmdline_handle_error): Likewise.
11139 * opts.c (print_filtered_help): Show valid interval in
11140 when --help is provided.
11141 * opts.h (struct cl_option): Add range_min and range_max fields.
11142 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11144 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11146 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11147 (x * C EQ/NE y * C): New transformation.
11149 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11151 * genmultilib (combination_space): Accept '+' in option names.
11153 2017-06-28 Martin Liska <mliska@suse.cz>
11156 * asan.c (instrument_derefs): Bail out inner references
11157 that are hard register variables.
11159 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11162 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11163 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11165 2017-06-28 Richard Biener <rguenther@suse.de>
11167 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11168 (vect_get_slp_defs): Adjust.
11169 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11171 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11173 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11174 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11175 (vectorizable_reduction): Adjust.
11176 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11178 (vect_get_slp_defs): Likewise.
11179 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11180 (vectorizable_bswap): Adjust.
11181 (vectorizable_call): Likewise.
11182 (vectorizable_conversion): Likewise.
11183 (vectorizable_assignment): Likewise.
11184 (vectorizable_shift): Likewise.
11185 (vectorizable_operation): Likewise.
11186 (vectorizable_store): Likewise.
11187 (vectorizable_condition): Likewise.
11188 (vectorizable_comparison): Likewise.
11190 2017-06-28 Michael Collison <michael.collison@arm.com>
11193 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11195 (arm_gen_movmemqi): Removed unused variable 'i'.
11196 Convert 'for' loop into 'while' loop.
11197 (arm_expand_prologue): Remove last unnecessary set of insn.
11198 (thumb_pop): Remove unused variable 'pushed_words'.
11199 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11201 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11203 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11204 * config/s390/s390-protos.h: Add prototype of
11205 s390_rel_address_ok_p.
11206 * config/s390/s390.c (s390_got_symbol): New function.
11207 (s390_rel_address_ok_p): New function.
11208 (legitimize_pic_address): Use s390_rel_address_ok_p.
11209 (s390_load_got): Use s390_got_symbol.
11210 (s390_option_override): Issue error if
11211 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11212 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11214 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11216 2017-06-27 Andrew Pinski <apinski@cavium.com>
11218 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11219 (X * copysign (1.0, X)): New pattern.
11220 (X * copysign (1.0, -X)): New pattern.
11221 (copysign (-1.0, CST)): New pattern.
11223 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11225 * genmultilib (combination_space): Remove variable.
11226 Validate reuse rules against regular expression for any sequence
11227 of multilib options in any order.
11229 2017-06-27 Michael Collison <michael.collison@arm.com>
11231 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11232 call aarch64_split_simd_combine.
11233 * (aarch64_combine_internal<mode>): Delete pattern.
11234 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11235 Allow register and subreg operands.
11237 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11239 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11240 specific need, just fallback on defaults.
11241 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11243 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11244 Olivier Hainque <hainque@adacore.com>
11246 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11248 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11249 targets. Pick a default if no particular attempt applied.
11250 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11253 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11255 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11256 (x86_64-wrs-vxworks7): Likewise.
11258 2017-06-27 Marek Polacek <polacek@redhat.com>
11261 * ubsan.c (instrument_null): Check get_base_address's result for null.
11263 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11265 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11267 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11269 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11270 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11271 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11272 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11273 New function types.
11274 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11275 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11276 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11277 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11278 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11279 BUILT_IN_FEUPDATEENV): New builtins.
11280 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11281 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11282 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11283 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11285 (builtin_structptr_types): Adjust size.
11286 * tree.c (builtin_structptr_types): Add four entries.
11288 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11289 Olivier Hainque <hainque@adacore.com>
11291 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11292 (TLS_SYM): New local macro, forcing reference to __tls__ on
11293 link command lines for VxWorks 7 RTPs, triggering initialization
11295 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11296 OS features TLS support, true for RTPs on VxWorks 7.
11297 * config/vxworks.c (vxworks_override_options): Setup emutls
11300 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11302 * predict.c (test_prediction_value_range): Use -1U instead of -1
11303 to avoid narrowing conversion warning.
11304 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11305 to avoid narrowing conversion warning.
11306 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11308 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11310 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11312 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11313 64bit configurations.
11314 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11315 (SIZE_TYPE): Likewise.
11316 * config/vxworks.c (vxworks_emutls_var_fields): Use
11317 long_unsigned_type_node instead of unsigned_type_node as the offset
11318 field type, which is "pointer" mode in emutls.c.
11320 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11323 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11325 PR middle-end/81207
11326 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11327 gimple_vuse copying separately from gimple_vdef copying.
11329 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11331 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11334 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11335 Olivier Hainque <hainque@adacore.com>
11337 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11338 for all vxworks7 targets.
11339 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11340 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11341 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11342 variations for VX6/VX7 and 32/64bits later on in ...
11343 (VXWORKS_LIB_SPEC): Leverage new macros.
11344 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11345 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11347 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11349 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11350 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11352 2017-06-26 Carl Love <cel@us.ibm.com>
11354 * config/rs6000/rs6000-c.c: Add support for built-in functions
11355 vector bool char vec_reve (vector bool char);
11356 vector signed char vec_reve (vector signed char);
11357 vector unsigned char vec_reve (vector unsigned char);
11358 vector bool int vec_reve (vector bool int);
11359 vector signed int vec_reve (vector signed int);
11360 vector unsigned int vec_reve (vector unsigned int);
11361 vector bool long long vec_reve (vector bool long long);
11362 vector signed long long vec_reve (vector signed long long);
11363 vector unsigned long long vec_reve (vector unsigned long long);
11364 vector bool short vec_reve (vector bool short);
11365 vector signed short vec_reve (vector signed short);
11366 vector double vec_reve (vector double);
11367 vector float vec_reve (vector float);
11368 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11369 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11370 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11371 (altivec_vreve): New pattern.
11372 * config/rs6000/altivec.h (vec_reve): New define.
11373 * doc/extend.texi (vec_rev): Update the built-in documentation file
11374 for the new built-in functions.
11376 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11378 PR tree-optimization/71815
11379 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11381 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11383 (slsr_process_phi): Likewise.
11384 (replace_uncond_cands_and_profitable_phis): Don't replace a
11385 multiply candidate with a stride of 1 (copy or cast).
11386 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11388 (lowest_cost_path): Likewise.
11389 (total_savings): Likewise.
11391 2017-06-26 Richard Biener <rguenther@suse.de>
11394 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11395 Use def_builtin_pure for all gather builtins.
11397 2017-06-26 Richard Biener <rguenther@suse.de>
11399 PR tree-optimization/81203
11400 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11401 non-dominating BBs.
11403 2017-06-26 Marek Polacek <polacek@redhat.com>
11406 * doc/invoke.texi: Document -Wmultistatement-macros.
11408 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11410 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11411 arm_neon_ok_no_float_abi effective target.
11413 2017-06-26 Richard Biener <rguenther@suse.de>
11415 PR tree-optimization/80928
11416 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11417 (copy_bbs): Set BB_DUPLICATED flag early.
11418 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11420 (execute_on_shrinking_pred): Likewise.
11421 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11422 BB_DUPLICATED blocks.
11423 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11424 iterate over all PHIs considering removal of *gsi.
11426 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11428 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11431 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11433 * config/rs6000/rs6000-string.c: (expand_block_clear,
11434 do_load_for_compare, select_block_compare_mode,
11435 compute_current_alignment, expand_block_compare,
11436 expand_strncmp_align_check, expand_strn_compare,
11437 expand_block_move, rs6000_output_load_multiple)
11438 Move functions related to string/block move/compare
11439 to a separate file.
11440 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11441 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11442 for this function which is now used in two files.
11443 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11444 * config.gcc: Add rs6000-string.o to extra_objs for
11445 targets powerpc*-*-* and rs6000*-*-*.
11447 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11450 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11451 32-bit, since indexed is not valid for DImode.
11452 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11453 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11454 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11455 (define_peephole2 for Altivec d-form store): Likewise.
11458 * multiple_target.c (create_dispatcher_calls): Only create the
11459 dispatcher call if the function is the default clone of a
11460 versioned function.
11462 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11464 PR middle-end/80902
11465 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11466 a call, force the call to not be a tail call.
11468 2017-06-23 Jeff Law <law@redhat.com>
11470 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11473 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11475 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11476 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11477 handling for early expansion of vector shifts (sl,sr,sra,rl).
11478 (builtin_function_type): Add vector shift right instructions
11479 to the unsigned argument list.
11481 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11483 rtl-optimizatoin/79286
11484 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11485 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11486 trap. PIC register plus a const unspec without offset can never trap.
11488 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11490 * tree.h (builtin_structptr_type): New type.
11491 (builtin_structptr_types): Declare new array.
11492 * tree.c (builtin_structptr_types): New array.
11493 (free_lang_data, build_common_tree_nodes): Use it.
11495 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11498 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11501 2017-06-22 Matt Turner <mattst88@gmail.com>
11503 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11504 Lake models to skylake case. Assume skylake for unknown
11505 models with clflushopt.
11507 2017-06-22 Jeff Law <law@redhat.com>
11509 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11510 frame sizes that do not satisfy aarch64_uimm12_shift.
11512 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11514 * profile-count.h (apply_probability,
11515 apply_scale, probability_in): Fix checks for zero.
11517 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11519 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11520 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11522 2016-06-22 Richard Biener <rguenther@suse.de>
11524 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11525 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11526 REDUC_MAX_EXPR support.
11527 (vectorizable_reduction): Likewise.
11528 (vect_create_epilog_for_reduction): Likewise.
11530 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11532 * match.pd (A / (1 << B) -> A >> B): New.
11533 * generic-match-head.c: Include optabs-tree.h.
11534 * gimple-match-head.c: Likewise.
11535 * optabs-tree.h (target_supports_op_p): New.
11536 * optabs-tree.c (target_supports_op_p): New.
11538 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11540 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11541 $gcc_cv_ld --help output.
11542 (gcc_cv_ld_demangle): Likewise.
11543 (gcc_cv_ld_eh_frame_hdr): Likewise.
11544 (gcc_cv_ld_pie): Likewise.
11545 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11546 (gcc_cv_ld_buildid): Likewise.
11547 (gcc_cv_ld_sysroot): Likewise.
11548 (ld_bndplt_support): Likewise.
11549 (ld_pushpopstate_support): Likewise.
11550 * configure: Regenerate.
11551 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11553 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11556 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11557 operands indexes to avoid gap between operands and match_dups.
11559 2017-06-21 Andrew Pinski <apinski@cavium.com>
11561 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11562 Increment Arith_shift and Arith_shift_reg by 1.
11563 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11565 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11566 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11567 (aarch64_strip_extend): Add new argument and test for it.
11568 (aarch64_cheap_mult_shift_p): New function.
11569 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11570 add a cost if it is true.
11571 Update calls to aarch64_strip_extend.
11572 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11574 2017-06-21 Andrew Pinski <apinski@cavium.com>
11576 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11578 (thunderxt88): Likewise.
11579 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11580 (thunderx_prefetch_tune): New variable.
11581 (thunderx2t99_prefetch_tune): Update for the correct values.
11582 (thunderxt88_tunings): New variable.
11583 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11584 generic_prefetch_tune.
11585 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11587 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11589 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11590 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11591 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11592 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11593 (aarch64_atomic_cas<mode>, GPI): Likewise.
11595 2017-06-21 Martin Liska <mliska@suse.cz>
11597 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11598 statements on cold and hot labels.
11599 * predict.c (tree_estimate_probability_bb): Remove the
11600 prediction from this place.
11602 2017-06-21 Martin Liska <mliska@suse.cz>
11604 PR tree-optimization/79489
11605 * gimplify.c (maybe_add_early_return_predict_stmt): New
11607 (gimplify_return_expr): Call the function.
11608 * predict.c (tree_estimate_probability_bb): Remove handling
11610 * predict.def: Update comment about early return predictor.
11611 * gimple-predict.h (is_gimple_predict): New function.
11612 * predict.def: Change default value of early return to 66.
11613 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11615 * passes.def: Put pass_strip_predict_hints to the beginning of
11618 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11620 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11621 FUNCTION_DECL declarations.
11622 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11624 (dwaf2out_decl): Likewise.
11625 * godump.c (go_early_global_decl): Skip call to the real debug hook
11626 for FUNCTION_DECL declarations.
11627 * passes.c (rest_of_decl_compilation): Skip call to the
11628 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11629 -fdump-go-spec is passed.
11631 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11633 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11634 Reorder for compactness.
11635 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11636 (def_builtin_pure, def_builtin_pure2): New functions.
11637 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11639 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11641 * match.pd (nop_convert): New predicate.
11642 ((A +- CST1) +- CST2): Allow some NOP conversions.
11644 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11647 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11648 with ctors/dtors if GOVD_SHARED is set.
11650 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11652 * config/aarch64/aarch64.md (movti_aarch64):
11653 Emit mov rather than orr.
11654 (movtf_aarch64): Likewise.
11655 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11656 Emit mov rather than orr.
11658 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11660 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11661 Swap alternatives, make integer dup more expensive.
11663 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11665 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11666 Return true for non-tls symbols.
11668 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11670 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11671 (cortex-a75): Likewise.
11672 (cortex-a75.cortex-a55): Likewise.
11673 * config/aarch64/aarch64-tune.md: Regenerate.
11674 * doc/invoke.texi (-mtune): Document new values for -mtune.
11676 2017-06-21 Tom de Vries <tom@codesourcery.com>
11678 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11679 stack_size feature.
11680 (Effective-Target Keywords, Other attributes): Suggest using
11681 dg-add-options stack_size feature to get stack limit in stack_size
11682 effective target documentation.
11684 2017-06-21 Julian Brown <julian@codesourcery.com>
11685 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11687 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11688 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11689 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11691 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11692 attribute type list for neon_multiply.
11693 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11694 attribute type list for neon_multiply.
11695 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11696 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11697 attribute type list for neon_multiply.
11698 * config/arm/types.md (crypto_pmull): Add.
11699 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11700 attribute type list.
11702 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11704 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11707 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11709 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11710 to make sure not to dereference a NULL cost_classes_ptr pointer.
11712 2017-06-20 Carl Love <cel@us.ibm.com>
11714 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11715 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11716 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11717 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11718 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11719 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11720 VMULOSW): New enum "unspec" values.
11721 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11722 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11723 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11724 altivec_vmulosw): New patterns.
11725 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11726 VMULOSW): Add definitions.
11728 2017-06-20 Julia Koval <julia.koval@intel.com>
11730 * config/i386/i386.c: Fix rounding expand for new pattern.
11731 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11733 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11735 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11736 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11738 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11740 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11743 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11745 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11746 architecture, then by implementer ID.
11747 * config/aarch64/aarch64-tune.md: Regenerate.
11749 2017-06-20 Richard Biener <rguenther@suse.de>
11751 PR middle-end/81097
11752 * fold-const.c (split_tree): Fold to type before negating.
11754 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11756 * diagnostic-show-locus.c
11757 (selftest::test_fixit_deletion_affecting_newline): New function.
11758 (selftest::diagnostic_show_locus_c_tests): Call it.
11760 2017-06-20 Andreas Schwab <schwab@suse.de>
11763 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11766 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11768 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11769 __ARM_FEATURE_COPROC according to support.
11771 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11773 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11774 Rewritten to avoid overflow for > 32-bit pointers.
11777 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11778 by removing enum keyword.
11779 (ubsan_type_descriptor): Likewise. Formatting fix.
11782 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11783 splitter): Require TARGET_SSE2 in the condition.
11785 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11788 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11789 for doing vector set of SFmode on ISA 3.0.
11790 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11791 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11793 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11794 SFmode value into a V4SF variable that was extracted from another
11795 V4SF variable without converting the element to double precision
11796 and back to single precision vector format.
11797 (vsx_insert_extract_v4sf_p9_2): Likewise.
11799 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11801 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11802 in UWHI to avoid undefined overflow.
11805 * ubsan.h (enum ubsan_encode_value_phase): New.
11806 (ubsan_encode_value): Change second argument to
11807 enum ubsan_encode_value_phase with default value of
11808 UBSAN_ENCODE_VALUE_GENERIC.
11809 * ubsan.c (ubsan_encode_value): Change second argument to
11810 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11811 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11812 create_tmp_var_raw instead of create_tmp_var and use a
11814 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11815 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11816 ubsan_encode_value callers.
11819 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11820 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11821 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11823 2017-06-19 Richard Biener <rguenther@suse.de>
11825 PR middle-end/81118
11826 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11827 estimates if we changed anything.
11829 2017-06-19 Richard Biener <rguenther@suse.de>
11831 PR tree-optimization/80887
11832 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11833 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11834 simplified lookups, then reset mprts_hook.
11835 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11837 (try_to_simplify): Likewise.
11839 2017-06-19 Martin Liska <mliska@suse.cz>
11842 * gimplify.c (gimplify_switch_expr):
11843 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11845 2017-06-19 Martin Liska <mliska@suse.cz>
11847 * doc/install.texi: Document that PGO runs in 4 stages.
11849 2017-06-19 Martin Liska <mliska@suse.cz>
11852 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11853 to dispatcher function name.
11854 * multiple_target.c (replace_function_decl): New function.
11855 (create_dispatcher_calls): Redirect both edges and references.
11857 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11859 * profile-count.c (profile_count::dump): Dump quality.
11860 (profile_count::differs_from_p): Update for unsigned val.
11861 * profile-count.h (profile_count_quality): New enum.
11862 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11864 2017-06-19 Richard Biener <rguenther@suse.de>
11866 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11867 struct function as arg.
11868 (estimate_numbers_of_iterations): Export overload with loop arg.
11869 (free_numbers_of_iterations_estimates_loop): Use an overload of
11870 free_numbers_of_iterations_estimates instead.
11871 * tree-cfg.c (remove_bb): Adjust.
11872 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11873 * tree-parloops.c (gen_parallel_loop): Likewise.
11874 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11876 (tree_unroll_loops_completely): Likewise.
11877 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11878 Use an overload instead and export.
11879 (estimated_loop_iterations): Adjust.
11880 (max_loop_iterations): Likewise.
11881 (likely_max_loop_iterations): Likewise.
11882 (estimate_numbers_of_iterations): Take struct function as arg
11884 (loop_exits_before_overflow): Adjust.
11885 (free_numbers_of_iterations_estimates_loop): Use an overload.
11886 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11887 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11889 2017-06-19 Richard Biener <rguenther@suse.de>
11892 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11893 RANGE_EXPR conservatively.
11895 2017-06-16 Carl Love <cel@us.ibm.com>
11897 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11898 definitions for vec_float, vec_float2, vec_floato,
11899 vec_floate built-ins.
11900 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11901 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11903 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11904 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11905 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11906 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11907 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11908 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11909 vec_floato): Add builtin defines.
11910 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11911 Update the built-in documentation file for the new built-in
11914 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11916 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11917 (mthumb): Mark as the negative of -marm.
11919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11921 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11923 (ARM Options, -mtune): Document that this accepts the same
11924 extension options as -mcpu.
11925 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11929 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11930 permitted extensions.
11932 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11934 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11935 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11936 (armv8-m.main): Add option +nodsp.
11937 * config/arm/arm-cpu-cdata.h: Regenerated.
11939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11941 * config/arm/t-fuchsia: New file.
11942 * config.gcc (arm*-*-fuchsia*): Use it.
11944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11946 * config/arm/t-symbian: Rewrite for new option infrastructure.
11948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11950 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11951 (MULTILIB_REQUIRED): Likewise.
11953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11955 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11956 (MULTILIB_RESUE): Likewise.
11957 (MULTILIB_MATCHES): Likewise.
11958 (MULTLIB_REQUIRED): Likewise.
11960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11962 * config/arm/t-rtems: Rewrite for new option framework.
11964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11966 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11967 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11968 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11969 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11970 * config/arm/t-multilib: ... here.
11971 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11972 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11973 armv7-a and armv8*-a when A-profile libraries have not been built.
11974 * config/arm/t-rmprofile: Rewrite.
11976 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11978 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11979 with a backslash. Remove the backslash after substituting unescaped
11981 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11985 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11986 the driver to pass through the default setting of -mfloat-abi.
11987 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11989 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11990 (all_feat_combs): New rule.
11991 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11993 * config/arm/t-aprofile: Rewrite.
11995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11997 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11998 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11999 fpu is not specified by the user/command-line.
12000 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12001 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12002 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12003 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12004 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12005 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12007 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12009 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12010 * config/arm/t-arm-elf: Rewritten.
12012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12014 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12015 have some floating-point instructions.
12016 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12017 (TARGET_MAYBE_HARD_FLOAT): New macro.
12018 * config/arm/arm-builtins.c (arm_init_builtins): Use
12019 TARGET_MAYBE_HARD_FLOAT.
12020 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12024 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12025 (configargs.h): Include it.
12026 (arm_print_hint_for_fpu_option): New function.
12027 (arm_parse_fpu_option): New function.
12028 (candidate_extension): New class.
12029 (arm_canon_for_multilib): New function.
12030 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12031 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12032 (ARCH_CANONICAL_SPECS): New macro.
12033 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12035 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12037 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12038 are set after handling multilib fragments. Set target_cpu_default2
12041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12043 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12045 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12046 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12048 (check_arch): Likewise.
12049 * config/arm/arm.c (arm_configure_build_target): Handle
12050 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12051 options in the default.
12053 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12055 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12056 when an option is an alias of another.
12057 * config/arm/parsecpu.awk (optalias): New parser token.
12058 (gen_comm_data): Mark non-alias options as such. Emit entries
12059 for extension aliases.
12060 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12061 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12062 (armv6kz, armv6zk, armv6t2): Likewise.
12063 (armv7): Make vfpv3-d16 an alias.
12064 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12066 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12067 Sort in canonical order.
12068 (armv8-a): Sort in canonical order.
12069 (armv8.1-a, armv8.2-a): Likewise.
12070 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12072 (cortex-a9): Sort in canonical order.
12073 * config/arm/arm.c (selftests.h): Include it.
12074 (arm_test_cpu_arch_data): New function.
12075 (arm_run_self_tests): New function.
12076 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12077 (targetm): Move declaration to the end of the file.
12078 * arm-cpu-cdata.h: Regenerated.
12080 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12082 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12083 call to target_mode_check describing the type of option passed.
12084 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12085 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12086 arm_parse_cpu_option_name to match parameters against list of
12088 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12089 arm_arch_core_flags data structure.
12090 * config/arm/arm-cpu_cdata.h: Regenerated.
12092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12094 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12096 (arm_print_hint_for_cpu_option): Likewise.
12097 (arm_print_hint_for_arch_option): Likewise.
12098 (arm_parse_cpu_option_name): Likewise.
12099 (arm_parse_arch_option_name): Likewise.
12100 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12101 of entries in the all_fpus list.
12102 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12103 (arm_parse_cpu_option_name): Declare.
12104 (arm_parse_arch_option_name): Declare.
12105 (arm_parse_option_features): Declare.
12106 (arm_intialize_isa): Declare.
12107 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12109 (gen_comm_data): ... here. Make definitions non-static.
12110 * config/arm/arm-cpu-data.h: Regenerated.
12111 * config/arm/arm-cpu-cdata.h: Regenerated.
12113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12115 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12116 (cpu_arch_extension): New structure.
12117 (cpu_arch_option, arch_option, cpu_option): New structures.
12118 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12119 architecture types.
12120 (gen_data): Generate new format data tables.
12121 * config/arm/arm.c (cpu_tune): New structure.
12122 (cpu_option, processors): Delete.
12123 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12124 (arm_print_hint_for_cpu_option): ... this and ...
12125 (arm_print_hint_for_arch_option): ... this.
12126 (arm_parse_arch_cpu_name): Delete. Replace with ...
12127 (arm_parse_cpu_option_name): ... this and ...
12128 (arm_parse_arch_option_name): ... this.
12129 (arm_unrecognized_feature): Change type of target parameter to
12131 (arm_parse_arch_cpu_features): Delete. Replace with ...
12132 (arm_parse_option_features): ... this.
12133 (arm_configure_build_target): Rework to use new configuration data
12135 (arm_print_tune_info): Rework for new configuration data tables.
12136 * config/arm/arm-cpu-data.h: Regenerated.
12137 * config/arm/arm-cpu.h: Regenerated.
12139 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12141 * Makefile.in (OBJS): Move sbitmap.o from here ...
12142 (OBJS-libcommon): ... to here.
12144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12146 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12147 (ISA_ALL_CRYPTO): New macro.
12148 (ISA_ALL_SIMD): New macro
12149 (ISA_ALL_FP): New macro.
12150 * config/arm/arm.c (fpu_bitlist): Update initializer.
12151 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12153 (arm9e): Add fpu. Add option for nofp
12154 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12155 (arm926ej-s, arm1026ej-s): Likewise.
12156 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12157 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12158 neon-fp16, neon-vfpv4, nofp and nosimd.
12159 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12160 (cortex-a8): Add fpu. Add option for nofp.
12161 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12162 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12163 (cortex-r4f): Add fpu.
12164 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12165 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12167 (cortex-r8): Likewise.
12168 (cortex-m4): Add fpu. Add option for nofp.
12169 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12170 (cortex-a17.cortex-a7): Likewise.
12171 (cortex-a32): Add fpu. Add options for crypto and nofp.
12172 (cortex-a35, cortex-a53): Likewise.
12173 (cortex-a57): Add fpu. Add option for crypto.
12174 (cortex-a72, cortex-a73): Likewise.
12175 (exynos-m1): Likewise.
12176 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12177 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12178 (cortex-m33): Add fpu. Add option for nofp.
12179 * config/arm/arm-cpu-cdata.h: Regenerated
12180 * config/arm/arm-cpu-data.h: Regenerated.
12182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12184 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12185 (armv5te, armv5tej): Likewise.
12186 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12187 (armv7): Add options fp and vfpv3-d16.
12188 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12189 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12191 (armv7ve): Likewise.
12192 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12193 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12194 (armv8-a): Add nocrypto option.
12195 (armv8.1-a, armv8.2-a): Likewise.
12196 (armv8-m.main): add options fp, fp.dp and nofp.
12198 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12200 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12202 (armv8-a+crc): Delete.
12203 (armv8.1-a): Add options simd, crypto and nofp.
12204 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12205 (armv8.2-a+fp16): Delete.
12206 (armv8-m.main): Add option dsp.
12207 (armv8-m.main+dsp): Delete.
12208 (cortex-a8): Add fpu. Add nofp option.
12209 (cortex-a9): Add fpu. Add nofp and nosimd options.
12210 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12211 link to main cpu and architecture data structures.
12212 (gen_comm_data): Only put isa attributes from the main architecture
12214 (option): New statement for architecture and CPU entries.
12215 * arm.c (struct cpu_option): New structure.
12216 (struct processors): Add entry for options.
12217 (arm_unrecognized_feature): New function.
12218 (arm_parse_arch_cpu_name): Ignore any characters after the first
12220 (arm_parse_arch_cpu_feature): New function.
12221 (arm_configure_build_target): Separate out any CPU and architecture
12222 features and parse separately. Don't error out if -mfpu=auto is
12223 used with only an architecture string.
12224 (arm_print_asm_arch_directives): New function.
12225 (arm_file_start): Call it.
12226 * config/arm/arm-cpu-cdata.h: Regenerated.
12227 * config/arm/arm-cpu-data.h: Likewise.
12228 * config/arm/arm-tables.opt: Likewise.
12230 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12232 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12233 assembler when it is not -mfpu=auto.
12235 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12237 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12238 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12239 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12240 (ASM_CPU_SPEC): Rewrite.
12241 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12242 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12243 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12244 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12245 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12246 copied string is NUL-terminated. Also strip any characters prefixed
12248 (arm_rewrite_selected_arch): New function.
12249 (arm_rewrite_march): New function.
12251 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12253 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12254 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12255 (march, mcpu, mtune): Convert to string-based options.
12256 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12257 (arm_parse_arch_cpu_name): New function.
12258 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12259 identify selected architecture or CPU.
12260 (arm_option_save): New function.
12261 (TARGET_OPTION_SAVE): Redefine.
12262 (arm_option_restore): Restore string options.
12263 (arm_option_print): Print string options.
12265 2017-06-16 Martin Sebor <msebor@redhat.com>
12267 PR tree-optimization/80933
12268 PR tree-optimization/80934
12269 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12270 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12271 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12272 (gimple_fold_builtin): Call them.
12274 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12276 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12277 as unlikely; update profile.
12279 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12281 * predict.c (force_edge_cold): Handle declaring edges impossible
12284 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12286 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12288 (try_unroll_loop_completely): Fix reporting.
12290 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12292 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12294 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12297 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12298 if given a non-constant argument for an intrinsic which requires a
12301 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12303 * profile.c (compare_freqs): New function.
12304 (branch_prob): Sort edge list.
12305 (find_spanning_tree): Assume that the list is priority sorted.
12307 2017-06-16 Richard Biener <rguenther@suse.de>
12309 PR tree-optimization/81090
12310 * passes.def (pass_record_bounds): Remove.
12311 * tree-pass.h (make_pass_record_bounds): Likewise.
12312 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12313 make_pass_record_bounds): Likewise.
12314 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12315 not free niter estimates at the beginning but at the end.
12316 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12318 2017-06-16 Richard Biener <rguenther@suse.de>
12320 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12321 initializer to workaround ICE in host GCC 4.8.
12323 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12325 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12327 (clone_inlined_nodes): Update.
12329 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12331 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12332 prefetch settings, and enable prefetching by default at -O3.
12334 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12336 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12337 Set flag_prefetch_loop_arrays according to tuning data.
12339 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12341 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12342 New tune structure.
12343 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12344 [Unrelated to main purpose of the patch] Place the pointer field last
12345 to enable type checking errors when tune structure are wrongly merged.
12346 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12347 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12348 (thunderx2t99_prefetch_tune): New tune constants.
12349 (tune_params *_tunings): Update all tunings (no functional change).
12350 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12351 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12352 from tunings structures.
12354 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12357 * ubsan.c (instrument_null): Add T argument, use it instead
12358 of computing it based on IS_LHS.
12359 (instrument_object_size): Likewise.
12360 (pass_ubsan::execute): Adjust instrument_null and
12361 instrument_object_size callers to pass gimple_get_lhs or
12362 gimple_assign_rhs1 result to it. Use instrument_null instead of
12363 calling get_base_address and instrument_mem_ref. Handle
12364 aggregate call arguments for object-size sanitization.
12366 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12368 PR tree-optimization/81089
12369 * tree-vrp.c (is_masked_range_test): Validate operands of
12372 2017-06-15 Martin Sebor <msebor@redhat.com>
12375 * dumpfile.c (dump_register): Avoid calling memset to initialize
12376 a class with a default ctor.
12377 * gcc.c (struct compiler): Remove const qualification.
12378 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12379 * hash-table.h: Ditto.
12380 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12382 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12383 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12385 * params.h (struct param_info): Make struct members non-const.
12386 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12387 with default initialization.
12388 * vec.h (vec_copy_construct, vec_default_construct): New helper
12390 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12391 with vec_copy_construct.
12392 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12393 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12394 * doc/invoke.texi (-Wclass-memaccess): Document.
12396 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12398 * emit-rtl.h (is_leaf): Update comment about local
12399 register allocator.
12401 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12404 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12405 for a variable to have a section before checking if the section has a
12407 Set section to.persistent if persistent attribute is set.
12408 Warn if .persistent attribute is used on an automatic variable.
12410 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12412 PR rtl-optimization/80474
12413 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12415 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12417 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12420 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12422 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12423 (casesi): Emit bounds checking as RTL.
12424 (casesi_internal_mips16_<mode>): Remove bounds checking.
12426 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12428 * config/xtensa/xtensa.c (xtensa_option_override): Append
12429 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12430 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12431 xtensa_doloop_hooks): Define unconditionally.
12432 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12434 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12435 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12436 for it in xtensa_option_override.
12437 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12438 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12440 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12442 * doc/cppopts.texi: Document '-' special value to -MF.
12444 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12446 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12447 (cortex_a53_fconst): Likewise.
12448 (cortex_a53_fpmul): Likewise.
12449 (cortex_a53_f_load_64): Likewise.
12450 (cortex_a53_f_load_many): Likewise.
12451 (cortex_a53_advsimd_alu): Likewise.
12452 (cortex_a53_advsimd_alu_q): Likewise.
12453 (cortex_a53_advsimd_mul): Likewise.
12454 (cortex_a53_advsimd_mul_q): Likewise.
12455 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12456 Add missing fmul, r2f_cvt and fconst cases.
12458 2017-06-14 Richard Biener <rguenther@suse.de>
12460 PR middle-end/81088
12461 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12463 (fold_binary_loc): When associating do not treat pre-existing
12464 TREE_OVERFLOW on literal constants as a reason to allow
12465 TREE_OVERFLOW on associated literal constants.
12467 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12469 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12470 (MASK_FEATURES): New macro.
12471 * config/sparc/sparc.c (sparc_option_override): Remove the special
12472 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12474 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12476 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12477 a division of 0 if non-call exceptions are enabled.
12479 2017-06-14 Andrew Pinski <apinski@cavium.com>
12480 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12483 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12484 Improve vector initialization code gen for only variable case.
12486 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12488 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12490 2017-06-14 Richard Biener <rguenther@suse.de>
12492 PR tree-optimization/81083
12493 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12496 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12498 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12499 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12500 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12501 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12502 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12503 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12505 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12507 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12508 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12510 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12512 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12514 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12516 * config/rs6000/t-rtems: Don't handle SPE.
12518 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12520 * config/rs6000/t-linux: Don't handle SPE.
12522 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12524 * config/rs6000/eabispe.h: Delete file.
12526 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12528 * config/rs6000/t-spe: Delete file.
12530 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12532 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12533 (rs6000_legitimate_offset_address_p): Return false for anything in
12534 V2SImode or V2SFmode.
12536 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12538 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12539 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12540 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12542 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12543 (rs6000_legitimate_offset_address_p): Ditto.
12544 (rs6000_emit_move): Ditto.
12545 (rs6000_init_builtins): Remove V4HI_type_node.
12547 2017-06-13 Martin Liska <mliska@suse.cz>
12550 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12551 (gate_asan): Likewise.
12552 * asan.h (asan_no_sanitize_address_p): Remove the function.
12553 (sanitize_flags_p): New function.
12554 * builtins.def: Fix coding style.
12555 * common.opt: Use renamed enum value.
12556 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12557 * doc/extend.texi: Document no_sanitize attribute.
12558 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12559 to SANITIZE_UNDEFINED_NONDEFAULT.
12560 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12561 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12562 Use sanitize_flags_p.
12563 * gimplify.c (gimplify_function_tree): Likewise.
12564 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12565 * opts.c (parse_no_sanitize_attribute): New function.
12566 (common_handle_option): Use renamed enum value.
12567 * opts.h (parse_no_sanitize_attribute): Declare.
12568 * tree.c (sanitize_flags_p): New function.
12569 * tree.h: Declared here.
12570 * tsan.c: Use sanitize_flags_p.
12571 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12572 (instrument_mem_ref): Likewise.
12573 (instrument_bool_enum_load): Likewise.
12574 (do_ubsan_in_current_function): Remove the function.
12575 (pass_ubsan::execute): Use sanitize_flags_p.
12576 * ubsan.h: Remove do_ubsan_in_current_function
12577 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12578 (dump_function_to_file): Use it here.
12580 2017-06-13 Martin Jambor <mjambor@suse.cz>
12582 PR tree-optimization/80803
12583 PR tree-optimization/81063
12584 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12585 (propagate_subaccesses_across_link): Enqueue subtree whenever
12586 necessary instead of relying on the caller.
12588 2017-06-13 Martin Jambor <mjambor@suse.cz>
12590 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12591 that have a first_link.
12592 (sort_and_splice_var_accesses): Do not check first_link before
12594 (subtree_mark_written_and_enqueue): Likewise.
12595 (propagate_all_subaccesses): Likewise and do not stop at first
12596 parent with a first_link.
12598 2017-06-13 Martin Jambor <mjambor@suse.cz>
12600 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12603 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12605 * match.pd: New pattern.
12607 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12609 * tree-vrp.c (is_masked_range_test): New function.
12610 (register_edge_assert_for): Determine ranges for
12613 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12615 PR tree-optimization/67328
12616 * fold-const.c (maskable_range_p): New function.
12617 (build_range_check): Generate bittests if possible.
12619 2017-06-13 Martin Liska <mliska@suse.cz>
12621 * gimple-pretty-print.c (dump_probability): Add new argument.
12622 (dump_edge_probability): Dump both probability and count.
12623 (dump_gimple_label): Likewise.
12624 (dump_gimple_bb_header): Likewise.
12626 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12629 * config/avr/avr-devices.c: Fix indentation.
12630 * config/avr/gen-avr-mmcu-specs.c: Dito.
12632 2017-06-13 Richard Biener <rguenther@suse.de>
12634 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12635 instead get vector type from stmt_info.
12636 (vectorizable_reduction): Adjust. Remove dead code.
12638 2017-06-13 Richard Biener <rguenther@suse.de>
12640 PR middle-end/81065
12641 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12642 case of C * (x * C2 + C3).
12643 (fold_addr_of_array_ref_difference): Properly fold index difference.
12645 2017-06-12 David S. Miller <davem@davemloft.net>
12648 * config/sparc/sparc.md (return expander): Emit frame blockage if
12649 function uses alloca.
12651 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12653 * combine.c (make_field_assignment): Check len rather than the mode
12654 precision when calling force_to_mode.
12656 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12658 Support multilibs and devices that see flash in RAM address range.
12661 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12662 (avr_mcu_t) <flash_pm_offset>: New field.
12663 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12664 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12665 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12666 (AVR_TINY_PM_OFFSET): Remove macro.
12667 * config/avr/avr.opt (-mshort-calls): New option.
12668 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12669 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12670 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12671 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12672 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12673 instead of avr_arch->have_jmp_call.
12674 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12675 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12676 avr_arch->flash_pm_offset to define.
12677 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12678 new field flash_pm_offset. Add entry for avrxmega3.
12679 (avr_texinfo): Add entry for avrxmega3.
12680 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12681 attiny212, attiny214,
12682 attiny412, attiny414, attiny416, attiny417,
12683 attiny814, attiny816, attiny817,
12684 attiny1614, attiny1616, attiny1617,
12685 attiny3214, attiny3216, attiny3217.
12686 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12687 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12688 (avr_print_operand_address) [AVR_TINY]: Same.
12689 (avr_asm_init_sections) <readonly_data_section>: Only patch
12690 callback if avr_arch->flash_pm_offset = 0.
12691 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12692 for rodata if avr_arch->flash_pm_offset != 0.
12693 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12694 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12695 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12696 (m_options): Append opt_rcall.
12697 (m_dirnames): Append dir_rcall.
12698 * config/avr/t-multilib: Regenerate.
12700 * configure.ac [target=avr]: Check whether avrxmega3 default
12701 linker description file works as needed.
12702 * configure: Regenerate.
12703 * doc/avr-mmcu.texi: Regenerate.
12704 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12705 <__AVR_ARCH__>: Document avrxmega3 and 103.
12706 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12707 <__AVR_SHORT_CALLS__>: Document it.
12708 <__AVR_PM_BASE_ADDRESS__>: Document it.
12709 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12710 (AVR Variable Attributes) <progmem>: Document this is
12711 not needed for avrxmega3.
12712 (AVR Named Address Spaces) <__flash>: Dito.
12714 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12716 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12718 2017-06-12 Doug Rupp <rupp@adacore.com>
12720 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12721 Append vxworks-stdint.h to the tm_file list.
12722 * config/vxworks-stdint.h: New file.
12724 2017-06-12 Martin Liska <mliska@suse.cz>
12726 PR tree-optimization/81041
12727 * tree-profile.c (gimple_gen_ic_func_profiler):
12728 Create an extra BB in profile-generate
12729 (gimple_gen_time_profiler): Likewise.
12731 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12733 PR tree-optimization/81003
12734 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12735 (update_range_test): Use it instead of force_gimple_operand_gsi.
12737 2017-06-12 Richard Biener <rguenther@suse.de>
12739 PR tree-optimization/81053
12740 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12741 with backedge value not defined in loop. Simplify def stmt
12744 2017-06-11 Tom de Vries <tom@codesourcery.com>
12747 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12749 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12750 nvptx_cannot_force_const_mem.
12752 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12754 * opts.c (finish_options): Move test for flag_split_stack after
12755 it has been initialized.
12757 2017-06-11 Jason Merrill <jason@redhat.com>
12759 * tree.h (id_equal): New.
12760 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12761 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12762 instead of strcmp of IDENTIFIER_POINTER.
12764 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12766 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12767 (mark_all_inlined_calls_cdtor): Fix formating.
12768 (inline_transform): Rescale profile before inlining.
12770 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12772 * cgraph.h (cgraph_edge::clone): Update prototype.
12773 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12774 (cgraph_node::create_clone): Update.
12775 (cgraph_node::create_version_clone): Update.
12776 * tree-inline.c (copy_bb): Update.
12777 (expand_call_inline): Update.
12779 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12781 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12782 factored out from ...
12783 (rs6000_emit_prologue): ... here.
12785 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12787 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12788 factored out from ...
12789 (rs6000_emit_prologue): ... here.
12791 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12793 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12795 (handle_missing_profiles): Fix computation of tp_first_run.
12796 (counts_to_freqs): Do not touch freqs when count is 0.
12798 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12800 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12803 2017-06-10 Tom de Vries <tom@codesourcery.com>
12805 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12806 attributes): Document signal effective target.
12808 2017-06-10 Tom de Vries <tom@codesourcery.com>
12810 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12811 Document effective target stack_size.
12813 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12815 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12816 to the edit_context if they can be auto-applied.
12818 2017-06-9 Ian Lance Taylor <iant@golang.org>
12820 * opts.c (finish_options): If -fsplit-stack, disable implicit
12821 -forder-blocks-and-partition.
12822 * doc/invoke.texi (Optimize Options): Document that when using
12823 -fsplit-stack -forder-blocks-and-partition is not implicitly
12826 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12828 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12829 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12830 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12831 * builtins.def (abort, trap, unreachable): Declare cold.
12832 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12833 * tree-core.h (ECF_COLD): New.
12834 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12835 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12837 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12839 * predict.c (unlikely_executed_stmt_p): Cleanup.
12841 2017-06-09 Richard Biener <rguenther@suse.de>
12843 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12844 model if the ref is always written to.
12846 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12848 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12850 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12852 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12855 2017-06-09 Tom de Vries <tom@codesourcery.com>
12858 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12859 "target cannot support label values" when encountering LABEL_REF.
12861 2017-06-09 Martin Liska <mliska@suse.cz>
12863 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12864 (gimple_gen_ic_func_profiler): Emit direct comparison
12865 of __gcov_indirect_call_callee with NULL.
12866 (gimple_gen_time_profiler): Change probability from
12867 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12869 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12871 * profile.c (edge_gcov_counts): Turn to pointer.
12872 (compute_branch_probabilities, compute_branch_probabilities): Update.
12873 (branch_prob): Do not clear edge_gcov_count.
12874 * profile.h (edge_gcov_counts): Turn to pointer.
12875 (edge_gcov_count): Update.
12877 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12879 * gimple.h (gimple_check_failed): Mark cold.
12881 2017-06-09 Richard Biener <rguenther@suse.de>
12883 PR tree-optimization/66623
12884 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12885 refactor check_reduction into two parts, properly computing
12886 whether we have to check reduction validity for outer loop
12889 2017-06-09 Richard Biener <rguenther@suse.de>
12891 PR tree-optimization/79483
12892 * graphite-scop-detection.c (order): New global.
12893 (get_order): Compute bb to order mapping that satisfies code
12894 generation constraints.
12895 (cmp_pbbs): New helper.
12896 (build_scops): Start domwalk at entry block, sort generated
12899 2017-06-09 Richard Biener <rguenther@suse.de>
12901 PR middle-end/81007
12902 * ipa-polymorphic-call.c
12903 (ipa_polymorphic_call_context::restrict_to_inner_class):
12904 Skip FIELD_DECLs with error_mark_node type.
12905 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12908 2017-06-09 Martin Liska <mliska@suse.cz>
12910 * predict.c (struct branch_predictor): New struct.
12911 (test_prediction_value_range): New test.
12912 (predict_c_tests): New function.
12913 * selftest-run-tests.c (selftest::run_tests): Run the function.
12914 * selftest.h: Declare new tests.
12916 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12919 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12920 gen_add3_insn did not fail.
12921 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12922 r0, construct that number in a temporary reg and add that reg to r0.
12923 If asked to put the result in r0 as well, fail.
12925 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12927 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12928 for early expansion of vec_eqv.
12930 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12932 PR middle-end/81005
12933 * ubsan.c (instrument_null): Avoid pointless code temporary.
12934 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12936 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12940 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12943 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12946 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12948 * predict.c (unlikely_executed_edge_p): Move ahead.
12949 (probably_never_executed_edge_p): Use it.
12951 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12953 PR middle-end/79988
12954 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12955 gimple_call_builtin_p call.
12957 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12959 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12960 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12961 rtl_check_failed_type2, rtl_check_failed_code1,
12962 rtl_check_failed_code2, rtl_check_failed_code_mode,
12963 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12964 rtvec_check_failed_bounds, rtl_check_failed_flag,
12965 _fatal_insn_not_found, _fatal_insn): Likewise.
12966 * tree.h (tree_contains_struct_check_failed,
12967 tree_check_failed, tree_not_check_failed,
12968 tree_class_check_failed, tree_range_check_failed,
12969 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12970 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12971 tree_operand_check_failed, omp_clause_check_failed,
12972 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12975 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12977 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12978 flag_branch_probabilities.
12979 * ipa-inline.c (edge_badness): Likewise.
12980 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12981 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12982 * predict.c (maybe_hot_frequency_p): Likewise.
12983 (probably_never_executed): Likewise.
12984 * sched-ebb.c (schedule_ebbs): Likewise.
12985 * sched-rgn.c (find_single_block_region): Likewise.
12986 * tracer.c (tail_duplicate): Likewise.
12988 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12990 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12991 longer requires x_flag_profile_use.
12993 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12995 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12996 instead of flag_reorder_blocks_and_partition.
12997 * dbxout.c (dbxout_function_end): Likewise.
12998 * dwarf2out.c (gen_subprogram_die): Likewise.
12999 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13000 * hw-doloop.c (reorg_loops): Likewise.
13001 * varasm.c (assemble_start_function,
13002 assemble_end_function): Likewise.
13003 (decide_function_section): Do not check for
13004 flag_reorder_blocks_and_partition.
13006 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13008 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13010 (chkp_get_hard_register_fake_addr_expr): Ditto.
13011 (chkp_build_addr_expr): Add check for hard reg case.
13012 (chkp_parse_array_and_component_ref): Ditto.
13013 (chkp_find_bounds_1): Ditto.
13014 (chkp_process_stmt): Don't generate bounds store for
13017 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13019 * predict.c (maybe_hot_bb_p): Do not check profile status.
13020 (maybe_hot_edge_p): Likewise.
13021 (probably_never_executed): Check for zero counts even if profile
13023 (unlikely_executed_edge_p): New function.
13024 (unlikely_executed_stmt_p): New function.
13025 (unlikely_executed_bb_p): New function.
13026 (set_even_probabilities): Use unlikely predicates.
13027 (combine_predictions_for_bb): Likewise.
13028 (predict_paths_for_bb): Likewise.
13029 (predict_paths_leading_to_edge): Likewise.
13030 (determine_unlikely_bbs): New function.
13031 (estimate_bb_frequencies): Use it.
13032 (compute_function_frequency): Use zero counts even if profile is
13034 * profile-count.h: Fix typo.
13036 2017-08-08 Julia Koval <julia.koval@intel.com>
13038 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13039 _mm512_mask_cvtsepi16_storeu_epi8,
13040 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13041 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13042 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13043 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13044 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13045 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13046 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13047 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13048 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13049 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13050 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13051 __builtin_ia32_pmovuswb256mem_mask,
13052 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13053 __builtin_ia32_pmovwb512mem_mask): New builtins.
13055 2017-08-08 Julia Koval <julia.koval@intel.com>
13057 PR target/73350,80862
13058 * config/i386/subst.md (round): Fix round pattern.
13059 * config/i386/i386.c (ix86_erase_embedded_rounding):
13060 Fix erasing rounding for the fixed pattern.
13062 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13064 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13066 2017-06-08 Martin Liska <mliska@suse.cz>
13068 PR gcov-profile/80911
13069 * gcov.c (block_info::block_info): New constructor.
13071 2017-06-07 Carl Love <cel@us.ibm.com>
13073 * config/rs6000/rs6000-c: The return type of the following
13074 built-in functions was implemented as int not long long. Fix sign
13075 of return value for the unsigned version of vec_mulo and vec_mule.
13076 vector unsigned long long vec_bperm (vector unsigned long long,
13077 vector unsigned char)
13078 vector signed long long vec_mule (vector signed int,
13080 vector unsigned long long vec_mule (vector unsigned int,
13081 vector unsigned int)
13082 vector signed long long vec_mulo (vector signed int,
13084 vector unsigned long long vec_mulo (vector unsigned int,
13085 vector unsigned int)
13086 * doc/extend.texi: Fix the documentation for the built-in
13089 2017-06-07 Carl Love <cel@us.ibm.com>
13092 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13095 2017-06-07 Carl Love <cel@us.ibm.com>
13097 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13098 support, Generate doublehv for signed int/float for BE case only.
13100 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13102 * doc/invoke.texi (mcx16): Rewrite.
13104 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13106 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13107 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13108 *mov<mode>_softfloat, and an anonymous splitter): Use
13109 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13111 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13113 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13115 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13116 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13117 (rs6000_debug_reg_global): Adjust.
13118 (rs6000_init_hard_regno_mode_ok): Adjust.
13119 (rs6000_dbx_register_number): Adjust.
13120 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13121 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13122 Remove SPE_ACC and SPEFSCR.
13123 (REG_ALLOC_ORDER): Ditto.
13124 (FRAME_POINTER_REGNUM): Change to 111.
13125 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13126 (REG_CLASS_NAMES): Ditto.
13127 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13128 (REGISTER_NAMES): Ditto.
13129 (ADDITIONAL_REG_NAMES): Ditto.
13130 (rs6000_reg_names): Ditto.
13131 * config/rs6000/rs6000.md: Renumber some register number
13134 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13136 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13138 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13139 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13141 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13142 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13143 Delete the SPE high registers.
13144 (REG_ALLOC_ORDER): Ditto.
13145 (enum reg_class): Remove SPE_HIGH_REGS.
13146 (REG_CLASS_NAMES): Ditto.
13147 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13148 (REGISTER_NAMES): Ditto.
13149 (rs6000_reg_names): Ditto.
13150 * doc/tm.texi.in: Remove SPE as example.
13151 * doc/tm.texi: Regenerate.
13153 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13155 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13156 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13157 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13158 * config/rs6000/rs6000.md (type): Remove "brinc".
13160 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13162 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13163 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13164 * config/rs6000/linuxspe.h: Delete file.
13165 * config/rs6000/rs6000.md: Don't include spe.md.
13166 * config/rs6000/spe.h: Delete file.
13167 * config/rs6000/spe.md: Delete file.
13168 * config/rs6000/t-rs6000: Remove spe.md.
13170 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13172 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13173 (reg_or_none500mem_operand): Delete.
13174 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13175 instead of reg_or_none500mem_operand.
13177 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13179 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13180 handling of SPE flags.
13181 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13185 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13187 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13188 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13189 paired_divv2sf3): Similar.
13190 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13191 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13192 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13194 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13195 Rename the paired_* instruction patterns.
13196 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13198 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13199 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13200 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13201 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13202 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13203 PAIRED_VECTOR_MODE.
13204 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13205 (spe_func_has_64bit_regs_p): Delete.
13206 (spe_expand_predicate_builtin): Delete.
13207 (spe_expand_evsel_builtin): Delete.
13208 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13209 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13210 (invalid_e500_subreg): Delete.
13211 (rs6000_legitimize_address): Always force_reg op2 as well, for
13212 paired single memory accesses.
13213 (rs6000_member_type_forces_blk): Delete.
13214 (rs6000_spe_function_arg): Delete.
13215 (rs6000_expand_unop_builtin): Delete SPE handling.
13216 (rs6000_expand_binop_builtin): Ditto.
13217 (spe_expand_stv_builtin): Delete.
13218 (bdesc_2arg_spe): Delete.
13219 (spe_expand_builtin): Delete.
13220 (spe_expand_predicate_builtin): Delete.
13221 (spe_expand_evsel_builtin): Delete.
13222 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13223 (spe_init_builtins): Delete.
13224 (spe_func_has_64bit_regs_p): Delete.
13225 (savres_routine_name): Delete "info" parameter. Adjust callers.
13226 (rs6000_emit_stack_reset): Ditto.
13227 (rs6000_dwarf_register_span): Delete.
13228 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13229 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13230 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13232 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13234 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13235 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13236 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13237 mulv2sf3, divv2sf3): Delete expanders.
13239 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13241 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13243 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13245 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13246 * config/rs6000/rs6000.c: Ditto.
13248 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13250 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13251 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13252 comparison_operator.
13254 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13256 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13257 * config/rs6000/rs6000.opt: Ditto.
13258 * config/rs6000/t-rtems: Ditto.
13260 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13262 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13263 TARGET_E500_SINGLE by 0, simplify.
13264 * config/rs6000/rs6000.c: Ditto.
13265 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13266 (spe_build_register_parallel): Delete.
13267 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13268 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13269 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13270 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13271 (E500_CONVERT): Delete.
13272 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13274 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13276 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13277 * config/rs6000/dfp.md: Ditto.
13278 (negdd2, *negdd2_fpr): Merge.
13279 (absdd2, *absdd2_fpr): Merge.
13280 (negtd2, *negtd2_fpr): Merge.
13281 (abstd2, *abstd2_fpr): Merge.
13282 * config/rs6000/e500.h: Delete file.
13283 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13284 TARGET_FPRS by 1 and simplify.
13285 * config/rs6000/rs6000-c.c: Ditto.
13286 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13287 TARGET_DF_SPE by 0.
13288 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13290 * config/rs6000/rs6000.md: Ditto.
13291 (floatdidf2, *floatdidf2_fpr): Merge.
13292 (move_from_CR_gt_bit): Delete.
13293 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13294 (E500_CR_IOR_COMPARE): Delete.
13295 (All patterns that require !TARGET_FPRS): Delete.
13296 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13298 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13300 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13302 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13304 * graphds.c (add_edge): Intitialize edge's attached data.
13305 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13306 pointer parameter. Call pointed function on each edge during
13307 graph traversing. Skip traversing the edge when the function
13309 (graphds_dfs, graphds_scc): Ditto.
13310 (for_each_edge): New parameter. Pass the new parameter to callback
13312 * graphds.h (skip_edge_callback): New function pointer type.
13313 (graphds_dfs, graphds_scc): New function pointer parameter.
13314 (graphds_edge_callback, for_each_edge): New parameter.
13316 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13318 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13319 out code checking if runtime alias check is possible to below ...
13320 Call the new function.
13321 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13322 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13324 2017-06-07 Marek Polacek <polacek@redhat.com>
13327 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13328 TYPE_OVERFLOW_WRAPS check.
13330 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13332 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13333 if versioning is required.
13334 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13335 peeling with the check for versioning.
13337 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13339 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13340 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13341 Set true to new parameter if new ssa variable is defined.
13342 (vect_gen_vector_loop_niters): Refactor. Set range information
13343 for the new vector loop bound variable.
13344 (vect_do_peeling): Ditto.
13346 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13348 * tree-affine.c (ssa.h): Include header file.
13349 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13350 has wrapping overflow behavior.
13352 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13354 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13356 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13358 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13359 (tree_to_aff_combination): ... here.
13361 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13363 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13364 reg_pressure model function.
13365 (ivopts_global_cost_for_size): Delete.
13366 (determine_set_costs, iv_ca_recount_cost): Call new model function
13367 ivopts_estimate_reg_pressure.
13369 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13371 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13372 expensive than udiv. Remove floating point cases from mod.
13374 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13376 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13377 Increase idiv cost.
13379 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13381 * config/aarch64/aarch64.md
13382 (copysignsf3): Fix mask generation.
13384 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13386 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13388 (class dump_manager): Add register_dumps method.
13389 * dumpfile.c: Include langhooks.h.
13390 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13391 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13392 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13393 (dump_manager::dump_register): Start with 512 entries instead of 32.
13394 (dump_manager::register_dumps): New method.
13395 * toplev.c (general_init): Instead of invoking register_dumps
13396 langhook, invoke register_dumps method on the dump manager.
13397 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13400 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13402 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13403 Say that # requires an associated define_split to exist, and that
13404 the define_split must be suitable for use after register allocation.
13406 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13408 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13409 (compute_outgoing_frequencies): Also initialize zero counts.
13410 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13411 around loops; preserve more of profile when nothing changes.
13413 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13415 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13417 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13418 * config/arm/arm-cpu-cdata.h: Regenerate.
13419 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13420 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13421 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13422 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13423 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13425 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13427 * config/arm/t-rmprofile: Likewise.
13428 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13430 2017-06-06 David S. Miller <davem@davemloft.net>
13433 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13434 blockage if function uses alloca.
13436 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13438 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13439 New "uid" fields to hold pretty-print IDs of group and ref.
13440 Memory references are now identified as <group_id>:<ref_id>
13441 instead of using [random] addresses.
13442 (dump_mem_details): Simplify, no functional change.
13443 (dump_mem_ref): Simplify and make output more concise.
13444 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13445 (find_or_create_group): Initialize group uid.
13446 (record_ref): Initialize ref uid. Improve debug output.
13447 (prune_group_by_reuse, should_issue_prefetch_p,)
13448 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13449 (mark_nontemporal_store, determine_loop_nest_reuse):
13450 Improve debug output.
13452 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13454 * dbgcnt.def (prefetch): New debug counter.
13455 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13456 (schedule_prefetches): Stop issueing prefetches if debug counter
13459 2017-06-06 Tom de Vries <tom@codesourcery.com>
13461 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13462 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13464 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13466 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13467 Use aarch64_reg_or_zero predicate for operand 4.
13468 (aarch64_compare_and_swap<mode> define_insn_and_split):
13469 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13470 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13472 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13474 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13475 (arm_compute_save_core_reg_mask): This.
13476 (thumb1_compute_save_reg_mask): Rename into ...
13477 (thumb1_compute_save_core_reg_mask): This.
13478 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13479 (arm_compute_frame_layout): Likewise.
13481 2017-06-06 Richard Biener <rguenther@suse.de>
13483 PR tree-optimization/80974
13484 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13485 keep or clear leaders SSA info.
13487 2017-06-06 Tom de Vries <tom@codesourcery.com>
13489 * config/nvptx/nvptx.c (split_mode_p): New function.
13490 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13492 2017-06-06 Tom de Vries <tom@codesourcery.com>
13494 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13496 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13499 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13502 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13504 * shrink-wrap.c (handle_simple_exit): Update profile.
13505 (try_shrink_wrapping): Upate profile.
13507 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13509 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13510 (tree_guess_outgoing_edge_probabilities): New.
13511 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13512 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13514 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13516 * ipa-split.c (split_function): Initialize return bb profile.
13518 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13520 * profile.c (compute_branch_probabilities): Also initialize
13521 EXIT_BLOCK profile.
13523 2017-06-06 Richard Biener <rguenther@suse.de>
13525 PR tree-optimization/80928
13526 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13527 (vect_analyze_loop_operations): Properly guard analysis for
13529 (vect_transform_loop): Likewise.
13530 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13531 (vect_model_induction_cost): Do not cost for pure SLP.
13532 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13533 of induction in inner loop vectorization.
13534 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13535 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13536 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13538 (vect_analyze_slp_cost_1): Cost induction.
13539 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13540 (vect_get_slp_vect_defs): Likewise.
13541 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13542 (vect_transform_stmt): Handle SLP reductions.
13543 * tree-vectorizer.h (vectorizable_induction): Adjust.
13545 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13547 * config/rs6000/rs6000.c (make_resolver_func): Update
13548 init_lowered_empty_function call.
13550 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13552 * doc/invoke.texi: Document the -fprofile-abs-path option.
13553 * common.opt (fprofile-abs-path): New option.
13554 * gcov-io.h (gcov_write_filename): Declare.
13555 * gcov-io.c (gcov_write_filename): New function.
13556 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13557 * profile.c (output_location): Likewise.
13559 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13561 * shring-wrap.c: Revert accidental commit.
13563 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13565 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13567 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13569 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13571 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13572 profile in callgraph edge.
13573 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13574 (apply_scale): Likewise.
13575 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13576 Also scale profile when inlining function with zero profile.
13577 (initialize_cfun): Update exit block profile even when it is zero.
13578 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13579 when profile is read.
13581 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13583 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13584 (CLONE_*): New constants to define the processors we can generate
13585 code for with the target_clone attribute.
13586 (rs6000_clone_map): New array to identify which clone processors
13587 the current program is running on.
13588 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13589 target_clone attribute.
13590 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13591 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13592 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13593 (cpu_expand_builtin): Add support for target_clone attribute.
13594 (rs6000_valid_attribute_p): Allow "default" attribute.
13595 (get_decl_name): New debug function to simplify printing the
13596 current function name in debugging statements.
13597 (rs6000_clone_priority): New functions to support the target_clone
13598 attribute, and be able to generate code to switch between ISA 2.05
13599 through ISA 3.0 (power6 through power9).
13600 (rs6000_compare_version_priority): Likewise.
13601 (rs6000_get_function_versions_dispatcher): Likewise.
13602 (make_resolver_func): Likewise.
13603 (add_condition_to_bb): Likewise.
13604 (dispatch_function_versions): Likewise.
13605 (rs6000_generate_version_dispatcher_body): Likewise.
13606 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13607 (fusion_gpr_load_p): Fix a spacing issue.
13608 * doc/extend.texi (Common Function Attributes): Document that the
13609 PowerPC supports the target_clone attribute.
13611 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13613 * config/arm/arm.h: explain F symbol found in description of ARM
13614 register allocation in its legend.
13616 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13618 * config/mips/frame-header-opt.c: Include profile-count.h.
13619 * config/riscv/riscv.c: Include profile-count.h
13621 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13623 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13625 (sm_set_flag_if_changed): Add bbs field.
13626 (execute_sm_if_changed_flag_set): Pass BBS.
13627 (execute_sm): Update.
13629 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13631 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13634 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13636 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13637 (peephole2): New peephole2 to emit the above.
13638 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13640 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13642 * config/aarch64/aarch64.c (define_peephole2 above
13643 *sub_<shift>_<mode>): New peephole.
13645 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13647 * config/i386/i386.c (make_resolver_func): Update.
13648 * Makefile.in: Add profile-count.h and profile-count.o
13649 * auto-profile.c (afdo_indirect_call): Update to new API.
13650 (afdo_set_bb_count): Update.
13651 (afdo_propagate_edge): Update.
13652 (afdo_propagate_circuit): Update.
13653 (afdo_calculate_branch_prob): Update.
13654 (afdo_annotate_cfg): Update.
13655 * basic-block.h: Include profile-count.h
13656 (struct edge_def): Turn count to profile_count.
13657 (struct basic_block_def): Likewie.
13658 (REG_BR_PROB_BASE): Move to profile-count.h
13659 (RDIV): Move to profile-count.h
13660 * bb-reorder.c (max_entry_count): Turn to profile_count.
13661 (find_traces): Update.
13662 (rotate_loop):Update.
13663 (connect_traces):Update.
13664 (sanitize_hot_paths):Update.
13665 * bt-load.c (migrate_btr_defs): Update.
13666 * cfg.c (RDIV): Remove.
13667 (init_flow): Use alloc_block.
13668 (alloc_block): Uninitialize count.
13669 (unchecked_make_edge): Uninitialize count.
13670 (check_bb_profile): Update.
13671 (dump_edge_info): Update.
13672 (dump_bb_info): Update.
13673 (update_bb_profile_for_threading): Update.
13674 (scale_bbs_frequencies_int): Update.
13675 (scale_bbs_frequencies_gcov_type): Update.
13676 (scale_bbs_frequencies_profile_count): New.
13677 * cfg.h (update_bb_profile_for_threading): Update.
13678 (scale_bbs_frequencies_profile_count): Declare.
13679 * cfgbuild.c (compute_outgoing_frequencies): Update.
13680 (find_many_sub_basic_blocks): Update.
13681 * cfgcleanup.c (try_forward_edges): Update.
13682 (try_crossjump_to_edge): Update.
13683 * cfgexpand.c (expand_gimple_tailcall): Update.
13684 (construct_exit_block): Update.
13685 * cfghooks.c (verify_flow_info): Update.
13686 (dump_bb_for_graph): Update.
13687 (split_edge): Update.
13688 (make_forwarder_block): Update.
13689 (duplicate_block): Update.
13690 (account_profile_record): Update.
13691 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13692 (get_estimated_loop_iterations): Update.
13693 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13694 (single_likely_exit): Update.
13695 * cfgloopmanip.c (scale_loop_profile): Update.
13697 (set_zero_probability): Update.
13698 (lv_adjust_loop_entry_edge): Update.
13699 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13700 (purge_dead_edges): Update.
13701 (rtl_account_profile_record): Update.
13702 * cgraph.c (cgraph_node::create): Uninitialize count.
13703 (symbol_table::create_edge): Uninitialize count.
13704 (cgraph_update_edges_for_call_stmt_node): Update.
13705 (cgraph_edge::dump_edge_flags): Update.
13706 (cgraph_node::dump): Update.
13707 (cgraph_edge::maybe_hot_p): Update.
13708 * cgraph.h: Include profile-count.h
13709 (create_clone), create_edge, create_indirect_edge): Update.
13710 (cgraph_node): Turn count to profile_count.
13711 (cgraph_edge0: Likewise.
13712 (make_speculative, clone): Update.
13713 (create_edge): Update.
13714 (init_lowered_empty_function): Update.
13715 * cgraphclones.c (cgraph_edge::clone): Update.
13716 (duplicate_thunk_for_node): Update.
13717 (cgraph_node::create_clone): Update.
13718 * cgraphunit.c (cgraph_node::analyze): Update.
13719 (cgraph_node::expand_thunk): Update.
13720 * final.c (dump_basic_block_info): Update.
13721 * gimple-streamer-in.c (input_bb): Update.
13722 * gimple-streamer-out.c (output_bb): Update.
13723 * graphite.c (print_global_statistics): Update.
13724 (print_graphite_scop_statistics): Update.
13725 * hsa-brig.c: Include basic-block.h.
13726 * hsa-dump.c: Include basic-block.h.
13727 * hsa-gen.c (T sum_slice): Update.
13728 (convert_switch_statements):Update.
13729 * hsa-regalloc.c: Include basic-block.h.
13730 * ipa-chkp.c (chkp_produce_thunks): Update.
13731 * ipa-cp.c (struct caller_statistics): Update.
13732 (init_caller_stats): Update.
13733 (gather_caller_stats): Update.
13734 (ipcp_cloning_candidate_p): Update.
13735 (good_cloning_opportunity_p): Update.
13736 (get_info_about_necessary_edges): Update.
13737 (dump_profile_updates): Update.
13738 (update_profiling_info): Update.
13739 (update_specialized_profile): Update.
13740 (perhaps_add_new_callers): Update.
13741 (decide_about_value): Update.
13742 (ipa_cp_c_finalize): Update.
13743 * ipa-devirt.c (struct odr_type_warn_count): Update.
13744 (struct decl_warn_count): Update.
13745 (struct final_warning_record): Update.
13746 (possible_polymorphic_call_targets): Update.
13747 (ipa_devirt): Update.
13748 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13749 * ipa-icf.c (sem_function::merge): Update.
13750 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13751 * ipa-inline.c (compute_uninlined_call_time): Update.
13752 (compute_inlined_call_time): Update.
13753 (want_inline_small_function_p): Update.
13754 (want_inline_self_recursive_call_p): Update.
13755 (edge_badness): Update.
13756 (lookup_recursive_calls): Update.
13757 (recursive_inlining): Update.
13758 (inline_small_functions): Update.
13759 (dump_overall_stats): Update.
13760 (dump_inline_stats): Update.
13761 * ipa-profile.c (ipa_profile_generate_summary): Update.
13762 (ipa_propagate_frequency): Update.
13763 (ipa_profile): Update.
13764 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13765 * ipa-utils.c (ipa_merge_profiles): Update.
13766 * loop-doloop.c (doloop_modify): Update.
13767 * loop-unroll.c (report_unroll): Update.
13768 (unroll_loop_runtime_iterations): Update.
13769 * lto-cgraph.c (lto_output_edge): Update.
13770 (lto_output_node): Update.
13771 (input_node): Update.
13772 (input_edge): Update.
13773 (merge_profile_summaries): Update.
13774 * lto-streamer-in.c (input_cfg): Update.
13775 * lto-streamer-out.c (output_cfg): Update.
13776 * mcf.c (create_fixup_graph): Update.
13777 (adjust_cfg_counts): Update.
13778 (sum_edge_counts): Update.
13779 * modulo-sched.c (sms_schedule): Update.
13780 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13781 * predict.c (maybe_hot_count_p): Update.
13782 (probably_never_executed): Update.
13783 (dump_prediction): Update.
13784 (combine_predictions_for_bb): Update.
13785 (propagate_freq): Update.
13786 (handle_missing_profiles): Update.
13787 (counts_to_freqs): Update.
13788 (rebuild_frequencies): Update.
13789 (force_edge_cold): Update.
13790 * predict.h: Include profile-count.h
13791 (maybe_hot_count_p, counts_to_freqs): UPdate.
13792 * print-rtl-function.c: Do not include cfg.h
13793 * print-rtl.c: Include basic-block.h
13794 * profile-count.c: New file.
13795 * profile-count.h: New file.
13796 * profile.c (is_edge_inconsistent): Update.
13797 (correct_negative_edge_counts): Update.
13798 (is_inconsistent): Update.
13799 (set_bb_counts): Update.
13800 (read_profile_edge_counts): Update.
13801 (compute_frequency_overlap): Update.
13802 (compute_branch_probabilities): Update; Initialize and deinitialize
13804 (branch_prob): Update.
13805 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13806 (edge_gcov_count): New.
13807 (bb_gcov_count): New.
13808 * shrink-wrap.c (try_shrink_wrapping): Update.
13809 * tracer.c (better_p): Update.
13810 * trans-mem.c (expand_transaction): Update.
13811 (ipa_tm_insert_irr_call): Update.
13812 (ipa_tm_insert_gettmclone_call): Update.
13813 * tree-call-cdce.c: Update.
13814 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13815 (gimple_duplicate_sese_tail): Update.
13816 (gimple_account_profile_record): Update.
13817 (execute_fixup_cfg): Update.
13818 * tree-inline.c (copy_bb): Update.
13819 (copy_edges_for_bb): Update.
13820 (initialize_cfun): Update.
13821 (freqs_to_counts): Update.
13822 (copy_cfg_body): Update.
13823 (expand_call_inline): Update.
13824 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13825 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13826 (try_unroll_loop_completely): Update.
13827 (try_peel_loop): Update.
13828 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13829 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13830 * tree-ssa-loop-split.c (connect_loops): Update.
13831 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13832 * tree-ssa-reassoc.c (branch_fixup): Update.
13833 * tree-ssa-tail-merge.c (replace_block_by): Update.
13834 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13835 (compute_path_counts): Update.
13836 (update_profile): Update.
13837 (recompute_probabilities): Update.
13838 (update_joiner_offpath_counts): Update.
13839 (estimated_freqs_path): Update.
13840 (freqs_to_counts_path): Update.
13841 (clear_counts_path): Update.
13842 (ssa_fix_duplicate_block_edges): Update.
13843 (duplicate_thread_path): Update.
13844 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13845 (struct switch_conv_info): Update.
13846 * tree-tailcall.c (decrease_profile): Update.
13847 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13848 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13849 * value-prof.c (check_counter): Update.
13850 (gimple_divmod_fixed_value): Update.
13851 (gimple_mod_pow2): Update.
13852 (gimple_mod_subtract): Update.
13853 (gimple_ic_transform): Update.
13854 (gimple_stringop_fixed_value): Update.
13855 * value-prof.h (gimple_ic): Update.
13857 2017-06-02 Carl Love <cel@us.ibm.com>
13859 * config/rs6000/rs6000-c: Add support for built-in functions
13860 vector double vec_doublee (vector signed int);
13861 vector double vec_doublee (vector unsigned int);
13862 vector double vec_doublee (vector float);
13863 vector double vec_doubleh (vector signed int);
13864 vector double vec_doubleh (vector unsigned int);
13865 vector double vec_doubleh (vector float);
13866 vector double vec_doublel (vector signed int);
13867 vector double vec_doublel (vector unsigned int);
13868 vector double vec_doublel (vector float);
13869 vector double vec_doubleo (vector signed int);
13870 vector double vec_doubleo (vector unsigned int);
13871 vector double vec_doubleo (vector float);.
13872 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13873 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13875 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13876 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13877 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13879 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13880 vec_doublel, vec_doubleh.
13881 * doc/extend.texi: Update the built-in documentation file for the
13882 new built-in functions.
13884 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13887 * ipa-inline-analysis.c (free_growth_caches): Set
13888 edge_removal_hook_holder to NULL after removing it.
13890 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13892 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13893 comparision with zero.
13895 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13896 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13897 for early expansion of vec_min and vec_max builtins.
13898 (builtin_function_type): Add min/max unsigned variants to those
13899 identified as having unsigned arguments.
13901 2017-06-02 Olivier Hainque <hainque@adacore.com>
13903 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13905 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13907 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13908 Use VALL_F16 iterator rather than VALL.
13910 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13912 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13913 Emit CBNZ inside loop when doing a strong exchange and comparing
13914 against zero. Generate the CC flags after the loop.
13916 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13918 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13919 (dl_section_ref): New.
13920 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13921 On AIX, append an expression to subtract the size of the
13922 section length to dl_section_ref.
13924 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13926 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13927 for early expansion of vector absolute builtins.
13929 2017-06-02 Richard Biener <rguenther@suse.de>
13931 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13932 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13934 2017-06-02 Richard Biener <rguenther@suse.de>
13936 PR tree-optimization/80948
13937 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13938 stmt order as well.
13940 2017-06-02 Richard Biener <rguenther@suse.de>
13942 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13944 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13945 for inductions as relevant.
13947 2017-06-02 Richard Biener <rguenther@suse.de>
13949 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13950 (vectorizable_induction): ... this. Remove dead code.
13952 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13954 * builtins. (expand_builtin_alloca): Remove second parameter and
13955 infer its value from the first parameter instead.
13956 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13958 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13960 PR rtl-optimization/80903
13961 * loop-doloop.c (add_test): Unshare sequence.
13963 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13965 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13967 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13969 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13971 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13972 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13973 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13974 (xlogue_layout::get_stub_rtx): Make static.
13975 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13976 (xlogue_layout::compute_stub_managed_regs): Rename to...
13977 (xlogue_layout::count_stub_managed_regs): ...this.
13978 (xlogue_layout::is_stub_managed_reg): New function.
13979 (xlogue_layout::m_stub_names): Rename to...
13980 (xlogue_layout::s_stub_names): ...this, make static.
13981 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13982 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13983 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13984 xlogue_layout::s_stub_names): Instantiate statics.
13985 (stub_managed_regs): Remove.
13986 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13987 (disable_call_ms2sysv_xlogues): Rename to...
13988 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13989 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13991 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13992 change after reload_completed.
13993 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13995 (ix86_expand_prologue): Likewise.
13996 (ix86_expand_epilogue): Likewise.
13997 (ix86_expand_split_stack_prologue): Likewise.
13998 (ix86_compute_frame_layout): Remove frame parameter ...
13999 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14000 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14002 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14003 (ix86_frame): Move from here ...
14004 * config/i386/i386.h (ix86_frame): ... to here.
14005 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14006 complete ix86_frame data structure instead. Remove some_ld_name.
14008 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14010 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14011 symbols that hold a DECL_VALUE_EXPR.
14013 2017-06-01 Martin Jambor <mjambor@suse.cz>
14015 PR tree-optimization/80898
14016 * tree-sra.c (process_subtree_disqualification): Removed.
14017 (disqualify_candidate): Do not acll
14018 process_subtree_disqualification.
14019 (subtree_mark_written_and_enqueue): New function.
14020 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14021 RHS has been disqualified and re-queue LHS if necessary. Apart
14022 from that, ignore disqualified RHS.
14024 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14026 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14027 address fetch for z10 or later.
14029 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14031 * config/arc/arc.md (tst_movb): Add guard when splitting.
14033 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14035 * config/arc/arc.c (arc_can_eliminate): Test against
14036 arc_frame_pointer_needed.
14038 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14040 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14041 to prevent store reordering.
14042 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14043 (type): Add block type.
14044 (stack_tie): Define special instruction to be used in
14047 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14049 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14050 constraint. It is not valid for the pattern.
14051 (noncommutative_binary_comparison): Likewise.
14053 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14055 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14058 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14060 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14061 be used by the reg-alloc.
14063 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14065 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14066 reg-alloc when having mul64 or mul32x16 instructions.
14067 (mulsidi3): Likewise.
14068 (umulsidi3): Likewise.
14069 (mulsi32x16): New pattern.
14070 (mulsi64): Likewise.
14071 (mulsidi64): Likewise.
14072 (umulsidi64): Likewise.
14073 (MUL32x16_REG): Define.
14074 (mul64_600): Use MUL32x16_REG.
14075 (mac64_600): Likewise.
14076 (umul64_600): Likewise.
14077 (umac64_600): Likewise.
14079 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14081 * config/arc/arc.md (mulsi3_700): Make it commutative.
14083 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14085 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14087 (*sign_extendsidi2_insn): Likewise for movstosw.
14089 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14091 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14092 the type of the input discriminant value. Convert the
14093 discriminant value of signedness vary.
14095 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14097 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14098 Add to -Wall section.
14100 2017-06-01 Richard Biener <rguenther@suse.de>
14102 PR middle-end/66313
14103 * fold-const.c (fold_plusminus_mult_expr): If the factored
14104 factor may be zero use a wrapping type for the inner operation.
14105 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14106 and handle moved defs.
14107 (process_assignment): Properly guard the unary op case. Return a
14108 tri-state indicating that moving the stmt before the call may allow
14109 to continue. Pass through to_move.
14110 (find_tail_calls): Handle moving unrelated defs before
14113 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14116 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14117 splitter result in the canonical way.
14119 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14121 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14122 also for 32bit target. Update insn attributes.
14123 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14125 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14127 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14128 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14129 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14130 (_mm_maskz_min_ss): New intrinsics.
14132 2017-05-31 Martin Liska <mliska@suse.cz>
14134 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14135 Change comment style to one we normally use.
14136 (vectorizable_reduction): Likewise.
14137 (vectorizable_induction): Likewise.
14138 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14139 (vectorizable_call): Likewise.
14140 (vectorizable_simd_clone_call): Likewise.
14141 (vectorizable_conversion): Likewise.
14142 (vectorizable_assignment): Likewise.
14143 (vectorizable_shift): Likewise.
14144 (vectorizable_operation): Likewise.
14145 (vectorizable_store): Likewise.
14146 (vectorizable_load): Likewise.
14147 * tree-vectorizer.h: Likewise.
14149 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14151 * passes.c (emergency_dump_function): New.
14152 * tree-pass.h (emergency_dump_function): Declare.
14153 * plugin.c (plugins_internal_error_function): Remove.
14154 * plugin.h (plugins_internal_error_function): Remove declaration.
14155 * toplev.c (internal_error_function): New static function. Use it...
14156 (general_init): ...here.
14158 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14160 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14161 (arc_rtx_costs): Add costs for new patterns.
14162 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14163 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14165 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14167 * tree-ssa-strlen.c (get_next_strinfo): New function.
14168 (get_stridx_plus_constant): Use it.
14169 (zero_length_string): Likewise.
14170 (adjust_related_strinfos): Likewise.
14171 (adjust_last_stmt): Likewise.
14173 2017-05-31 Richard Biener <rguenther@suse.de>
14176 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14177 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14179 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14181 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14182 loop_vinfo argument and use of dependence distance vectors.
14183 Check instead whether the two references differ only in their
14184 initial value and assume that they have the same alignment if the
14185 difference is a multiple of the vector alignment.
14186 (vect_analyze_data_refs_alignment): Update call accordingly.
14188 2017-05-31 Martin Liska <mliska@suse.cz>
14191 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14193 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14195 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14196 (create_intersect_range_checks): Move from ...
14197 * tree-data-ref.c (create_intersect_range_checks_index)
14198 (create_intersect_range_checks): ... to here.
14199 (create_runtime_alias_checks): New function factored from ...
14200 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14201 here. Call above function.
14202 * tree-data-ref.h (create_runtime_alias_checks): New function.
14204 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14206 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14207 segment length for dr_b and compute it in wide_int.
14209 2017-05-31 Richard Biener <rguenther@suse.de>
14211 PR tree-optimization/80906
14212 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14213 and pass through iv_map.
14214 (copy_bb_and_scalar_dependences): Adjust.
14215 (translate_pending_phi_nodes): Likewise.
14216 (copy_loop_close_phi_args): Handle code-generating IVs instead
14219 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14221 * diagnostic-color.c (color_dict): Add "type-diff".
14222 (parse_gcc_colors): Update comment.
14223 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14224 -fdiagnostics-show-template-tree and -fno-elide-type.
14225 (GCC_COLORS): Add type-diff to example.
14227 (-fdiagnostics-show-template-tree): New.
14228 (-fno-elide-type): New.
14229 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14230 the pp_format_decoder callback. Call any m_format_postprocessor's
14232 (pretty_printer::pretty_printer): Initialize
14233 m_format_postprocessor.
14234 (pretty_printer::~pretty_printer): Delete any
14235 m_format_postprocessor.
14236 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14237 (class format_postprocessor): New class.
14238 (struct pretty_printer::format_decoder): Document the new parameters.
14239 (struct pretty_printer::m_format_postprocessor): New field.
14240 * tree-diagnostic.c (default_tree_printer): Update for new
14241 bool and const char ** params.
14242 * tree-diagnostic.h (default_tree_printer): Likewise.
14244 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14246 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14247 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14248 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14249 rs6000_gen_cell_microcode code.
14250 (rs6000_final_prescan_insn): Delete.
14251 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14252 "warn-cell-microcode" entries.
14253 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14254 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14255 throughout. Change cc_reg_not_micro_cr0_operand to
14256 cc_reg_not_cr0_operand throughout.
14257 (*extendhi<mode>2_noload): Delete.
14258 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14259 (mwarn-cell-microcode): Delete.
14260 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14261 -mgen-cell-microcode and -mwarn-cell-microcode.
14263 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14266 * config/i386/constraints.md (Yd): New constraint.
14268 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14269 and (?Yd, r) alternatives. Update insn attributes.
14270 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14271 and (?*Yd, r) alternatives. Update insn attributes.
14272 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14274 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14276 * gimplify.c (gimplify_modify_expr): Don't create a
14277 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14280 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14282 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14284 2017-05-30 Richard Biener <rguenther@suse.de>
14286 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14287 and reduc_def fields.
14288 (STMT_VINFO_REDUC_TYPE): New define.
14289 (STMT_VINFO_REDUC_DEF): Likewise.
14290 (vect_force_simple_reduction): Adjust prototype.
14291 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14292 (vect_is_simple_reduction): Remove check_reduction argument.
14293 (vect_force_simple_reduction): Adjust and set
14294 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14295 (vectorizable_reduction): Do not re-do reduction analysis
14296 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14297 * tree-parloops.c (gather_scalar_reductions): Adjust.
14299 2017-05-30 Richard Biener <rguenther@suse.de>
14301 PR middle-end/80901
14302 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14305 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14307 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14308 Introduce unknown_misalignment parameter and remove vf.
14309 (vect_peeling_hash_get_lowest_cost):
14310 Pass unknown_misalignment parameter.
14311 (vect_enhance_data_refs_alignment):
14312 Fix unsupportable data ref treatment.
14314 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14316 * tree-vect-data-refs.c (vect_get_data_access_cost):
14317 Workaround for SLP handling.
14318 (vect_enhance_data_refs_alignment):
14319 Compute costs for doing no peeling at all, compare to the best
14320 peeling costs so far and avoid peeling if cheaper.
14322 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14324 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14325 Return peeling info and set costs to zero for unlimited cost
14327 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14328 with unknown misalignment. Compute and costs for unknown
14329 misalignment, compare them to the costs for known misalignment
14330 and choose the cheapest for peeling.
14332 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14334 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14335 (vect_get_peeling_costs_all_drs): Create function.
14336 (vect_peeling_hash_get_lowest_cost):
14337 Use vect_get_peeling_costs_all_drs.
14338 (vect_peeling_supportable): Create function.
14339 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14341 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14343 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14344 DR_HAS_NEGATIVE_STEP.
14345 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14346 (vect_enhance_data_refs_alignment): Use.
14347 (vect_duplicate_ssa_name_ptr_info): Use.
14348 * tree-vectorizer.h (dr_misalignment): Use.
14349 (known_alignment_for_access_p): Use.
14351 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14354 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14356 (has_section_name): New function.
14358 2017-05-30 Martin Liska <mliska@suse.cz>
14361 * auto-profile.c (get_function_decl_from_block): Fix
14364 2017-05-30 Richard Biener <rguenther@suse.de>
14366 PR middle-end/80876
14367 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14369 2017-05-30 Martin Liska <mliska@suse.cz>
14371 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14372 * dumpfile.h (struct dump_file_info): Remove ctors.
14374 2017-05-30 Martin Liska <mliska@suse.cz>
14376 * predict.def: Fix GNU coding style.
14378 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14380 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14381 Mark 'to' argument with ATTRIBUTE_UNUSED.
14383 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14385 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14386 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14387 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14390 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14392 * doc/install.texi (Options specification): Restore entry of
14393 --enable-sjlj-exceptions.
14395 2017-05-27 Michael Eager <eager@eagercon.com>
14398 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14400 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14402 * config/microblaze/microblaze.h
14403 (FIXED_REGISTERS): Update in macro.
14404 (CALL_USED_REGISTERS): Update in macro.
14406 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14408 * doc/install.texi: Add links to macOS binary distributions.
14410 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14414 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14416 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14418 2017-05-26 Martin Liska <mliska@suse.cz>
14420 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14422 2017-05-26 Martin Liska <mliska@suse.cz>
14424 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14425 always leading ';; '.
14426 (dump_bb_info): Likewise.
14427 (brief_dump_cfg): Likewise.
14428 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14429 * dumpfile.c: Remove usage of TDF_VERBOSE.
14430 * dumpfile.h (enum dump_kind): Likewise.
14431 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14432 * print-tree.c (debug_verbose): Remove.
14433 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14434 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14435 * tree-diagnostic.c (default_tree_printer): Replace
14436 TDF_DIAGNOSTIC with TDF_SLIM.
14438 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14440 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14441 in parameter loop, rather than loop_vinfo.
14442 (create_intersect_range_checks): Ditto.
14443 (vect_create_cond_for_alias_checks): Update call to above functions.
14445 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14447 PR tree-optimization/80815
14448 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14449 for merging runtime alias checks. Handle negative DR_STEPs.
14451 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14453 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14455 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14456 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14457 out code pruning runtime alias checks.
14458 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14459 factored out from above.
14460 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14462 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14464 (prune_runtime_alias_test_list): New decalaration.
14466 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14468 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14469 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14470 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14471 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14472 (operator==, comp_dr_with_seg_len_pair): Ditto.
14473 (vect_prune_runtime_alias_test_list): Ditto.
14475 2017-05-26 Martin Liska <mliska@suse.cz>
14478 * params.def: Bound partial-inlining-entry-probability param.
14480 2017-05-26 Marek Polacek <polacek@redhat.com>
14483 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14486 2017-05-26 Richard Biener <rguenther@suse.de>
14488 PR tree-optimization/80842
14489 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14492 2017-05-26 Richard Biener <rguenther@suse.de>
14494 PR tree-optimization/80844
14495 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14497 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14499 * doc/md.texi (Machine Constraints): Update x86 family
14500 machine constraints section to match 'config/i386/constraints.md'.
14502 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14504 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14506 2017-05-25 Nathan Sidwell <nathan@acm.org>
14508 * doc/invoke.texi (--enable-languages): Update documentation.
14510 2017-05-25 Martin Liska <mliska@suse.cz>
14512 * dumpfile.c: Add TDF_FOLDING.
14513 * dumpfile.h (enum dump_kind): Likewise.
14514 * genmatch.c (dt_simplify::gen_1): Use it.
14516 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14518 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14520 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14522 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14523 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14525 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14527 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14528 * match.pd (X == C): Rewrite it here.
14529 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14530 with_certain_nonzero_bits2): New predicates.
14531 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14533 2017-05-24 Nathan Sidwell <nathan@acm.org>
14535 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14538 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14541 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14543 * config/powerpcspe: New port. Files are copied from the rs6000
14544 port, with "rs6000" in filenames replaced by "powerpcspe".
14546 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14548 PR rtl-optimization/80754
14549 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14551 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14553 * config/sparc/sparc.md (length): Return the correct value for -mflat
14554 sibcalls to match output_sibcall.
14556 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14560 * config/rs6000/rs6000.c (struct machine_function): Add new field
14562 (rs6000_get_separate_components): Init that field, use it.
14563 (rs6000_components_for_bb): Use the field.
14565 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14567 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14569 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14571 PR middle-end/80823
14572 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14574 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14577 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14578 address against address_operand predicate.
14579 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14581 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14583 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14584 parameters passed indirectly.
14586 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14588 * config/i386/i386.md (*movdi_internal): Remove SSE4
14589 alternative 18 (?r, *v). Update insn attributes.
14590 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14591 Update insn attributes.
14592 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14593 Update insn attributes.
14594 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14595 alternative 1 (r, v). Remove isa attribute.
14596 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14597 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14598 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14600 2017-05-23 Tom de Vries <tom@codesourcery.com>
14602 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14605 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14607 * cgraphunit.c (symbol_table::process_new_functions): Update.
14608 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14609 (inline_generate_summary): Rename to ...
14610 (ipa_fn_summary_generate): ... this one.
14611 (inline_read_summary): Rename to ...
14612 (ipa_fn_summary_read): ... this one.
14613 (inline_write_summary): Rename to ...
14614 (ipa_fn_summary_write): ... this one.
14615 (inline_free_summary): Rename to ...
14616 (ipa_free_fn_summary): ... this one.
14617 (pass_data_local_fn_summary, pass_local_fn_summary,
14618 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14619 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14620 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14621 make_pass_ipa_fn_summary): New.
14622 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14623 inline_write_summary, inline_free_summary): Remove.
14624 (ipa_free_fn_summary) : New.
14625 * ipa-inline.c (ipa_inline): Update.
14626 (pass_ipa_inline): Do not generate summaries.
14627 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14629 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14630 and add pass_ipa_fn_summary.
14631 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14633 (make_pass_inline_parameters): Remove.
14635 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14637 * omp-low.c (struct omp_context): Remove "default_kind" member.
14640 * omp-offload.c (execute_oacc_device_lower): Remove the
14641 parallelism dimensions function attributes for unparallelized
14642 OpenACC kernels constructs.
14644 2017-05-23 Martin Liska <mliska@suse.cz>
14646 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14648 (cgraph_edge::make_speculative): Likewise.
14649 (cgraph_edge::resolve_speculation): Likewise.
14650 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14651 (cgraph_node::dump): Likewise.
14652 * cgraph.h: Likewise.
14653 * cgraphunit.c (analyze_functions): Likewise.
14654 (symbol_table::compile): Likewise.
14655 * ipa-cp.c (print_all_lattices): Likewise.
14656 (determine_versionability): Likewise.
14657 (initialize_node_lattices): Likewise.
14658 (ipcp_verify_propagated_values): Likewise.
14659 (estimate_local_effects): Likewise.
14660 (update_profiling_info): Likewise.
14661 (create_specialized_node): Likewise.
14662 (perhaps_add_new_callers): Likewise.
14663 (decide_about_value): Likewise.
14664 (decide_whether_version_node): Likewise.
14665 (identify_dead_nodes): Likewise.
14666 (ipcp_store_bits_results): Likewise.
14667 * ipa-devirt.c (dump_targets): Likewise.
14668 (ipa_devirt): Likewise.
14669 * ipa-icf.c (sem_item::dump): Likewise.
14670 (sem_function::equals): Likewise.
14671 (sem_variable::equals): Likewise.
14672 (sem_item_optimizer::read_section): Likewise.
14673 (sem_item_optimizer::execute): Likewise.
14674 (congruence_class::dump): Likewise.
14675 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14676 (dump_inline_summary): Likewise.
14677 (estimate_node_size_and_time): Likewise.
14678 (inline_analyze_function): Likewise.
14679 * ipa-inline-transform.c (inline_call): Likewise.
14680 * ipa-inline.c (report_inline_failed_reason): Likewise.
14681 (want_early_inline_function_p): Likewise.
14682 (edge_badness): Likewise.
14683 (update_edge_key): Likewise.
14684 (inline_small_functions): Likewise.
14685 * ipa-profile.c (ipa_profile): Likewise.
14686 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14687 (ipa_make_edge_direct_to_target): Likewise.
14688 (remove_described_reference): Likewise.
14689 (ipa_impossible_devirt_target): Likewise.
14690 (propagate_controlled_uses): Likewise.
14691 (ipa_print_node_params): Likewise.
14692 (ipcp_transform_function): Likewise.
14693 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14694 (propagate_pure_const): Likewise.
14695 * ipa-reference.c (generate_summary): Likewise.
14696 (read_write_all_from_decl): Likewise.
14697 (propagate): Likewise.
14698 (ipa_reference_read_optimization_summary): Likewise.
14699 * ipa-utils.c (ipa_merge_profiles): Likewise.
14700 * ipa.c (walk_polymorphic_call_targets): Likewise.
14701 (symbol_table::remove_unreachable_nodes): Likewise.
14702 (ipa_single_use): Likewise.
14703 * passes.c (execute_todo): Likewise.
14704 * predict.c (drop_profile): Likewise.
14705 * symtab.c (symtab_node::get_dump_name): New function.
14706 (symtab_node::dump_name): Likewise.
14707 (symtab_node::dump_asm_name): Likewise.
14708 (symtab_node::dump_references): Likewise.
14709 (symtab_node::dump_referring): Likewise.
14710 (symtab_node::dump_base): Likewise.
14711 (symtab_node::debug_symtab): Likewise.
14712 * tree-sra.c (convert_callers_for_node): Likewise.
14713 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14714 * value-prof.c (init_node_map): Likewise.
14716 2017-05-23 Martin Liska <mliska@suse.cz>
14718 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14719 and symtab_node::debug_symtab to symbol_table::debug.
14720 * cgraphunit.c (analyze_functions): Use the renamed function.
14721 (symbol_table::compile): Likewise.
14722 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14723 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14724 * passes.c (execute_todo): Likewise.
14725 * symtab.c (symbol_table::dump): New function.
14726 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14728 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14730 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14731 that nonconst implies exec.
14733 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14735 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14736 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14737 (inline_edge_summary_vec): Turn into ...
14738 (ipa_call_summaries): ... this one.
14739 (redirect_to_unreachable, edge_set_predicate,
14740 evaluate_properties_for_edge, inline_summary_alloc,
14741 reset_ipa_call_summary, reset_inline_summary,
14742 inline_summary_t::duplicate): Update.
14743 (inline_edge_duplication_hook): Turn to ...
14744 (ipa_call_summary_t::duplicate): ... this one.
14745 (inline_edge_removal_hook): Turn to ...
14746 (ipa_call_summary_t::remove): ... this one.
14747 (dump_inline_edge_summary): Turn to ...
14748 (dump_ipa_call_summary): ... this one.
14749 (estimate_function_body_sizes): Update.
14750 (inline_update_callee_summaries): Update.
14751 (remap_edge_change_prob): Update.
14752 (remap_edge_summaries): Update.
14753 (inline_merge_summary): Update.
14754 (do_estimate_edge_time): Update.
14755 (inline_generate_summary): Update.
14756 (inline_read_section): Update.
14757 (inline_read_summary): Update.
14758 (inline_free_summary): Update.
14759 * ipa-inline.c (can_inline_edge_p): Update.
14760 (compute_inlined_call_time): Update.
14761 (want_inline_small_function_p): Update.
14762 (edge_badness): Update.
14763 (early_inliner): Update.
14764 * ipa-inline.h (inline_edge_summary): Turn to ...
14765 (ipa_call_summary): ... this one.
14766 (ipa_call_summary_t): New class.
14767 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14768 (ipa_call_summaries): New.
14769 (inline_edge_summary): Remove.
14770 (estimate_edge_growth): Update.
14771 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14772 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14773 * ipa-split.c (execute_split_functions): Update.
14774 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14776 2017-05-23 Tom de Vries <tom@codesourcery.com>
14778 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14779 attributes): Document rdrand effective target.
14781 2017-05-23 Tom de Vries <tom@codesourcery.com>
14783 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14784 attributes): Sort alphabetically.
14786 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14788 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14790 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14793 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14794 V2DF/V2DI splat into two separate patterns, one that handles
14795 registers, and the other that only handles memory. Drop support
14796 for splatting from a GPR on ISA 2.07 and then splitting the
14797 splat into direct move and splat.
14798 (vsx_splat_<mode>_reg): Likewise.
14799 (vsx_splat_<mode>_mem): Likewise.
14801 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14803 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14805 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14807 PR middle-end/80809
14808 * omp-low.c (finish_taskreg_remap): New function.
14809 (finish_taskreg_scan): If unit size of ctx->record_type
14810 is non-constant, unshare the size expression and replace
14811 decls in it with possible outer var refs.
14813 PR middle-end/80809
14814 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14815 GOVD_SHARED rather than GOVD_PRIVATE with it.
14816 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14817 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14819 PR middle-end/80853
14820 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14821 as last argument to build_outer_var_ref for pointer bases of array
14822 section reductions.
14824 2017-05-19 Martin Sebor <msebor@redhat.com>
14826 * print-tree.c (print_node): Print DECL_READ_P flag.
14828 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14830 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14831 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14832 * cgraph.c: Likewise.
14833 * cgraphunit.c: Likewise.
14834 * gengtype.c: Likewise.
14835 * ipa-cp.c: Likewise.
14836 * ipa-devirt.c: Likewise.
14837 * ipa-icf.c: Likewise.
14838 * ipa-predicate.c: Likewise.
14839 * ipa-profile.c: Likewise.
14840 * ipa-prop.c: Likewise.
14841 * ipa-split.c: Likewise.
14843 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14844 edge_predicate_pool, dump_inline_hints,
14845 inline_summary::account_size_time, redirect_to_unreachable,
14846 edge_set_predicate, set_hint_predicate,
14847 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14848 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14849 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14850 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14851 ipa_call_summary_t::remove, initialize_growth_caches,
14852 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14853 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14854 mark_modified, unmodified_parm_1, unmodified_parm,
14855 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14856 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14857 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14858 will_be_nonconstant_predicate, record_modified_bb_info,
14859 get_minimal_bb, record_modified, param_change_prob,
14860 phi_result_unknown_predicate, predicate_for_phi_result,
14861 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14862 estimate_function_body_sizes, compute_inline_parameters,
14863 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14864 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14865 inline_update_callee_summaries, remap_edge_change_prob,
14866 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14867 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14868 inline_analyze_function, inline_summary_t::insert,
14869 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14870 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14871 inline_free_summary): Move to ipa-fnsummary.h
14872 (predicate_t): Remove.
14873 * ipa-fnsummary.c: New file.
14874 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14875 (enum inline_hints_vals, inline_hints, agg_position_info,
14876 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14877 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14878 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14879 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14880 inline_read_summary, inline_write_summary, inline_free_summary,
14881 inline_analyze_function, initialize_inline_failed,
14882 inline_merge_summary, inline_update_overall_summary,
14883 compute_inline_parameters): Move to ipa-fnsummary.h
14884 * ipa-fnsummary.h: New file.
14885 * ipa-inline-transform.h: Include ipa-inline.h.
14886 * ipa-inline.c: LIkewise.
14888 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14890 * ipa-inline.c (edge_badness): Use inlined_time instead of
14891 inline_summaries->get.
14893 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14895 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14897 2017-05-22 Nathan Sidwell <nathan@acm.org>
14899 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14900 (fdump-lang): Document 'raw' option.
14901 * dumpfile.h (TDI_tu): Delete.
14902 * dumpfile.c (dump_files): Remove translation-unit.
14903 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14905 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14907 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14908 command option from $(AWK) call.
14909 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14910 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14911 [FORMAT]: Remove handling of variable.
14912 * config/avr/t-multilib: Regenerate.
14914 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14916 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14918 (dump_inline_summary): Do not print self_time.
14919 (estimate_function_body_sizes): Do not set self_time.
14920 (compute_inline_parameters): Likewise.
14921 (inline_read_section, inline_write_summary): Do not stream self_time.
14922 * ipa-inline.h (inline_summary): Drop self_time.
14924 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14926 * ipa-inline-analysis.c (account_size_time): Rename to ...
14927 (inline_summary::account_size_time): ... this one.
14928 (reset_ipa_call_summary): Turn to ...
14929 (ipa_call_summary::reset): ... this one.
14930 (reset_inline_summary): Turn to ...
14931 (inline_summary::reset): ... this one.
14932 (inline_summary_t::remove): Update.
14933 (inline_summary_t::duplicate): Update.
14934 (ipa_call_summary_t::remove): Update.
14935 (dump_inline_summary): Update.
14936 (estimate_function_body_sizes): Update.
14937 (compute_inline_parameters): Update.
14938 (estimate_node_size_and_time): Update.
14939 (inline_merge_summary): Update.
14940 (inline_update_overall_summary): Update.
14941 (inline_read_section): Update.
14942 (inline_write_summary): Update.
14943 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14944 add account_size_time and reset member functions.
14945 (ipa_call_summary): Add reset function.
14946 * ipa-predicate.h (predicate::operator &): Constify.
14948 2017-05-22 Richard Biener <rguenther@suse.de>
14950 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14952 2017-05-19 Jason Merrill <jason@redhat.com>
14954 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14956 2017-05-19 Marek Polacek <polacek@redhat.com>
14959 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14960 TYPE_OVERFLOW_WRAPS checks.
14962 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14964 * tree-core.h (enum omp_clause_default_kind): Add
14965 "OMP_CLAUSE_DEFAULT_PRESENT".
14966 * tree-pretty-print.c (dump_omp_clause): Handle it.
14967 * gimplify.c (enum gimplify_omp_var_data): Add
14968 "GOVD_MAP_FORCE_PRESENT".
14969 (gimplify_adjust_omp_clauses_1): Map it to
14970 "GOMP_MAP_FORCE_PRESENT".
14971 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14973 * gimplify.c (oacc_default_clause): Clarify.
14975 2017-05-19 Nathan Sidwell <nathan@acm.org>
14977 LANG_HOOK_REGISTER_DUMPS
14978 * toplev.c (general_init): Call register dump lang hook.
14979 * doc/invoke.texi: Document -fdump-lang option family.
14980 * dumpfile.c (dump_files): Remove class dump here.
14981 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14982 * dumpfile.h (tree_dump_index): Remove TDI_class.
14983 * langhooks-def.h (lhd_register_dumps): Declare.
14984 (LANG_HOOKS_REGISTER_DUMPS): Define.
14985 (LANG_HOOKS_INITIALIZER): Add it.
14986 * langhooks.c (lhd_register_dumps): Define.
14987 * langhooks.h (struct lang_hooks): Add register_dumps.
14989 2017-05-19 Nathan Sidwell <nathan@acm.org>
14991 * context.h (context::set_passes): New.
14992 * context.c (context::context): Do not create pass manager.
14993 * toplev.c (general_init): Create pass manager here.
14995 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14997 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14998 use this splitter if two add or or instructions would also work for
14999 the constant we want to generate.
15001 2017-05-19 Richard Biener <rguenther@suse.de>
15004 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15005 predicate evaluation.
15007 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15009 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15011 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15014 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15016 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15017 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15018 (inline_edge_summary_vec): Turn into ...
15019 (ipa_call_summaries): ... this one.
15020 (redirect_to_unreachable, edge_set_predicate,
15021 evaluate_properties_for_edge, inline_summary_alloc,
15022 reset_ipa_call_summary, reset_inline_summary,
15023 inline_summary_t::duplicate): Update.
15024 (inline_edge_duplication_hook): Turn to ...
15025 (ipa_call_summary_t::duplicate): ... this one.
15026 (inline_edge_removal_hook): Turn to ...
15027 (ipa_call_summary_t::remove): ... this one.
15028 (dump_inline_edge_summary): Turn to ...
15029 (dump_ipa_call_summary): ... this one.
15030 (estimate_function_body_sizes): Update.
15031 (inline_update_callee_summaries): Update.
15032 (remap_edge_change_prob): Update.
15033 (remap_edge_summaries): Update.
15034 (inline_merge_summary): Update.
15035 (do_estimate_edge_time): Update.
15036 (inline_generate_summary): Update.
15037 (inline_read_section): Update.
15038 (inline_read_summary): Update.
15039 (inline_free_summary): Update.
15040 * ipa-inline.c (can_inline_edge_p): Update.
15041 (compute_inlined_call_time): Update.
15042 (want_inline_small_function_p): Update.
15043 (edge_badness): Update.
15044 (early_inliner): Update.
15045 * ipa-inline.h (inline_edge_summary): Turn to ...
15046 (ipa_call_summary): ... this one.
15047 (ipa_call_summary_t): New class.
15048 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15049 (ipa_call_summaries): New.
15050 (inline_edge_summary): Remove.
15051 (estimate_edge_growth): Update.
15052 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15053 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15054 * ipa-split.c (execute_split_functions): Update.
15055 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15057 2017-05-19 Richard Biener <rguenther@suse.de>
15059 PR middle-end/80764
15060 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15062 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15064 * config/rs6000/rs6000.c (struct machine_function): Add field
15065 fpr_is_wrapped_separately.
15066 (rs6000_get_separate_components): Use 64 components. Handle the
15067 new FPR components.
15068 (rs6000_components_for_bb): Handle the FPR components.
15069 (rs6000_emit_prologue_components): Handle the FPR components.
15070 (rs6000_emit_epilogue_components): Handle the FPR components.
15071 (rs6000_set_handled_components): Handle the FPR components.
15072 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15073 that are already separately shrink-wrapped.
15074 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15075 that are already separately shrink-wrapped.
15077 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15080 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15083 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15084 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15085 cases where the register allocator uses a move and an offsettable
15086 memory operation to/from a FPR register on ISA 2.06/2.07.
15087 (define_peephole2 for Altivec d-form store): Likewise.
15089 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15092 * config/i386/mmx.md (*mov<mode>_internal): Enable
15093 alternatives 11, 12, 13 and 14 also for 32bit targets.
15094 Remove alternatives 15, 16, 17 and 18.
15095 * config/i386/sse.md (vec_concatv2di): Change
15096 alternative (!x, *y) to (x, ?!*Yn).
15098 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15100 * dumpfile.h (enum dump_kind): Remove stray comma.
15102 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15104 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15105 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15106 predicate::num_conditions
15107 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15108 (CHANGED): turn into predicate::changed.
15109 (agg_position_info): Move to ipa-predicate.h
15110 (add_condition, predicate::add_clause, predicate::operator &=,
15111 predicate::or_with, predicate::evaluate, predicate::probability,
15112 dump_condition, dump_clause, predicate::dump,
15113 predicate::remap_after_duplication, predicate::remap_after_inlining,
15114 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15115 (evaluate_conditions_for_known_args): Update.
15116 (set_cond_stmt_execution_predicate): Update.
15117 * ipa-inline.h: Include ipa-predicate.h
15118 (condition, inline_param_summary, conditions, agg_position_info,
15119 predicate): Move to ipa-predicate.h
15120 * ipa-predicate.c: New file.
15121 * ipa-predicate.h: New file.
15123 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15125 * final.c (leaf_function_p): Check we are not in a sequence.
15127 2017-05-18 Martin Liska <mliska@suse.cz>
15129 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15130 * dumpfile.c (dump_register): Use new enum dump_kind.
15131 (get_dump_file_name): Likewise.
15132 (dump_enable_all): Likewise.
15133 (dump_switch_p_1): Likewise.
15134 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15135 * dumpfile.h (enum dump_kind): New enum type.
15136 (struct dump_file_info): Create constructor and
15137 format fields and comments.
15138 * passes.c (pass_manager::register_one_dump_file):
15140 * statistics.c (statistics_early_init): Likewise.
15141 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15142 TDF_TREE with TDF_SLIM.
15143 (gather_memory_references_ref): Likewise.
15145 2017-05-18 Martin Liska <mliska@suse.cz>
15147 * vec.h (struct vnull): Use it.
15149 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15151 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15152 (true_predicate, false_predicate, true_predicate_p,
15153 false_predicate_p): Remove.
15154 (single_cond_predicate, not_inlined_predicate): Turn to member function
15156 (add_condition): Update.
15157 (add_clause): Turn to...
15158 (predicate::add_clause): ... this one; update; allow passing NULL
15160 (and_predicates): Turn to ...
15161 (predicate::operator &=): ... this one.
15162 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15163 (or_predicates): Turn to ...
15164 (predicate::or_with): ... this one.
15165 (evaluate_predicate): Turn to ...
15166 (predicate::evaluate): ... this one.
15167 (predicate_probability): Turn to ...
15168 (predicate::probability): ... this one.
15169 (dump_condition): Update.
15170 (dump_predicate): Turn to ...
15171 (predicate::dump): ... this one.
15172 (account_size_time): Update.
15173 (edge_set_predicate): Update.
15174 (set_hint_predicate): UPdate.
15175 (evaluate_conditions_for_known_args): Update.
15176 (evaluate_properties_for_edge): Update.
15177 (remap_predicate_after_duplication): Turn to...
15178 (predicate::remap_after_duplication): ... this one.
15179 (remap_hint_predicate_after_duplication): Update.
15180 (inline_summary_t::duplicate): UPdate.
15181 (dump_inline_edge_summary): Update.
15182 (dump_inline_summary): Update.
15183 (set_cond_stmt_execution_predicate): Update.
15184 (set_switch_stmt_execution_predicate): Update.
15185 (compute_bb_predicates): Update.
15186 (will_be_nonconstant_expr_predicate): Update.
15187 (will_be_nonconstant_predicate): Update.
15188 (phi_result_unknown_predicate): Update.
15189 (predicate_for_phi_result): Update.
15190 (array_index_predicate): Update.
15191 (estimate_function_body_sizes): Update.
15192 (estimate_node_size_and_time): Update.
15193 (estimate_ipcp_clone_size_and_time): Update.
15194 (remap_predicate): Rename to ...
15195 (predicate::remap_after_inlining): ... this one.
15196 (remap_hint_predicate): Update.
15197 (inline_merge_summary): Update.
15198 (inline_update_overall_summary): Update.
15199 (estimate_size_after_inlining): Update.
15200 (read_predicate): Rename to ...
15201 (predicate::stream_in): ... this one.
15202 (read_inline_edge_summary): Update.
15203 (write_predicate): Rename to ...
15204 (predicate::stream_out): ... this one.
15205 (write_inline_edge_summary): Update.
15206 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15207 (clause_t): Turn to uint32_t
15208 (predicate): Turn to class; implement constructor and operators
15210 (size_time_entry): Update.
15211 (inline_summary): Update.
15212 (inline_edge_summary): Update.
15214 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15216 * fold-const.c (fold_binary_loc): Move transformation...
15217 * match.pd (C - X CMP X): ... here.
15219 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15221 * config/sparc/sparc.c (sparc_option_override): Set function
15222 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15223 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15225 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15227 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15229 2017-05-18 Marek Polacek <polacek@redhat.com>
15232 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15233 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15234 gimple_assign_load_p.
15236 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15238 PR middle-end/80692
15239 * real.c (do_compare): Give decimal_do_compare preference over
15240 comparing just the signs.
15242 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15244 * doc/md.texi (Canonicalization of Instructions): Describe the
15245 canonical form of instructions that inherently set a condition
15248 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15250 PR middle-end/80775
15251 * tree-cfg.c: Move deletion of unreachable case statements to after
15252 the merging of consecutive case labels.
15254 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15256 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15257 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15258 restoring of callee-saved registers.
15260 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15262 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15263 * config/visium/visium.c (single_set_and_flags): Likewise.
15264 * config/visium/visium.md (Substitutions): Likewise.
15266 2017-05-17 Martin Liska <mliska@suse.cz>
15268 * cfg.c: Introduce dump_flags_t type and
15269 use it instead of int type.
15271 * cfghooks.c: Likewise.
15272 * cfghooks.h (struct cfg_hooks): Likewise.
15273 * cfgrtl.c: Likewise.
15274 * cfgrtl.h: Likewise.
15275 * cgraph.c (cgraph_node::get_body): Likewise.
15276 * coretypes.h: Likewise.
15277 * domwalk.c: Likewise.
15278 * domwalk.h: Likewise.
15279 * dumpfile.c (struct dump_option_value_info): Likewise.
15280 (dump_enable_all): Likewise.
15281 (dump_switch_p_1): Likewise.
15282 (opt_info_switch_p): Likewise.
15283 * dumpfile.h (enum tree_dump_index): Likewise.
15284 (struct dump_file_info): Likewise.
15285 * genemit.c: Likewise.
15286 * generic-match-head.c: Likewise.
15287 * gengtype.c (open_base_files): Likewise.
15288 * gimple-pretty-print.c: Likewise.
15289 * gimple-pretty-print.h: Likewise.
15290 * graph.c (print_graph_cfg): Likewise.
15291 * graphite-scop-detection.c (dot_all_sese): Likewise.
15292 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15293 * loop-unroll.c (report_unroll): Likewise.
15294 * passes.c (pass_manager::register_one_dump_file): Likewise.
15295 * print-tree.c: Likewise.
15296 * statistics.c: Likewise.
15297 * tree-cfg.c: Likewise.
15298 * tree-cfg.h: Likewise.
15299 * tree-dfa.c: Likewise.
15300 * tree-dfa.h: Likewise.
15301 * tree-dump.c (dump_function): Likewise.
15302 * tree-dump.h (struct dump_info): Likewise.
15303 * tree-pretty-print.c: Likewise.
15304 * tree-pretty-print.h: Likewise.
15305 * tree-ssa-live.c: Likewise.
15306 * tree-ssa-live.h: Likewise.
15307 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15308 * tree-vect-loop.c: Likewise.
15309 * tree-vect-slp.c: Likewise.
15311 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15312 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15314 PR tree-optimization/80457
15315 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15316 of all arguments to a statement as scalar_to_vec operations.
15317 (vectorizable_call): Adjust call to vect_model_simple_cost for
15319 (vectorizable_conversion): Likewise.
15320 (vectorizable_assignment): Likewise.
15321 (vectorizable_shift): Likewise.
15322 (vectorizable_operation): Likewise.
15323 (vectorizable_comparison): Likewise.
15324 (vect_is_simple_cond): Record the def types for operands.
15325 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15326 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15327 for statement argument count.
15329 2017-05-16 Carl Love <cel@us.ibm.com>
15331 * config/rs6000/rs6000-c: Add support for built-in functions
15332 vector unsigned long long vec_bperm (vector unsigned long long,
15333 vector unsigned char)
15334 vector signed long long vec_mule (vector signed int,
15336 vector unsigned long long vec_mule (vector unsigned int,
15337 vector unsigned int)
15338 vector signed long long vec_mulo (vector signed int,
15340 vector unsigned long long vec_mulo (vector unsigned int,
15341 vector unsigned int)
15342 vector signed char vec_sldw (vector signed char,
15343 vector signed char,
15345 vector unsigned char vec_sldw (vector unsigned char,
15346 vector unsigned char,
15348 vector signed short vec_sldw (vector signed short,
15349 vector signed short,
15351 vector unsigned short vec_sldw (vector unsigned short,
15352 vector unsigned short,
15354 vector signed int vec_sldw (vector signed int,
15357 vector unsigned int vec_sldw (vector unsigned int,
15358 vector unsigned int,
15360 vector signed long long vec_sldw (vector signed long long,
15361 vector signed long long,
15363 vector unsigned long long vec_sldw (vector unsigned long long,
15364 vector unsigned long long,
15366 * config/rs6000/rs6000-c: Add support for built-in functions
15367 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15368 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15369 * doc/extend.texi: Update the built-in documentation for the
15370 new built-in functions.
15372 2017-05-16 Marek Polacek <polacek@redhat.com>
15376 * tree.c (save_expr): Don't fold the expression.
15378 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15380 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15381 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15382 and (?*y,m). Update insn attributes.
15384 2017-05-16 Martin Liska <mliska@suse.cz>
15386 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15387 flags argument of print_gimple_stmt, print_gimple_expr,
15388 print_generic_stmt and print_generic_expr.
15389 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15390 * coretypes.h: Likewise.
15391 * except.c (dump_eh_tree): Likewise.
15392 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15393 * gimple-pretty-print.h: Likewise.
15394 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15395 (backprop::push_to_worklist): Likewise.
15396 (backprop::pop_from_worklist): Likewise.
15397 (backprop::process_use): Likewise.
15398 (backprop::intersect_uses): Likewise.
15399 (note_replacement): Likewise.
15400 * gimple-ssa-store-merging.c
15401 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15402 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15403 (pass_store_merging::execute): Likewise.
15404 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15405 (ssa_base_cand_dump_callback): Likewise.
15406 (dump_incr_vec): Likewise.
15407 (replace_refs): Likewise.
15408 (replace_mult_candidate): Likewise.
15409 (create_add_on_incoming_edge): Likewise.
15410 (create_phi_basis): Likewise.
15411 (insert_initializers): Likewise.
15412 (all_phi_incrs_profitable): Likewise.
15413 (introduce_cast_before_cand): Likewise.
15414 (replace_one_candidate): Likewise.
15415 * gimplify.c (gimplify_expr): Likewise.
15416 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15417 (set_rename): Likewise.
15418 (rename_uses): Likewise.
15419 (copy_loop_phi_nodes): Likewise.
15420 (add_close_phis_to_merge_points): Likewise.
15421 (copy_loop_close_phi_args): Likewise.
15422 (copy_cond_phi_args): Likewise.
15423 (graphite_copy_stmts_from_block): Likewise.
15424 (translate_pending_phi_nodes): Likewise.
15425 * graphite-poly.c (print_pdr): Likewise.
15426 (dump_gbb_cases): Likewise.
15427 (dump_gbb_conditions): Likewise.
15428 (print_scop_params): Likewise.
15429 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15430 (build_cross_bb_scalars_use): Likewise.
15431 (gather_bbs::before_dom_children): Likewise.
15432 * hsa-dump.c (dump_hsa_immed): Likewise.
15433 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15434 (get_replacement_map): Likewise.
15435 * ipa-inline-analysis.c (dump_condition): Likewise.
15436 (estimate_function_body_sizes): Likewise.
15437 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15438 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15439 * ipa-prop.c (ipa_dump_param): Likewise.
15440 (ipa_print_node_jump_functions_for_edge): Likewise.
15441 (ipa_modify_call_arguments): Likewise.
15442 (ipa_modify_expr): Likewise.
15443 (ipa_dump_param_adjustments): Likewise.
15444 (ipa_dump_agg_replacement_values): Likewise.
15445 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15446 * ipa-pure-const.c (check_stmt): Likewise.
15447 (pass_nothrow::execute): Likewise.
15448 * ipa-split.c (execute_split_functions): Likewise.
15449 * omp-offload.c (dump_oacc_loop_part): Likewise.
15450 (dump_oacc_loop): Likewise.
15451 * trans-mem.c (tm_log_emit): Likewise.
15452 (tm_memopt_accumulate_memops): Likewise.
15453 (dump_tm_memopt_set): Likewise.
15454 (dump_tm_memopt_transform): Likewise.
15455 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15456 (print_loop): Likewise.
15457 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15458 (chkp_gather_checks_info): Likewise.
15459 (chkp_get_check_result): Likewise.
15460 (chkp_remove_check_if_pass): Likewise.
15461 (chkp_use_outer_bounds_if_possible): Likewise.
15462 (chkp_reduce_bounds_lifetime): Likewise.
15463 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15464 (chkp_mark_completed_bounds): Likewise.
15465 (chkp_register_incomplete_bounds): Likewise.
15466 (chkp_mark_invalid_bounds): Likewise.
15467 (chkp_maybe_copy_and_register_bounds): Likewise.
15468 (chkp_build_returned_bound): Likewise.
15469 (chkp_get_bound_for_parm): Likewise.
15470 (chkp_build_bndldx): Likewise.
15471 (chkp_get_bounds_by_definition): Likewise.
15472 (chkp_generate_extern_var_bounds): Likewise.
15473 (chkp_get_bounds_for_decl_addr): Likewise.
15474 * tree-chrec.c (chrec_apply): Likewise.
15475 * tree-data-ref.c (dump_data_reference): Likewise.
15476 (dump_subscript): Likewise.
15477 (dump_data_dependence_relation): Likewise.
15478 (analyze_overlapping_iterations): Likewise.
15479 * tree-inline.c (expand_call_inline): Likewise.
15480 (tree_function_versioning): Likewise.
15481 * tree-into-ssa.c (dump_defs_stack): Likewise.
15482 (dump_currdefs): Likewise.
15483 (dump_names_replaced_by): Likewise.
15484 (dump_update_ssa): Likewise.
15485 (update_ssa): Likewise.
15486 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15487 * tree-parloops.c (build_new_reduction): Likewise.
15488 (try_create_reduction_list): Likewise.
15489 (ref_conflicts_with_region): Likewise.
15490 (oacc_entry_exit_ok_1): Likewise.
15491 (oacc_entry_exit_single_gang): Likewise.
15492 * tree-pretty-print.h: Likewise.
15493 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15494 (get_scalar_evolution): Likewise.
15495 (add_to_evolution): Likewise.
15496 (get_loop_exit_condition): Likewise.
15497 (analyze_evolution_in_loop): Likewise.
15498 (analyze_initial_condition): Likewise.
15499 (analyze_scalar_evolution): Likewise.
15500 (instantiate_scev): Likewise.
15501 (number_of_latch_executions): Likewise.
15502 (gather_chrec_stats): Likewise.
15503 (final_value_replacement_loop): Likewise.
15504 (scev_const_prop): Likewise.
15505 * tree-sra.c (dump_access): Likewise.
15506 (disqualify_candidate): Likewise.
15507 (create_access): Likewise.
15508 (reject): Likewise.
15509 (maybe_add_sra_candidate): Likewise.
15510 (create_access_replacement): Likewise.
15511 (analyze_access_subtree): Likewise.
15512 (analyze_all_variable_accesses): Likewise.
15513 (sra_modify_assign): Likewise.
15514 (initialize_constant_pool_replacements): Likewise.
15515 (find_param_candidates): Likewise.
15516 (decide_one_param_reduction): Likewise.
15517 (replace_removed_params_ssa_names): Likewise.
15518 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15519 * tree-ssa-copy.c (dump_copy_of): Likewise.
15520 (copy_prop_visit_cond_stmt): Likewise.
15521 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15522 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15523 (record_equivalences_from_stmt): Likewise.
15524 * tree-ssa-dse.c (compute_trims): Likewise.
15525 (delete_dead_call): Likewise.
15526 (delete_dead_assignment): Likewise.
15527 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15528 (forward_propagate_into_cond): Likewise.
15529 (pass_forwprop::execute): Likewise.
15530 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15531 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15533 (move_computations_worker): Likewise.
15534 (execute_sm): Likewise.
15535 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15536 (remove_exits_and_undefined_stmts): Likewise.
15537 (remove_redundant_iv_tests): Likewise.
15538 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15539 (adjust_iv_update_pos): Likewise.
15540 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15541 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15542 (value_replacement): Likewise.
15543 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15544 * tree-ssa-pre.c (print_pre_expr): Likewise.
15545 (get_representative_for): Likewise.
15546 (create_expression_by_pieces): Likewise.
15547 (insert_into_preds_of_block): Likewise.
15548 (eliminate_insert): Likewise.
15549 (eliminate_dom_walker::before_dom_children): Likewise.
15550 (eliminate): Likewise.
15551 (remove_dead_inserted_code): Likewise.
15552 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15553 * tree-ssa-reassoc.c (get_rank): Likewise.
15554 (eliminate_duplicate_pair): Likewise.
15555 (eliminate_plus_minus_pair): Likewise.
15556 (eliminate_not_pairs): Likewise.
15557 (undistribute_ops_list): Likewise.
15558 (eliminate_redundant_comparison): Likewise.
15559 (update_range_test): Likewise.
15560 (optimize_range_tests_var_bound): Likewise.
15561 (optimize_vec_cond_expr): Likewise.
15562 (rewrite_expr_tree): Likewise.
15563 (rewrite_expr_tree_parallel): Likewise.
15564 (linearize_expr): Likewise.
15565 (break_up_subtract): Likewise.
15566 (linearize_expr_tree): Likewise.
15567 (attempt_builtin_powi): Likewise.
15568 (attempt_builtin_copysign): Likewise.
15569 (transform_stmt_to_copy): Likewise.
15570 (transform_stmt_to_multiply): Likewise.
15571 (dump_ops_vector): Likewise.
15572 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15573 (print_scc): Likewise.
15574 (set_ssa_val_to): Likewise.
15575 (visit_reference_op_store): Likewise.
15576 (visit_use): Likewise.
15577 (sccvn_dom_walker::before_dom_children): Likewise.
15578 (run_scc_vn): Likewise.
15579 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15581 (expr_hash_elt::print): Likewise.
15582 (const_and_copies::pop_to_marker): Likewise.
15583 (const_and_copies::record_const_or_copy_raw): Likewise.
15584 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15585 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15586 (dump_predicates): Likewise.
15587 (find_uninit_use): Likewise.
15588 (warn_uninitialized_phi): Likewise.
15589 (pass_late_warn_uninitialized::execute): Likewise.
15590 * tree-ssa.c (verify_vssa): Likewise.
15591 (verify_ssa): Likewise.
15592 (maybe_optimize_var): Likewise.
15593 * tree-vrp.c (dump_value_range): Likewise.
15594 (dump_all_value_ranges): Likewise.
15595 (dump_asserts_for): Likewise.
15596 (register_edge_assert_for_2): Likewise.
15597 (vrp_visit_cond_stmt): Likewise.
15598 (vrp_visit_switch_stmt): Likewise.
15599 (vrp_visit_stmt): Likewise.
15600 (vrp_visit_phi_node): Likewise.
15601 (simplify_cond_using_ranges_1): Likewise.
15602 (fold_predicate_in): Likewise.
15603 (evrp_dom_walker::before_dom_children): Likewise.
15604 (evrp_dom_walker::push_value_range): Likewise.
15605 (evrp_dom_walker::pop_value_range): Likewise.
15606 (execute_early_vrp): Likewise.
15608 2017-05-16 Richard Biener <rguenther@suse.de>
15610 * dwarf2out.c (loc_list_from_tree_1): Do not create
15611 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15613 2017-05-16 Richard Biener <rguenther@suse.de>
15615 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15617 (note_variable_value_in_expr): If we resolved the decl ref
15618 do not push to the stack.
15620 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15622 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15623 operations in fast-math mode.
15624 (vaddq_f16): Likewise.
15625 (vmul_f16): Likewise.
15626 (vmulq_f16): Likewise.
15627 (vsub_f16): Likewise.
15628 (vsubq_f16): Likewise.
15629 * config/arm/neon.md (add<mode>3): New.
15631 (fma:<VH:mode>3): New. Also remove outdated comment.
15634 2017-05-16 Martin Liska <mliska@suse.cz>
15638 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15639 (odr_types_equivalent_p): Likewise.
15641 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15643 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15645 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15648 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15649 non-interunit SSE move alternatives with '?'.
15650 (zero-extendsidi peephole2): New peephole to skip intermediate
15651 general register in SSE zero-extend sequence.
15653 2017-05-15 Jeff Law <law@redhat.com>
15655 * reorg.c (relax_delay_slots): Create a new variable to hold
15656 the temporary target rather than clobbering TARGET_LABEL.
15658 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15659 missing argument to extract_bit_field call.
15660 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15662 2017-05-15 Martin Liska <mliska@suse.cz>
15665 * gcc.c (process_command): Do not allow empty argument of -o option.
15667 2017-05-15 Renlin Li <renlin.li@arm.com>
15669 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15670 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15671 * config/aarch64/constraints.md (Usf): Add long call check.
15672 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15673 (call_value): Likewise.
15674 (sibcall): Likewise.
15675 (sibcall_value): Likewise.
15677 (call_value_insn): New.
15678 (sibcall_insn): Update rtx pattern.
15679 (sibcall_value_insn): Likewise.
15680 (call_internal): Remove.
15681 (call_value_internal): Likewise.
15682 (sibcall_internal): Likewise.
15683 (sibcall_value_internal): Likewise.
15684 (call_reg): Likewise.
15685 (call_symbol): Likewise.
15686 (call_value_reg): Likewise.
15687 (call_value_symbol): Likewise.
15689 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15692 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15694 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15696 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15697 compatible with CCGOCmode and with CCZmode.
15699 2017-05-14 Martin Sebor <msebor@redhat.com>
15701 PR middle-end/77671
15702 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15703 (gimple_fold_builtin_snprintf): Same.
15704 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15705 (gimple_fold_builtin_snprintf): Same.
15706 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15707 of character types.
15708 (is_call_safe): New function.
15709 (try_substitute_return_value): Call it.
15710 (try_simplify_call): New function.
15711 (pass_sprintf_length::handle_gimple_call): Call it.
15713 2017-05-14 Martin Sebor <msebor@redhat.com>
15715 PR middle-end/80669
15716 * builtins.c (expand_builtin_stpncpy): Simplify.
15718 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15720 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15721 * config/i386/i386.h
15722 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15723 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15724 (struct machine_function): Add new members call_ms2sysv,
15725 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15726 (struct machine_frame_state): New fields sp_realigned and
15727 sp_realigned_offset.
15728 * config/i386/i386.c
15729 (enum xlogue_stub): New enum.
15730 (enum xlogue_stub_sets): New enum.
15731 (class xlogue_layout): New class.
15732 (struct ix86_frame): New fields stack_realign_allocate_offset,
15733 stack_realign_offset and outlined_save_offset. Modify comments to
15734 detail stack layout when using out-of-line stubs.
15735 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15736 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15737 -mcall-ms2sysv-xlogues.
15738 (stub_managed_regs): New static variable.
15739 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15740 registers managed by out-of-line stub.
15741 (disable_call_ms2sysv_xlogues): New function.
15742 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15743 m->call_ms2sysv when appropriate and compute frame layout for
15745 (sp_valid_at, fp_valid_at): New inline functions.
15746 (choose_basereg): New function.
15747 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15749 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15750 Use align parameter of choose_baseaddr to generated aligned SSE movs
15752 (pro_epilogue_adjust_stack): Modify to track
15753 machine_frame_state::sp_realigned.
15754 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15755 (ix86_nsaved_sseregs): Likewise.
15756 (ix86_emit_save_regs): Likewise.
15757 (ix86_emit_save_regs_using_mov): Likewise.
15758 (ix86_emit_save_sse_regs_using_mov): Likewise.
15759 (get_scratch_register_on_entry): Likewise.
15760 (gen_frame_set): New function.
15761 (gen_frame_load): Likewise.
15762 (gen_frame_store): Likewise.
15763 (emit_outlined_ms2sysv_save): Likewise.
15764 (emit_outlined_ms2sysv_restore): Likewise.
15765 (ix86_expand_prologue): Modify stack re-alignment code and call
15766 emit_outlined_ms2sysv_save when appropriate.
15767 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15768 parameter rtx_insn *insn, which allows the function to be used to only
15769 generate the notes.
15770 (ix86_expand_epilogue): Modify validity checks of frame and stack
15771 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15772 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15773 * config/i386/predicates.md
15774 (save_multiple): New predicate.
15775 (restore_multiple): Likewise.
15776 * config/i386/sse.md
15777 (save_multiple<mode>): New pattern.
15778 (save_multiple_realign<mode>): Likewise.
15779 (restore_multiple<mode>): Likewise.
15780 (restore_multiple_and_return<mode>): Likewise.
15781 (restore_multiple_leave_return<mode>): Likewise.
15782 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15784 2017-05-14 Julia Koval <julia.koval@intel.com>
15786 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15787 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15788 (__builtin_ia32_xsetbv): New builtins.
15789 * config/i386/i386.c (ix86_expand_special_args_builtin):
15791 (ix86_expand_builtin): Special expand for new intrinsics.
15792 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15793 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15794 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15796 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15798 * cfganal.c (inverted_post_order_compute): Change argument type
15800 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15801 * df-core.c (rest_of_handle_df_initialize): Adjust.
15802 (rest_of_handle_df_finish): Likewise.
15803 (df_analyze_1): Likewise.
15804 (df_analyze): Likewise.
15805 (loop_inverted_post_order_compute): Change argument to be a vec *.
15806 (df_analyze_loop): Adjust.
15807 (df_get_n_blocks): Likewise.
15808 (df_get_postorder): Likewise.
15809 * df.h (struct df_d): Change field to be a vec.
15810 * lcm.c (compute_laterin): Adjust.
15811 (compute_available): Likewise.
15812 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15813 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15814 * tree-ssa-pre.c (compute_antic): Likewise.
15816 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15818 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15819 (depth_first_search::depth_first_search): Change structure init
15820 function to this constructor.
15821 (depth_first_search::add_bb): Rename function to this member.
15822 (depth_first_search::execute): Likewise.
15823 (flow_dfs_compute_reverse_finish): Adjust.
15825 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15827 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15828 (longest_simple_path): Likewise.
15829 * shrink-wrap.c (spread_components): Likewise.
15830 (disqualify_problematic_components): Likewise.
15831 (emit_common_heads_for_components): Likewise.
15832 (emit_common_tails_for_components): Likewise.
15833 (insert_prologue_epilogue_for_components): Likewise.
15835 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15837 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15840 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15842 * df-core.c (df_set_blocks): Start using auto_bitmap.
15843 (df_compact_blocks): Likewise.
15844 * df-problems.c (df_rd_confluence_n): Likewise.
15845 * df-scan.c (df_insn_rescan_all): Likewise.
15846 (df_process_deferred_rescans): Likewise.
15847 (df_update_entry_block_defs): Likewise.
15848 (df_update_exit_block_uses): Likewise.
15849 (df_entry_block_bitmap_verify): Likewise.
15850 (df_exit_block_bitmap_verify): Likewise.
15851 (df_scan_verify): Likewise.
15852 * lra-constraints.c (lra_constraints): Likewise.
15853 (undo_optional_reloads): Likewise.
15854 (lra_undo_inheritance): Likewise.
15855 * lra-remat.c (calculate_gen_cands): Likewise.
15856 (do_remat): Likewise.
15857 * lra-spills.c (assign_spill_hard_regs): Likewise.
15858 (spill_pseudos): Likewise.
15859 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15860 (bitmap_set_subtract_values): Likewise.
15862 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15864 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15865 management with auto_bitmap.
15866 (fix_inter_tick): Likewise.
15867 (fix_recovery_deps): Likewise.
15868 * ira.c (add_store_equivs): Likewise.
15869 (find_moveable_pseudos): Likewise.
15870 (split_live_ranges_for_shrink_wrap): Likewise.
15871 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15872 (rtx_reuse_manager::seen_def_p): Likewise.
15873 (rtx_reuse_manager::set_seen_def): Likewise.
15874 * print-rtl.h (class rtx_reuse_manager): Likewise.
15876 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15878 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15880 (migrate_btr_def): Likewise.
15881 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15882 * df-core.c (loop_post_order_compute): Likewise.
15883 (loop_inverted_post_order_compute): Likewise.
15884 * hsa-common.h: Likewise.
15885 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15886 * init-regs.c (initialize_uninitialized_regs): Likewise.
15887 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15888 (inline_small_functions): Likewise.
15889 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15890 * ira.c (combine_and_move_insns): Likewise.
15891 (build_insn_chain): Likewise.
15892 * loop-invariant.c (find_invariants): Likewise.
15893 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15894 * predict.c (tree_predict_by_opcode): Likewise.
15895 (predict_paths_leading_to): Likewise.
15896 (predict_paths_leading_to_edge): Likewise.
15897 (estimate_loops_at_level): Likewise.
15898 (estimate_loops): Likewise.
15899 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15900 (spread_components): Likewise.
15901 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15902 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15903 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15904 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15905 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15906 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15907 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15908 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15909 (mark_threaded_blocks): Likewise.
15910 (thread_through_all_blocks): Likewise.
15911 * tree-ssa.c (verify_ssa): Likewise.
15912 (execute_update_addresses_taken): Likewise.
15913 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15915 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15917 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15919 (post_order_compute): Likewise.
15920 (inverted_post_order_compute): Likewise.
15921 (pre_and_rev_post_order_compute_fn): Likewise.
15923 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15925 * genrecog.c (int_set::int_set): Explicitly construct our
15926 auto_vec base class.
15927 * vec.h (auto_vec::auto_vec): New constructor.
15929 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15931 * bitmap.h (class auto_bitmap): New constructor taking
15932 bitmap_obstack * argument.
15934 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15936 * bitmap.h (class auto_bitmap): Change type of m_bits to
15937 bitmap_head, and adjust ctor / dtor and member operators.
15939 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15941 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15942 when returned register mode doesn't match original mode.
15944 2017-05-12 Jeff Law <law@redhat.com>
15945 Jakub Jelinek <jakub@redhat.com>
15947 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15948 we look for cc setter after the compare-elim changes.
15949 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15950 within the vector to match what compare-elim now expects.
15951 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15952 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15954 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15955 after the compare-elim changes.
15956 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15957 the vector to match what compare-elim now expects.
15958 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15959 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15960 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15961 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15962 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15964 * config/visium/visium.c (single_set_and_flags): Fix where
15965 we look for cc setter after the compare-elim changes.
15966 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15967 with the vector to match what compare-elim now expects.
15968 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15969 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15970 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15971 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15972 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15973 (neg<mode>2_insn_set_overflow): Likewise.
15975 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15977 PR middle-end/79794
15978 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15979 maybe_expand_insn call, set ops[0].target. If still set after call,
15980 set alt_rtl. Add extra arg to recursive calls.
15981 (extract_bit_field): Add alt_rtl argument. Pass to
15983 * expmed.h (extract_bit_field): Fix prototype.
15984 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15985 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15986 to extract_bit_field_calls.
15987 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15988 Pass alt_rtl to extract_bit_field calls.
15989 * calls.c (store_unaligned_arguments_into_psuedos)
15990 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15991 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15993 * optabs.h (struct expand_operand): Add target bitfield.
15995 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15997 * compare-elim.c (try_eliminate_compare): Canonicalize
15998 operation with embedded compare to
15999 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16000 (set (reg) (operation)].
16002 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16004 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16007 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16008 cost of adding a carry flag for ADC instruction.
16009 [case MINUS]: Ignore the cost of subtracting a carry flag
16010 for SBB instruction.
16012 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16014 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16016 * config/rs6000/bmiintrin.h: New file.
16017 * config/rs6000/bmi2intrin.h: New file.
16018 * config/rs6000/x86intrin.h: New file.
16020 2017-05-12 Jeff Law <law@redhat.com>
16022 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16025 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16027 PR middle-end/80707
16028 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16030 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16033 early expansion of vector divide builtins.
16034 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16035 builtins identified as having unsigned arguments.
16037 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16039 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16040 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16041 expansion of vector logical operations (and, andc, or, xor,
16044 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16046 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16047 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16049 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16051 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16052 early GIMPLE expansion of vector multiplies.
16054 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16056 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16057 TARGET_HAVE_MOVT conditional.
16058 (movt splitter): Likewise.
16060 2017-05-12 Richard Biener <rguenther@suse.de>
16062 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16063 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16064 Fold all stmts not inplace.
16066 2017-05-12 Richard Biener <rguenther@suse.de>
16068 PR tree-optimization/80713
16069 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16070 inserted_exprs bit for not removed stmts.
16072 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16074 PR middle-end/69921
16075 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16076 parallelized" attribute for parallelized OpenACC kernels.
16077 * omp-offload.c (execute_oacc_device_lower): Use it.
16079 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16080 Set "oacc kernels" attribute.
16081 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16082 parameter. Adjust all users.
16083 (oacc_fn_attrib_kernels_p): Remove function.
16084 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16085 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16086 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16087 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16088 assert "oacc kernels" attribute is set.
16090 2017-05-11 Carl Love <cel@us.ibm.com>
16092 * config/rs6000/rs6000-c: Add support for built-in functions
16093 vector unsigned char vec_popcnt (vector signed char)
16094 vector unsigned char vec_popcnt (vector unsigned char)
16095 vector unsigned short vec_popcnt (vector signed short)
16096 vector unsigned short vec_popcnt (vector unsigned short)
16097 vector unsigned int vec_popcnt (vector signed int)
16098 vector unsigned int vec_popcnt (vector unsigned int)
16099 vector unsigned long long vec_popcnt (vector signed long long)
16100 vector unsigned long long vec_popcnt (vector unsigned long long)
16101 vector signed long long vec_slo (vector signed long long,
16102 vector signed char)
16103 vector signed long long vec_slo (vector signed long long,
16104 vector unsigned char)
16105 vector unsigned long long vec_slo (vector unsigned long long,
16106 vector signed char)
16107 vector unsigned long long vec_slo (vector unsigned long long,
16108 vector unsigned char)
16109 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16110 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16111 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16112 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16113 * doc/extend.texi: Update the built-in documentation file for the
16114 new built-in functions.
16116 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16118 * attribs.h (sorted_attr_string): Move machine independent
16119 functions for target clone support from the i386 port to common
16120 code. Rename ix86_function_versions to common_function_versions.
16121 Rename make_name to make_unique_name.
16122 (common_function_versions): Likewise.
16123 (make_unique_name): Likewise.
16124 (make_dispatcher_decl): Likewise.
16125 (is_function_default_version): Likewise.
16126 * attribs.c (attr_strcmp): Likewise.
16127 (sorted_attr_string): Likewise.
16128 (common_function_versions): Likewise.
16129 (make_unique_name): Likewise.
16130 (make_dispatcher_decl): Likewise.
16131 (is_function_default_version): Likewise.
16132 * config/i386/i386.c (attr_strcmp): Likewise.
16133 (sorted_attr_string): Likewise.
16134 (ix86_function_versions): Likewise.
16135 (make_name): Likewise.
16136 (make_dispatcher_decl): Likewise.
16137 (is_function_default_version): Likewise.
16138 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16140 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16143 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16144 Account for direct move costs for vec_construct of integer
16147 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16150 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16151 (UNSPEC_STX_ATOMIC): Ditto.
16152 (loaddi_via_sse): New insn.
16153 (storedi_via_sse): Ditto.
16154 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16155 Update corresponding peephole2 patterns.
16156 (atomic_storedi_fpu): Ditto.
16158 2017-05-11 Julia Koval <julia.koval@intel.com>
16160 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16161 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16163 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16164 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16165 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16167 2017-05-11 Nathan Sidwell <nathan@acm.org>
16169 * graphite-poly.c: Include dumpfile.h.
16171 * dumpfle.h (dump_function): Declare here ...
16172 * tree-dump.h (dump_function): ... not here.
16173 * dumpfile.c: #include tree-cfg.h.
16174 (dump_function): Move here from ...
16175 * tree-dump.c (dump_function): ... here.
16176 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16177 * graphite-poly.c: Don't include tree-dump.h.
16178 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16179 * print-tree.c: Likewise.
16180 * stor-layout.c: Likewise.
16181 * tree-nested.c: Likewise.
16183 * dumpfile.c (dump_start): Use TDF_FLAGS.
16184 (dump_enable_all): Fix TDF_KIND check thinko.
16186 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16188 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16189 array entries to represent two legal parameterizations of the
16190 overloaded __builtin_cmpb function, as represented by the
16191 P6_OV_BUILTIN_CMPB constant.
16192 (altivec_resolve_overloaded_builtin): Add special case handling
16193 for the __builtin_cmpb function, as represented by the
16194 P6_OV_BUILTIN_CMPB constant.
16195 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16196 (BU_P6_64BIT_2): New macro.
16197 (BU_P6_OVERLOAD_2): New macro
16198 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16199 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16200 (CMPB): Add overload support to represent both 32-bit and 64-bit
16201 compare-bytes function.
16202 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16203 support for TARGET_CMPB.
16204 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16205 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16206 documentation of the __builtin_cmpb overloaded built-in function.
16208 2017-05-11 Richard Biener <rguenther@suse.de>
16210 PR tree-optimization/80705
16211 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16212 bases are not vectorizable.
16214 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16216 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16217 when counting register pressure.
16219 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16221 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16222 (iv_ca_more_deps): Renamed to ...
16223 (iv_ca_compare_deps): ... this.
16224 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16226 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16228 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16230 (determine_group_iv_costs): ... here.
16231 (find_inv_vars_cb): Record inv var if it's not recorded before.
16233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16235 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16236 (get_shiftadd_cost): Ditto.
16238 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16240 * tree-ssa-address.c: Include header file.
16241 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16243 (add_to_parts): Refactor.
16244 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16245 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16248 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16250 PR tree-optimization/53090
16251 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16253 (extract_cond_operands): Detect condition with IV on both sides
16254 and return COMP_IV_EXPR_2.
16255 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16256 (rewrite_use_compare): Simplify by removing call to function
16257 extract_cond_operands.
16259 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16261 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16262 (extract_cond_operands): Detect condition comparing against non-
16263 invariant bound and return appropriate enum value.
16264 (find_interesting_uses_cond): Update use of extract_cond_operands.
16265 Handle its return value accordingly.
16266 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16268 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16270 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16271 nonlinear iv_use computation in loop invariant sensitive way.
16273 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16275 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16276 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16278 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16280 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16281 (dump_cand): Support iv_cand.inv_exprs.
16282 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16284 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16287 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16289 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16291 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16292 as local function. Include necessary header files.
16293 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16297 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16299 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16301 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16302 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16303 RSHIFT_EXPR and BIT_NOT_EXPR.
16305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16307 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16308 (adjust_setup_cost): New parameter supporting round up adjustment.
16309 (struct address_cost_data): Delete.
16310 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16311 (split_address_cost, ptr_difference_cost): Delete.
16312 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16313 (struct ainc_cost_data): New struct.
16314 (get_address_cost_ainc): New function.
16315 (get_address_cost, get_computation_cost): Reimplement.
16316 (determine_group_iv_cost_address): Record inv_expr for all uses of
16318 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16319 (iv_ca_has_deps): Reimplemented to ...
16320 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16322 (iv_ca_extend): Call iv_ca_more_deps.
16324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16326 * tree-ssa-address.c (struct mem_address): Move to header file.
16327 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16328 * tree-ssa-address.h (struct mem_address): Move from C file.
16329 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16331 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16333 * tree-affine.h (aff_combination_type): New interface.
16334 (aff_combination_zero_p): Remove static.
16335 (aff_combination_const_p): New interface.
16336 (aff_combination_singleton_var_p): New interfaces.
16338 2017-05-11 Richard Biener <rguenther@suse.de>
16340 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16341 Skip unreachable blocks and destinations.
16342 (eliminate): Move stmt removal and fixup ...
16343 (fini_eliminate): ... here. Skip inserted exprs.
16344 (pass_pre::execute): Move fini_pre after fini_eliminate.
16345 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16346 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16347 PRE to get rid of dead code that has invalid SSA form and
16348 split critical edges again.
16350 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16352 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16354 2017-05-11 Richard Biener <rguenther@suse.de>
16356 * passes.c (execute_function_todo): Verify loops if they are
16357 said to be up-to-date.
16358 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16359 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16361 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16364 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16365 handle calling assemble_external ourself.
16368 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16369 modes with zero size. Enhance comment.
16371 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16373 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16374 built-ins for vec_xl and vec_xst with short and char pointer
16377 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16379 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16380 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16381 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16382 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16383 (_mm_maskz_min_round_ss): New intrinsics.
16384 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16385 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16386 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16387 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16388 (__builtin_ia32_minss_mask_round): New builtins.
16389 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16390 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16391 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16393 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16394 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16396 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16399 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16401 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16402 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16403 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16404 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16405 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16406 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16407 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16408 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16409 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16410 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16411 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16412 (__builtin_ia32_mulss_mask_round): New builtins.
16413 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16414 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16415 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16417 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16418 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16420 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16423 2017-05-10 Julia Koval <julia.koval@intel.com>
16425 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16426 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16427 (_mm256_setr_m128i): New intrinsics.
16429 2017-05-10 Julia Koval <julia.koval@intel.com>
16431 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16432 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16433 (_mm_maskz_rcp14_ss): New intrinsics.
16434 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16435 (__builtin_ia32_rcp14ss_mask): New builtins.
16436 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16438 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16440 PR tree-optimization/51513
16441 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16442 (assert_unreachable_fallthru_edge_p): Use it.
16443 (group_case_labels_stmt): Likewise.
16444 * tree-cfg.h: Prototype it.
16445 * stmt.c: Include cfghooks.h and tree-cfg.h.
16446 (emit_case_dispatch_table) <gap_label>: New local variable.
16447 Use it to fill dispatch table gaps.
16448 Test for default_label before updating probabilities.
16449 (expand_case) <default_label>: Remove unneeded initialization.
16450 Test for unreachable default case statement and remove its edge.
16451 Set default_label accordingly.
16452 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16454 2017-05-10 Carl Love <cel@us.ibm.com>
16456 * config/rs6000/rs6000-c: Add support for built-in functions
16457 vector signed char vec_neg (vector signed char)
16458 vector signed short int vec_neg (vector short int)
16459 vector signed int vec_neg (vector signed int)
16460 vector signed long long vec_neg (vector signed long long)
16461 vector float vec_neg (vector float)
16462 vector double vec_neg (vector double)
16463 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16465 * config/rs6000/altivec.h: Add define for vec_neg
16466 * doc/extend.texi: Update the built-in documentation for the
16467 new built-in functions.
16469 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16471 PR tree-optimization/77644
16472 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16474 2017-05-10 Nathan Sidwell <nathan@acm.org>
16476 * dumpfile.h (TDI_lang_all): New.
16477 (TDF_KIND): New. Renumber others
16478 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16480 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16482 (get_dump_file_name): Adjust suffix generation.
16483 (dump_enable_all): Use TDF_KIND.
16484 * doc/invoke.texi (-fdump-lang-all): Document.
16486 * dumpfile.h: Tabify.
16488 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16491 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16492 Move member access before delete.
16494 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16496 * tree-inline.c (expand_call_inline): Split block at stmt
16499 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16502 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16503 are now unused after splitting mov{sf,sd}_hardfloat.
16504 (f32_lr2): Likewise.
16505 (f32_lm): Likewise.
16506 (f32_lm2): Likewise.
16507 (f32_li): Likewise.
16508 (f32_li2): Likewise.
16509 (f32_lv): Likewise.
16510 (f32_sr): Likewise.
16511 (f32_sr2): Likewise.
16512 (f32_sm): Likewise.
16513 (f32_sm2): Likewise.
16514 (f32_si): Likewise.
16515 (f32_si2): Likewise.
16516 (f32_sv): Likewise.
16517 (f32_dm): Likewise.
16518 (f32_vsx): Likewise.
16519 (f32_av): Likewise.
16520 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16521 For movsf, order stores so the VSX stores occur before the GPR
16522 store which encourages the register allocator to use a traditional
16523 FPR instead of a GPR. For movsd, order the stores so that the GPR
16524 store comes before the VSX stores to allow the power6 to work.
16525 This is due to the power6 not having a 32-bit integer store
16526 instruction from a FPR.
16527 (movsf_hardfloat): Likewise.
16528 (movsd_hardfloat): Likewise.
16530 2017-05-09 Martin Sebor <msebor@redhat.com>
16532 PR translation/80280
16533 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16536 PR translation/80280
16537 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16538 data member added in r247778.
16539 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16541 2017-05-09 Nathan Sidwell <nathan@acm.org>
16543 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16545 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16548 2017-05-09 Marek Polacek <polacek@redhat.com>
16550 * doc/invoke.texi: Fix typo.
16552 2017-05-09 Richard Biener <rguenther@suse.de>
16554 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16555 (vrp_val_is_min): Likewise.
16556 (set_value_range_to_value): Likewise.
16557 (set_value_range_to_nonnegative): Likewise.
16558 (gimple_assign_nonzero_p): Likewise.
16559 (gimple_stmt_nonzero_p): Likewise.
16560 (vrp_int_const_binop): Likewise. Remove unreachable case.
16561 (adjust_range_with_scev): Adjust comments.
16562 (compare_range_with_value): Likewise.
16563 (extract_range_from_phi_node): Likewise.
16564 (test_for_singularity): Likewise.
16566 2017-05-09 Richard Biener <rguenther@suse.de>
16568 * tree-vrp.c (get_single_symbol): Add assert that we don't
16569 get overflowed constants as invariant part.
16570 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16571 checks. Use wi::cmp instead of recursing for integer constants.
16572 (compare_values): Just ignore whether we assumed undefined
16573 overflow instead of failing the compare.
16574 (extract_range_for_var_from_comparison_expr): Add comment before the
16575 TREE_NO_WARNING sets.
16576 (test_for_singularity): Likewise.
16577 (extract_range_from_comparison): Do not disable optimization
16578 when we assumed undefined overflow.
16579 (extract_range_basic): Remove init of unused var.
16581 2017-05-09 Richard Biener <rguenther@suse.de>
16583 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16584 (extract_range_from_multiplicative_op_1): Adjust.
16585 (extract_range_from_binary_expr_1): Use int_const_binop.
16587 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16590 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16591 rs6000_store_data_bypass_p in seven define_bypass directives and
16592 in several comments.
16593 * config/rs6000/rs6000-protos.h: Add prototype for
16594 rs6000_store_data_bypass_p function.
16595 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16596 function implements slightly different (rs6000-specific) semantics
16597 than store_data_bypass_p, returning false rather than aborting
16598 with assertion error when arguments do not satisfy the
16599 requirements of store data bypass.
16600 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16601 rs6000_store_data_bypass_p.
16603 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16605 * config/xtensa/xtensa-protos.h
16606 (xtensa_initial_elimination_offset): New declaration.
16607 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16608 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16609 macro definition, add case for FRAME_POINTER_REGNUM when
16610 FRAME_GROWS_DOWNWARD.
16611 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16612 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16613 xtensa_initial_elimination_offset.
16615 2017-05-08 Nathan Sidwell <nathan@acm.org>
16617 * doc/invoke.texi: Alphabetize -fdump options.
16619 2017-05-08 Martin Sebor <msebor@redhat.com>
16621 PR translation/80280
16622 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16624 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16626 * target.def (compute_frame_layout): New optional target hook.
16627 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16628 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16629 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16631 * reload1.c (verify_initial_elim_offsets): Likewise.
16632 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16633 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16634 (arm_get_frame_offsets): Split up into this ...
16635 (arm_compute_frame_layout): ... and this function.
16637 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16639 * config/aarch64/constraints.md (Usa): New constraint.
16640 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16642 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16644 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16645 with_multilib_list after it has been checked.
16647 2017-05-08 Richard Biener <rguenther@suse.de>
16649 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16650 (bitmap_set_subtract_values): Likewise.
16652 2017-05-08 Richard Biener <rguenther@suse.de>
16654 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16655 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16657 (gimple_stmt_nonzero_warnv_p): Rename to ...
16658 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16660 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16661 (extract_range_basic): Adjust, do not disable propagation on
16662 strict overflow sensitive simplification.
16663 (vrp_visit_cond_stmt): Likewise.
16665 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16667 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16668 body size unconditionally.
16670 2017-05-07 Jeff Law <law@redhat.com>
16673 2017-05-06 Jeff Law <law@redhat.com>
16674 PR tree-optimization/78496
16675 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16678 PR tree-optimization/78496
16679 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16680 (simplify_stmt_using_ranges): Call it.
16681 (vrp_dom_walker::before_dom_children): Extract equivalences
16682 from an ASSERT_EXPR with an equality comparison against a
16685 2017-05-06 Jeff Law <law@redhat.com>
16687 PR tree-optimization/78496
16688 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16691 PR tree-optimization/78496
16692 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16693 (simplify_stmt_using_ranges): Call it.
16694 (vrp_dom_walker::before_dom_children): Extract equivalences
16695 from an ASSERT_EXPR with an equality comparison against a
16698 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16700 * lra-constraints.c (lra_copy_reg_equiv): New function.
16701 (split_reg): Use it to copy equivalence information from the
16702 original register to the spill register.
16704 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16706 PR rtl-optimization/75964
16707 * simplify-rtx.c (simplify_const_relational_operation): Remove
16708 invalid handling of comparisons of integer ABS.
16710 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16712 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16713 initialize to zero.
16714 (init_regs): Remove declaration.
16715 (function_arg_advance_32): Initialize error_p as boolean variable.
16717 2017-05-05 Nathan Sidwell <nathan@acm.org>
16719 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16720 lines. Use for (;;).
16722 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16724 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16725 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16726 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16727 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16728 VF=2 that require versioning.
16730 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16732 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16735 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16737 * diagnostic.h (diagnostic_override_option_index): Convert from
16738 macro to inline function.
16740 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16742 * diagnostic.c (last_module_changed_p): New function.
16743 (set_last_module): New function.
16744 (diagnostic_report_current_module): Convert macro usage to
16745 the above functions.
16746 * diagnostic.h (diagnostic_context::last_module): Strengthen
16747 from const line_map * to const line_map_ordinary *.
16748 (diagnostic_last_module_changed): Delete macro.
16749 (diagnostic_set_last_module): Delete macro.
16751 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16753 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16754 with diagnostic_report_diagnostic.
16755 (diagnostic_n_impl_richloc): Likewise.
16756 * diagnostic.h (report_diagnostic): Delete macro.
16757 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16758 with diagnostic_report_diagnostic.
16759 * substring-locations.c (format_warning_va): Likewise.
16761 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16763 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16764 save/restor of format_spec. Move option-printing code to...
16765 (print_option_information): ...this new function, and
16766 reimplement by simply printing to the pretty_printer,
16767 rather than appending to the format string.
16769 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16771 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16772 handling logic into...
16773 (update_effective_level_from_pragmas): ...this new function.
16775 2017-05-04 Andrew Waterman <andrew@sifive.com>
16777 * config/riscv/riscv.opt (mstrict-align): New option.
16778 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16779 (SLOW_UNALIGNED_ACCESS): Define.
16780 (riscv_slow_unaligned_access): Declare.
16781 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16783 (riscv_slow_unaligned_access): New variable.
16784 (rocket_tune_info): Set slow_unaligned_access to true.
16785 (optimize_size_tune_info): Set slow_unaligned_access to false.
16786 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16787 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16788 (riscv_option_override): Set riscv_slow_unaligned_access.
16789 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16791 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16793 * config/riscv/riscv.md: Unify indentation.
16795 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16800 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16802 (extendsi<mode>2): Add support for doing sign extension via
16803 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16804 don't have ISA 3.0 instructions.
16805 (extendsi<mode>2 splitter): Likewise.
16806 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16807 generate the normal insns since SImode can now go in vector
16808 registers. Disallow the special UNSPECs needed for previous
16809 machines to hide SImode being used. Add new insns
16810 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16811 (fix_trunc<mode>si2_stfiwx): Likewise.
16812 (fix_trunc<mode>si2_internal): Likewise.
16813 (fixuns_trunc<mode>si2): Likewise.
16814 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16815 (fctiw<u>z_<mode>_smallint): Likewise.
16816 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16817 of floating point to 32-bit integer from doing a direct move to
16818 the GPR registers to do a store.
16819 (fctiwz_<mode>): Break long line.
16821 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16823 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16824 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16825 (addr_list, addr_offset_valid_p): New.
16826 (split_address_groups): Check offset validity with above function.
16827 (gt-tree-ssa-loop-ivopts.h): Include header file.
16829 2017-05-05 Nathan Sidwell <nathan@acm.org>
16831 * config.gcc (arm*-*-*): Add missing 'fi'.
16833 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16835 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16836 included in -fopt-info does not matter.
16837 * doc/optinfo.texi (-fopt-info): Fix description of default
16838 behavour. Explicitly say order of options included in -fopt-info
16841 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16843 * config.gcc: Allow combinations of aprofile and rmprofile values for
16844 --with-multilib-list.
16845 * config/arm/t-multilib: New file.
16846 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16847 variables. Remove setting of ISA and floating-point ABI in
16848 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16849 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16850 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16851 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16853 * config/arm/t-rmprofile: Likewise except for the matches changes.
16854 * doc/install.texi (--with-multilib-list): Document the combination of
16855 aprofile and rmprofile values and warn about pitfalls in doing that.
16857 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16859 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16860 (movdi_aarch64): Likewise.
16862 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16864 PR tree-optimization/80632
16865 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16867 (build_arrays): Initialize it for virtual phis.
16868 (fix_phi_nodes): Use it for virtual phis.
16870 PR tree-optimization/80558
16871 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16872 [x, y] op z into [x op, y op z] for op & or | if conditions
16875 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16876 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16879 * config/arm/arm.md (use_literal_pool): Remove.
16880 (64-bit immediate split): No longer takes cost into consideration
16881 if arm_disable_literal_pool is enabled.
16882 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16883 used when arm_disable_literal_pool is enabled.
16884 (arm_max_const_double_inline_cost): Remove use of
16885 arm_disable_literal_pool.
16886 (push_minipool_fix): Add assert.
16887 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16888 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16889 (no_literal_pool_sf_immediate): New.
16891 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16893 PR tree-optimization/80613
16894 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16895 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16897 2017-05-05 Richard Biener <rguenther@suse.de>
16899 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16901 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16903 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16904 of this flag from insn conditions due to removal from r247495.
16906 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16908 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16910 (arm_early_store_addr_dep_ptr): Likewise.
16911 * config/arm/aarch-common-protos.h
16912 (arm_early_load_addr_dep_ptr): Add prototype.
16913 (arm_early_store_addr_dep_ptr): Likewise.
16914 * config/arm/cortex-a53.md: Add new bypasses.
16916 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16918 * tree.c (next_type_uid): Change type to unsigned.
16919 (type_hash_canon): Decrement back next_type_uid if
16920 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16921 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16924 2017-05-04 Martin Sebor <msebor@redhat.com>
16926 * builtins.c: Fix a trivial typo in a comment.
16928 PR middle-end/79234
16929 * builtins.c (check_sizes): Adjust to handle reading past the end.
16930 Avoid printing excessive upper bound of ranges. Use %E to print
16931 tree nodes instead of converting them to %wu.
16932 (expand_builtin_memchr): New function.
16933 (compute_dest_size): Rename...
16934 (compute_objsize): ...to this.
16935 (expand_builtin_memcpy): Adjust.
16936 (expand_builtin_mempcpy): Adjust.
16937 (expand_builtin_strcat): Adjust.
16938 (expand_builtin_strcpy): Adjust.
16939 (check_strncat_sizes): Adjust.
16940 (expand_builtin_strncat): Adjust.
16941 (expand_builtin_strncpy): Adjust and simplify.
16942 (expand_builtin_memset): Adjust.
16943 (expand_builtin_bzero): Adjust.
16944 (expand_builtin_memcmp): Adjust.
16945 (expand_builtin): Handle memcmp.
16946 (maybe_emit_chk_warning): Check strncat just once.
16948 2017-05-04 Martin Sebor <msebor@redhat.com>
16950 PR preprocessor/79214
16951 PR middle-end/79222
16952 PR middle-end/79223
16953 * builtins.c (check_sizes): Add inlining context and issue
16954 warnings even when -Wno-system-headers is set.
16955 (check_strncat_sizes): Same.
16956 (expand_builtin_strncat): Same.
16957 (expand_builtin_memmove): New function.
16958 (expand_builtin_stpncpy): Same.
16959 (expand_builtin): Handle memmove and stpncpy.
16961 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16963 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16964 which is not used any more.
16966 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16968 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16970 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16972 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16973 (cortexa53_tunings): Likewise.
16974 (cortexa57_tunings): Likewise.
16975 (cortexa72_tunings): Likewise.
16976 (cortexa73_tunings): Likewise.
16978 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16980 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16981 Set loop alignment to 8.
16983 2017-05-04 Martin Sebor <msebor@redhat.com>
16985 PR translation/80280
16986 * builtins.c (expand_builtin_object_size): Add missing quoting to
16987 %D and like directives.
16988 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16989 (hsa_type_for_tree_type): Same.
16990 (verify_function_arguments): Same.
16991 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16992 * varasm.c (get_section): Same.
16995 2017-05-04 Martin Sebor <msebor@redhat.com>
16997 PR translation/80280
16998 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17000 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17002 * config/aarch64/aarch64.c (generic_addrcost_table):
17003 Change HI/TI mode setting.
17005 2017-05-04 Martin Jambor <mjambor@suse.cz>
17007 PR tree-optimization/80622
17008 * tree-sra.c (comes_initialized_p): New function.
17009 (build_accesses_from_assign): Only set write lazily when
17010 comes_initialized_p is false.
17011 (analyze_access_subtree): Use comes_initialized_p.
17012 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17013 instead of testing for PARM_DECL.
17015 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17017 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17018 constraint on operand 0 to allow more general addressing modes.
17019 Adjust output template.
17020 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17022 * config/aarch64/aarch64-protos.h
17023 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17024 * config/aarch64/constraints.md (Dp): New address constraint.
17025 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17028 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17030 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17031 update use of estimate_ipcp_clone_size_and_time.
17032 (estimate_local_effects): Update use of
17033 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17034 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17035 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17036 Return nonspecialized time.
17038 2017-05-04 Richard Biener <rguenther@suse.de>
17040 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17041 for the last VUSE which def dominates the PHI. Directly call
17043 (get_continuation_for_phi_1): Remove.
17045 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17047 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17048 to explain the use of truncating division. Cap the number of
17049 iterations to the maximum given by nb_iterations_upper_bound,
17052 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17054 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17055 * configure: Regenerate.
17056 * config.in: Regenerate.
17057 * config/i386/driver-mingw32.c: new file.
17058 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17059 * config.host: Link driver-mingw32.o on MinGW host.
17060 * doc/install.texi: Document new --enable-mingw-wildcard configure
17063 2017-05-04 Marek Polacek <polacek@redhat.com>
17065 PR tree-optimization/80612
17066 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17068 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17069 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17071 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17072 (movt splitter): Likewise.
17073 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17074 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17075 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17076 block for Thumb-1 with MOVT.
17077 (thumb2_legitimate_address_p): Move code block ...
17078 (can_avoid_literal_pool_for_label_p): ... into this new function.
17079 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17081 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17082 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17083 "M-profile targets with the MOVT instruction".
17085 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17087 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17088 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17089 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17091 2017-05-04 Martin Liska <mliska@suse.cz>
17093 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17094 variable cond_code.
17096 2017-05-04 Richard Biener <rguenther@suse.de>
17098 * tree.c (array_at_struct_end_p): Handle arrays at struct
17099 end with flexarrays more conservatively. Refactor and treat
17100 arrays of arrays or aggregates more strict. Fix
17101 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17102 * tree.c (array_at_struct_end_p): Adjust prototype.
17103 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17104 * gimple-fold.c (get_range_strlen): Likewise.
17105 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17107 2017-05-04 Richard Biener <rguenther@suse.de>
17109 PR tree-optimization/31130
17110 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17112 (supports_overflow_infinity): Likewise.
17113 (is_negative_overflow_infinity): Likewise.
17114 (is_positive_overflow_infinity): Likewise.
17115 (is_overflow_infinity): Likewise.
17116 (stmt_overflow_infinity): Likewise.
17117 (overflow_infinity_range_p): Likewise.
17118 (usable_range_p): Remove as always returning true.
17119 (make_overflow_infinity): Remove.
17120 (negative_overflow_infinity): Likewise.
17121 (positive_overflow_infinity): Likewise.
17122 (avoid_overflow_infinity): Likewise.
17123 (set_value_range): Adjust accordingly.
17124 (set_value_range_to_nonnegative): Likewise, remove now unused
17125 overflow_infinity arg.
17126 (vrp_operand_equal_p): Adjust.
17127 (update_value_range): Likewise.
17128 (range_int_cst_singleton_p): Likewise.
17129 (operand_less_p): Likewise.
17130 (compare_values_warnv): Likewise.
17131 (extract_range_for_var_from_comparison_expr): Likewise.
17132 (vrp_int_const_binop): Likewise.
17133 (zero_nonzero_bits_from_vr): Likewise.
17134 (extract_range_from_multiplicative_op_1): Likewise.
17135 (extract_range_from_binary_expr_1): Likewise.
17136 (extract_range_from_unary_expr): Likewise.
17137 (extract_range_from_comparison): Likewise.
17138 (extract_range_basic): Likewise.
17139 (adjust_range_with_scev): Likewise.
17140 (compare_ranges): Likewise.
17141 (compare_range_with_value): Likewise.
17142 (dump_value_range): Likewise.
17143 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17145 (simplify_cond_using_ranges): Adjust.
17147 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17149 * brig-builtins.def: Added a builtin for class_f64.
17150 * builtin-types.def: Added a builtin type needed by class_f64.
17152 2017-05-03 Jason Merrill <jason@redhat.com>
17154 * timevar.def: Add TV_CONSTEXPR.
17156 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17158 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17160 2017-05-03 Martin Jambor <mjambor@suse.cz>
17162 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17163 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17164 * ipa-cp.c (ipcp_propagate_stage): Do not call
17165 ipa_update_after_lto_read.
17166 * ipa-inline.c (ipa_inline): Likewise.
17168 2017-05-03 Martin Jambor <mjambor@suse.cz>
17170 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17171 tag. Added a default constructor and a destructor.
17172 (ipa_edge_args_sum_t): New class;
17173 (ipa_edge_args_sum): Declare.
17174 (ipa_edge_args_vector): Remove declaration.
17175 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17176 (ipa_free_edge_args_substructures): Remove declaration.
17177 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17178 (ipa_edge_args_info_available_for_edge_p): Likewise.
17179 * ipa-prop.c (ipa_edge_args_vector): Removed.
17180 (edge_removal_hook_holder): Likewise.
17181 (edge_duplication_hook_holder): Likewise.
17182 (ipa_edge_args_sum): New variable.
17183 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17184 ipa_edge_args_vector.
17185 (ipa_free_edge_args_substructures): Likewise.
17186 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17187 ipa_edge_args_vector.
17188 (ipa_edge_removal_hook): Turned into method
17189 ipa_edge_args_sum_t::remove.
17190 (ipa_edge_duplication_hook): Turned into method
17191 ipa_edge_args_sum_t::duplicate.
17192 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17193 registering edge hooks.
17194 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17195 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17196 ipa_edge_args_sum instead of ipa_edge_args_vector.
17197 * ipa-profile.c (ipa_profile): Likewise.
17199 2017-05-03 Martin Jambor <mjambor@suse.cz>
17201 * symbol-summary.h (function_summary): New method exists.
17202 (function_summary::symtab_removal): Deallocate through release.
17203 (call_summary): New class.
17204 (gt_ggc_mx): New overload.
17205 (gt_pch_nx): Likewise.
17206 (gt_pch_nx): Likewise.
17208 2017-05-03 Jeff Law <law@redhat.com>
17210 PR tree-optimization/78496
17211 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17212 from simplify_cond_using_ranges. Split off code to walk
17213 backwards through casts into ...
17214 (simplify_cond_using_ranges_2): New function.
17215 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17216 (execute_vrp): After identifying jump threads, call
17217 simplify_cond_using_ranges_2.
17219 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17222 * ipa-inline.h (inline_summary): Add ctor.
17223 (create_ggc): Do not use ggc_cleared_alloc.
17225 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17226 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17228 * gcc.c (handle_braces): Support escaping in switch matching
17230 * doc/invoke.texi (Spec Files): Document it.
17231 Remove superfluous @code markup in items.
17233 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17235 * diagnostic-show-locus.c (struct column_range): New struct.
17236 (get_affected_columns): New function.
17237 (get_printed_columns): New function.
17238 (struct correction): New struct.
17239 (correction::ensure_capacity): New function.
17240 (correction::ensure_terminated): New function.
17241 (struct line_corrections): New struct.
17242 (line_corrections::~line_corrections): New dtor.
17243 (line_corrections::add_hint): New function.
17244 (layout::print_trailing_fixits): Reimplement in terms of the new
17246 (selftest::test_overlapped_fixit_printing): New function.
17247 (selftest::diagnostic_show_locus_c_tests): Call it.
17249 2017-05-03 Nathan Sidwell <nathan@acm.org>
17251 Canonicalize canonical type hashing
17252 * tree.h (type_hash_canon_hash): Declare.
17253 * tree.c (type_hash_list, attribute_hash_list): Move into
17254 type_hash_canon_hash.
17255 (build_type_attribute_qual_variant): Break out hash code calc into
17256 type_hash_canon_hash.
17257 (type_hash_canon_hash): New. Generic type hash computation.
17258 (build_range_type_1, build_array_type_1, build_function_type,
17259 build_method_type_directly, build_offset_type, build_complex_type,
17260 make_vector_type): Call it.
17262 2017-05-03 Richard Biener <rguenther@suse.de>
17264 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17265 When all DRs have unknown misaligned do not always peel
17266 when there is a store but apply the same costing model as if
17267 there were only loads.
17269 2017-05-03 Richard Biener <rguenther@suse.de>
17272 PR tree-optimization/80492
17273 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17274 compare_base_decls returning dont-know properly.
17276 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17278 * config/arm/iterators.md (CCSI): New mode iterator.
17279 (arch): New mode attribute.
17280 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17281 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17282 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17283 code iterator for success result mode.
17284 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17285 the corresponding new insn generators.
17287 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17290 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17291 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17293 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17295 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17296 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17297 (DDR_A): Wrap DDR argument in brackets.
17298 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17299 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17300 (DDR_REVERSED_P): Likewise.
17302 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17304 PR tree-optimization/79472
17305 * tree-switch-conversion.c (struct switch_conv_info): Add
17306 contiguous_range and default_case_nonstandard fields.
17307 (collect_switch_conv_info): Compute contiguous_range and
17308 default_case_nonstandard fields, don't clear final_bb if
17309 contiguous_range and only the default case doesn't have the required
17311 (check_all_empty_except_final): Set default_case_nonstandard instead
17312 of failing if contiguous_range and the default case doesn't have empty
17314 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17315 and only the default case doesn't have the required constants. Skip
17317 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17318 if default_case_nonstandard.
17319 (build_constructors): Build constant 1 just once. Assert that default
17320 values aren't inserted in between cases if contiguous_range. Skip
17322 (build_arrays): Skip virtual phis.
17323 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17324 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17325 Handle virtual phis.
17326 (gen_inbound_check): Handle default_case_nonstandard case.
17327 (process_switch): Adjust check_final_bb caller. Call
17328 gather_default_values with the first non-default case instead of
17329 default case if default_case_nonstandard.
17331 2017-05-02 Nathan Sidwell <nathan@acm.org>
17333 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17334 check. Fix formatting.
17336 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17338 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17339 errors when comparing specialized and unspecialized times.
17341 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17343 * diagnostic-show-locus.c
17344 (layout::should_print_annotation_line_p): Make private.
17345 (layout::print_annotation_line): Make private.
17346 (layout::annotation_line_showed_range_p): Make private.
17347 (layout::show_ruler): Make private.
17348 (layout::print_source_line): Make private. Pass in line and
17349 line_width, rather than calling location_get_source_line. Drop
17351 (layout::print_leading_fixits): New method.
17352 (layout::print_any_fixits): Rename to...
17353 (layout::print_trailing_fixits): ...this, and make private.
17354 Don't print newline fixits.
17355 (diagnostic_show_locus): Move logic for printing one row into...
17356 (layout::print_line): ...this new function. Move the
17357 location_get_source_line call and error-handling from
17358 print_source_line to here. Call print_leading_fixits, and rename
17359 print_any_fixits to print_trailing_fixits.
17360 (selftest::test_fixit_insert_containing_newline): Update now that
17361 newlines are partially supported.
17362 (selftest::test_fixit_insert_containing_newline_2): New test.
17363 (selftest::test_fixit_replace_containing_newline): Update comments.
17364 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17365 * edit-context.c (class added_line): New class.
17366 (class edited_line): Describe newline handling in comment.
17367 (edited_line::actually_edited_p): New method.
17368 (edited_line::print_content): Delete redundant decl.
17369 (edited_line::m_predecessors): New field.
17370 (edited_file::print_content): Call edited_line::print_content.
17371 (edited_file::print_diff): Update to support newlines.
17372 (edited_file::print_diff_hunk): Likewise.
17373 (edited_file::print_run_of_changed_lines): New function.
17374 (edited_file::print_diff_line): Convert to...
17375 (print_diff_line): ...this.
17376 (edited_file::get_effective_line_count): New function.
17377 (edited_line::edited_line): Initialize new field m_predecessors.
17378 (edited_line::~edited_line): Clean up m_predecessors.
17379 (edited_line::apply_fixit): Handle newlines.
17380 (edited_line::get_effective_line_count): New function.
17381 (edited_line::print_content): New function.
17382 (edited_line::print_diff_lines): New function.
17383 (selftest::test_applying_fixits_insert_containing_newline): New
17385 (selftest::test_applying_fixits_replace_containing_newline): New
17387 (selftest::insert_line): New function.
17388 (selftest::test_applying_fixits_multiple_lines): Add example of
17390 (selftest::edit_context_c_tests): Call the new tests.
17392 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17394 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17395 parameter cand. Update dump information.
17396 (get_computation_cost): Update uses.
17398 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17400 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17401 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17402 (get_computation_at, rewrite_use_address): Update use of
17403 get_computation_aff.
17405 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17407 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17408 (get_computation): Delete.
17409 (get_computation_cost): Implement like get_computation_cost_at.
17410 Use get_computation_at.
17411 (get_computation_cost_at): Delete.
17412 (rewrite_use_nonlinear_expr): Use get_computation_at.
17413 (rewrite_use_compare, remove_unused_ivs): Ditto.
17415 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17417 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17419 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17421 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17422 (ivopts_global_cost_for_size): Rename parameter and update uses.
17423 (iv_ca_recount_cost): Update uses.
17424 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17425 candidates seperately in n_invs and n_cands.
17426 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17430 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17431 (find_inv_vars_cb): New.
17432 (find_depends): Renamed to ...
17433 (find_inv_vars): ... this.
17434 (add_candidate_1, force_var_cost): Call find_inv_vars.
17435 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17437 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17439 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17440 inv_vars. Add inv_exprs.
17441 (struct iv_cand): Rename depends_on to inv_vars.
17442 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17443 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17444 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17445 (dump_cand): Dump inv_vars.
17446 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17447 (record_invariant, find_depends, add_candidate_1): Ditto.
17448 (set_group_iv_cost, force_var_cost): Ditto.
17449 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17450 (get_computation_cost_at, get_computation_cost): Ditto.
17451 (determine_group_iv_cost_generic): Ditto.
17452 (determine_group_iv_cost_address): Ditto.
17453 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17454 (determine_group_iv_costs): Ditto.
17455 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17456 (iv_ca_set_remove_invariants): Renamed to ...
17457 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17458 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17459 (iv_ca_set_add_invariants): Renamed to ...
17460 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17461 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17462 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17463 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17464 (create_new_ivs): Remove useless dump.
17466 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17468 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17470 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17471 (iv_ca_set_no_cp, create_new_iv): Ditto.
17473 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17475 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17477 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17479 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17480 function tree_check2.
17482 2017-05-02 Martin Liska <mliska@suse.cz>
17484 * doc/gcov.texi: Add missing preposition.
17485 * gcov.c (function_info::function_info): Properly fill up
17486 all member variables.
17488 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17490 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17492 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17494 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17496 2017-05-02 Martin Liska <mliska@suse.cz>
17499 * lto-streamer-in.c (lto_read_tree_1): Remove
17500 LTO_STREAMER_DEBUG.
17501 * lto-streamer.c (struct tree_hash_entry): Likewise.
17502 (struct tree_entry_hasher): Likewise.
17503 (tree_entry_hasher::hash): Likewise.
17504 (tree_entry_hasher::equal): Likewise.
17505 (lto_streamer_init): Likewise.
17506 (lto_orig_address_map): Likewise.
17507 (lto_orig_address_get): Likewise.
17508 (lto_orig_address_remove): Likewise.
17509 * lto-streamer.h: Likewise.
17510 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17511 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17513 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17515 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17516 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17517 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17518 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17519 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17520 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17521 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17522 (mm_maskz_sub_ss): New intrinsics.
17523 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17524 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17525 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17526 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17527 (__builtin_ia32_subss_mask_round): New builtins.
17528 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17529 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17530 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17532 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17533 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17535 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17538 2017-05-02 Martin Jambor <mjambor@suse.cz>
17540 PR tree-optimization/78687
17541 * tree-sra.c (access): New field parent.
17542 (process_subtree_disqualification): New function.
17543 (disqualify_candidate): Call it.
17544 (build_accesses_from_assign): Reset write flag if creating an
17546 (build_access_subtree): Fill in parent field and also prpagate
17547 down grp_write flag.
17548 (create_artificial_child_access): New parameter set_grp_write, set
17549 grp_write to its value.
17550 (propagate_subaccesses_across_link): Also propagate grp_write flag
17552 (propagate_all_subaccesses): Push the closest parent back to work
17553 queue if add_access_to_work_queue returned true.
17555 2017-05-02 Richard Biener <rguenther@suse.de>
17557 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17558 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17559 -fstrict-overflow documentation.
17560 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17561 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17562 flag_strict_overflow.
17563 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17564 * lto-opts.c (lto_write_options): Do not stream it.
17565 * lto-wrapper.c (merge_and_complain): Do not handle it.
17566 * opts.c (default_options_table): Do not set -fstrict-overflow.
17567 (finish_options): Likewise do not clear it when sanitizing.
17568 * simplify-rtx.c (simplify_const_relational_operation): Do not
17569 test flag_strict_overflow.
17571 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17573 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17574 using enabled attribute.
17575 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17576 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17577 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17578 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17579 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17580 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17581 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17582 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17583 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17584 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17586 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17588 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17590 2017-05-02 Richard Biener <rguenther@suse.de>
17592 PR tree-optimization/80591
17594 2017-04-10 Richard Biener <rguenther@suse.de>
17596 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17599 2017-05-02 Richard Biener <rguenther@suse.de>
17601 PR tree-optimization/80549
17602 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17603 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17604 headers if they do not have a preheader.
17606 2017-05-02 Martin Liska <mliska@suse.cz>
17609 * common.opt: Fix typo.
17610 * doc/invoke.texi: Likewise.
17612 2017-05-01 Jan Beulich <jbeulich@suse.com>
17614 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17615 swapping, add (x,x,m,x,n) alternative.
17617 2017-05-01 Nathan Sidwell <nathan@acm.org>
17619 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17620 unnecessary unadjusted_alignment check.
17622 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17625 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17627 * gimplify.c (gimplify_cilk_detach): New function.
17628 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17629 * tree-core.h: Document EXPR_CILK_SPAWN.
17630 * tree.h (EXPR_CILK_SPAWN): Define.
17632 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17634 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17635 to use new fixit_hint representation, using the "replace" logic.
17636 (get_line_span_for_fixit_hint): Likewise.
17637 (layout::print_any_fixits): Likewise.
17638 (selftest::test_one_liner_many_fixits): Rename to...
17639 (selftest::test_one_liner_many_fixits_1): ...this, and update
17640 comment and expected output to reflect that the multiple fix-it
17641 hints are now consolidated into one insertion.
17642 (selftest::test_one_liner_many_fixits_2): New test.
17643 (selftest::test_diagnostic_show_locus_one_liner): Update for
17645 (selftest::test_fixit_consolidation): Update for fix-it API
17647 * diagnostic.c (print_parseable_fixits): Likewise.
17648 * edit-context.c (edited_line::m_line_events): Convert from
17649 auto_vec <line_event *> to auto_vec <line_event>.
17650 (class line_event): Convert from abstract base class to a concrete
17651 class, taking over the role of replace_event.
17652 (class insert_event): Delete.
17653 (class replace_event): Rename to class line_event. Convert to
17655 (edit_context::add_fixits): Reimplement.
17656 (edit_context::apply_insert): Delete.
17657 (edit_context::apply_replace): Rename to...
17658 (edit_context::apply_fixit): ...this. Convert to half-open range.
17659 (edited_file::apply_insert): Delete.
17660 (edited_file::apply_replace): Rename to...
17661 (edited_file::apply_fixit): ...this.
17662 (edited_line::~edited_line): Drop deletion of events.
17663 (edited_line::apply_insert): Delete.
17664 (edited_line::apply_replace): Rename to...
17665 (edited_line::apply_fixit): ...this. Convert to half-open range.
17666 Update for change to type of m_line_events.
17667 * edit-context.h (edit_context::apply_insert): Delete.
17668 (edit_context::apply_replace): Rename to...
17669 (edit_context::apply_fixit): ...this.
17671 2017-05-01 Martin Sebor <msebor@redhat.com>
17673 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17676 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17679 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17680 __get_cpuid_max returns 0.
17681 (__get_cpuid_count): Ditto.
17683 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17685 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17686 replacement expression is another instance of one of its arguments.
17688 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17691 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17692 check for stack push/pop autoinc.
17693 * config/i386/i386.c (ix86_agi_dependent): Return false
17694 if the only reason why modified_in_p returned true is that
17695 addr is SP based and set_insn is a push or pop.
17697 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17699 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17702 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17705 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17706 (account_size_time): Use two predicates - exec_pred and
17708 (evaluate_conditions_for_known_args): Compute both clause and
17710 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17711 (inline_summary_t::duplicate): Update.
17712 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17714 (compute_inline_parameters): Likewise.
17715 (estimate_edge_size_and_time): Update caluclation of time.
17716 (estimate_node_size_and_time): Compute both time and nonspecialized
17718 (estimate_ipcp_clone_size_and_time): Update.
17719 (inline_merge_summary): Update.
17720 (do_estimate_edge_time): Update.
17721 (do_estimate_edge_size): Update.
17722 (do_estimate_edge_hints): Update.
17723 (inline_read_section, inline_write_summary): Stream both new predicates.
17724 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17726 (compute_inlined_call_time): Cleanup.
17727 (big_speedup_p): Update.
17728 (edge_badness): Update.
17729 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17730 (size_time_entry): Replace predicate by exec_predicate and
17731 nonconst_predicate.
17732 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17733 (estimate_edge_time): Return also nonspec_time.
17734 (reset_edge_growth_cache): Update.
17736 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17738 PR rtl-optimization/80491
17739 * ifcvt.c (noce_process_if_block): When looking for x setter
17740 with missing else_bb, don't check only the insn right before
17741 cond_earliest, but look for the last insn that x is modified in
17742 within the same bb.
17744 PR rtl-optimization/80491
17745 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17747 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17749 PR tree-optimization/80487
17750 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17752 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17754 PR tree-optimization/79697
17755 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17756 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17757 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17759 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17760 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17762 2017-04-28 Martin Sebor <msebor@redhat.com>
17764 PR tree-optimization/80523
17765 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17766 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17768 (maybe_warn, format_directive, parse_directive): Use new functions.
17769 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17771 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17773 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17775 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17777 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17778 target_header_dir): Set correctly.
17779 * configure: Regenerated.
17780 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17781 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17782 instead of SYSTEM_HEADER_DIR.
17784 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17786 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17787 (estimate_local_effects): Likewise.
17788 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17789 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17790 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17791 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17792 do_estimate_edge_time, estimate_edge_time): Likewise.
17793 * ipa-inline-analysis.c (estimate_node_size_and_time,
17794 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17795 (estimate_time_after_inlining): Remove.
17797 2017-04-28 Martin Liska <mliska@suse.cz>
17799 * doc/gcov.texi: Enhance documentation of gcov.
17801 2017-04-28 Martin Liska <mliska@suse.cz>
17803 * doc/gcov.texi: Sort options in alphabetic order.
17804 * doc/gcov-dump.texi: Likewise.
17805 * doc/gcov-tool.texi: Likewise.
17806 * gcov.c (print_usage): Likewise.
17807 * gcov-dump.c (print_usage): Likewise.
17808 * gcov-tool.c (print_merge_usage_message): Likewise.
17809 (print_rewrite_usage_message): Likewise.
17810 (print_overlap_usage_message): Likewise.
17812 2017-04-28 Martin Liska <mliska@suse.cz>
17814 PR gcov-profile/53915
17815 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17817 2017-04-28 Martin Liska <mliska@suse.cz>
17819 PR gcov-profile/79891
17820 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17821 is marked by compiler as living on a line.
17822 (get_cycles_count): Remove usage of the union.
17823 (output_intermediate_file): Likewise.
17824 (find_source): Fix GNU coding style.
17825 (accumulate_line_counts): Remove old non-all block mode.
17826 (output_lines): Remove usage of the union.
17827 * profile.c (output_location): Include all BBs, even if
17828 belonging to a same line (and file) as a previous BB.
17830 2017-04-28 Martin Liska <mliska@suse.cz>
17832 * gcov.c (process_args): Handle new argument 'w'.
17833 (read_graph_file): Assign ID to BBs.
17834 (output_branch_count): Display BB # if verbose flag is set.
17835 (output_lines): Likewise for arcs.
17836 (print_usage): Add '--verbose' option help.
17837 * doc/gcov.texi: Document --verbose (-w) option.
17839 2017-04-28 Martin Liska <mliska@suse.cz>
17841 * gcov.c (struct block_location_info): New struct.
17842 (process_file): Fill up the new structure.
17843 (read_graph_file): Replace usage of encoding by the newly added
17845 (add_line_counts): Likewise.
17846 (accumulate_line_counts): Remove usage of the union.
17847 (function_info::function_info): New function.
17848 (function_info::~function_info): Likewise.
17849 (process_file): Call delete instead of release_function.
17850 (release_function): Release the function.
17851 (release_structures): Call delete instead of release_function.
17852 (solve_flow_graph): Replace usage of num_blocks.
17853 (find_exception_blocks): Likewise.
17854 (output_lines): Fix GNU coding style.
17856 2017-04-28 Martin Liska <mliska@suse.cz>
17859 * coverage.c (coverage_remove_note_file): New function.
17860 * coverage.h: Declare the function.
17861 * toplev.c (finalize): Clean if an error has been seen.
17863 2017-04-28 Martin Liska <mliska@suse.cz>
17865 PR gcov-profile/80031
17866 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17867 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17868 * gcov.c (read_graph_file): Read just number of blocks.
17869 * profile.c (branch_prob): Do not stream 0 flags per a basic
17872 2017-04-28 Martin Liska <mliska@suse.cz>
17874 * gcov-dump.c (tag_*): Add new argument to declarations.
17875 (dump_gcov_file): Likewise.
17876 (tag_blocks): Add and use new argument depth.
17877 (tag_arcs): Likewise.
17878 (tag_lines): Likewise.
17879 (tag_counters): Likewise.
17880 (tag_summary): Likewise.
17881 (dump_working_sets): Use depth to do a proper indentation.
17883 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17886 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17887 * symtab.c (symtab_node::debug_symtab): Move definition here.
17889 2017-04-28 Richard Biener <rguenther@suse.de>
17891 * lto-streamer.h (LTO_major_version): Bump to 7.
17893 2017-04-28 Richard Biener <rguenther@suse.de>
17895 * tree-vrp.c (assert_info): New struct.
17896 (add_assert_info): New helper.
17897 (register_edge_assert_for_2): Refactor to add asserts to a vector
17899 (register_edge_assert_for_1): Likewise.
17900 (register_edge_assert_for): Likewise.
17901 (finish_register_edge_assert_for): New helper actually registering
17902 asserts where live on edge.
17903 (find_conditional_asserts): Adjust.
17904 (find_switch_asserts): Likewise.
17905 (evrp_dom_walker::try_find_new_range): Generalize.
17906 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17908 2017-04-27 Marek Polacek <polacek@redhat.com>
17911 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17912 arg10 and arg11 to itype.
17914 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17916 * doc/extend.texi (Object Size Checking): Improve grammar.
17918 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17921 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17922 that the logic for permitting reciprocal estimates matches that
17925 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17928 * tree.c (type_cache_hasher::equal): Only compare
17929 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17930 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17931 non-aggregate element types.
17932 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17933 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17935 2017-04-27 Richard Biener <rguenther@suse.de>
17937 PR middle-end/80533
17938 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17939 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17940 keeping a reference to a trailing array.
17942 2017-04-27 Richard Biener <rguenther@suse.de>
17944 PR middle-end/80539
17945 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17946 being in loop-closed SSA form conservatively.
17947 (chrec_fold_multiply_poly_poly): Likewise.
17949 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17951 PR middle-end/79665
17952 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17953 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17955 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17958 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17959 (aarch64_function_arg_alignment): Return unsigned int again, but still
17960 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17961 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17962 Don't emit -Wpsabi note.
17963 (aarch64_function_arg_boundary): Likewise.
17964 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17967 2017-04-26 Nathan Sidwell <nathan@acm.org>
17969 * tree.h (crc32_unsigned_n): Declare.
17970 (crc32_unsigned, crc32_unsigned): Make inline.
17971 * tree.c (crc32_unsigned_bits): Replace with ...
17972 (crc32_unsigned_n): ... this.
17973 (crc32_unsigned, crc32_byte): Remove.
17974 (crc32_string): Remove unnecessary braces.
17976 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17978 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17979 * ipa-inline-analysis.c (MAX_TIME): Remove.
17980 (account_size_time): Use sreal for time.
17981 (dump_inline_summary): Update.
17982 (estimate_function_body_sizes): Update.
17983 (estimate_edge_size_and_time): Update.
17984 (estimate_calls_size_and_time): Update.
17985 (estimate_node_size_and_time): Update.
17986 (inline_merge_summary): Update.
17987 (inline_update_overall_summary): Update.
17988 (estimate_time_after_inlining): Update.
17989 (inline_read_section): Update.
17990 (inline_write_summary): Update.
17991 * ipa-inline.c (compute_uninlined_call_time): Update.
17992 (compute_inlined_call_time): Update.
17993 (recursive_inlining): Update.
17994 (inline_small_functions): Update.
17995 (dump_overall_stats): Update.
17996 * ipa-inline.h: Include sreal.h.
17997 (size_time_entry): Turn time to sreal.
17998 (inline_summary): Turn self_time nad time to sreal.
18000 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18002 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18004 (sreal::stream_out, sreal::stream_in): New.
18005 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18007 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18009 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18012 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18015 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18017 (dimode_scalar_chain::compute_convert_gain): Ditto.
18018 (dimode_scalar_chain::make_vector_copies): Ditto.
18019 (dimode_scalar_chain::convert_reg): Ditto.
18020 (dimode_scalar_chain::convert_insn): Ditto.
18021 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18022 (VI248_AVX512BW_1): New mode iterator.
18023 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18024 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18027 2017-04-25 Martin Sebor <msebor@redhat.com>
18029 PR tree-optimization/80497
18030 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18031 constants are representable in HOST_WIDE_INT.
18032 (parse_directive): Ditto.
18034 2017-04-25 Martin Sebor <msebor@redhat.com>
18037 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18038 (new_zero_array): Adjust signature.
18039 (dom_info::dom_init): Used unsigned rather that size_t.
18040 (dom_info::dom_info): Same.
18042 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18043 Jakub Jelinek <jakub@redhat.com>
18046 * config/arm/arm.c: Include gimple.h.
18047 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18048 returns negative, increment ncrn only if it returned positive.
18049 (arm_needs_doubleword_align): Return int instead of bool,
18050 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18051 members, but if there is any such non-FIELD_DECL
18052 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18053 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18054 returns negative, increment nregs only if it returned positive.
18055 (arm_setup_incoming_varargs): Likewise.
18056 (arm_function_arg_boundary): Emit -Wpsabi note if
18057 arm_needs_doubleword_align returns negative, return
18058 DOUBLEWORD_ALIGNMENT only if it returned positive.
18060 2017-04-25 Marek Polacek <polacek@redhat.com>
18063 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18064 first argument to type.
18066 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18069 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18070 type checks to test for compatibility instead of equality.
18072 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18073 Jakub Jelinek <jakub@redhat.com>
18076 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18078 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18079 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18080 the alignment computation, but return their maximum in warn_alignment.
18081 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18082 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18084 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18085 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18088 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18090 * config/arc/simdext.md (dmpyh): Fix typo.
18092 2017-04-25 Richard Biener <rguenther@suse.de>
18094 PR tree-optimization/80492
18095 * alias.c (compare_base_decls): Handle registers with asm
18096 specification conservatively.
18097 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18098 compare_base_decls returning dont-know properly.
18100 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18102 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18103 (legitimate_offset_address_p): New function.
18104 (arc_legitimate_address_p): Use above function.
18106 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18108 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18110 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18112 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18113 ACCH registers whenever they are available.
18115 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18117 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18118 double regs fix when not used.
18120 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18122 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18124 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18125 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18127 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18129 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18130 when emitting short ADD instructions.
18132 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18134 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18136 (cmpsi_cc_c_insn): Likewise.
18137 (cbranchsi4_scratch): Compute proper instruction length using
18138 compact_hreg_operand.
18139 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18141 2017-04-25 Richard Biener <rguenther@suse.de>
18143 PR middle-end/80509
18144 * passes.c (pass_manager::pass_manager): Initialize
18145 m_name_to_pass_map.
18147 2017-04-25 Richard Biener <rguenther@suse.de>
18149 PR tree-optimization/79201
18150 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18152 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18155 * config/s390/vector.md: Split MEM->GPR vector moves for
18156 non-s_operand addresses.
18158 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18161 * config/s390/predicates.md (reload_const_wide_int_operand): New
18163 * config/s390/s390.md ("movti"): Remove d/P alternative.
18164 ("movti_bigconst"): New pattern definition.
18166 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18169 * s390-protos.h (s390_expand_cs_hqi): Removed.
18170 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18171 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18172 modes as well as CCZ1mode and CCZmode.
18173 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18174 signature of s390_emit_compare_and_swap.
18175 (s390_expand_cs_hqi): Likewise, make static.
18176 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18177 compare-and-swap, in some cases.
18178 (s390_expand_cs): Wrapper function.
18179 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18181 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18182 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18183 patterns for small and large integers. Forbid symref memory operands.
18184 Move expander to s390.c. Require cc register.
18185 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18186 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18187 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18188 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18189 symref memory operands. Remove CC mode and call s390_match_ccmode
18191 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18193 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18195 * config/s390/s390.md (define_peephole2): New peephole to help
18196 combining the load-and-test pattern with volatile memory.
18198 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18200 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18201 with CCZmode for TARGET_Z196.
18203 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18205 PR rtl-optimization/80501
18206 * combine.c (make_compound_operation_int): Set subreg_code to SET
18207 even for AND with mask of the sign bit of mode.
18209 PR rtl-optimization/80500
18210 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18211 sum's initial value.
18213 2017-04-25 Julian Brown <julian@codesourcery.com>
18214 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18216 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18218 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18220 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18222 2017-04-25 Julian Brown <julian@codesourcery.com>
18223 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18225 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18226 (thunderx2t99_sha): New Reservation.
18228 2017-04-25 Julian Brown <julian@codesourcery.com>
18229 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18231 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18232 type for 1-element load.
18234 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18236 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18238 2017-04-24 Martin Jambor <mjambor@suse.cz>
18240 PR tree-optimization/80293
18241 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18242 char arrays not totally scalarizable if it is false.
18243 (analyze_all_variable_accesses): Pass correct value in the new
18244 parameter. Add a statistics counter.
18246 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18248 PR middle-end/79931
18249 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18251 2017-04-24 Richard Biener <rguenther@suse.de>
18253 PR tree-optimization/80494
18254 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18255 out for complex types.
18257 2017-04-24 Richard Biener <rguenther@suse.de>
18259 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18260 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18261 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18262 (DFS): Adjust and never fail.
18263 (sccvn_dom_walker::fail): Remove.
18264 (sccvn_dom_walker::before_dom_children): Adjust.
18265 (run_scc_vn): Likewise and never fail.
18266 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18267 (pass_fre::execute): Likewise.
18269 2017-04-24 Richard Biener <rguenther@suse.de>
18271 PR tree-optimization/79725
18272 * tree-ssa-sink.c (statement_sink_location): Return whether
18273 failure reason was zero uses. Move that check later.
18274 (sink_code_in_bb): Deal with zero uses by removing the stmt
18277 2017-04-24 Richard Biener <rguenther@suse.de>
18280 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18281 pointer-based references.
18283 2017-04-24 Richard Biener <rguenther@suse.de>
18286 * pass_manager.h (pass_manager::operator new): Remove.
18287 (pass_manager::operator delete): Likewise.
18288 * passes.c (pass_manager::operator new): Remove.
18289 (pass_manager::operator delete): Likewise.
18290 (pass_manager::pass_manager): Zero individual pass members.
18292 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18295 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18296 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18297 Check "XEXP (src, 1)" operand here.
18298 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18299 Check "XEXP (src, 1)" operand here.
18300 (dimode_scalar_chain::make_vector_copies): Detect count register
18301 of a shift instruction. Zero extend count register from QImode
18302 to DImode to satisfy vector shift pattern count operand predicate.
18303 Substitute vector shift count operand with a DImode copy.
18304 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18307 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18309 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18310 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18311 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18312 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18313 (UNSPEC_NOREX_MEM): Remove definition.
18315 2017-04-21 Richard Biener <rguenther@suse.de>
18317 PR tree-optimization/79547
18318 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18319 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18320 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18321 without any constraints.
18323 2017-04-21 Richard Biener <rguenther@suse.de>
18325 PR tree-optimization/78847
18326 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18328 2017-04-21 Richard Biener <rguenther@suse.de>
18330 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18331 (build_distinct_type_copy): Likewise.
18332 (build_variant_type_copy): Likewise.
18333 * tree.c (build_qualified_type): Pass down mem-stat info.
18334 (build_distinct_type_copy): Likewise.
18335 (build_variant_type_copy): Likewise.
18337 2017-04-21 Richard Biener <rguenther@suse.de>
18339 PR tree-optimization/80237
18340 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18342 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18343 for a simplified result.
18345 2016-04-21 Richard Biener <rguenther@suse.de>
18347 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18348 sth as strict as a simple_iv but a chrec without symbols and an
18349 operand defined in the loop we are peeling (and not some subloop).
18350 (propagate_constants_for_unrolling): Propagate all constants.
18352 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18355 * config/i386/i386.c (print_reg): Remove assert for disalowed
18356 regno values, call output_operand_lossage instead.
18358 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18361 * config/i386/constraints.md (Yc): New register constraint.
18362 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18363 Use Yc constraint for alternative 2 of operand 0. Remove
18364 preferred_for_speed attribute.
18366 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18368 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18369 lastprivate clauses in SIMT case.
18371 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18373 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18375 2017-04-20 Richard Biener <rguenther@suse.de>
18377 PR tree-optimization/57796
18378 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18379 as N scalar stores.
18380 (vect_model_load_cost): Cost gathers as N scalar loads.
18382 2017-04-20 Richard Biener <rguenther@suse.de>
18384 * ggc-page.c (ggc_allocated_p): Rename to ...
18385 (safe_lookup_page_table_entry): ... this and return the lookup
18387 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18389 2017-04-20 Richard Biener <rguenther@suse.de>
18391 PR tree-optimization/80453
18392 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18393 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18394 from the conditions.
18395 (vn_phi_eq): Pass them down.
18396 (vn_phi_lookup): Record them.
18397 (vn_phi_insert): Likewise.
18399 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18401 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18402 uninitialized variable warning to avoid buffer overrun.
18404 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18407 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18408 is suppressed for '{ 0 }' in C.
18410 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18412 * BASE-VER: Set to 8.0.0.
18414 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18416 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18417 priority .init_array and .fini_array section with SECTION_NOTYPE
18420 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18422 PR middle-end/80423
18423 * tree.h (build_array_type): Add typeless_storage default argument.
18424 * tree.c (type_cache_hasher::equal): Also compare
18425 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18426 (build_array_type): Add typeless_storage argument, set
18427 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18429 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18430 (build_array_type): Likewise. Add typeless_storage argument.
18432 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18433 Jakub Jelinek <jakub@redhat.com>
18435 PR tree-optimization/80426
18436 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18437 operation on symbolic operands, also compute the overflow for the
18438 invariant part when the operation degenerates into a negation.
18440 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18443 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18444 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18447 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18449 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18452 * config/avr/avr.c (tree.h): Include it.
18453 (cgraph.h): Include it.
18454 (avr_encode_section_info): Don't warn for uninitialized progmem
18455 variable if it's just an alias.
18457 2017-04-19 Richard Biener <rguenther@suse.de>
18460 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18461 when needed by AutoPGO.
18463 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18466 * doc/lto.texi: Remove an extra 'that'.
18468 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18470 PR rtl-optimization/80429
18471 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18472 are only used in debug insns.
18474 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18475 Vladimir Makarov <vmakarov@redhat.com>
18477 * config/sparc/predicates.md (input_operand): Add comment. Return
18478 true for any memory operand when LRA is in progress.
18479 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18481 2017-04-18 Jeff Law <law@redhat.com>
18484 * mips.md ({return,simple_return}_internal): Do not overwrite
18487 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18489 PR tree-optimization/80443
18490 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18491 instead of adding 1, subtract -1 and similarly instead of subtracting
18494 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18496 PR rtl-optimization/80357
18497 * haifa-sched.c (tmp_bitmap): New variable.
18498 (model_recompute): Handle duplicate use records.
18499 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18500 (free_global_sched_pressure_data): Free it.
18502 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18505 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18506 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18507 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18508 instead of SYSTEM_HEADER_DIR.
18510 2017-04-18 Jeff Law <law@redhat.com>
18512 PR middle-end/80422
18513 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18514 predecessors after walking up the insn chain.
18516 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18519 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18520 sizetype type into debug info.
18522 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18525 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18526 unneeded test for TARGET_UPPER_REGS_SF.
18527 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18529 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18532 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18533 instead of gsi_after_labels.
18535 2017-04-18 Jeff Law <law@redhat.com>
18537 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18541 2017-04-13 Jeff Law <law@redhat.com>
18542 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18543 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18545 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18548 * config/avr/avr.c (intl.h): Include it.
18549 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18551 2017-04-18 Martin Liska <mliska@suse.cz>
18553 PR gcov-profile/78783
18554 * gcov-tool.c (gcov_output_files): Validate that destination
18555 file is either removed by the tool or by a user.
18557 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18558 Guy Benyei <guybe@mellanox.com>
18560 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18561 block, and do not negate it, the stored id is already negative.
18563 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18565 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18567 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18570 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18571 masks of options that should be turned off if the VSX vector
18572 options are turned off.
18573 (OTHER_P8_VECTOR_MASKS): Likewise.
18574 (OTHER_VSX_VECTOR_MASKS): Likewise.
18575 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18576 rs6000_disable_incompatible_switches to validate no type switches
18578 (rs6000_incompatible_switch): New function to disallow turning on
18579 other vector options if -mno-vsx, -mno-power8-vector, or
18580 -mno-power9-vector are specified.
18582 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18584 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18586 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18588 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18589 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18590 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18591 (ARG_POINTER_CFA_OFFSET): Likewise.
18593 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18595 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18596 conditions to take advantage of various optimizations.
18598 2017-04-13 Jeff Law <law@redhat.com>
18600 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18601 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18602 (zero_extendsidi2_dext): Likewise.
18604 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18607 * fold-const.c (fold_ternary_loc): Revert
18608 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18611 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18613 PR rtl-optimization/80343
18614 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18615 new scratch pseudo.
18617 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18620 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18621 to ubsan_encode_value.
18623 2017-04-13 Jeff Law <law@redhat.com>
18625 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18626 appearing in DEBUG_INSNs.
18628 2017-04-13 Martin Liska <mliska@suse.cz>
18630 PR gcov-profile/80413
18631 * gcov-io.c (gcov_write_string): Copy to buffer just when
18632 allocated size is greater than zero.
18634 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18637 * dwarf2out.c (decls_for_scope): Ignore declarations of
18638 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18640 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18643 * ipa-visibility.c (non_local_p): Fix typos.
18644 (localize_node): When localizing symbol in same comdat group,
18645 dissolve the group only when we know external symbols are going
18647 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18649 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18651 PR tree-optimization/79390
18652 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18653 order does not result in usable sequence, retry with reversed operand
18659 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18660 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18661 op0 instead of fold_convert_loc (loc, type, arg0).
18663 2017-04-12 Jeff Law <law@redhat.com>
18665 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18666 has a delay slot in the generated code.
18668 * config/cris/cris.md (cris_preferred_reload_class): Return
18669 GENNONACR_REGS rather than GENERAL_REGS.
18671 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18674 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18675 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18676 signedness of the result type.
18678 2017-04-12 Richard Biener <rguenther@suse.de>
18679 Jeff Law <law@redhat.com>
18681 PR tree-optimization/80359
18682 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18683 trim stores to TARGET_MEM_REFs.
18685 2017-04-12 Richard Biener <rguenther@suse.de>
18687 PR tree-optimization/79390
18688 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18689 threading case even more.
18691 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18694 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18695 for quad_address_p for TImode, instead of just not indexed_address.
18697 2017-04-12 Richard Biener <rguenther@suse.de>
18698 Bernd Edlinger <bernd.edlinger@hotmail.de>
18700 PR middle-end/79671
18701 * alias.c (component_uses_parent_alias_set_from): Handle
18702 TYPE_TYPELESS_STORAGE.
18703 (get_alias_set): Likewise.
18704 * tree-core.h (tree_type_common): Add typeless_storage flag.
18705 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18706 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18707 for types containing members with TYPE_TYPELESS_STORAGE.
18708 (place_field): Likewise.
18709 (layout_type): Likewise for ARRAY_TYPE.
18710 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18711 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18712 TYPE_TYPELESS_STORAGE.
18713 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18715 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18718 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18719 first argument to type.
18721 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18725 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18726 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18727 (rs6000_expand_binop_builtin): Likewise.
18728 (rs6000_expand_ternop_builtin): Likewise; also add missing
18729 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18730 vshasigma built-ins.
18731 * doc/extend.texi: Document that vec_xxpermdi's third argument
18732 must be a constant.
18734 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18736 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18737 Use shift_const cost parameter when calculating gain of STV shifts.
18739 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18741 PR rtl-optimization/70478
18742 * lra-constraints.c (process_alt_operands): Check memory for
18743 disfavoring memory insn operand.
18745 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18747 PR middle-end/80100
18748 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18749 left shift in unsigned HOST_WIDE_INT type.
18751 PR rtl-optimization/80385
18752 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18753 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18756 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18757 if they have any depend clauses.
18759 2017-04-11 Martin Liska <mliska@suse.cz>
18762 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18763 * ipa-split.c (split_function): Create a local comdat symbol
18764 if caller is in a comdat group.
18766 2017-04-11 Martin Liska <mliska@suse.cz>
18769 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18772 2017-04-11 Martin Sebor <msebor@redhat.com>
18774 PR middle-end/80364
18775 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18776 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18778 (directive::set_width, directive::set_precision, format_character):
18780 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18783 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18786 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18787 conflict, set target->arch_name instead of target->cpu_name.
18789 2017-04-11 Richard Biener <rguenther@suse.de>
18791 PR tree-optimization/80374
18792 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18793 build_zero_cst, remove fold_convertible_p check again.
18795 2017-04-11 Martin Liska <mliska@suse.cz>
18798 * ubsan.c (instrument_object_size): Do not instrument register
18801 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18804 * config/i386/i386-builtin-types.def
18805 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18806 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18807 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18808 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18809 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18810 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18811 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18812 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18813 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18814 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18815 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18816 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18817 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18818 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18819 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18820 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18821 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18822 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18823 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18824 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18825 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18826 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18827 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18828 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18829 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18830 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18831 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18832 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18833 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18834 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18835 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18836 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18837 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18838 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18839 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18840 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18841 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18842 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18843 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18844 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18845 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18846 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18847 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18848 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18849 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18850 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18852 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18853 flag to second_arg_count, handle 4 argument function type _COUNT
18854 aliases, handle second_arg_count on second argument rather than last.
18856 2017-04-10 Jeff Law <law@redhat.com>
18858 PR tree-optimization/80374
18859 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18860 record anything if we can not convert integer_zero_node to the
18863 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18866 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18867 Enhance special handling given to the TARGET_P9_MINMAX option in
18868 relation to certain other options.
18870 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18872 PR tree-optimization/80153
18873 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18874 remove POINTER_PLUS_EXPR's base part directly, rather than through
18877 2017-04-10 Richard Biener <rguenther@suse.de>
18878 Bin Cheng <bin.cheng@arm.com>
18880 PR tree-optimization/80153
18881 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18882 the first element of pointer type aff_tree. Build result expr in
18884 (add_elt_to_tree): Convert to type unconditionally. Remove other
18885 fold_convert calls.
18886 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18887 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18889 2017-04-10 Richard Biener <rguenther@suse.de>
18891 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18894 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18896 PR rtl-optimization/70478
18897 * lra-constraints.c (curr_small_class_check): New.
18898 (update_and_check_small_class_inputs): New.
18899 (process_alt_operands): Update curr_small_class_check. Disfavor
18900 alternative insn memory operands. Check available regs for small
18903 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18906 * config/mips/mips.opt (-mvirt): Update description.
18907 * doc/invoke.texi (-mvirt): Likewise.
18909 2017-04-10 Richard Biener <rguenther@suse.de>
18911 PR middle-end/80362
18912 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18913 looking for NEGATE_EXPR in -A / -B to A / B folding.
18915 2017-04-10 Martin Liska <mliska@suse.cz>
18917 PR gcov-profile/80224
18918 * gcov.c (print_usage): Fix usage string.
18919 (get_gcov_intermediate_filename): Remove.
18920 (output_gcov_file): Use both for normal and intermediate format.
18921 (generate_results): Do not initialize special file for
18922 intermediate format.
18924 2017-04-10 Richard Biener <rguenther@suse.de>
18926 PR tree-optimization/80304
18927 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18930 2017-04-10 Nathan Sidwell <nathan@acm.org>
18933 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18934 (rs6000_init_builtins): Use it.
18936 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18938 * config/arm/arm.md (<mrc>): Add mode to SET source.
18939 (<mrrc>): Likewise.
18941 2017-04-10 Richard Biener <rguenther@suse.de>
18943 PR middle-end/80344
18944 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18946 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18949 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18950 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18951 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18952 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18953 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18954 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18955 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18956 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18957 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18958 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18959 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18960 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18961 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18962 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18963 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18964 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18965 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18966 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18967 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18968 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18969 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18970 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18971 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18973 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18975 PR rtl-optimization/70478
18976 * lra-constraints.c: Reverse the last patch.
18978 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18980 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18981 Add comment for WCHAR_T.
18983 2017-04-08 Martin Liska <mliska@suse.cz>
18986 2017-04-07 Martin Liska <mliska@suse.cz>
18989 * ipa-split.c (split_function): Add function part to a same comdat
18992 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18995 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18997 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18999 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19000 (vec_pairing): ...to this.
19001 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19002 (rs6000_sched_init): Adjust for name change.
19003 (struct rs6000_sched_context): Likewise.
19004 (rs6000_init_sched_context): Likewise.
19005 (rs6000_set_sched_context): Likewise.
19007 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19013 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19015 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19016 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19017 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19019 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19021 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19023 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19025 PR rtl-optimization/70703
19026 * ira-color.c (update_conflict_hard_regno_costs): Use
19027 int64_t instead of HOST_WIDE_INT.
19029 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19031 PR rtl-optimization/70478
19032 * lra-constraints.c (process_alt_operands): Disfavor alternative
19033 insn memory operands.
19035 2017-04-07 Jeff Law <law@redhat.com>
19037 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19038 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19040 2017-04-07 Martin Liska <mliska@suse.cz>
19043 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19044 Show error message instead of an ICE.
19046 2017-04-07 Martin Liska <mliska@suse.cz>
19049 * ipa-split.c (split_function): Add function part to a same comdat
19052 2017-04-07 Richard Biener <rguenther@suse.de>
19054 PR middle-end/80341
19055 * tree.c (get_unwidened): Also handle ! for_type case for
19057 * convert.c (do_narrow): Split out from ...
19058 (convert_to_integer_1): ... here. Do not pass final truncation
19059 type to get_unwidened for TRUNC_DIV_EXPR.
19061 2017-04-07 Richard Biener <rguenther@suse.de>
19063 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19065 (aff_combination_const): Adjust.
19066 (aff_combination_scale): Likewise.
19067 (aff_combination_add_elt): Likewise.
19068 (aff_combination_add_cst): Likewise.
19069 (aff_combination_convert): Likewise.
19070 (add_elt_to_tree): Likewise. Remove unused argument.
19071 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19073 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19075 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19077 * config/arm/arm.c (arm_default_short_enums): Use
19078 ARM_DEFAULT_SHORT_ENUMS.
19079 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19081 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19084 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19085 members with redundant out-of-class redeclaration.
19087 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19090 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19091 * config/i386/i386.md (*zero_extendsidi2):
19092 Add (?*x,*x) and (?*v,*v) alternatives.
19094 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19097 * config/i386/i386.c (ix86_expand_builtin)
19098 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19099 mode from insn data. Convert operands to insn operand mode.
19100 Copy operands that don't satisfy insn predicate to a register.
19102 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19104 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19107 2017-04-06 Richard Biener <rguenther@suse.de>
19109 PR tree-optimization/80334
19110 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19111 preserve alignment of accesses.
19113 2017-04-06 Richard Biener <rguenther@suse.de>
19115 PR tree-optimization/80262
19116 * tree-sra.c (build_ref_for_offset): Preserve address-space
19118 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19119 Drop useless address-space information on MEM_REF offsets.
19121 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19123 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19125 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19127 PR rtl-optimization/70703
19128 * ira-color.c (update_conflict_hard_regno_costs): Use
19129 HOST_WIDE_INT instead of long.
19131 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19134 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19135 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19137 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19138 for x86_64 target. Handle -m3dnowa option.
19140 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19142 PR rtl-optimization/70703
19143 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19144 (update_conflict_hard_regno_costs): Use long instead of unsigned
19145 arithmetic for cost calculation.
19147 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19148 Bernd Edlinger <bernd.edlinger@hotmail.de>
19151 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19154 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19157 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19158 ptr_mode with Pmode throughout.
19159 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19160 into probe_stack_range and use DImode.
19162 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19165 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19166 call_eh_return is true.
19168 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19170 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19171 Initialize last_match_fntype_index.
19173 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19176 * tree-nvr.c: Include internal-fn.h.
19177 (pass_return_slot::execute): Ignore internal calls without
19180 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19181 Richard Biener <rguenther@suse.de>
19184 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19185 captures used multiple times, except for the last use.
19186 * generic-match-head.c: Include gimplify.h.
19188 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19190 PR tree-optimization/79390
19191 * target.h (struct noce_if_info): Declare.
19192 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19193 * target.def (noce_conversion_profitable_p): New target hook.
19194 * ifcvt.h (struct noce_if_info): New type, moved from ...
19195 * ifcvt.c (struct noce_if_info): ... here.
19196 (noce_conversion_profitable_p): Renamed to ...
19197 (default_noce_conversion_profitable_p): ... this. No longer
19199 (noce_try_store_flag_constants, noce_try_addcc,
19200 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19201 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19202 instead of noce_conversion_profitable_p.
19203 * config/i386/i386.c: Include ifcvt.h.
19204 (ix86_option_override_internal): Don't override
19205 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19206 (ix86_noce_conversion_profitable_p): New function.
19207 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19208 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19209 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19210 * doc/tm.texi: Regenerated.
19212 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19214 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19217 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19220 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19221 instructions for small multiply cores.
19223 2017-04-04 Jeff Law <law@redhat.com>
19225 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19227 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19228 that are not set by the loop over the elements.
19230 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19233 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19234 int mode, convert_modes it to mode as unsigned, otherwise use
19235 lowpart_subreg to mode rather than SImode.
19236 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19237 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19238 Use DImode instead of SImode for the shift count operand.
19239 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19242 2017-04-04 Richard Biener <rguenther@suse.de>
19244 PR middle-end/80281
19245 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19246 arithmetic done for the negate or the plus. Simplify.
19247 (A - (-B) -> A + B): Likewise.
19248 * fold-const.c (split_tree): Make sure to not negate pointers.
19250 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19252 PR rtl-optimization/60818
19253 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19254 a compare of comparisons with the thing compared if this results
19255 in a different machine mode.
19257 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19259 * alias.c (base_alias_check): Fix typo in comment.
19260 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19261 * cgraphunit.c (symbol_table::compile): Likewise.
19262 * collect2.c (maybe_run_lto_and_relink): Likewise.
19263 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19264 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19265 * config/avr/avr.c (avr_map_op_t): Likewise.
19266 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19267 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19268 * config/epiphany/epiphany.md (movcc): Likewise.
19269 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19270 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19272 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19273 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19274 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19275 * config/sh/sh.c (sh_rtx_costs): Likewise.
19276 * fold-const.c (fold_truth_andor): Likewise.
19277 * genautomata.c (collapse_flag): Likewise.
19278 * gengtype.h (struct type::u::s): Likewise.
19279 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19280 * input.c (FORMAT_AMOUNT): Likewise.
19281 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19282 (known_aggs_to_agg_replacement_list): Likewise.
19283 * ipa-inline-analysis.c: Likewise.
19284 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19285 * ipa-polymorphic-call.c
19286 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19287 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19288 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19290 * modulo-sched.c (apply_reg_moves): Likewise.
19291 * omp-expand.c (build_omp_regions_1): Likewise.
19292 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19293 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19294 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19295 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19296 * value-prof.c: Likewise.
19297 * var-tracking.c (val_reset): Likewise.
19299 2017-04-03 Richard Biener <rguenther@suse.de>
19301 PR tree-optimization/80275
19302 * fold-const.c (split_address_to_core_and_offset): Handle
19305 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19307 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19308 descriptors is at least equal to that of functions.
19310 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19312 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19314 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19317 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19318 (mov<IMOD4:mode>): New expander.
19319 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19321 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19323 PR rtl-optimization/79405
19324 * fwprop.c (propagations_left): New variable.
19325 (forward_propagate_into): Decrement it.
19326 (fwprop_init): Initialize it.
19327 (fw_prop): If the variable has reached zero, stop propagating.
19328 (fwprop_addr): Ditto.
19330 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19333 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19334 a FUNCTION_DECL, pass it as decl instead of origin to
19337 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19339 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19342 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19345 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19346 TARGET_VSX_SMALL_INTEGER.
19348 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19350 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19351 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19353 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19355 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19356 extraction from odd-numbered MSA register.
19358 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19360 PR middle-end/80173
19361 * expmed.c (store_bit_field_1): Don't attempt to create
19362 a word subreg out of hard registers wider than word if they
19363 have HARD_REGNO_NREGS of 1 for their mode.
19365 PR middle-end/80163
19366 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19367 conversions to integer types wider than word and pointer.
19370 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19371 (rtx_equal_for_cselib_p): Pass 0 to it.
19372 * cselib.c (cselib_hasher::equal): Likewise.
19373 (rtx_equal_for_cselib_1): Add depth argument. If depth
19374 is 128, don't look up VALUE locs and punt. Increment
19375 depth in recursive calls when walking VALUE locs.
19377 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19379 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19380 (make_gcov_file_name): Use the canonical path name for generating
19382 (read_line): Fix handling of files with ascii null bytes.
19384 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19386 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19387 to initialise a vector register instead
19388 of using a const_int.
19390 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19392 PR translation/80189
19393 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19394 diagnostic messages.
19396 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19399 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19400 (dfp_diex_<mode>): Update mode of operand 1.
19401 * doc/extend.texi (dxex, dxexq): Document change to return type.
19402 (diex, diexq): Document change to argument type.
19404 2017-03-30 Martin Jambor <mjambor@suse.cz>
19407 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19408 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19409 it reflects the signature changes performed at the callee side.
19410 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19411 to cgraph_build_function_type_skip_args.
19412 (build_function_decl_skip_args): Adjust call to the above function.
19414 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19417 * config/i386/sse.md
19418 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19419 register as dest whenever it is a MEM not rtx_equal_p to the
19420 corresponding dup operand, and when forcing into reg move the
19421 reg into the memory afterwards.
19422 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19423 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19424 for the force_reg mode.
19425 (avx512vl_vextractf128<mode>): Use register as dest either
19426 always when a MEM, or when it is a MEM not rtx_equal_p to the
19427 corresponding dup operand, or even not when it is a CONST_VECTOR
19428 depending on the mode and lo vs. hi.
19429 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19431 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19432 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19433 Likewise. Require that operands[2] is even.
19434 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19435 Remove extraneous parens. Require that operands[2] is a multiple
19437 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19438 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19439 disallow memory then.
19441 2017-03-30 Richard Biener <rguenther@suse.de>
19443 PR tree-optimization/77498
19444 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19445 to non-constants over backedges.
19447 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19449 PR rtl-optimization/80233
19450 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19451 as last_combined_insn. Do not test for BARRIER_P separately.
19453 2017-03-29 Andreas Schwab <schwab@suse.de>
19456 * calls.c (prepare_call_address): Convert funexp to Pmode before
19457 copying to temp reg.
19459 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19461 PR tree-optimization/80158
19462 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19463 Handle possible future case of more than one alternate
19465 (replace_rhs_if_not_dup): Likewise.
19466 (replace_one_candidate): Likewise.
19468 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19470 PR rtl-optimization/80193
19471 * ira.c (ira): Do not check allocation for LRA.
19473 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19475 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19476 (nvptx_output_simt_exit): Declare.
19477 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19478 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19479 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19480 (nvptx_declare_function_name): ...here. Emit declaration of local
19481 memory space buffer for omp_simt_enter insn.
19482 (nvptx_output_unisimt_switch): New.
19483 (nvptx_output_softstack_switch): New.
19484 (nvptx_output_simt_enter): New.
19485 (nvptx_output_simt_exit): New.
19486 * config/nvptx/nvptx.h (struct machine_function): New fields
19487 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19488 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19489 (UNSPECV_SIMT_EXIT): Ditto.
19490 (omp_simt_enter_insn): New insn.
19491 (omp_simt_enter): New expansion.
19492 (omp_simt_exit): New insn.
19493 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19495 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19496 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19497 (expand_GOMP_SIMT_EXIT): New.
19498 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19499 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19500 (GOMP_SIMT_EXIT): Ditto.
19501 * target-insns.def (omp_simt_enter): New insn.
19502 (omp_simt_exit): Ditto.
19503 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19505 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19506 (lower_rec_input_clauses): Likewise.
19507 (lower_lastprivate_clauses): Handle SIMT privatization.
19509 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19510 (ompdevlow_adjust_simt_enter): New.
19511 (find_simtpriv_var_op): New.
19512 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19513 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19515 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19516 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19517 (copy_decl_for_dup_finish): Ditto.
19519 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19521 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19524 * config/i386/i386.c (ix86_option_override_internal): Always
19525 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19527 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19529 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19531 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19533 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19534 mark new edge's irreducible flag accordign to it.
19535 (vect_do_peeling): Check loop preheader edge's irreducible flag
19536 and pass it to function slpeel_add_loop_guard.
19538 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19540 PR tree-optimization/80218
19541 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19542 Update block frequencies and counts.
19544 2017-03-28 Richard Biener <rguenther@suse.de>
19546 PR tree-optimization/78644
19547 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19548 of a simplification result we may not use it at all.
19550 2017-03-28 Richard Biener <rguenther@suse.de>
19553 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19554 without arguments, generate default definition of a SSA name.
19556 2017-03-28 Richard Biener <rguenther@suse.de>
19558 PR middle-end/80222
19559 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19560 TYPE_REF_CAN_ALIAS_ALL references.
19561 * fold-const.c (fold_indirect_ref_1): Likewise.
19563 2017-03-28 Martin Liska <mliska@suse.cz>
19566 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19567 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19569 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19570 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19572 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19573 (EXTRA_SPECS): Define.
19574 (SUBTARGET_EXTRA_SPECS): Likewise.
19575 (SUBTARGET_CPP_SPEC): Likewise.
19576 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19577 SUBTARGET_EXTRA_SPECS.
19578 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19580 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19582 * config/arc/simdext.md (vst64_insn): Update pattern.
19583 (vld32wh_insn): Likewise.
19584 (vld32wl_insn): Likewise.
19585 (vld64_insn): Likewise.
19586 (vld32_insn): Likewise.
19588 2017-03-28 Marek Polacek <polacek@redhat.com>
19591 * fold-const.c (fold_comparison): Use protected_set_expr_location
19592 instead of SET_EXPR_LOCATION.
19594 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19596 * tree.c (add_expr): Avoid name lookup warning.
19598 2017-03-27 Jeff Law <law@redhat.com>
19600 PR tree-optimization/80216
19601 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19602 function name. Limit recursion depth.
19603 (record_temporary_equivalences): Corresponding changes.
19605 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19607 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19610 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19613 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19615 * cfgcleanup.c (reg_note_cfa_p): New array.
19616 (insns_have_identical_cfa_notes): New function.
19617 (old_insns_match_p): Don't cross-jump in between /f
19618 and non-/f instructions. If both i1 and i2 are frame related,
19619 verify all CFA notes, their order and content.
19621 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19624 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19625 HImode and SImode with zero extend to DImode to one insn.
19626 (bswap<mode>2_extenddi): Likewise.
19627 (bswapsi2_extenddi): Likewise.
19628 (bswaphi2_extendsi): Likewise.
19629 (bswaphi2): Combine bswap HImode and SImode into one insn.
19630 Separate memory insns from swapping register.
19631 (bswapsi2): Likewise.
19632 (bswap<mode>2): Likewise.
19633 (bswaphi2_internal): Delete, no longer used.
19634 (bswapsi2_internal): Likewise.
19635 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19636 store, and gpr<-gpr swap insns.
19637 (bswap<mode>2_store): Likewise.
19638 (bswaphi2_reg): Register only splitter, combine with the splitter.
19639 (bswaphi2 splitter): Likewise.
19640 (bswapsi2_reg): Likewise.
19641 (bswapsi2 splitter): Likewise.
19642 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19643 the insns into load, store, and register/register insns.
19644 (bswapdi2_ldbrx): Likewise.
19645 (bswapdi2_load): Likewise.
19646 (bswapdi2_store): Likewise.
19647 (bswapdi2_reg): Likewise.
19649 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19651 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19652 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19654 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19657 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19659 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19660 special handling for target option conflicts between dform
19661 options (-mpower9-dform, -mpower9-dform-vector,
19662 -mpower9-dform-scalar) and -mno-direct-move.
19664 2017-03-27 Richard Biener <rguenther@suse.de>
19666 PR tree-optimization/80181
19667 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19669 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19671 * config/arc/predicates.md (move_double_src_operand): Replace the
19672 call to move_double_src_operand with a call to address_operand.
19674 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19676 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19677 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19678 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19680 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19682 * config/arc/predicates.md (long_immediate_loadstore_operand):
19683 Consider scaled addresses cases.
19685 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19687 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19688 restored when in interrupt.
19689 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19690 doesn't have delay slot.
19692 2017-03-27 Richard Biener <rguenther@suse.de>
19695 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19696 inlined thunk clones.
19698 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19701 * asan.c (instrument_derefs): Copy over last operand from
19702 original COMPONENT_REF to the new COMPONENT_REF with
19703 DECL_BIT_FIELD_REPRESENTATIVE.
19704 * ubsan.c (instrument_object_size): Likewise.
19706 2017-03-27 Richard Biener <rguenther@suse.de>
19708 PR tree-optimization/80170
19709 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19710 sure DR/SCEV didnt fold in constants we do not see when looking
19711 at the reference base alignment.
19713 2017-03-27 Richard Biener <rguenther@suse.de>
19715 PR middle-end/80171
19716 * gimple-fold.c (fold_ctor_reference): Properly guard against
19717 NULL return value from canonicalize_constructor_val.
19719 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19722 * config/i386/i386.c (ix86_expand_builtin)
19723 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19724 flags reg setting and flags reg using instructions.
19725 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19726 clobbering instructions to zero extend op2.
19728 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19730 * doc/install.texi (Configuration) <--with-aix-soname>:
19731 Update link to AIX ld.
19733 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19735 PR rtl-optimization/80160
19736 PR rtl-optimization/80159
19737 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19738 reg_alternate_class into account.
19740 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19743 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19744 to consider in curr_insn_transform.
19746 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19748 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19749 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19750 and emit_mode_inner.
19752 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19754 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19755 argument to the overloaded builtin variants. Use the new flag to
19756 deprecate certain builtin variants.
19757 * config/s390/s390-builtin-types.def: Add new builtin types.
19758 * config/s390/s390-builtins.h: Support new flags field for
19759 overloaded builtins.
19760 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19761 (s390_macro_to_expand): Enable vector float data type.
19762 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19763 builtins by incrementing the __VEC__ version number.
19764 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19766 (s390_resolve_overloaded_builtin): Emit error messages depending
19767 on the builtin flags.
19768 * config/s390/s390.c (s390_expand_builtin): Support additional
19769 flags argument. Change error message to match the messages
19770 emitted in s390-c.c.
19771 * config/s390/s390.md: New UNSPEC_* constants.
19772 (op_type): Add new instruction types.
19773 * config/s390/vecintrin.h: Add new builtins and test data class
19775 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19776 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19777 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19778 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19779 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19780 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19782 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19783 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19784 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19785 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19787 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19788 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19789 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19790 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19792 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19793 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19794 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19795 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19796 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19797 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19798 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19800 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19801 ("vec_scatter_element<V_HW_4:mode>_DI")
19802 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19803 ("vec_fpint<mode>", "vflls")
19804 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19805 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19806 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19807 ("*vec_cmphe<mode>_cc"): ... these.
19809 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19810 mode constant instead of magic value.
19812 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19814 * config/s390/s390.c (s390_expand_vec_compare): Support other
19815 vector floating point modes than just V2DF.
19816 (s390_expand_vcond): Likewise.
19817 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19818 (s390_cannot_change_mode_class): Prevent mode changes between TF
19819 and V1TF in vector registers.
19820 * config/s390/s390.md (DF, SF): New mode attributes.
19821 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19822 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19823 SFmode support for VRs.
19824 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19826 (VFT, VF_HW): New mode iterators.
19827 (vw, sdx): New mode attributes.
19828 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19829 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19830 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19831 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19832 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19833 also the new vector floating point modes. Renaming to ...
19835 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19836 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19837 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19838 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19839 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19840 ("vec_unordered<mode>"): ... these.
19842 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19843 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19844 ("*vec_extendv2df"): New insn definitions.
19846 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19848 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19849 ("mulditi3_2", "*muldi3_sign"): New patterns.
19850 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19851 rename the pattern definition.
19853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19855 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19857 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19859 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19861 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19862 instruction if possible.
19863 * config/s390/vector.md (vec_halfnumelts): New mode
19865 ("*vec_vllezlf<mode>"): New pattern.
19867 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19869 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19870 ("popcountv4si2", "popcountv2di2"): Rename to ...
19871 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19872 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19874 ("popcount<mode>2_vxe"): New pattern.
19876 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19878 * common/config/s390/s390-common.c (processor_flags_table): Add
19880 * config.gcc: Add arch12.
19881 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19882 Default to arch12 for unknown CPU model numbers.
19883 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19884 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19885 PROCESSOR_max sanity check.
19886 * config/s390/s390-opts.h (enum processor_type): Add
19888 * config/s390/s390.c (processor_table): Add arch12.
19889 (s390_expand_builtin): Add check for B_VXE flag.
19890 (s390_issue_rate): Add PROCESSOR_ARCH12.
19891 (s390_get_sched_attrmask): Likewise.
19892 (s390_get_unit_mask): Likewise.
19893 (s390_sched_score): Enable z13 scheduling for arch12.
19894 (s390_sched_reorder): Likewise.
19895 (s390_sched_variable_issue): Likewise.
19896 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19898 (s390_tune_attr): Use z13 scheduling also for arch12.
19899 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19900 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19901 (TARGET_VXE_P): New macros.
19902 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19903 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19904 * config/s390/s390.opt: Add arch12 as processor_type.
19906 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19908 * config/s390/s390.md
19909 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19910 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19911 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19913 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19914 Rename expanders to ...
19916 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19917 ("fixuns_truncdddi2_emu"): ... these.
19919 ("fixuns_trunc<mode>si2_emu"): New expander.
19921 ("*fixuns_truncdfdi2_z13"): Rename to ...
19922 ("*fixuns_truncdfdi2_vx"): ... this.
19924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19926 * config/s390/2964.md: Remove the single element vector compare
19927 instructions which are no longer used.
19928 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19930 (s390_canonicalize_comparison): Remove handling of DFmode
19932 (s390_expand_vec_compare_scalar): Remove function.
19933 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19934 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19936 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19938 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19940 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19942 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19943 will anyway by matched by mov<mode>_64dfp.
19945 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19947 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19948 vlef/vstef. Add missing operand to vleif.
19950 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19952 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19953 pair for all vector types with 64 bit elements.
19954 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19955 * config/s390/vector.md (V_HW_64): ... here.
19956 (V_128_NOSINGLE): New mode iterator.
19957 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19958 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19959 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19960 ("*vec_load_pairv2di"): Change to ...
19961 ("*vec_load_pair<mode>"): ... this one.
19963 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19965 * config/s390/constraints.md: Add comments.
19966 (jKK): Reject element sizes > 8 bytes.
19967 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19969 * config/s390/s390.md: Add the s_operand checks formerly in
19970 s390_split_ok_p to various splitters where they are still
19972 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19973 for 128 bit vectors. Plus two splitters.
19975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19977 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19980 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19983 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19984 error if the boundary argument is not constant.
19986 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19988 PR rtl-optimization/80112
19989 * loop-doloop.c (doloop_condition_get): Don't check condition
19990 if cmp isn't SET with IF_THEN_ELSE src.
19992 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19994 PR tree-optimization/80158
19995 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19996 replacing a candidate statement, also replace it for the
19997 candidate's alternate interpretation.
19998 (replace_rhs_if_not_dup): Likewise.
19999 (replace_one_candidate): Likewise.
20001 2017-03-24 Richard Biener <rguenther@suse.de>
20003 PR tree-optimization/80167
20004 * graphite-isl-ast-to-gimple.c
20005 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20007 (translate_isl_ast_to_gimple::get_rename): Likewise.
20009 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20011 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20012 handling of certain combinations of target options, including the
20013 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20014 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20016 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20019 * config/arm/arm.md (*load_multiple): Add reload_completed to
20020 matching condition.
20022 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20023 Richard Biener <rguenth@suse.de>
20025 PR tree-optimization/79908
20026 PR tree-optimization/80136
20027 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20028 been cast away, gimplify_and_add suffices.
20030 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20032 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20034 2017-03-23 Richard Biener <rguenther@suse.de>
20036 PR tree-optimization/80032
20037 * gimplify.c (gimple_push_cleanup): Forced unconditional
20038 cleanups still have to go to the conditional_cleanups
20041 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20043 PR tree-optimization/80072
20044 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20046 (next_operand_entry_id): Change type to unsigned int.
20047 (sort_by_operand_rank): Make sure to return the right return value
20048 even if unsigned fields are bigger than INT_MAX.
20049 (struct oecount): Change cnt and id type to unsigned int.
20050 (oecount_hasher::equal): Formatting fix.
20051 (oecount_cmp): Make sure to return the right return value
20052 even if unsigned fields are bigger than INT_MAX.
20053 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20056 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20057 TREE_READONLY on result if writing it more than once.
20060 * doc/invoke.texi (-fsanitize=thread): Document that with
20061 -fnon-call-exceptions atomics are not able to throw
20065 * tsan.c: Include tree-eh.h.
20066 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20067 maybe_clean_or_replace_eh_stmt where needed.
20068 (instrument_memory_accesses): Add cfg_changed argument.
20069 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20070 if it returned true.
20071 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20073 PR rtl-optimization/63191
20074 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20075 wrapper function, moved the whole old content into ...
20076 (ix86_delegitimize_address_1): ... this. New inline function.
20077 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20078 true as last argument instead of ix86_delegitimize_address.
20080 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20082 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20083 cortexa57_branch_cost.
20085 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20087 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20089 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20092 * doc/md.texi (Constraints): Document wA constraint.
20093 * config/rs6000/constraints.md (wA): New.
20094 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20095 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20096 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20097 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20099 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20102 * gimplify.c (is_oacc_declared): New function.
20103 (oacc_default_clause): Use it to set default flags for acc declared
20104 variables inside parallel regions.
20105 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20106 declared variables.
20107 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20108 declare attribute to any decl as necessary.
20110 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20113 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20114 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20115 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20116 (arm_arch_lpae): This.
20117 * config/arm/arm.c (arm_arch7ve): Rename into ...
20118 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20119 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20122 2017-03-22 Martin Liska <mliska@suse.cz>
20125 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20126 error message instead of an ICE.
20128 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20130 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20132 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20134 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20136 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20139 2017-03-21 Martin Sebor <msebor@redhat.com>
20141 * doc/extend.texi: Use "cannot" instead of "can't."
20142 * doc/hostconfig.texi: Same.
20143 * doc/install.texi: Same.
20144 * doc/invoke.texi: Same.
20145 * doc/loop.texi: Same.
20146 * doc/md.texi: Same.
20147 * doc/objc.texi: Same.
20148 * doc/rtl.texi: Same.
20149 * doc/tm.texi: Same.
20150 * doc/tm.texi.in: Same.
20151 * doc/trouble.texi: Same.
20153 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20156 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20157 (collect_checksum_attributes): Set it.
20158 (die_checksum_ordered): Use it.
20160 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20162 PR tree-optimization/79908
20163 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20164 change: For a VA_ARG whose LHS has been cast away, use
20165 force_gimple_operand to construct the side effects.
20167 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20169 PR translation/80001
20170 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20171 more amenable to translation.
20172 (oacc_loop_auto_partitions): Likewise.
20174 2017-03-21 Marek Polacek <polacek@redhat.com>
20175 Martin Sebor <msebor@redhat.com>
20177 PR tree-optimization/80109
20178 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20179 on INTEGRAL_TYPE_P.
20181 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20182 Segher Boessenkool <segher@kernel.crashing.org>
20185 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20186 check reg_used_between_p between insn and one of succ or succ2
20187 depending on if succ is artificial insn not inserted into insn
20190 2017-03-21 Martin Liska <mliska@suse.cz>
20192 PR gcov-profile/80081
20193 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20194 * doc/gcc.texi: Include gcov-dump stuff.
20195 * doc/gcov-dump.texi: New file.
20197 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20199 PR rtl-optimization/79150
20200 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20201 conditional jump, if the jump is the last insn of the loop.
20203 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20204 Richard Biener <rguenth@suse.de>
20206 PR tree-optimization/79908
20207 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20208 been cast away, use force_gimple_operand to construct the side
20211 2017-03-21 Martin Liska <mliska@suse.cz>
20213 PR libfortran/79956
20214 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20217 2017-03-21 Brad Spengler <spender@grsecurity.net>
20220 * plugin.c (htab_hash_plugin): New function.
20221 (add_new_plugin): Use it and adjust.
20222 (parse_plugin_arg_opt): Adjust.
20223 (init_one_plugin): Likewise.
20225 2017-03-21 Richard Biener <rguenther@suse.de>
20227 PR tree-optimization/80032
20228 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20229 if set force the cleanup to happen unconditionally.
20230 (gimplify_target_expr): Push inserted clobbers with force_uncond
20231 to avoid them being removed by control-dependent DCE.
20233 2017-03-21 Richard Biener <rguenther@suse.de>
20235 PR tree-optimization/80122
20236 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20237 va_arg_pack_len when the inlined call stmt requires pack
20239 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20241 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20244 * tsan.c (instrument_builtin_call): If the memory model argument
20245 is not a constant, assume it is valid.
20248 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20251 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20253 PR rtl-optimization/79910
20254 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20255 if its dest is used by an insn before I2 (other than the combined
20256 insns themselves, which are properly handled already).
20258 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20261 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20263 * combine.c (record_used_regs): New static function.
20264 (try_combine): Handle situations where there is an additional
20265 instruction between I2 and I3 which needs to have a LOG_LINK
20269 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20271 * combine.c (try_combine): Delete redundant i1 test. Call
20272 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20274 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20277 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20278 alternatives 13/14.
20280 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20282 PR tree-optimization/80054
20283 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20284 the optimization if a PHI or any of its arguments is not dominated
20285 by the candidate's basis. Use gphi* rather than gimple* as
20287 (replace_profitable_candidates): Clean up a gimple* variable that
20288 should be a gphi* variable.
20290 2017-03-20 Martin Sebor <msebor@redhat.com>
20293 * doc/extend.texi (attribute constructor): Document present limitation.
20295 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20298 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20299 __POWER9_VECTOR__ #ifdef control, change template definition to
20300 use Power9-specific built-in function.
20301 (vec_any_eq): Likewise.
20302 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20303 to control outcomes from this test.
20304 (vector_ae_<mode>p): For VEC_F modes, likewise.
20306 2017-03-20 Ian Lance Taylor <iant@google.com>
20308 * config/i386/i386.c (ix86_function_regparm): Save an extra
20309 register for -fsplit-stack with DECL_STATIC_CHAIN.
20311 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20314 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20315 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20317 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20319 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20321 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20324 2017-03-20 Marek Polacek <polacek@redhat.com>
20327 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20329 2017-03-20 Richard Biener <rguenther@suse.de>
20331 PR tree-optimization/80113
20332 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20333 allocate extra SSA name for PHI def.
20334 (add_close_phis_to_outer_loops): Likewise.
20335 (add_close_phis_to_merge_points): Likewise.
20336 (copy_loop_close_phi_args): Likewise.
20337 (copy_cond_phi_nodes): Likewise.
20339 2017-03-20 Martin Liska <mliska@suse.cz>
20341 PR middle-end/79753
20342 * tree-chkp.c (chkp_build_returned_bound): Do not build
20343 returned bounds for a LHS that's not a BOUNDED_P type.
20345 2017-03-20 Martin Liska <mliska@suse.cz>
20349 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20350 COMPLEX_CST and VECTOR_CST.
20352 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20355 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20356 target operand. A new splitter adds the clobber statement in case
20357 the target operand is dead anyway.
20359 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20361 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20362 to age-old versions of binutils and glibc.
20364 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20366 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20368 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20370 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20372 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20374 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20375 requirement for binutils 2.13.
20377 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20379 * combine.c (try_combine): Delete redundant i1 test. Call
20380 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20382 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20384 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20385 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20387 <riscv64-*-elf>: Re-arrange section
20388 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20389 <riscv32-*-linux>: Likewise.
20390 <riscv64-*-elf>: Likewise
20391 <riscv64-*-linux>: Likewise.
20393 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20396 * aarch64.opt(verbose-cost-dump): Fix typo.
20398 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20401 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20402 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20404 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20406 * reload.c (find_reloads): When reloading a nonoffsettable address,
20407 use RELOAD_OTHER for it and its address reloads.
20409 PR rtl-optimization/79910
20410 * combine.c (record_used_regs): New static function.
20411 (try_combine): Handle situations where there is an additional
20412 instruction between I2 and I3 which needs to have a LOG_LINK
20415 2017-03-17 Jeff Law <law@redhat.com>
20417 PR tree-optimization/71437
20418 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20419 conditional in the hash table first.
20420 (vrp_dom_walker::before_dom_children): Extract condition from
20421 ASSERT_EXPR. Record condition, its inverion and any implied
20422 conditions as well.
20424 2017-03-17 Marek Polacek <polacek@redhat.com>
20425 Markus Trippelsdorf <markus@trippelsdorf.de>
20427 PR tree-optimization/80079
20428 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20431 2017-03-17 Richard Biener <rguenther@suse.de>
20433 PR middle-end/80075
20434 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20435 Properly verify the LHS before the RHS possibly claims to be
20437 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20440 2017-03-17 Martin Jambor <mjambor@suse.cz>
20442 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20443 (List of -O2 options): Likewise.
20444 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20447 2017-03-17 Tom de Vries <tom@codesourcery.com>
20449 * gcov-dump.c (print_usage): Print bug_report_url.
20451 2017-03-17 Richard Biener <rguenther@suse.de>
20453 PR middle-end/80050
20454 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20455 (parser::peek): Likewise.
20457 2017-03-17 Richard Biener <rguenther@suse.de>
20459 PR tree-optimization/80048
20460 * sese.c (free_sese_info): Properly release rename_map and
20461 copied_bb_map elements.
20463 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20465 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20466 Add linked-list forward and backlinks. Insert on
20467 construction, remove on destruction.
20468 (class pass_store_merging): Add m_stores_head field.
20469 (pass_store_merging::terminate_and_process_all_chains):
20470 Iterate over m_stores_head list.
20471 (pass_store_merging::terminate_all_aliasing_chains):
20473 (pass_store_merging::execute): Check for debug stmts first.
20474 Push new chains onto the m_stores_head stack.
20476 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20479 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20480 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20481 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20483 2017-03-16 Jeff Law <law@redhat.com>
20485 PR tree-optimization/71437
20486 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20487 member function. Implementation moved into after_dom_children
20488 member function and into the threader's thread_outgoing_edges
20490 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20491 some code into new thread_outgoing_edges.
20492 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20493 definition. Simplify marker handling (do it here). Assume we always
20494 have the available expression and the const/copies tables.
20495 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20497 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20498 * tree-vrp.c (equiv_stack): No longer file scoped.
20499 (vrp_dom_walker): New class.
20500 (vrp_dom_walker::before_dom_children): New member function.
20501 (vrp_dom_walker::after_dom_children): Likewise.
20502 (identify_jump_threads): Setup domwalker. Use it rather than
20503 walking edges in a random order by hand. Simplify setup/finalization.
20504 (finalize_jump_threads): Remove.
20505 (vrp_finalize): Do not call identify_jump_threads here.
20506 (execute_vrp): Do it here instead and call thread_through_all_blocks
20509 PR tree-optimization/71437
20510 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20512 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20514 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20515 (dom_opt_dom_walker::thread_across_edge): Remove
20516 handle_dominating_asserts argument. All callers changed.
20517 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20518 changes. Remove calls to lhs_of_dominating_assert. Other
20519 uses of handle_dominating_asserts turn into unconditional code
20520 (simplify_control_stmt_condition_1): Likewise.
20521 (simplify_control_stmt_condition): Likewise.
20522 (thread_through_normal_block, thread_across_edge): Likewise.
20523 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20524 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20525 object if it is not an SSA_NAME.
20526 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20527 before calling into the VRP specific simplifiers.
20528 (identify_jump_threads): Remove handle_dominating_asserts
20531 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20534 * tree-diagnostic.c (default_tree_printer): No longer static.
20535 * tree-diagnostic.h (default_tree_printer): New prototype.
20537 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20539 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20540 Change ins into fmov.
20542 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20544 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20545 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20546 Use h_con constraint for operand 1.
20547 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20548 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20550 2017-03-15 Jeff Law <law@redhat.com>
20552 PR tree-optimization/71437
20553 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20554 (record_temporary_equivalences): Use it.
20556 PR tree-optimization/71437
20557 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20558 tree-ssa-scopedtables.
20559 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20560 (record_conditions, record_cond, vuse_eq): Likewise.
20561 (record_edge_info): Adjust to API tweak of record_conditions.
20562 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20563 (record_temporary_equivalences, optimize_stmt): Likewise.
20564 (eliminate_redundant_computations): Likewise.
20565 (record_equivalences_from_stmt): Likewise.
20566 * tree-ssa-scopedtables.c: Include options.h and params.h.
20567 (vuse_eq): New function, moved from tree-ssa-dom.c
20568 (build_and_record_new_cond): Likewise.
20569 (record_conditions): Likewise. Accept vector of conditions rather
20570 than edge_equivalence structure for first argument.
20571 for the first argument.
20572 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20573 from tree-ssa-dom.c.
20574 (avail_exprs_stack::record_cond): Likewise.
20575 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20576 from tree-ssa-dom.c.
20577 (avail_exprs_stack): Add new member functions lookup_avail_expr
20579 (record_conditions): Declare.
20581 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20584 * lra-constraints.c (process_alt_operands): Increase reject for
20585 reloading an input/output operand.
20587 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20590 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20591 insns to convert from signed/unsigned char/short to IEEE 128-bit
20593 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20595 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20598 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20599 subreg of inner mode for values already in registers.
20601 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20603 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20604 iteration reg is used after the loop.
20606 2017-03-14 Martin Sebor <msebor@redhat.com>
20608 PR tree-optimization/79800
20609 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20610 precision in negative-positive range.
20611 (format_floating): Call non-const overload with adjusted precision.
20613 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20616 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20619 2017-03-14 Martin Sebor <msebor@redhat.com>
20621 PR middle-end/80020
20622 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20623 * builtins.def (aligned_alloc): Use it.
20626 * Makefile.in (GTFILES): Add calls.c.
20627 * calls.c: Include "gt-calls.h".
20629 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20631 PR rtl-optimization/79728
20632 * regs.h (struct target_regs): New field
20633 x_contains_allocatable_regs_of_mode.
20634 (contains_allocatable_regs_of_mode): New macro.
20635 * reginfo.c (init_reg_sets_1): Initialize it, and change
20636 contains_reg_of_mode so it includes global regs as well.
20637 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20638 rather than contains_regs_of_mode.
20640 2017-03-14 Martin Liska <mliska@suse.cz>
20642 * doc/invoke.texi: Document options that can't be combined with
20643 -fcheck-pointer-bounds.
20645 2017-03-14 Martin Liska <mliska@suse.cz>
20647 PR middle-end/79831
20648 * doc/invoke.texi (-Wchkp): Document the option.
20650 2017-03-14 Martin Liska <mliska@suse.cz>
20652 * Makefile.in: Install gcov-dump.
20654 2017-03-14 Martin Liska <mliska@suse.cz>
20656 * multiple_target.c (expand_target_clones): Bail out for
20657 an invalid attribute.
20659 2017-03-14 Richard Biener <rguenther@suse.de>
20661 * alias.c (struct alias_set_entry): Pack properly.
20662 * cfgloop.h (struct loop): Likewise.
20663 * cse.c (struct set): Likewise.
20664 * ipa-utils.c (struct searchc_env): Likewise.
20665 * loop-invariant.c (struct invariant): Likewise.
20666 * lra-remat.c (struct cand): Likewise.
20667 * recog.c (struct change_t): Likewise.
20668 * rtl.h (struct address_info): Likewise.
20669 * symbol-summary.h (function_summary): Likewise.
20670 * tree-loop-distribution.c (struct partition): Likewise.
20671 * tree-object-size.c (struct object_size_info): Likewise.
20672 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20673 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20674 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20675 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20676 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20677 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20678 (struct _stmt_vec_info): Likewise.
20680 2017-03-14 Martin Liska <mliska@suse.cz>
20683 * multiple_target.c (create_dispatcher_calls): Check that
20684 a target can create a function dispatcher.
20686 2017-03-14 Martin Liska <mliska@suse.cz>
20689 * multiple_target.c (expand_target_clones): Drop local.local
20690 flag for default implementation.
20692 2017-03-14 Richard Biener <rguenther@suse.de>
20694 PR tree-optimization/80030
20695 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20697 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20699 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20700 gcc_fallthrough() instead of __attribute__((fallthrough));
20702 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20704 * doc/gcc.texi: Remove "up" link to (DIR).
20705 * doc/gccint.texi: Ditto.
20707 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20709 * doc/install.texi (Specific) <avr>: Remove reference to
20712 2017-03-13 Jeff Law <law@redhat.com>
20714 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20715 attribute rather than comments.
20717 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20718 match_scratch operand is highest.
20720 2017-03-13 Martin Liska <mliska@suse.cz>
20722 PR middle-end/78339
20723 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20724 is a CHKP clone, use original declaration.
20726 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20728 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20729 (arc_conditional_register_usage): Use a different allocation order
20730 when optimizing for size.
20731 * common/config/arc/arc-common.c (arc_option_optimization_table):
20732 Section anchors default on when optimizing for size.
20734 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20736 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20738 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20740 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20741 * config/arc/arc.md (cpu_facility): Add cd variant.
20742 (*movqi_insn): Add code density variant.
20743 (*movhi_insn): Likewise.
20744 (*movqi_insn): Likewise.
20745 (*addsi3_mixed): Likewise.
20746 (subsi3_insn): Likewise.
20748 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20750 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20752 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20754 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20755 expressions with MINUS and UNARY ops.
20757 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20760 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20762 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20763 between vec_select and vector argument.
20764 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20765 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20766 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20767 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20768 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20769 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20771 2017-03-13 Richard Biener <rguenther@suse.de>
20774 * params.def (vect-max-peeling-for-alignment): Fix typo.
20776 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20778 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20779 issue that only occurred with binutils below 2.18.
20781 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20783 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20784 refer to binutils 2.11/2.12 minimum.
20786 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20788 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20789 ftp.kernel.org and simplify binutils requirement.
20791 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20793 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20795 (Optimize Options): Ditto. Also remove redundancy.
20797 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20799 PR translation/79848
20800 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20802 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20803 to G_ to avoid double translation.
20805 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20807 PR translation/79923
20808 * auto-profile.c (get_combined_location): Convert leading
20809 character of diagnostics to lower case and remove trailing period.
20810 (read_profile): Likewise for various diagnostics.
20811 * config/arm/arm.c (arm_option_override): Remove trailing period
20812 from various diagnostics.
20813 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20814 (msp430_expand_delay_cycles): Likewise.
20816 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20819 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20820 full command-line argument, rather than just "str".
20821 (aarch64_validate_march): Likewise.
20822 (aarch64_validate_mtune): Likewise.
20824 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20826 PR rtl-optimization/78911
20827 * lra-assigns.c (must_not_spill_p): New function.
20828 (spill_for): Use it.
20830 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20832 PR tree-optimization/79981
20833 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20834 ATOMIC_COMPARE_EXCHANGE ifn result.
20835 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20836 IFN_ATOMIC_COMPARE_EXCHANGE.
20838 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20841 * opts.c (parse_sanitizer_options): Add missing question mark to
20842 "did you mean" message.
20844 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20846 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20848 (VMULEUH_UNS): Likewise.
20849 (VMULOUB_UNS): Likewise.
20850 (VMULOUH_UNS): Likewise.
20851 * config/rs6000/rs6000.c (builtin_function_type): Remove
20852 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20854 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20857 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20858 x with result of extra_parsing_for_operand_code_0.
20859 (function_reader::extra_parsing_for_operand_code_0): Convert
20860 return type from void to rtx, returning x. When reading
20861 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20862 larger size containing struct block_symbol.
20864 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20866 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20867 -mfloat128-hardware without -m64.
20869 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20872 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20873 entries to the case statement that marks unsigned arguments to
20874 overloaded functions.
20876 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20879 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20881 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20884 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20885 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20887 2017-03-10 Martin Liska <mliska@suse.cz>
20891 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20892 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20895 2017-03-10 Olivier Hainque <hainque@adacore.com>
20897 * tree-switch-conversion (array_value_type): Start by resetting
20898 candidate type to it's main variant.
20900 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20902 PR rtl-optimization/79909
20903 * combine.c (try_combine): Use simplify_replace_rtx on individual
20904 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20905 of the whole CALL_INSN_FUNCTION_USAGE.
20907 PR tree-optimization/79972
20908 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20909 get_range_info on SSA_NAMEs. Formatting fixes.
20911 2017-03-10 Richard Biener <rguenther@suse.de>
20912 Jakub Jelinek <jakub@redhat.com>
20914 PR tree-optimization/77975
20915 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20916 edge to be constant.
20917 (get_val_for): For constant x return it. Formatting fix.
20918 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20919 would use the same bases as the current one.
20921 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20923 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20924 instead of vec_select for V1TImode.
20925 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20927 (VSX_LE_128): Add V1TI to this mode iterator.
20928 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20929 (*vsx_le_perm_store_<mode>): Likewise.
20930 (pre-reload splitter for VSX stores): Likewise.
20931 (post-reload splitter for VSX stores): Likewise.
20932 (*vsx_xxpermdi2_le_<mode>): Likewise.
20933 (*vsx_lxvd2x2_le_<mode>): Likewise.
20934 (*vsx_stxvd2x2_le_<mode>): Likewise.
20936 2017-03-09 Michael Eager <eager@eagercon.com>
20938 Correct failures with --enable-checking=yes,rtl.
20940 * config/microblaze/microblaze.c (microblaze_expand_shift):
20941 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20942 test for const0_rtx.
20943 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20944 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20946 2017-03-09 Richard Biener <rguenther@suse.de>
20948 PR tree-optimization/79977
20949 * graphite-scop-detection.c (scop_detection::merge_sese):
20950 Handle the case of extra exits to blocks dominating the entry.
20952 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20954 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20957 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20959 PR rtl-optimization/79949
20960 * lra-constraints.c (process_alt_operands): Check memory when
20961 trying to predict a cycle. Print about the overall increase.
20963 2017-03-09 Richard Biener <rguenther@suse.de>
20965 PR middle-end/79971
20966 * gimple-expr.c (useless_type_conversion_p): Preserve
20967 TYPE_SATURATING for fixed-point types.
20969 2017-03-09 Richard Biener <rguenther@suse.de>
20972 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20973 alignment of BLKmode params.
20975 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20978 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20979 (VALL_NO_V2Q): Likewise.
20981 * config/aarch64/aarch64-simd.md
20982 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20984 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20985 VALL_NO_V2Q mode iterator.
20986 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20988 2017-03-09 Martin Liska <mliska@suse.cz>
20990 PR tree-optimization/79631
20991 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20992 tree_int_cst_sign_bit just for INTEGER constants.
20994 2017-03-09 Martin Liska <mliska@suse.cz>
20998 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21001 2017-03-09 Marek Polacek <polacek@redhat.com>
21004 * tree.c (inchash::add_expr): Handle TREE_VEC.
21006 2017-03-09 Martin Liska <mliska@suse.cz>
21009 (chkp_narrow_size_and_offset): New function.
21010 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21011 (void chkp_parse_bit_field_ref): New function.
21012 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21013 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21015 2017-03-09 Martin Liska <mliska@suse.cz>
21018 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21019 (chkp_find_bounds_1): Remove gcc_unreachable.
21021 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21024 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21025 BUILT_IN_SYNC*, determine the access type from the size suffix and
21026 always build a MEM_REF with that type. Handle forgotten
21027 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21030 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21031 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21032 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21033 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21034 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21035 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21036 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21037 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21038 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21039 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21040 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21041 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21042 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21043 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21044 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21045 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21046 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21047 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21048 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21049 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21050 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21051 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21052 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21053 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21054 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21055 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21056 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21057 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21058 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21059 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21060 definitions outside of __OPTIMIZE__ guarded section.
21063 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21064 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21065 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21066 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21069 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21071 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21072 ("vfenez<mode>"): Add missing constraints.
21074 2017-03-08 Martin Sebor <msebor@redhat.com>
21077 * config/nds32/nds32.c (nds32_option_override):
21078 Fix misspelled diagnostic.
21080 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21083 * gimplify.c (gimplify_omp_for): Replace index var in outer
21084 taskloop statement with an artificial variable and add
21085 OMP_CLAUSE_PRIVATE clause for it.
21087 2017-03-08 Richard Biener <rguenther@suse.de>
21089 PR tree-optimization/79955
21090 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21091 for accesses that are completely outside of the variable.
21093 2017-03-08 Andrew Haley <aph@redhat.com>
21095 PR tree-optimization/79943
21096 * tree-ssa-loop-split.c (compute_new_first_bound): When
21097 calculating the new upper bound, (END-BEG) should be added, not
21100 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21102 * config/avr/avr.md (setmemhi): Make sure match_dup
21103 operand number comes before match_scratch.
21105 2017-03-08 Richard Biener <rguenther@suse.de>
21107 PR tree-optimization/79920
21108 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21109 with ncopies == 1 to ...
21110 (vect_transform_slp_perm_load): ... here. Properly compute
21111 all element loads by iterating VF times over the group. Do
21112 not handle ncopies (computed in a broken way) in
21113 vect_create_mask_and_perm.
21115 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21118 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21119 is a uniform vector, use uniform_vector_p return value instead of
21120 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21122 2017-03-07 Marek Polacek <polacek@redhat.com>
21124 PR middle-end/79809
21125 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21126 (alloca_call_type): Likewise.
21128 2017-03-07 Martin Liska <mliska@suse.cz>
21130 * gcov.c (process_args): Put comment to correct location.
21132 2017-03-07 Martin Liska <mliska@suse.cz>
21134 PR middle-end/68270
21135 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21136 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21137 (chkp_narrow_bounds_for_field): Likewise.
21138 (chkp_parse_array_and_component_ref): Pass one more argument to
21141 2017-03-07 Richard Biener <rguenther@suse.de>
21143 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21146 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21148 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21149 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21151 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21154 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21155 to end of description.
21156 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21158 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21160 PR rtl-optimization/79901
21161 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21163 (*avx512f_<code><mode>3<mask_name>): ... this.
21164 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21165 iterator instead of VI8_AVX2_AVX512BW.
21167 PR rtl-optimization/79901
21168 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21169 min/max expander, expand it using expand_vec_cond_expr.
21172 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21175 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21178 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21179 to void * for PCH reasons.
21180 * dwarf2out.c (output_loc_operands, output_die): Cast
21181 v.val_vec.array to unsigned char *.
21183 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21186 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21189 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21191 PR rtl-optimization/79571
21192 * lra-constraints.c (process_alt_operands): Calculate static
21193 reject and subtract it from overall when only addresses will be
21196 2017-03-06 Julia Koval <julia.koval@intel.com>
21199 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21200 incoming stack boundary to 128 for 64-bit targets.
21202 2017-03-06 Richard Biener <rguenther@suse.de>
21204 PR tree-optimization/79894
21205 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21206 to NULL after folding it.
21208 2017-03-06 Richard Biener <rguenther@suse.de>
21210 PR tree-optimization/79824
21211 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21212 check disabling peeling for gaps.
21214 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21216 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21217 attributes): Document gettimeofday.
21219 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21221 * config/s390/s390.c (s390_option_override_internal): Set
21222 PARAM_MIN_VECT_LOOP_BOUND
21224 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21226 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21227 * config/s390/s390.md: Likewise.
21229 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21232 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21233 (<avx2_avx512>_perm<mode>): Rename to ...
21234 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21236 (<avx512>_perm<mode>_mask): Rename to ...
21237 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21239 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21240 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21241 instead of VI8F_256_512.
21242 (avx512f_perm<mode>): New define_expand.
21243 (avx512f_perm<mode>_mask): Likewise.
21244 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21245 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21247 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21249 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21250 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21252 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21254 2017-03-06 Martin Liska <mliska@suse.cz>
21257 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21258 when having a SSA NAME w/o VAR_DECL assigned to it.
21260 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21262 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21263 msa_dpsub_<su>_d): Fix MODE for vec_select.
21265 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21267 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21269 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21271 2017-03-06 Richard Biener <rguenther@suse.de>
21273 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21274 * plugin.c (register_plugin_info): Likewise.
21275 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21277 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21279 * config/i386/sse.md (sse_storehps, sse_storelps,
21280 avx_<castmode><avxsizesuffix>_<castmode>,
21281 avx512f_<castmode><avxsizesuffix>_<castmode>,
21282 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21283 in condition that at least one operand is not a MEM.
21285 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21287 PR middle-end/79805
21288 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21289 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21291 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21292 gimple_call_nothrow_p flag based on whether original builtin can throw.
21293 If it can, emit following stmts on the fallthrough edge.
21294 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21295 don't create new bb if inserting just debug stmts on the edge, try to
21296 insert them on the fallthru bb or just reset debug stmts.
21298 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21301 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21302 restore recog_data (including the operand rtxes inside it) around
21303 the call to get_insn_template.
21305 2017-03-03 Martin Sebor <msebor@redhat.com>
21307 PR tree-optimization/79699
21308 * context.c (context::~context): Free MPFR caches to avoid
21309 a memory leak on program exit.
21311 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21313 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21314 Use wide_int::ulow () instead of .elt (0).
21316 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21318 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21319 (*pushxf): Limit oF constraint to 32bit targets and add oC
21320 constraint for 64bit targets.
21321 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21322 (*pushdf): Change rmF constraint to rmC.
21324 2017-03-03 Martin Liska <mliska@suse.cz>
21326 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21327 Remove unused variable.
21329 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21332 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21333 is a memory operand, increase num_memory.
21334 (ix86_expand_args_builtin): Likewise.
21336 2017-03-03 Jan Hubicka <jh@suse.cz>
21339 * ipa-devirt.c (maybe_record_node): Properly handle
21340 __cxa_pure_virtual visibility.
21342 2017-03-03 Martin Liska <mliska@suse.cz>
21344 PR tree-optimization/79803
21345 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21347 (pass_loop_prefetch::execute): Disabled optimization if an
21348 assumption about L1 cache size is not met.
21350 2017-03-03 Martin Liska <mliska@suse.cz>
21352 PR rtl-optimization/79574
21353 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21354 (hash_scan_set): Likewise.
21355 (dump_hash_table): Likewise.
21356 (hoist_code): Likewise.
21358 2017-03-03 Richard Biener <rguenther@suse.de>
21360 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21361 in place of uhigh ().
21362 (fixed_convert_from_real): Likewise.
21364 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21367 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21369 2017-03-03 Richard Biener <rguenther@suse.de>
21371 PR middle-end/79818
21372 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21373 TYPE_OVERFLOW_UNDEFINED check.
21375 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21377 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21379 (vector_ae_<mode>_p): Likewise.
21380 (vector_nez_<mode>_p): Likewise.
21381 (vector_ne_v2di_p): Likewise.
21382 (vector_ae_v2di_p): Likewise.
21383 (vector_ne_<mode>_p): Likewise.
21384 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21386 (vsx_tsqrt<mode>2_fe): Likewise.
21388 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21391 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21393 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21395 PR rtl-optimization/79780
21396 * cprop.c (one_cprop_pass): When second and further conditional trap
21397 in a single basic block is turned into an unconditional trap, turn it
21398 into a deleted note to avoid RTL verification failures.
21400 2017-03-02 Richard Biener <rguenther@suse.de>
21402 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21404 2017-03-02 Richard Biener <rguenther@suse.de>
21406 PR tree-optimization/79345
21408 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21409 param and abort the walk, returning -1 if it is hit.
21410 (walk_aliased_vdefs): Take a limit param and pass it on.
21411 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21412 defaulting to 0 and return a signed int.
21413 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21414 (check_defs): New helper.
21415 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21416 about uninitialized memory.
21417 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21418 bogus uninitialized warning.
21419 (fixed_convert_from_real): Likewise.
21421 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21423 PR tree-optimization/66768
21424 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21425 iv_use if base object can't be determined.
21427 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21429 PR tree-optimization/79345
21430 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21431 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21432 (get_pattern_stats): Initialize it.
21433 * genemit.c (gen_expand): Verify match_scratch numbers come after
21434 match_operand/match_dup numbers.
21435 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21436 match_scratch numbers.
21437 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21439 * config/s390/s390.md (trunctdsd2): Likewise.
21441 2017-03-02 Richard Biener <rguenther@suse.de>
21443 * wide-int.h (wide_int_storage::operator=): Implement in terms
21446 2017-03-02 Richard Biener <rguenther@suse.de>
21448 PR tree-optimization/79777
21449 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21450 the to insert expression to sth existing.
21452 2017-03-01 Martin Sebor <msebor@redhat.com>
21454 PR middle-end/79692
21455 * gimple-ssa-sprintf.c
21456 (directive::known_width_and_precision): New function.
21457 (format_integer): Use it.
21458 (get_mpfr_format_length): Consider the full range of precision
21459 when computing %g output with the # flag. Set the likely byte
21460 count to 3 rather than 1 when precision is indeterminate.
21461 (format_floating): Correct the lower bound of precision.
21463 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21465 * doc/invoke.texi: Document default code model for 64-bit Linux.
21467 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21470 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21471 udiv rather than div since input pattern is unsigned.
21473 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21475 * config/i386/i386.c (print_reg): Warn for values of
21476 unsupported size in integer register.
21478 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21481 * config/rs6000/predicates.md (current_file_function_operand): Do
21482 not allow self calls to be local if the function is replaceable.
21484 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21487 * config/rs6000/altivec.h (vec_ctz and others): Change the
21488 preprocessor macro that controls conditional compilation from
21489 _ARCH_PWR9 to __POWER9_VECTOR__.
21490 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21491 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21492 control (instead of _ARCH_PWR9 control) so that template
21493 definition uses power9-specific function.
21494 (vec_any_eq): Likewise.
21495 (vec_all_ne): Change macro definition to use a power9-specific
21496 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21497 _ARCH_PWR9 control).
21498 (vec_any_eq) Likewise.
21499 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21500 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21501 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21502 support for xvcmpnedp instruction.
21503 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21504 macro expansion so that Power9 implementation of vec_all_ne does
21505 not use the AltiVec predicate framework.
21506 (VCMPNEH_P): Likewise.
21507 (VCMPNEW_P): Likewise.
21508 (VCMPNED_P): Likewise.
21509 (VCMPNEFP_P): Likewise.
21510 (VCMPNEDP_P): Likewise.
21511 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21512 implementation of vec_any_eq to not use AltiVec predicate
21514 (VCMPAEH_P): Likewise.
21515 (VCMPAEW_P): Likewise.
21516 (VCMPAED_P): Likewise.
21517 (VCMPAEFP_P): Likewise.
21518 (VCMPAEDP_P): Likewise.
21519 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21520 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21521 not use the AltiVec predicate framework.
21522 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21523 of vec_any_eq to not use AltiVec predicate framework.
21524 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21525 support for predefined __POWER9_VECTOR__ macro to indicate that
21526 Power9 instruction selection is enabled.
21527 (altivec_overloaded_builtins): Remove extraneous
21528 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21529 function argument types RS6000_BTI_bool_V16QI and
21530 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21531 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21532 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21533 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21534 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21535 Power9 for implementations of vec_cmpne. Change the signature for
21536 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21537 (representing vec_all_ne) to remove the previously described first
21538 argument of type RS6000_BTI_INTSI, as this was an artifact of
21539 reliance on the AltiVec predicate framework, which is no longer
21540 used in the implementation of these functions. Add
21541 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21542 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21543 since, unlike the AltiVec predicate framework implementation, we
21544 do not share function descriptors between vec_alle and vec_anyeq.
21545 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21546 set of modes that receive special treatment even when
21547 TARGET_P9_VECTOR is true. The special treatment emits code that
21548 does not depend on Power9 instructions.
21549 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21550 define_expand to not rely on AltiVec predicate framework.
21551 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21553 (vector_ne_v2di_p): Change this define_expand to not rely on
21554 AltiVec predicate framework.
21555 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21557 (vector_ne_<mode>_p): Change this define_expand to not rely on
21558 AltiVec predicate framework.
21559 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21561 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21562 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21563 define_insn pattern.
21564 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21565 define_insn pattern because the xvcmpne<VSs>. instruction is not
21567 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21568 instruction is not supported.
21570 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21572 * config/nvptx/nvptx.c: Include intl.h.
21574 2017-03-01 Martin Jambor <mjambor@suse.cz>
21577 * ipa-prop.h (ipa_bits): Removed field known.
21578 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21579 to pointers. Adjusted their comments to warn about their sharing.
21580 (ipcp_transformation_summary): Change bits to a vector of pointers.
21581 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21582 (ipa_get_ipa_bits_for_value): Declare.
21583 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21584 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21585 (ipa_bits_hash_table): Likewise.
21586 (ipa_vr_ggc_hash_traits): Likewise.
21587 (ipa_vr_hash_table): Likewise.
21588 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21589 being pointers and vr_known being removed.
21590 (ipa_set_jf_unknown): Likewise.
21591 (ipa_get_ipa_bits_for_value): New function.
21592 (ipa_set_jfunc_bits): Likewise.
21593 (ipa_get_value_range): New overloaded functions.
21594 (ipa_set_jfunc_vr): Likewise.
21595 (ipa_compute_jump_functions_for_edge): Use the above functions to
21596 construct bits and vr parts of jump functions.
21597 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21598 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21600 (ipcp_grow_transformations_if_necessary): Also allocate
21601 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21603 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21604 them. Fix too long lines.
21605 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21606 vr_known being removed.
21607 (ipa_read_jump_function): Use new setter functions to construct bits
21608 and vr parts of jump functions or set them to NULL.
21609 (write_ipcp_transformation_info): Adjust for bits being pointers.
21610 (read_ipcp_transformation_info): Likewise.
21611 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21613 Include gt-ipa-prop.h.
21614 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21616 (ipcp_store_bits_results): Likewise.
21617 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21618 Do not write to existing jump functions but use a temporary instead.
21620 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21623 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21624 attempt to use its first operand as BIT_FIELD_REF base.
21626 2017-03-01 Richard Biener <rguenther@suse.de>
21628 PR middle-end/79721
21629 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21630 interpolating formula in wrapping arithmetic.
21631 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21633 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21635 PR tree-optimization/79734
21636 * tree-vect-generic.c (expand_vector_condition): Optimize
21637 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21638 Handle VEC_COND_EXPR where comparison has different inner width from
21639 type's inner width.
21641 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21643 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21644 markup, and similar issues. Remove @opindex entries for things
21645 that aren't options. Add missing -mmpy-option entries.
21647 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21649 PR tree-optimization/79737
21650 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21651 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21652 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21653 instead of byte_size. Formatting fix.
21654 (shift_bytes_in_array_right): Formatting fix.
21656 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21659 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21660 condition on optimize for the leaf function test.
21662 2017-02-28 Martin Liska <mliska@suse.cz>
21665 * read-rtl-function.c (function_reader::handle_unknown_directive):
21666 Bail out when one uses -flto.
21668 2017-02-28 Martin Liska <mliska@suse.cz>
21670 * common.opt: Replace space with tabular for options of <number>
21672 * config/i386/i386.opt: Show <number> value for
21673 -mlarge-data-threshold.
21674 * opts.c (print_filtered_help): Do not display number in hexadecimal
21677 2017-02-28 Martin Liska <mliska@suse.cz>
21679 * common.opt: Fix --help=option -Q for options which are of
21682 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21684 * config/i386/i386.c (print_reg): Error out for values
21685 of 8-bit size in invalid integer register.
21687 2017-02-28 Martin Sebor <msebor@redhat.com>
21689 PR tree-optimization/79691
21690 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21692 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21695 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21696 gcc_unreachable with output_operand_lossage.
21698 2017-02-28 Richard Biener <rguenther@suse.de>
21700 PR tree-optimization/79740
21701 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21703 (visit_nary_op): Insert the nary into the hashtable if we
21704 pattern-matched sth.
21705 * tree-ssa-pre.c (eliminate_insert): Robustify.
21707 2017-02-28 Richard Biener <rguenther@suse.de>
21709 PR middle-end/79731
21710 * fold-const.c (decode_field_reference): Reject out-of-bound
21713 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21715 * config/i386/i386.c: Include intl.h.
21716 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21717 instead of just cond ? "..." : "...".
21718 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21719 * coverage.c (read_counts_file): Likewise.
21720 * omp-offload.c: Include intl.h.
21721 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21722 of just cond ? "..." : "...".
21723 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21724 of just cond ? "..." : "...".
21726 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21729 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21731 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21732 'tune for' CPU name.
21733 * config/arm/arm-cpu-data.h: Regenerated.
21735 2017-02-28 Richard Biener <rguenther@suse.de>
21737 PR tree-optimization/79732
21738 * tree-inline.c (expand_call_inline): Do not shadow var.
21740 2017-02-28 Richard Biener <rguenther@suse.de>
21742 PR tree-optimization/79723
21743 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21744 address-space properly.
21746 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21748 * doc/optinfo.texi (Optimization groups): Fix option used for
21750 * doc/invoke.texi (-fopt-info): Document "omp".
21751 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21752 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21753 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21754 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21755 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21757 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21759 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21760 OPTGROUP_OMP with "omp".
21762 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21765 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21766 for arithmetic shift of unsigned V2DI.
21768 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21770 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21771 arc/linux.h headers.
21772 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21773 (LINK_SPEC): Likewise.
21774 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21775 (EXTRA_SPECS): Likewise.
21776 (STARTFILE_SPEC): Likewise.
21777 (ENDFILE_SPEC): Likewise.
21778 (LIB_SPEC): Likewise.
21779 (TARGET_SDATA_DEFAULT): Likewise.
21780 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21781 (MULTILIB_DEFAULTS): Likewise.
21782 (DWARF2_UNWIND_INFO): Likewise.
21783 * config/arc/big.h: New file.
21784 * config/arc/elf.h: Likewise.
21785 * config/arc/linux.h: Likewise.
21786 * config/arc/t-uClibc: Remove.
21788 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21790 PR tree-optimization/77536
21791 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21792 (tree_transform_and_unroll_loop): Use above function to compute the
21793 estimated niter of unrolled loop and use it when scaling profile.
21794 Also use count info rather than frequency if it's non-zero.
21795 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21796 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21797 (vect_transform_loop): Call above function.
21799 2017-02-27 Richard Biener <rguenther@suse.de>
21801 PR tree-optimization/45397
21802 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21803 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21804 (visit_nary_op): Add pattern matching for CSEing sign-changed
21805 or truncated operations with wider ones.
21807 2017-02-27 Richard Biener <rguenther@suse.de>
21809 PR tree-optimization/79690
21810 * tree-vect-stmts.c (vectorizable_store): Use vector type
21811 built from the DR with address-space.
21813 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21815 * doc/invoke.texi (Optimize Options): Refine the description
21816 of asan-use-after-return.
21818 2017-02-25 Alan Modra <amodra@gmail.com>
21820 PR rtl-optimization/79584
21821 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21822 base, not ad->base_term, the reg within base. Remove assertion
21823 that ad->base == ad->base_term. Replace gen_int_mode using
21824 bogus mode with const0_rtx.
21826 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21828 PR middle-end/79396
21829 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21830 FMA_EXPR like tcc_binary or tcc_unary.
21832 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21835 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21837 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21838 (output_loc_operands): Handle DW_OP_call_ref and
21839 DW_OP_GNU_variable_value.
21840 (struct variable_value_struct): New type.
21841 (struct variable_value_hasher): Likewise.
21842 (variable_value_hash): New variable.
21843 (string_types): Remove.
21844 (copy_loc_descr): New function.
21845 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21846 (prepend_loc_descr_to_each): New function.
21847 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21848 instead of add_loc_descr_to_each if the first argument is single
21849 location list and the second has multiple.
21850 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21851 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21852 when looking for variable value which doesn't have other location info.
21853 (loc_list_from_tree): Formatting fix.
21854 (gen_array_type_die): Simplify DW_AT_string_length handling.
21855 (adjust_string_types): Remove.
21856 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21857 string_types. Call resolve_variable_values.
21858 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21859 (resolve_addr_in_expr): Likewise. Add A argument.
21860 (copy_deref_exprloc): Remove deref argument. Adjust for the
21861 original expression being DW_OP_GNU_variable_value with optionally
21862 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21863 optionally after it.
21864 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21865 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21866 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21867 (variable_value_hasher::hash, variable_value_hasher::equal): New
21869 (resolve_variable_value_in_expr, resolve_variable_value,
21870 resolve_variable_values, note_variable_value_in_expr,
21871 note_variable_value): New functions.
21872 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21875 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21878 * opts.h (handle_generated_option): Add GENERATED_P argument.
21879 * opts-common.c (handle_option): Adjust function comment.
21880 (handle_generated_option): Add GENERATED_P argument, pass it to
21882 (control_warning_option): Pass false to handle_generated_option
21884 * opts.c (maybe_default_option): Pass true to handle_generated_option
21886 * optc-gen.awk: Likewise.
21888 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21890 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21891 a REG, look at the REG it is a SUBREG of.
21892 (splitter for cmpeqsi_t): Ditto.
21894 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21896 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21897 the special USEs with the pattern of the insn, not the insn itself.
21899 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21902 * doc/invoke.texi: Document -mload-store-pairs.
21904 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21905 Sandra Loosemore <sandra@codesourcery.com>
21907 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21908 argument isn't a CONST_INT.
21909 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21910 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21911 (nios2_validate_compare): Bypass alternate compare logic if *op2
21912 is not a CONST_INT.
21913 (ldstwm_operation_p): Return false if first_base is not a REG or
21914 if first_offset is not a CONST_INT.
21916 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21918 * config/cris/cris.md: Use correct operand in a define_peephole2.
21920 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21922 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21924 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21926 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21927 this_insn if it is an INSN or JUMP_INSN.
21928 (force_offsettable): Look at base, not at addr.
21929 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21930 on things that aren't necessarily CONST_INTs.
21932 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21934 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21935 -mfpmath=sse is the default also for x86-32 targets with SSE2
21936 instruction set when @option{-ffast-math} is enabled
21938 2017-02-24 Jeff Law <law@redhat.com>
21940 PR rtl-optimizatoin/79286
21941 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21944 2017-02-24 Richard Biener <rguenther@suse.de>
21946 PR tree-optimization/79389
21947 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21950 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21952 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21953 function comment to reflect reality.
21954 (loop_exits_before_overflow): Fix typo in function description.
21956 2017-02-24 Richard Biener <rguenther@suse.de>
21958 PR tree-optimization/79389
21959 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21960 properly that a threading opportunity exists. Detect conditional
21961 copy/constant propagation opportunities.
21963 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21965 * config/visium/visium.md (type): Add trap.
21966 (b): New mode attribute.
21967 (*btst): Rename into...
21968 (*btst<mode>): ...this and adjust.
21969 (*cbranchsi4_btst_insn): Rename into...
21970 (*cbranch<mode>4_btst_insn): ...this and adjust.
21971 (trap): New define_insn.
21973 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21975 PR tree-optimization/79389
21976 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21977 (noce_reversed_cond_code): New function.
21978 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21979 reversed_comparison_code. Formatting fix.
21980 (noce_try_store_flag): Test rev_cond != NULL in addition to
21981 reversed_comparison_code.
21982 (noce_try_store_flag_constants): Likewise.
21983 (noce_try_store_flag_mask): Likewise.
21984 (noce_try_addcc): Use rev_cond if non-NULL instead of
21985 reversed_comparison_code.
21986 (noce_try_cmove_arith): Likewise. Formatting fixes.
21987 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21988 (noce_find_if_block): Initialize rev_cond.
21989 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21990 instead of false as last argument never attempt to reverse it
21993 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21995 PR tree-optimization/79663
21996 * tree-predcom.c (combine_chains): Process refs in reverse order
21997 only for ZERO length chains, and add explaining comment.
21999 2017-02-23 Jeff Law <law@redhat.com>
22001 PR tree-optimization/79578
22002 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22003 in call to operand_equal_p.
22005 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22008 * config/i386/cpuid.h: Fix another undefined behavior.
22010 2017-02-23 Richard Biener <rguenther@suse.de>
22012 PR tree-optimization/79683
22013 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22014 vector types for data-refs.
22016 2017-02-23 Martin Liska <mliska@suse.cz>
22018 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22020 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22022 PR middle-end/79665
22023 * internal-fn.c (get_range_pos_neg): Moved to ...
22024 * tree.c (get_range_pos_neg): ... here. No longer static.
22025 * tree.h (get_range_pos_neg): New prototype.
22026 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22027 are known to be in between 0 and signed maximum inclusive, try to
22028 expand both unsigned and signed divmod and use the cheaper one from
22031 2017-02-22 Jeff Law <law@redhat.com>
22033 PR tree-optimization/79578
22034 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22035 to compare base operands.
22037 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22040 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22041 gpc_reg_operand instead of fpr_reg_operand.
22043 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22045 * config/mips/mips.c (mips_return_in_memory): Force FP
22046 vector types to be returned in memory for o32 ABI.
22048 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22050 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22051 instead of DW_TAG_member for static data member declarations and don't
22052 set no_linkage_name for static inline data members.
22053 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22056 2017-02-22 Martin Liska <mliska@suse.cz>
22058 * doc/invoke.texi: Replace inequality signs with square brackets
22061 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22063 PR tree-optimization/68644
22064 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22066 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22069 * lra-constraints.c (simplify_operand_subreg): Handle
22070 WORD_REGISTER_OPERATIONS targets.
22072 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22075 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22076 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22077 elimination by swapping fld*.
22079 2017-02-22 Richard Biener <rguenther@suse.de>
22081 PR tree-optimization/79673
22082 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22083 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22084 irrelevant address-space qualifiers and avoiding a
22085 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22087 2017-02-22 Richard Biener <rguenther@suse.de>
22089 PR tree-optimization/79666
22090 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22091 to not symbolically negate if that may introduce undefined
22094 2017-02-22 Martin Liska <mliska@suse.cz>
22097 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22098 * data-streamer-out.c (streamer_write_gcov_count_stream):
22100 * value-prof.c (stream_out_histogram_value): Make assert more
22101 precise based on type of counter.
22103 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22106 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22107 Use nonimmediate_operand instead of memory_operand for operand 1.
22108 (float-extend standard_x87sse_constant_load splitter): Ditto.
22110 2017-02-21 Jeff Law <law@redhat.com>
22112 PR tree-optimization/79621
22113 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22114 blocks with edges to themselves.
22116 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22119 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22120 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22121 Use gimple_call_builtin_p.
22124 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22125 on temporarily removed DEBUG_INSNs.
22127 PR tree-optimization/79649
22128 * tree-loop-distribution.c (classify_partition): Give up on
22129 non-generic address space loads/stores.
22131 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22133 * doc/loop.texi (Loop manipulation): Remove nonexistent
22134 tree_ssa_loop_version from the documentation.
22135 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22137 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22140 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22141 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22142 * config/rs6000/rs6000.c: Include except.h.
22143 (rs6000_expand_split_stack_prologue): Call
22144 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22146 2017-02-21 Martin Jambor <mjambor@suse.cz>
22149 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22150 have been analyzed.
22152 2017-02-21 Martin Jambor <mjambor@suse.cz>
22154 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22155 for backward compatibility only.
22156 * doc/invoke.texi (Option Summary): Remove all references to
22157 -fipa-cp-alignment.
22159 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22163 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22165 * lra-constraints.c (curr_insn_transform): Handle
22166 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22168 2017-02-21 Martin Liska <mliska@suse.cz>
22170 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22172 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22174 PR translation/79638
22175 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22177 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22180 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22181 (arm_function_ok_for_sibcall): Return false for an indirect call by
22182 descriptor if all the argument registers are used.
22183 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22184 alignment of the function.
22186 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22188 PR tree-optimization/61441
22189 * simplify-rtx.c (simplify_const_unary_operation): For
22190 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22191 the sNaN unmodified.
22193 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22195 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22196 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22197 instead of SYSTEM_HEADER_DIR.
22199 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22200 Martin Liška <mliska@suse.cz>
22202 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22203 Fix typos and grammar, use active voice, and clarify.
22205 2017-02-20 Marek Polacek <polacek@redhat.com>
22207 PR middle-end/79537
22208 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22211 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22213 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22216 * config/i386/i386.c (ix86_expand_builtin): Handle
22217 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22218 ix86_builtins_isa[fcode].isa as a requirement of those
22219 flags and any other flag in the bitmask.
22220 (ix86_init_mmx_sse_builtins): Use 0 instead of
22221 ~OPTION_MASK_ISA_64BIT as mask.
22222 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22223 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22224 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22225 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22227 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22230 * lra-constraints.c (split_reg): Check requested split mode
22231 is supported by the register.
22233 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22235 * lra-constraints.c (simplify_operand_subreg): Remove early
22238 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22241 * lra-constraints.c (curr_insn_transform): Tighten condition
22242 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22244 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22247 * lra-constraints.c (curr_insn_transform): Handle
22248 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22250 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22253 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22255 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22257 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22260 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22263 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22265 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22268 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22270 * final.c (last_columnnum, override_columnnum): New variables.
22271 (final_start_function): Set last_columnnum, pass it to begin_prologue
22272 hook and pass 0 to dwarf2out_begin_prologue.
22273 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22274 to source_line debug hook.
22275 (notice_source_line): Compute last_columnnum and for debug_column_info
22276 return true on column changes.
22277 * debug.h (struct gcc_debug_hooks): Add column argument to
22278 source_line and begin_prologue hooks.
22279 (debug_nothing_int_charstar_int_bool): Remove prototype.
22280 (debug_nothing_int_int_charstar,
22281 debug_nothing_int_int_charstar_int_bool): New prototypes.
22282 (dwarf2out_begin_prologue): Add column argument.
22283 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22284 begin_prologue hooks.
22285 (debug_nothing_int_charstar_int_bool): Remove.
22286 (debug_nothing_int_int_charstar,
22287 debug_nothing_int_int_charstar_int_bool): New functions.
22288 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22289 through to dwarf2out_source_line.
22290 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22291 (dwarf2out_source_line): Add column argument, emit it if requested.
22292 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22294 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22295 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22296 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22297 through to dwarf2out_begin_prologue.
22298 (vmsdbgout_source_line): Add column argument, pass it through to
22299 dwarf2out_source_line.
22300 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22301 dbxout_source_line caller.
22302 (dbxout_source_line): Add column argument.
22304 * common.opt (gno-column-info, gcolumn-info): New options.
22305 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22306 (check_die): Also test for multiple DW_AT_decl_column attributes.
22307 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22308 DW_AT_decl_column if requested.
22309 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22311 (gen_variable_die): Likewise.
22312 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22313 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22316 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22317 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22318 (ix86_handle_option): Handle OPT_m3dnowa.
22319 * doc/invoke.texi (-m3dnowa): Document.
22320 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22321 -m3dnowa instead of -m3dnow -march=athlon.
22324 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22325 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22327 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22330 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22331 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22332 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22333 generator for vsx_xxpermdi_<mode>_be.
22334 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22335 force big-endian semantics.
22336 (vsx_xxpermdi_<mode>_be): New define_expand with same
22337 implementation as previous version of vsx_xxpermdi_<mode>.
22339 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22341 PR tree-optimization/79327
22342 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22343 variable, its initialization and use.
22345 2017-02-17 Julia Koval <julia.koval@intel.com>
22347 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22348 (OPTION_MASK_ISA_PKU_UNSET): New.
22349 (ix86_handle_option): Handle -mrdpid.
22350 * config/i386/cpuid.h (bit_RDPID): New.
22351 * config/i386/driver-i386.c (host_detect_local_cpu):
22352 Detect RDPID feature.
22353 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22354 * config/i386/i386-c.c (ix86_target_macros_internal):
22356 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22357 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22358 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22359 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22360 * config/i386/i386.md (define_insn "rdpid"): New.
22361 * config/i386/i386.opt Add -mrdpid.
22362 * config/i386/immintrin.h (_rdpid_u32): New.
22364 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22366 PR rtl-optimization/79541
22367 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22368 instead of transforming it into USE.
22370 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22372 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22373 If HONOR_SNANS (SFmode) force the input to a register.
22374 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22375 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22376 an frsp or similar insn.
22378 2017-02-17 Martin Liska <mliska@suse.cz>
22380 PR rtl-optimization/79577
22381 * params.def (selsched-max-sched-times): Increase minimum to 1.
22383 2017-02-17 Martin Liska <mliska@suse.cz>
22385 PR rtl-optimization/79574
22386 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22388 2017-02-17 Martin Liska <mliska@suse.cz>
22390 PR tree-optimization/79529
22391 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22392 ssa_defined_default_def_p to handle cases which are implicitly
22394 * tree-ssa.c (ssa_defined_default_def_p): New function.
22395 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22396 which are implicitly defined.
22397 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22399 2017-02-17 Richard Biener <rguenther@suse.de>
22401 PR middle-end/79576
22402 * params.def (max-ssa-name-query-depth): Limit to 10.
22404 2017-02-17 Richard Biener <rguenther@suse.de>
22406 PR tree-optimization/79552
22407 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22410 2017-02-17 Richard Biener <rguenther@suse.de>
22413 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22415 2017-02-17 Marek Polacek <polacek@redhat.com>
22417 PR middle-end/79536
22418 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22419 (fold_negate_expr): New wrapper.
22421 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22423 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22424 Correct terminology and de-emphasize pre-standard behavior.
22426 2017-02-16 Alan Modra <amodra@gmail.com>
22428 PR rtl-optimization/79286
22429 * ira.c (def_dominates_uses): New function.
22430 (update_equiv_regs): Don't create an equivalence for insns that
22431 may trap where the register def does not dominate the use.
22433 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22435 PR rtl-optimization/78127
22436 * lra.c (lra): Call lra_eliminate before finish the loop after
22439 2017-02-16 Richard Biener <rguenther@suse.de>
22441 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22443 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22444 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22445 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22446 (isl_val_int_from_wi): New function.
22447 (extract_affine_gmp): Rename to ...
22448 (extract_affine_wi): ... this, take a widest_int.
22449 (extract_affine_int): Just wrap extract_affine_wi.
22450 (add_param_constraints): Use isl_val_int_from_wi.
22451 (add_loop_constraints): Likewise, and extract_affine_wi.
22453 2017-02-15 Jeff Law <law@redhat.com>
22455 PR middle-end/79521
22456 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22457 ira_init_register_move_cost_if_necessary.
22459 2017-02-15 Martin Sebor <msebor@redhat.com>
22461 PR middle-end/32003
22462 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22463 removed in a prior commit.
22465 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22467 PR tree-optimization/79347
22468 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22469 counters during peeling.
22471 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22473 * Makefile.in (site.exp): Remove "set ISLVER".
22475 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22478 * real.c (real_from_integer): Call real_convert even for decimal.
22480 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22483 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22485 2017-02-14 Andrew Pinski <apinski@cavium.com>
22487 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22488 cores and change the partno/implementer to be correct.
22489 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22490 the 'B" as the implementer.
22491 * config/aarch64/aarch64-tune.md: Regenerate.
22493 2017-02-14 Carl Love <cel@us.ibm.com>
22495 * config/rs6000/rs6000.c: Add case statement entry to make the
22496 xvcvuxdsp built-in argument unsigned.
22497 * config/rs6000/vsx.md: Fix the source and return operand types so they
22498 match the instruction definitions from the ISA document. Fix typo
22499 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22502 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22505 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22506 member early_clobber_alts.
22507 * lra-lives.c (reg_early_clobber_p): New.
22508 (process_bb_lives): Use it.
22509 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22510 (debug_operand_data): Initialize early_clobber_alts.
22511 (setup_operand_alternative): Set up early_clobber_alts.
22512 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22513 alternatives to new_insn_reg.
22514 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22516 (lra_update_insn_regno_info): Pass the new arg.
22518 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22520 PR middle-end/79505
22521 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22522 (new_oacc_loop_raw): Don't clear already cleared fields.
22525 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22526 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22527 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22529 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22532 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22534 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22537 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22538 the extra instruction to the right place to store 128-bit constant
22541 2017-02-14 Martin Sebor <msebor@redhat.com>
22543 PR middle-end/79448
22544 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22545 warning for strings of unknown length.
22547 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22549 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22551 2017-02-14 Jeff Law <law@redhat.com>
22554 * ira-costs.c (scan_one_insn): Initialize register move costs
22555 for pseudos seen in USE/CLOBBER insns.
22557 PR tree-optimization/79095
22558 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22559 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22560 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22561 if the operands are known to be not equal, then the resulting range
22563 (intersect_ranges): If the new range is ~[0,0] and the old range is
22564 wide, then prefer ~[0,0].
22565 * tree-vrp.c (overflow_comparison_p_1): New function.
22566 (overflow_comparison_p): New function.
22567 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22568 if NAME is used in an overflow test.
22569 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22570 overflow check that can be expressed as an equality test, then adjust
22571 ops to be that equality test.
22573 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22575 * config/s390/s390-builtin-types.def: Remove flags argument.
22576 * config/s390/s390.c (s390_init_builtins): Likewise.
22578 2017-02-14 Martin Liska <mliska@suse.cz>
22580 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22581 vector. Fix trailing white spaces.
22583 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22585 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22588 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22590 PR rtl-optimization/68664
22591 * config/arm/arm.c (arm_sched_can_speculate_insn):
22592 New function. Declare prototype.
22593 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22595 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22597 PR rtl-optimization/68664
22598 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22600 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22602 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22604 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22605 max skip bytes for function, loop and jump.
22607 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22609 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22610 ABS_EXPR for gimple dump.
22612 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22615 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22617 PR tree-optimization/79408
22618 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22619 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22620 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22621 also if rhs1 is INTEGER_CST.
22623 2017-02-14 Richard Biener <rguenther@suse.de>
22625 PR middle-end/79432
22626 * tree-into-ssa.c (insert_phi_nodes): When the function can
22627 have abnormal edges rewrite SSA names with broken use-def
22628 dominance out of SSA and register them for PHI insertion.
22630 2017-02-13 Martin Sebor <msebor@redhat.com>
22632 PR middle-end/79496
22633 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22634 clearing info.nowrite flag when snprintf size argument is a range.
22636 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22638 * cprop.c (cprop_jump): Add missing space in string literal.
22639 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22640 (get_constraint_for_component_ref): Likewise.
22641 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22642 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22643 * lra-constraints.c (process_alt_operands): Likewise.
22644 * ipa-inline.c (inline_small_functions): Likewise.
22645 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22646 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22647 * trans-mem.c (diagnose_tm_1_op): Likewise.
22648 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22649 (grid_parallel_clauses_gridifiable): Likewise.
22651 * config/nvptx/mkoffload.c (process): Add space in between
22654 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22655 "MOD4_SSE_REGS" and "ALL_REGS".
22657 * spellcheck.c (test_data): Add , in between "foo" and "food".
22659 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22662 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22663 boundary crossing check and subsequent code generation agree.
22665 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22667 * config/aarch64/aarch64.c (has_memory_op): Delete.
22668 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22671 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22673 PR rtl-optimization/79388
22674 PR rtl-optimization/79450
22675 * combine.c (distribute_notes): When removing TEM_INSN for which
22676 corresponding dest has last value recorded, invalidate that last
22679 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22681 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22682 of explicit '@'. Add missing assembly comment marker on branch costs
22685 2017-02-13 Nathan Sidwell <nathan@acm.org>
22687 * gengtype-lex.l (<in_struct>): Add '/'.
22689 2017-02-13 Martin Liska <mliska@suse.cz>
22692 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22694 2017-02-13 Richard Biener <rguenther@suse.de>
22696 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22698 * configure: Re-generate.
22699 * config.in: Likewise.
22700 * graphite-dependences.c: Simplify as if
22701 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22702 * graphite-isl-ast-to-gimple.c: Likewise.
22703 * graphite-optimize-isl.c: Likewise.
22704 * graphite-poly.c: Likewise.
22705 * graphite-sese-to-poly.c: Likewise.
22706 * graphite.h: Likewise.
22707 * toplev.c: Include isl/version.h and use isl_version () for
22708 printing the ISL version.
22709 * doc/install.texi: Update ISL requirement.
22711 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22713 * doc/standards.texi (Standards): Update reference to
22716 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22718 * doc/extend.texi (Named Address Spaces): sourceware.org now
22720 * doc/install.texi (Binaries): Ditto.
22723 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22725 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22726 standard terminology "stringize"/"stringizing" throughout.
22727 * doc/cppinternals.texi: Likewise.
22729 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22731 * doc/extend.texi: Fix some spelling mistakes and typos.
22732 * doc/invoke.texi: Likewise.
22734 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22737 * params.def (inline-min-speedup) Change from 10 to 8.
22739 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22741 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22744 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22747 * ipa-inline-analysis.c (get_minimal_bb): New function.
22748 (record_modified): Use it.
22749 (remap_edge_change_prob): Handle also ancestor functions.
22751 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22753 * doc/contrib.texi (Contributors): Remove broken link into
22754 the Mauve CVS repository.
22756 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22758 PR middle-end/79454
22759 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22760 result computation whenever lhs doesn't have vector mode, not
22761 just when it has BLKmode.
22763 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22765 * doc/makefile.texi (profiledbootstrap): Refer to the
22766 installation instructions only in textual form.
22768 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22771 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22773 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22775 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22776 (Specific): Update mingw-w64 reference.
22778 (Specific): Remove broken link to Renesas RX processor.
22780 2017-02-10 Richard Biener <rguenther@suse.de>
22782 * toplev.c (process_options): Do not mention obsolete graphite
22783 options when printing sorry message about missing graphite support.
22784 Mention -floop-nest-optimize.
22786 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22788 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22789 (vtst_p16): Likewise.
22790 (vtstq_p8): Likewise.
22791 (vtstq_p16): Likewise.
22793 (vtstq_p64): Likewise.
22794 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22795 (vset_lane_p64): New.
22796 (vsetq_lane_p64): New.
22798 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22800 PR tree-optimization/79411
22801 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22802 stmt operands are SSA_NAMEs used in abnormal phis.
22803 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22806 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22809 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22812 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22814 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22816 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22818 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22822 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22823 not arbitrary TREE_CONSTANT.
22826 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22827 "omp declare target link" attribute unless is_global_var.
22828 * omp-offload.c (find_link_var_op): Likewise.
22830 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22831 Chung-Lin Tang <cltang@codesourcery.com>
22833 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22835 (gimplify_adjust_omp_clauses): Don't delete TILE.
22836 (gimplify_omp_for): Deal with TILE.
22837 * internal-fn.c (expand_GOACC_TILE): New function.
22838 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22840 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22841 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22843 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22844 avoid DIV for outermost collapse var.
22845 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22846 Remove out of date comments, fix whitespace.
22847 * omp-general.c (omp_extract_for_data): Deal with tiling.
22848 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22849 adjust OLF_DIM_BASE value.
22850 (struct omp_for_data): Add tiling field.
22851 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22852 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22853 for auto loops. Remove default auto determining, moved to
22854 oacc_loop_fixed_partitions.
22855 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22856 stmts, add e_mask field.
22857 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22858 (oacc_thread_numbers): Use oacc_dim_call.
22859 (oacc_xform_tile): New.
22860 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22861 (finish_oacc_loop): Adjust for ifns vector.
22862 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22863 add case for GOACC_TILE fns.
22864 (oacc_loop_xform_loop): Delete.
22865 (oacc_loop_process): Iterate over call list directly, and add
22866 handling for GOACC_TILE fns.
22867 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22869 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22870 vector partitioning to outer loops. Assign 2 partitions to loops
22871 when available. Add TILE handling.
22872 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22873 (execite_oacc_device_lower): Process GOACC_TILE fns,
22874 ignore unknown specs.
22875 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22876 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22877 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22879 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22881 * configure.ac (ACX_BUGURL): Update.
22882 * configure: Regenerate.
22884 2017-02-09 Richard Biener <rguenther@suse.de>
22886 PR tree-optimization/69823
22887 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22888 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22890 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22892 * config/arc/arc-c.def: Add __NPS400__ definition.
22893 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22894 (TARGET_NPS400): Define.
22896 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22898 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22900 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22901 pointer, arch_info.
22902 (arc_cpu_types): Fill the arch_info field with a pointer into the
22903 arc_arch_types table.
22904 (arc_selected_cpu): Declare.
22905 * config/arc/arc.c (arc_selected_cpu): Make global.
22906 (arc_selected_arch): Delete.
22907 (arc_base_cpu): Delete.
22908 (arc_override_options): Remove references to deleted variables,
22909 update access to arch information.
22910 (ARC_OPT): Update access to arch information.
22911 (ARC_OPTX): Likewise.
22912 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22913 (TARGET_ARC600): Update access to arch information.
22914 (TARGET_ARC601): Likewise.
22915 (TARGET_ARC700): Likewise.
22916 (TARGET_EM): Likewise.
22917 (TARGET_HS): Likewise.
22918 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22921 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22924 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22925 condition/operands for integer GE/LE/GEU/LEU operations.
22927 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22929 PR translation/79397
22930 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22933 2017-02-08 Martin Jambor <mjambor@suse.cz>
22936 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22937 whether allocation happened.
22938 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22939 nothing was allocated.
22941 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22943 PR tree-optimization/79408
22944 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22945 constant, but SSA_NAME with a known integer range, use the minimum
22946 of that range instead of op1 to determine if modulo can be replaced
22947 with its first operand.
22949 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22951 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22953 2017-02-08 Richard Biener <rguenther@suse.de>
22955 PR tree-optimization/71824
22956 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22957 Check all loops contained in the merged region.
22959 2017-02-07 Andrew Pinski <apinski@cavium.com>
22961 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22963 2017-02-07 Andrew Pinski <apinski@cavium.com>
22965 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22966 (thunderxt88): Likewise.
22967 (thunderxt81): Disable LSE and change v8.1 to v8.
22968 (thunderxt83): Likewise.
22970 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22971 Richard Biener <rguenther@suse.de>
22973 PR middle-end/79399
22974 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22975 type from int to size_t.
22976 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22978 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22980 PR rtl-optimization/79386
22981 * cprop.c (bypass_conditional_jumps): Initialize
22982 bypass_last_basic_block already before splitting bbs after
22983 unconditional traps...
22984 (bypass_conditional_jumps): ... rather than here.
22987 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22988 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22989 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22990 fix -masm=intel patterns.
22992 2017-02-07 Richard Biener <rguenther@suse.de>
22994 PR tree-optimization/79256
22995 PR middle-end/79278
22996 * builtins.c (get_object_alignment_2): Use min_align_of_type
22997 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22998 and ADJUST_FIELD_ALIGN.
23000 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23002 * doc/tm.texi: Regenerate.
23003 * stor-layout.c (layout_decl): Adjust.
23004 (update_alignment_for_field): Likewise.
23005 (place_field): Likewise.
23006 (min_align_of_type): Likewise.
23007 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23008 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23009 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23010 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23011 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23012 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23013 * config/i386/i386.c (x86_field_alignment): Likewise.
23014 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23015 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23016 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23017 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23018 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23019 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23023 2017-01-30 Richard Biener <rguenther@suse.de>
23025 PR tree-optimization/79256
23026 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23027 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23030 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23032 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23033 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23034 builtins to SImode and emit a zero-extend, if necessary.
23036 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23038 * docs/invoke.texi (RISC-V Options): Alphabetize.
23040 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23042 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23045 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23047 * config/riscv/riscv.c: New file.
23048 * common/config/riscv/riscv-common.c: Likewise.
23049 * config.gcc: Likewise.
23050 * config/riscv/constraints.md: Likewise.
23051 * config/riscv/elf.h: Likewise.
23052 * config/riscv/generic.md: Likewise.
23053 * config/riscv/linux.h: Likewise.
23054 * config/riscv/multilib-generator: Likewise.
23055 * config/riscv/peephole.md: Likewise.
23056 * config/riscv/pic.md: Likewise.
23057 * config/riscv/predicates.md: Likewise.
23058 * config/riscv/riscv-builtins.c: Likewise.
23059 * config/riscv/riscv-c.c: Likewise.
23060 * config/riscv/riscv-ftypes.def: Likewise.
23061 * config/riscv/riscv-modes.def: Likewise.
23062 * config/riscv/riscv-opts.h: Likewise.
23063 * config/riscv/riscv-protos.h: Likewise.
23064 * config/riscv/riscv.h: Likewise.
23065 * config/riscv/riscv.md: Likewise.
23066 * config/riscv/riscv.opt: Likewise.
23067 * config/riscv/sync.md: Likewise.
23068 * config/riscv/t-elf-multilib: Likewise.
23069 * config/riscv/t-linux: Likewise.
23070 * config/riscv/t-linux-multilib: Likewise.
23071 * config/riscv/t-riscv: Likewise.
23072 * configure.ac: Likewise.
23073 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23074 Waterman as RISC-V maintainers.
23075 * doc/install.texi: Add RISC-V entries.
23076 * doc/invoke.texi: Add RISC-V options section.
23077 * doc/md.texi: Add RISC-V constraints section.
23078 * configure: Regenerated.
23080 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23083 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23084 false values to be constant vectors with all 0 or all 1 bits set.
23085 (vcondu<mode><mode>): Likewise.
23086 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23088 (fpmask_comparison_operator): Update comment.
23089 (vecint_comparison_operator): New predicate.
23090 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23091 vector conditionals when the true and false values are constant
23092 vectors with all 0 bits or all 1 bits set.
23094 2017-02-06 Martin Sebor <msebor@redhat.com>
23096 PR tree-optimization/79376
23097 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23099 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23101 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23102 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23103 to simplify split condition.
23105 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23107 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23108 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23111 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23113 PR rtl-optimization/68664
23114 * target.def (can_speculate_insn): New hook.
23115 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23116 * doc/tm.texi: Regenerate.
23117 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23118 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23119 (rs6000_sched_can_speculate_insn): New function.
23121 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23123 PR tree-optimization/79284
23124 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23125 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23126 vectorizable_mask_load_store, vectorizable_operation,
23127 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23128 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23129 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23130 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23131 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23132 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23133 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23134 is_gimple_assign (stmt). Replace another such test with
23135 is_gimple_assign (stmt).
23137 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23140 * config/avr/avr.c (rtl-iter.h): Include it.
23141 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23142 (avr_legitimate_combined_insn): ...and implementation.
23144 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23146 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23147 * config/s390/s390.c (s390_const_operand_ok)
23148 (s390_canonicalize_comparison, s390_extract_part)
23149 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23150 (s390_contiguous_bitmask_p, s390_rtx_costs)
23151 (legitimize_pic_address): Likewise.
23152 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23153 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23154 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23155 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23156 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23158 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23160 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23161 REGNO($0) == REGNO($1).
23163 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23165 * config/s390/linux.h(SIZE_TYPE): Add comment.
23167 2017-02-06 Julian Brown <julian@codesourcery.com>
23168 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23169 Virendra Pathak <virendra.pathak@broadcom.com>
23171 * config/aarch64/aarch64-cores.def: Change the scheduler
23173 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23174 * config/aarch64/thunderx2t99.md: New file.
23176 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23178 * doc/standards.texi (Go Language): Update link to language
23181 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23183 * tree-eh.c (lower_resx): Sanitize profile.
23184 (cleanup_empty_eh_move_lp): Likewise.
23186 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23189 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23191 * cfgloopmanip.h (loop_version): Update prototype.
23192 * modulo-sched.c (sms_schedule): Update call of loop_version.
23193 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23194 * tree-parloops.c (gen_parallel_loop): Likewise.
23195 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23196 * tree-ssa-loop-split.c (split_loop): Likewise.
23197 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23198 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23200 2017-02-05 Martin Liska <mliska@suse.cz>
23203 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23205 (print_operand_address): Initialize a struct to zero.
23207 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23209 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23210 garbage collector only in textual form.
23212 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23214 * doc/extend.texi (x86 specific memory model extensions for
23215 transactional memory): Simplify a phrase.
23217 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23220 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23221 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23222 (atomic_storedi_1): Likewise.
23224 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23226 PR tree-optimization/79338
23227 * tree-parloops.c (gather_scalar_reductions): Don't call
23228 vect_analyze_loop_form for loop->inner before destroying loop's
23231 2017-02-03 Martin Sebor <msebor@redhat.com>
23233 PR tree-optimization/79327
23234 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23235 when precision has resulted in leading zeros.
23236 (format_integer): Adjust the likely counter to assume an unknown
23237 argument that may be zero is non-zero.
23239 2017-02-03 Jason Merrill <jason@redhat.com>
23242 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23243 avoid copying non-taken branch.
23245 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23247 PR tree-optimization/79340
23248 * tree-vect-loop.c (vectorizable_reduction): Release
23249 vec_defs elements after safe_splicing them into other vectors.
23252 PR tree-optimization/79327
23253 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23254 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23256 (format_integer): Use wide_int_to_tree instead of build_int_cst
23257 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23258 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23259 of shortest and longest sequence.
23261 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23263 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23264 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23266 2017-02-03 Walter Lee <walt@tilera.com>
23269 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23270 after initial stackframe link reg save.
23271 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23273 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23276 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23277 wu for stxssp alternative.
23279 2017-02-03 Martin Sebor <msebor@redhat.com>
23281 PR tree-optimization/79352
23282 * gimple-fold.c (get_range_strlen): Add argument.
23283 (get_range_strlen): Change return type to bool.
23284 (get_maxval_strlen): Pass in a dummy argument.
23285 * gimple-fold.h (get_range_strlen): Change return type to bool.
23286 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23287 * tree.h (array_at_struct_end_p): Add argument.
23288 * tree.c (array_at_struct_end_p): Handle it.
23290 2017-02-03 Martin Liska <mliska@suse.cz>
23293 * multiple_target.c (create_dispatcher_calls): Redirect edge
23294 from a caller of a dispatcher.
23295 (expand_target_clones): Make the clones local.
23296 (ipa_target_clone): Do both target clones and resolvers.
23297 (ipa_dispatcher_calls): Remove the pass.
23298 (pass_dispatcher_calls::gate): Likewise.
23299 (make_pass_dispatcher_calls): Likewise.
23300 * passes.def (pass_target_clone): Put as very first IPA early
23303 2017-02-03 Martin Liska <mliska@suse.cz>
23305 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23306 in case of a function with ifunc attribute.
23308 2017-02-03 Martin Liska <mliska@suse.cz>
23310 * cgraph.c (cgraph_node::dump): Dump function version info.
23311 * symtab.c (symtab_node::dump_base): Add missing new line.
23313 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23315 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23316 (ifcombine_ifandif): Use it.
23318 2017-02-03 Martin Liska <mliska@suse.cz>
23320 * doc/invoke.texi: Document default value for
23321 use-after-scope-direct-emission-threshold.
23323 2017-02-03 Martin Liska <mliska@suse.cz>
23325 PR tree-optimization/79339
23326 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23327 (format_floating): Likewise.
23329 2017-02-03 Martin Liska <mliska@suse.cz>
23332 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23334 (ipa_node_params_t::remove): Likewise.
23335 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23336 initialization from removed ipa_node_params_t::insert.
23337 (ipa_node_params::~ipa_node_params): Move from removed
23338 ipa_node_params_t::release.
23339 * symbol-summary.h (symbol_summary::m_released): New member.
23340 Do not release a summary twice. Do not allow to call finalizer
23341 for types of a summary that live in GGC memory.
23343 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23345 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23348 2017-02-02 Martin Sebor <msebor@redhat.com>
23350 PR middle-end/79275
23351 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23352 (format_string): Tighten up the range of output for non-constant
23353 strings and correct the expected range for wide non-constant strings.
23355 2017-02-02 Martin Sebor <msebor@redhat.com>
23357 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23359 PR middle-end/32003
23360 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23362 (-fdump-tree-@var): Add to index and document how to come up
23363 with pass-specific option and dump file names.
23364 (-fdump-passes): Clarify where to look for output.
23366 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23368 PR middle-end/77445
23369 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23370 statistics of the analyzed path; allow threading for speed when
23371 any of BBs along the path are optimized for speed.
23373 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23375 PR middle-end/78468
23376 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23377 settings of the virtual registers.
23380 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23382 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23383 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23386 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23388 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23389 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23391 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23393 * config/s390/s390.md: Add missing comments with the expanded
23395 * config/s390/vector.md: Likewise.
23396 * config/s390/vx-builtins.md: Likewise.
23398 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23401 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23402 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23403 conditions on a single line.
23405 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23407 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23408 __S390_VX__ to __VX__.
23410 2017-02-01 Andrew Pinski <apinski@cavium.com>
23412 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23413 stmt_info to record_stmt_cost.
23414 (vect_get_known_peeling_cost): Pass stmt_info if known to
23416 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23417 cpu_vector_cost field into
23418 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23419 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23420 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23421 splitting of scalar_stmt_cost and vec_stmt_cost.
23422 (thunderx_vector_cost): Likewise.
23423 (cortexa57_vector_cost): LIkewise.
23424 (exynosm1_vector_cost): Likewise.
23425 (xgene1_vector_cost): Likewise.
23426 (thunderx2t99_vector_cost): Improve after the splitting of the two
23428 (aarch64_builtin_vectorization_cost): Update for the splitting of
23429 scalar_stmt_cost and vec_stmt_cost.
23431 2017-02-01 Torvald Riegel <triegel@redhat.com>
23432 Richard Henderson <rth@redhat.com>
23434 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23435 conditional on existance of a fast atomic load.
23436 * optabs-query.c (can_atomic_load_p): New function.
23437 * optabs-query.h (can_atomic_load_p): Declare it.
23438 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23439 no fast atomic load is available for the particular size of access.
23440 (expand_atomic_compare_and_swap): Likewise.
23441 (expand_atomic_load): Likewise.
23442 (expand_atomic_store): Likewise.
23443 (expand_atomic_fetch_op): Likewise.
23444 * testsuite/lib/target-supports.exp
23445 (check_effective_target_sync_int_128): Remove x86 because it provides
23446 no fast atomic load.
23447 (check_effective_target_sync_int_128_runtime): Likewise.
23449 2017-02-01 Richard Biener <rguenther@suse.de>
23451 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23452 (graphite_transform_loops): Provide opt-info for optimized nests.
23453 * tree-parloop.c (parallelize_loops): Provide opt-info for
23454 parallelized loops.
23456 2017-02-01 Richard Biener <rguenther@suse.de>
23458 PR middle-end/79315
23459 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23460 was not set before.
23462 2017-02-01 Richard Biener <rguenther@suse.de>
23464 PR tree-optimization/71824
23465 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23466 Verify the loops are valid in the merged SESE region.
23467 (scop_detection::can_represent_loop_1): Check analyzing the
23468 evolution of the number of iterations in the region succeeds.
23470 2017-01-31 Ian Lance Taylor <iant@golang.org>
23472 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23473 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23475 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23477 PR preprocessor/79210
23478 * input.c (get_substring_ranges_for_loc): Replace line_width
23479 assertion with error-handling.
23481 2017-01-31 Richard Biener <rguenther@suse.de>
23483 PR tree-optimization/77318
23484 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23485 (create_pw_aff_from_tree): Take loop parameter.
23486 (add_condition_to_pbb): Pass loop of the condition to
23487 create_pw_aff_from_tree.
23489 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23491 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23492 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23494 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23498 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23500 (convert_int_to_float128): Likewise.
23501 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23502 (convert_int_to_float128): Likewise.
23503 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23504 (UNSPEC_IEEE128_CONVERT): Likewise.
23505 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23506 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23507 Use local variables for IBM extended format.
23508 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23509 (fix_trunc<mode>si2_fprs): Likewise.
23510 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23511 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23512 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23513 to know that we can now have integers of all sizes in vector
23515 (fix<uns>_<mode>di2_hw): Likewise.
23516 (float<uns>_<mode>si2_hw): Likewise.
23517 (fix_<mode>si2_hw): Likewise.
23518 (fixuns_<mode>si2_hw): Likewise.
23519 (float<uns>_<mode>di2_hw): Likewise.
23520 (float_<mode>di2_hw): Likewise.
23521 (float_<mode>si2_hw): Likewise.
23522 (floatuns_<mode>di2_hw): Likewise.
23523 (floatuns_<mode>si2_hw): Likewise.
23524 (xscvqp<su>wz_<mode>): Delete, no longer used.
23525 (xscvqp<su>dz_<mode>): Likewise.
23526 (xscv<su>dqp_<mode>): Likewise.
23527 (ieee128_mfvsrd_64bit): Likewise.
23528 (ieee128_mfvsrd_32bit): Likewise.
23529 (ieee128_mfvsrwz): Likewise.
23530 (ieee128_mtvsrw): Likewise.
23531 (ieee128_mtvsrd_64bit): Likewise.
23532 (ieee128_mtvsrd_32bit): Likewise.
23534 2017-01-31 Martin Liska <mliska@suse.cz>
23537 * ipa-prop.c (ipa_free_all_node_params): Call release method
23538 instead of ~sumbol_summary to not to trigger double times
23541 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23543 PR tree-optimization/71691
23544 * bitmap.h (class auto_bitmap): New.
23545 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23546 is_maybe_undefined instead of ssa_undefined_value_p.
23548 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23550 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23551 __S390_ARCH_LEVEL__ to __ARCH__.
23553 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23555 PR tree-optimization/79267
23556 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23557 if should_remove_lhs_p is true.
23559 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23562 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23563 (add_alignment_attribute): New.
23564 (base_type_die): Add alignment attribute.
23565 (subrange_type_die): Likewise.
23566 (modified_type_die): Likewise.
23567 (gen_array_type_die): Likewise.
23568 (gen_descr_array_type_die: Likewise.
23569 (gen_enumeration_type_die): Likewise.
23570 (gen_subprogram_die): Likewise.
23571 (gen_variable_die): Likewise.
23572 (gen_field_die): Likewise.
23573 (gen_ptr_to_mbr_type_die): Likewise.
23574 (gen_struct_or_union_type_die): Likewise.
23575 (gen_subroutine_type_die): Likewise.
23576 (gen_typedef_die): Likewise.
23577 (base_type_cmp): Compare alignment attribute.
23579 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23582 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23583 (setb_unsigned) New pattern for setb with CCUNS.
23584 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23585 subfc./subfe sequence to avoid overflow problems. Generate a
23586 shorter sequence with cmpld/setb for power9.
23587 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23588 for generating subfc. instruction.
23589 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23590 now uses this instruction.
23592 2017-01-30 Ian Lance Taylor <iant@google.com>
23595 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23596 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23598 2017-01-30 Martin Sebor <msebor@redhat.com>
23600 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23601 Move constant to the right of a relational operator.
23602 (get_mpfr_format_length, format_character, format_string): Ditto.
23603 (should_warn_p, maybe_warn): Same.
23605 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23607 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23610 * asan.c (get_translation_unit_decl): Remove function.
23611 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23613 2017-01-30 Martin Liska <mliska@suse.cz>
23615 PR gcov-profile/79259
23616 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23617 -fprofile-generate.
23619 2017-01-30 Martin Liska <mliska@suse.cz>
23622 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23623 Initialize variables with NULL value.
23625 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23628 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23630 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23632 2017-01-30 Richard Biener <rguenther@suse.de>
23634 PR tree-optimization/79276
23635 * tree-vrp.c (process_assert_insertions): Properly adjust common
23636 when removing a duplicate.
23638 * gcc.dg/torture/pr79276.c: New testcase.
23640 2017-01-30 Richard Biener <rguenther@suse.de>
23642 PR tree-optimization/79256
23643 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23644 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23646 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23648 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23651 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23652 ("*r<noxa>sbg_<mode>_sll_bitmask")
23653 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23654 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23655 Use contiguous_bitmask_nowrap_operand.
23657 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23660 * config/rs6000/altivec.h (vec_xl): Revise #define.
23661 (vec_xst): Likewise.
23663 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23665 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23667 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23669 PR rtl-optimization/79194
23670 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23671 traps before call to bypass_conditional_jumps.
23673 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23675 PR tree-optimization/71374
23676 * lra-constraints.c (check_conflict_input_operands): New.
23677 (match_reload): Use it.
23679 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23682 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23683 account to calculate conflict_set.
23685 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23687 PR rtl-optimization/78559
23688 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23689 other_insn in combine.
23691 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23693 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23694 uint16_type_node for BT_UINT16.
23696 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23698 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23699 "RTL Tests" to menu.
23700 (GIMPLE Tests): New node.
23701 (RTL Tests): New node.
23703 2017-01-27 Richard Biener <rguenther@suse.de>
23705 PR tree-optimization/79245
23706 * tree-loop-distribution.c (distribute_loop): Apply cost
23707 modeling also to detected patterns.
23709 2017-01-27 Richard Biener <rguenther@suse.de>
23711 PR tree-optimization/71433
23712 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23713 (compare_assert_loc): New function.
23714 (process_assert_insertions): Sort and optimize assert locations
23715 to remove duplicates and push down identical assertions on
23716 edges to their destination block.
23718 2017-01-27 Richard Biener <rguenther@suse.de>
23720 PR tree-optimization/79244
23721 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23722 out SSA names even if abnormal.
23724 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23726 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23727 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23728 instead of MPFR_RNDN.
23730 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23733 * arm.c (arm_option_override): Don't call build_target_option_node
23734 until after doing all option overrides.
23735 (arm_valid_target_attribute_tree): Likewise.
23737 2017-01-27 Martin Liska <mliska@suse.cz>
23739 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23740 for {cd}tors and C++ {cd}tors.
23742 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23744 * config/s390/s390.md ("*setmem_long_and")
23745 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23747 2017-01-26 Martin Sebor <msebor@redhat.com>
23749 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23752 2017-01-26 Martin Sebor <msebor@redhat.com>
23754 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23755 HAVE_DFmode before using XFmode or DFmode.
23756 (parse_directive): Avoid using the z length modifier to avoid
23757 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23759 PR middle-end/78703
23760 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23761 to accept adjustment as an array.
23762 (get_int_range): New function.
23763 (struct directive): Make width and prec arrays.
23764 (directive::set_width, directive::set_precision): Call get_int_range.
23765 (format_integer, format_floating): Handle width and precision ranges.
23766 (format_string, parse_directive): Same.
23768 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23771 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23772 comdat_type_p set, just clone them, but keep the children in the
23776 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23777 which have direct callers with -fvar-tracking-assignments enabled
23779 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23780 inside of type units.
23782 2017-01-26 Martin Sebor <msebor@redhat.com>
23784 PR middle-end/78703
23785 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23787 (struct format_result): Replace number_chars, number_chars_min,
23788 and number_chars_max with a single member of struct result_range.
23790 (format_result::operator+=): Adjust.
23791 (struct fmtresult): Remove bounded. Handle likely and unlikely
23793 (fmtresult::adjust_for_width_or_precision): New function.
23794 (fmtresult:type_max_digits): New function.
23795 (bytes_remaining): Handle likely and unlikely counters.
23796 (min_bytes_remaining): Remove.
23797 (format_percent): Simplify.
23798 (format_integer, format_floating): Set likely and unlikely counters.
23799 (get_string_length, format_character, format_string): Same.
23800 (format_plain, should_warn_p): New function.
23801 (maybe_warn): Call should_warn_p. Update diagnostic messages
23802 and handle those for all directives, including plain strings.
23803 (format_directive): Handle likely and unlikely counters.
23804 Remove unnecessary quoting from diagnostics. Add an informational
23806 (add_bytes): Remove.
23807 (pass_sprintf_length::compute_format_length): Simplify.
23808 (try_substitute_return_value): Handle likely and unlikely counters.
23810 2017-01-26 Carl Love <cel@us.ibm.com>
23812 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23813 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23815 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23818 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23819 endianess for subregs into account.
23820 * lra-constraints.c (lra_constraints): Do risky transformations
23821 always on the first iteration.
23822 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23823 last_call_used_reg_set.
23824 (process_bb_lives): Define and use last_call_used_reg_set.
23825 * lra.c (lra): Always continue after lra_constraints on the first
23828 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23830 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23832 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23834 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23836 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23837 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23838 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23839 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23840 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23841 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23842 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23843 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23844 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23846 2017-01-26 Marek Polacek <polacek@redhat.com>
23849 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23850 for the third operand.
23852 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23854 PR middle-end/79236
23855 * omp-low.c (struct omp_context): Add simt_stmt field.
23856 (scan_omp_for): Return omp_context *.
23857 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23858 context to the _simt_ SIMD stmt.
23859 (lower_omp_for): For combined SIMD with sibling _simt_
23860 SIMD, make sure to use the same decls in _looptemp_
23861 clauses as in the sibling.
23863 2017-01-26 David Sherwood <david.sherwood@arm.com>
23865 PR middle-end/79212
23866 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23869 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23872 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23873 emit fld b; fld a; if possible.
23875 * brig-builtins.def: Update copyright years.
23876 * config/arm/arm_acle_builtins.def: Update copyright years.
23878 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23881 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23882 constraint instead of o for the stxsd instruction.
23884 2017-01-25 Carl Love <cel@us.ibm.com>
23886 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23887 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23889 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23891 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23893 2017-01-25 Richard Biener <rguenther@suse.de>
23895 PR tree-optimization/69264
23896 * target.def (vector_alignment_reachable): Improve documentation.
23897 * doc/tm.texi: Regenerate.
23898 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23900 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23901 earlier changes with respect to TYPE_USER_ALIGN.
23902 (vector_alignment_reachable_p): Likewise. Improve dumping.
23904 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23907 * config/arm/arm.md (xordi3): Force constant operand into a register
23910 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23912 * doc/invoke.texi (-fstore-merging): Correct default optimization
23913 levels at which it is enabled.
23914 (-O): Move -fstore-merging from list to...
23917 2017-01-25 Richard Biener <rguenther@suse.de>
23920 * omp-expand.c: Include debug.h.
23921 (expand_omp_taskreg): Make sure to generate early debug before
23922 outlining anything from a function.
23923 (expand_omp_target): Likewise.
23924 (grid_expand_target_grid_body): Likewise.
23926 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23929 * asan.c (get_translation_unit_decl): New function.
23930 (asan_add_global): Extract modules file name from globals
23931 TRANSLATION_UNIT_DECL name.
23933 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23936 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23937 for long calls with APCS frame and VFP.
23939 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23941 * cfg.c (original_copy_tables_initialized_p): New function.
23942 * cfg.h (original_copy_tables_initialized_p): New decl.
23943 * cfgrtl.c (relink_block_chain): Guard the call to
23944 free_original_copy_tables with a call to
23945 original_copy_tables_initialized_p.
23946 * cgraph.h (symtab_node::native_rtl_p): New decl.
23947 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23948 (symtab_node::needed_p): Don't assert for early assembly output
23949 for __RTL functions.
23950 (cgraph_node::finalize_function): Set "force_output" for __RTL
23952 (cgraph_node::analyze): Bail out early for __RTL functions.
23953 (analyze_functions): Update assertion to support __RTL functions.
23954 (cgraph_node::expand): Bail out early for __RTL functions.
23955 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23957 * function.h (struct function): Update comment for field
23959 * gimple-expr.c: Include "tree-pass.h".
23960 (gimple_has_body_p): Return false for __RTL functions.
23961 * Makefile.in (OBJS): Add run-rtl-passes.o.
23962 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23964 (gcc::pass_manager::get_clean_slate): New accessor.
23965 * passes.c: Include "insn-addr.h".
23966 (should_skip_pass_p): Add logging. Update logic for running
23967 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23968 property-provider override so it is only done for gimple passes.
23970 (skip_pass): New function.
23971 (execute_one_pass): Call skip_pass when skipping passes.
23972 * read-md.c (md_reader::read_char): Support filtering
23973 the input to a subset of line numbers.
23974 (md_reader::md_reader): Initialize fields
23975 m_first_line and m_last_line.
23976 (md_reader::read_file_fragment): New function.
23977 * read-md.h (md_reader::read_file_fragment): New decl.
23978 (md_reader::m_first_line): New field.
23979 (md_reader::m_last_line): New field.
23980 * read-rtl-function.c (function_reader::create_function): Only
23981 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23982 curr_properties. Set DECL_INITIAL to a dummy block.
23983 (read_rtl_function_body_from_file_range): New function.
23984 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23986 * run-rtl-passes.c: New file.
23987 * run-rtl-passes.h: New file.
23989 2017-01-24 Jeff Law <law@redhat.com>
23991 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23994 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23996 PR tree-optimization/79159
23997 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23998 (record_nonwrapping_iv): Improve boundary using above function if no
23999 value range information.
24001 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24002 Martin Jambor <mjambor@suse.cz>
24004 * brig-builtins.def: New file.
24005 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24006 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24007 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24008 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24009 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24010 * builtin-types.def (BT_INT8): New.
24011 (BT_INT16): Likewise.
24012 (BT_UINT8): Likewise.
24013 (BT_UINT16): Likewise.
24014 (BT_FN_ULONG): Likewise.
24015 (BT_FN_UINT_INT): Likewise.
24016 (BT_FN_UINT_ULONG): Likewise.
24017 (BT_FN_UINT_LONG): Likewise.
24018 (BT_FN_UINT_PTR): Likewise.
24019 (BT_FN_ULONG_PTR): Likewise.
24020 (BT_FN_INT8_FLOAT): Likewise.
24021 (BT_FN_INT16_FLOAT): Likewise.
24022 (BT_FN_UINT32_FLOAT): Likewise.
24023 (BT_FN_UINT16_FLOAT): Likewise.
24024 (BT_FN_UINT8_FLOAT): Likewise.
24025 (BT_FN_UINT64_FLOAT): Likewise.
24026 (BT_FN_UINT16_UINT32): Likewise.
24027 (BT_FN_UINT32_UINT16): Likewise.
24028 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24029 (BT_FN_INT_PTR_INT): Likewise.
24030 (BT_FN_UINT_PTR_UINT): Likewise.
24031 (BT_FN_LONG_PTR_LONG): Likewise.
24032 (BT_FN_ULONG_PTR_ULONG): Likewise.
24033 (BT_FN_VOID_UINT64_UINT64): Likewise.
24034 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24035 (BT_FN_INT8_INT8_INT8): Likewise.
24036 (BT_FN_INT16_INT16_INT16): Likewise.
24037 (BT_FN_INT_INT_INT): Likewise.
24038 (BT_FN_UINT_FLOAT_UINT): Likewise.
24039 (BT_FN_FLOAT_UINT_UINT): Likewise.
24040 (BT_FN_ULONG_UINT_UINT): Likewise.
24041 (BT_FN_ULONG_UINT_PTR): Likewise.
24042 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24043 (BT_FN_UINT_UINT_UINT): Likewise.
24044 (BT_FN_VOID_UINT_PTR): Likewise.
24045 (BT_FN_UINT_UINT_PTR: Likewise.
24046 (BT_FN_UINT32_UINT64_PTR): Likewise.
24047 (BT_FN_INT_INT_UINT_UINT): Likewise.
24048 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24049 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24050 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24051 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24052 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24053 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24054 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24055 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24056 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24057 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24058 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24059 * doc/frontends.texi: List BRIG FE.
24060 * doc/install.texi (Testing): Add BRIG tesring requirements.
24061 * doc/invoke.texi (Overall Options): Mention BRIG.
24062 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24064 2017-01-24 Richard Biener <rguenther@suse.de>
24066 PR translation/79208
24067 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24069 2017-01-24 Martin Jambor <mjambor@suse.cz>
24072 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24073 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24074 and known_contexts.
24076 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24078 PR middle-end/79123
24079 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24080 casts from signed to unsigned really don't have a range.
24082 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24084 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24085 GMP_RNDx for compatiblity.
24087 2017-01-24 Martin Liska <mliska@suse.cz>
24090 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24091 that would prevent us to call alloca with -1 as argument.
24093 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24095 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24096 Avoid -Wformat-security warning.
24098 2017-01-23 Andrew Pinski <apinski@cavium.com>
24100 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24103 2017-01-23 Martin Sebor <msebor@redhat.com>
24105 PR middle-end/78703
24106 * gimple-ssa-sprintf.c (warn_level): New global.
24107 (format_integer): Use it here and throughout the rest of the file.
24108 Use the same switch to compute sign as base.
24109 (maybe_warn): New function.
24110 (format_directive): Factor out warnings into maybe_warn.
24111 Add debugging output. Use warn_level.
24112 (add_bytes): Use warn_level.
24113 (pass_sprintf_length::compute_format_length): Add debugging output.
24114 (try_substitute_return_value): Same.
24115 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24117 PR middle-end/78703
24118 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24119 (struct fmtresult, format_integer, format_floating): Adjust.
24120 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24121 (get_string_length, format_string,format_directive): Same.
24122 (pass_sprintf_length::compute_format_length): Same.
24123 (try_substitute_return_value): Simplify slightly.
24125 PR middle-end/78703
24126 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24127 (fmtresult::operator+=): Outlined.
24128 (struct fmtresult): Add ctors.
24129 (struct conversion_spec): Rename...
24130 (struct directive): ...to this. Add and remove data members.
24131 (directive::set_width, directive::set_precision): New functions.
24132 (format_percent): Use fmtresult ctor.
24133 (get_width_and_precision): Remove.
24134 (format_integer): Make naming changes. Avoid computing width and
24136 (format_floating): Same. Adjust indentation.
24137 (format_character, format_none): New functions.
24138 (format_string): Moved character handling to format_character.
24139 (format_directive): Remove arguments, change return type.
24140 (parse_directive): New function.
24141 (pass_sprintf_length::compute_format_length): Move directive
24142 parsing to parse_directive.
24144 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24146 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24147 (assign_assembler_name_if_needed): ... this.
24148 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24149 (assign_assembler_name_if_needed): ... this.
24150 (free_lang_data_in_cgraph): Adjust callers.
24151 * cgraphunit.c (cgraph_node::analyze): Likewise.
24152 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24155 2017-01-23 Richard Biener <rguenther@suse.de>
24157 PR tree-optimization/79088
24158 PR tree-optimization/79188
24159 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24160 resetting loop bounds after last path deletion. Reset loop
24161 bounds of the target loop, make code match the comments.
24162 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24163 Make sure loops need no fixups.
24165 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24167 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24168 exponent support with double type for first argument.
24169 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24170 type returned by __builtin_vec_extract_sig,
24171 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24172 functions from "vector int" to "vector unsigned int" or from
24173 "vector long long int" to "vector unsigned long long int".
24174 Changed type returned by __builtin_vec_extract_exp,
24175 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24176 functions from "vector int" to "vector unsigned int" or from
24177 "vector long long int" to "vector unsigned long long int".
24178 Changed return type of __builtin_vec_test_data_class,
24179 __builtin_vec_test_data_class_sp, and
24180 __builtin_vec_test_data_class_dp from "vector int" to
24181 "vector bool int" or from "vector long long int" to "vector bool
24182 long long int" and changed second argument type from "unsigned
24183 int" to "int". Added new overloaded function forms "vector float
24184 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24185 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24186 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24187 double, vector unsigned long long int)" and "vector double
24188 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24189 long int)". Changed return type of
24190 __builtin_scalar_test_data_class and
24191 __builtin_scalar_test_data_class_sp and
24192 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24193 int" and changed second argument from "unsigned int" to "int".
24194 Changed type returned by __builtin_scalar_test_neg,
24195 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24196 from "int" to "bool int". Added new overloaded function form
24197 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24198 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24199 exponent double-precision with floating point first argument.
24200 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24201 documentation of scalar_test_data_class, scalar_test_neg,
24202 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24203 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24204 vec_test_data_class built-in functions to reflect refinements in
24205 their type signatures.
24207 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24209 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24211 (aarch64_elf_asm_destructor): Likewise.
24213 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24215 PR rtl-optimization/78634
24216 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24217 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24218 * ifcvt.c (noce_try_cmove): Add missing cost check.
24220 PR rtl-optimization/71724
24221 * combine.c (if_then_else_cond): Look for situations where it is
24222 beneficial to undo the work of one of the recursive calls.
24224 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24226 PR tree-optimization/70754
24227 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24228 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24229 combined stmt before it if not NULL.
24230 (combine_chains): Process refs reversely and compute dominance point
24233 2017-01-23 Martin Liska <mliska@suse.cz>
24235 PR tree-optimization/79196
24236 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24237 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24239 (strlen_optimize_stmt): Call the renamed function.
24241 2017-01-23 Michael Matz <matz@suse.de>
24243 PR tree-optimization/78384
24244 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24246 2017-01-23 Richard Biener <rguenther@suse.de>
24248 PR tree-optimization/79186
24249 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24250 both incoming edges before moving an assert.
24252 2017-01-23 Martin Jambor <mjambor@suse.cz>
24254 * ipa-prop.c (load_from_param_1): Removed.
24255 (load_from_unmodified_param): Bits from load_from_param_1 put back
24257 (load_from_param): Removed.
24258 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24259 with stmt. Reverted back to use of load_from_unmodified_param.
24261 2017-01-23 Martin Jambor <mjambor@suse.cz>
24264 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24265 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24266 field a pointer to garbage collected vector, mark lattices and
24267 ipcp_orig_node with GTY((skip)).
24268 (ipa_get_param_count): Adjust to descriptors being a pointer.
24269 (ipa_get_param): Likewise.
24270 (ipa_get_type): Likewise.
24271 (ipa_get_param_move_cost): Likewise.
24272 (ipa_set_param_used): Likewise.
24273 (ipa_get_controlled_uses): Likewise.
24274 (ipa_set_controlled_uses): Likewise.
24275 (ipa_is_param_used): Likewise.
24276 (ipa_node_params_t): Move into garbage collector. New methods insert
24278 (ipa_node_params_sum): Annotate wth GTY(()).
24279 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24281 (ipa_load_from_parm_agg): Adjust declaration.
24282 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24283 * ipa-profile.c (ipa_profile): Likewise.
24284 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24285 (ipa_populate_param_decls): Make descriptors parameter garbage
24287 (ipa_dump_param): Adjust to descriptors being a pointer.
24288 (ipa_alloc_node_params): Likewise.
24289 (ipa_initialize_node_params): Likewise.
24290 (load_from_param_1): Make descriptors parameter garbage collected.
24291 (load_from_unmodified_param): Likewise.
24292 (load_from_param): Likewise.
24293 (ipa_load_from_parm_agg): Likewise.
24294 (ipa_node_params::~ipa_node_params): Removed.
24295 (ipa_free_all_node_params): Remove call to delete operator.
24296 (ipa_node_params_t::insert): New.
24297 (ipa_node_params_t::remove): Likewise.
24298 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24299 copy known_csts and known_contexts vectors.
24300 (ipa_read_node_info): Adjust to descriptors being a pointer.
24301 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24303 (ipcp_transform_function): Make descriptors variable garbage
24306 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24308 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24309 * config/i386/avx512dqintrin.h: Ditto.
24310 * config/i386/avx512fintrin.h: Ditto.
24311 * config/i386/i386.c: Handle new builtins.
24312 * config/i386/i386-builtin.def: Add new builtins.
24313 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24314 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24316 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24317 Martin Liska <mliska@suse.cz>
24319 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24320 * asan.c (asan_expand_poison_ifn): Support stores and use
24321 appropriate ASAN report function.
24322 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24323 * internal-fn.def (ASAN_POISON_USE): Declare.
24324 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24325 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24326 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24327 ASAN_POISON calls w/o LHS.
24328 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24329 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24330 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24331 * gimplify.c (asan_poison_variables): Add attribute
24332 use_after_scope_memory to variables that really needs to live
24334 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24335 having the attribute.
24337 2017-01-23 Martin Liska <mliska@suse.cz>
24339 * asan.c (create_asan_shadow_var): New function.
24340 (asan_expand_poison_ifn): Likewise.
24341 * asan.h (asan_expand_poison_ifn): New declaration.
24342 * internal-fn.c (expand_ASAN_POISON): Likewise.
24343 * internal-fn.def (ASAN_POISON): New builtin.
24344 * sanopt.c (pass_sanopt::execute): Expand
24345 asan_expand_poison_ifn.
24346 * tree-inline.c (copy_decl_for_dup_finish): Make function
24348 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24349 * tree-ssa.c (is_asan_mark_p): New function.
24350 (execute_update_addresses_taken): Rewrite local variables
24351 (identified just by use-after-scope as addressable) into SSA.
24353 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24355 * doc/install.texi (Specific): opensource.apple.com uses https
24356 now. Remove trailing slash.
24358 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24360 * README.Portability: Remove note on an Irix compatibility issue.
24362 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24364 * gcov.c (INCLUDE_ALGORITHM): Define.
24365 (INCLUDE_VECTOR): Define.
24366 No longer include <vector> and <algorithm> directly.
24368 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24370 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24372 * doc/invoke.texi (Code Gen Options): Ditto.
24374 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24377 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24379 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24381 rtl-optimization/79125
24382 * cprop.c (local_cprop_pass): Handle cases where we make an
24383 unconditional trap.
24385 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24389 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24390 read from, for big endian.
24392 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24394 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24395 register pauth builtins for LP64 only.
24397 2017-01-20 Marek Polacek <polacek@redhat.com>
24400 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24403 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24405 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24407 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24408 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24409 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24411 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24414 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24415 in big-endian mode when they are not a single duplicated value.
24417 2017-01-20 Richard Biener <rguenther@suse.de>
24419 * BASE-VER: Bump to 7.0.1.
24421 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24423 * omp-low.c (omplow_simd_context): New struct. Use it...
24424 (lower_rec_simd_input_clauses): ...here and...
24425 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24426 references to idx, lane, max_vf, is_simt.
24428 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24430 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24433 2017-01-20 Martin Jambor <mjambor@suse.cz>
24435 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24436 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24438 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24439 (GTFILES): Rename hsa.c to hsa-common.c.
24440 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24441 * hsa-dump.c: Likewise.
24442 * hsa-gen.c: Likewise.
24443 * hsa-regalloc.c: Likewise.
24444 * ipa-hsa.c: Likewise.
24445 * omp-expand.c: Likewise.
24446 * omp-low.c: Likewise.
24447 * toplev.c: Likewise.
24449 2017-01-20 Marek Polacek <polacek@redhat.com>
24452 * doc/invoke.texi: Document -Wduplicated-branches.
24453 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24454 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24455 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24456 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24457 return 0 only when not OEP_LEXICOGRAPHIC.
24458 (fold_build_cleanup_point_expr): Use the expression
24459 location when building CLEANUP_POINT_EXPR.
24460 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24461 * tree.c (add_expr): Handle error_mark_node.
24463 2017-01-20 Martin Liska <mliska@suse.cz>
24466 * tree-profile.c (init_ic_make_global_vars): Do not call
24468 (gimple_init_gcov_profiler): Likewise.
24470 2017-01-20 Martin Liska <mliska@suse.cz>
24473 * cgraph.h (maybe_create_reference): Remove argument and
24475 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24477 * ipa-cp.c (create_specialized_node): Likewise.
24478 * symtab.c (symtab_node::maybe_create_reference): Handle
24479 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24481 2017-01-20 Martin Liska <mliska@suse.cz>
24483 * read-rtl-function.c (function_reader::create_function): Use
24484 build_decl instread of build_decl_stat.
24486 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24488 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24489 * config/i386/avx512dqintrin.h: Ditto.
24490 * config/i386/avx512fintrin.h: Ditto.
24491 * config/i386/i386-builtin-types.def: Add new types.
24492 * config/i386/i386.c: Handle new types.
24493 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24494 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24495 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24496 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24497 (__builtin_ia32_kshiftridi): New.
24498 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24500 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24504 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24505 define to rs6000_init_stack_protect_guard.
24506 (rs6000_init_stack_protect_guard): New function.
24508 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24509 Yunqiang Su <yunqiang.su@imgtec.com>
24511 * config.gcc (supported_defaults): Add madd4.
24512 (with_madd4): Add validation.
24513 (all_defaults): Add madd4.
24514 * config/mips/mips.opt (mmadd4): New option.
24515 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24517 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24519 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24520 (ISA_HAS_FUSED_MADD4): Likewise.
24521 * doc/invoke.texi (-mmadd4): Document the new option.
24522 * doc/install.texi (--with-madd4): Document the new option.
24524 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24526 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24527 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24528 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24529 (aarch64_init_pauth_hint_builtins): New.
24530 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24531 (aarch64_expand_builtin): Expand new builtins.
24533 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24535 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24536 * combine-stack-adj.c (no_unhandled_cfa): Handle
24537 REG_CFA_TOGGLE_RA_MANGLE.
24538 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24539 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24540 info for return address signing.
24541 (aarch64_expand_epilogue): Likewise.
24543 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24545 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24546 * config/aarch64/aarch64-protos.h
24547 (aarch64_return_address_signing_enabled): New declaration.
24548 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24550 (aarch64_expand_prologue): Sign return address before it's pushed onto
24552 (aarch64_expand_epilogue): Authenticate return address fetched from
24554 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24555 (aarch64_attributes): New function attributes for "sign-return-address".
24556 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24557 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24558 ("*do_return"): Generate combined instructions according to key index.
24559 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24560 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24562 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24563 * config/aarch64/aarch64.opt (msign-return-address=): New.
24564 * doc/extend.texi (AArch64 Function Attributes): Documents
24565 "sign-return-address=".
24566 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24568 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24570 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24571 overall option summary.
24573 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24575 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24576 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24577 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24578 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24580 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24582 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24583 -mpower9-minmax by default for -mcpu=power9.
24584 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24585 128-bit floating point.
24587 2017-01-20 Alan Modra <amodra@gmail.com>
24589 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24590 optimizing for size.
24592 2017-01-20 Alan Modra <amodra@gmail.com>
24595 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24596 for strcmp and strncmp from corresponding builtin decl.
24598 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24600 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24601 instead of i386/rtems-64.h.
24602 * config/i386/rtems-64.h: Remove.
24604 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24608 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24610 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24612 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24614 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24615 Change int to HOST_WIDE_INT.
24616 * config/aarch64/aarch64-protos.h
24617 (aarch64_simd_gen_const_vector_dup): Likewise.
24618 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24620 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24622 * langhooks-def.h (lhd_type_for_size): New decl.
24623 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24624 * langhooks.c (lhd_type_for_size): New function, taken from
24627 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24629 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24630 define_bypass for CR latency.
24631 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24632 (power9-alu2): Add define_bypass for CR latency.
24634 (power9-mul): Update insn latency.
24635 (power9-mul-compare): Update insn latency, bypass latency and remove
24638 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24640 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24642 * config/aarch64/aarch64.md
24643 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24644 aarch64_nopcrelative_literal_loads.
24645 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24647 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24649 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24650 TARGET_LOONGSON_3A.
24651 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24653 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24656 * config.gcc (supported_defaults): Add lxc1-sxc1.
24657 (with_lxc1_sxc1): Add validation.
24658 (all_defaults): Add lxc1-sxc1.
24659 * config/mips/mips.opt (mlxc1-sxc1): New option.
24660 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24662 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24663 __mips_no_lxc1_sxc1.
24664 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24665 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24666 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24668 2017-01-19 Richard Biener <rguenther@suse.de>
24670 PR tree-optimization/72488
24671 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24672 sure to restore SSA info.
24673 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24675 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24677 PR rtl-optimization/79121
24678 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24679 of the inner type when shifting an extended value.
24681 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24684 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24685 interposable aliases.
24687 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24690 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24691 Use the evmergelohi instruction.
24692 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24693 (mov_sitf_e500_subreg8_2_be): Likewise.
24694 (mov_sitf_e500_subreg12_2_le): Likewise.
24695 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24696 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24697 (mov_sitf_e500_subreg8_2_le): Likewise.
24698 (mov_sitf_e500_subreg12_2_be): Likewise.
24700 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24702 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24703 attribute from vecsimple to vecperm.
24704 (altivec_vbpermq2): Likewise.
24706 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24709 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24711 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24712 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24713 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24714 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24715 case where N arg is SIZE_MAX.
24716 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24717 (cmpstrsi): Add pattern.
24719 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24721 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24722 __builtin_vec_revb builtins.
24723 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24724 built-in functions to support generation of the ISA 3.0 XXBR<x>
24725 vector byte reverse instructions.
24726 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24727 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24728 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24729 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24730 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24731 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24732 (P9V_BUILTIN_VEC_REVB): Likewise.
24733 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24734 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24735 (p9_xxbrq_v16qi): Likewise.
24736 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24737 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24738 (p9_xxbrh_v8hi): Likewise.
24739 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24740 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24741 vec_revb built-in functions.
24743 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24745 PR rtl-optimization/78952
24746 * config/i386/i386.md (any_extract): New code iterator.
24747 (*insvqi_2): Use any_extract for source operand.
24748 (*insvqi_3): Use any_shiftrt for source operand.
24750 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24752 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24754 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24756 2017-01-18 Matthias Klose <doko@ubuntu.com>
24758 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24760 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24762 * config/rs6000/altivec.h (vec_bperm): Change #define.
24763 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24764 (altivec_vbpermq2): New define_insn.
24765 (altivec_vbpermd): Likewise.
24766 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24767 function interface.
24768 (VBPERMD): Likewise.
24769 (VBPERM): New polymorphic function interface.
24770 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24771 Add entries for P9V_BUILTIN_VEC_VBPERM.
24772 * doc/extend.texi: Add interfaces for vec_bperm.
24774 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24776 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24777 first letter of error messages.
24778 (s390_resolve_overloaded_builtin): Likewise.
24779 * config/s390/s390.c (s390_expand_builtin): Likewise.
24780 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24781 (s390_valid_target_attribute_inner_p): Likewise.
24782 * config/s390/s390.md ("tabort"): Likewise.
24784 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24786 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24787 (ISA_AVOID_DIV_HILO): New macro.
24788 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24789 (ISA_HAS_DDIV): Likewise.
24791 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24793 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24795 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24797 * doc/invoke.texi (fabi-version): Spelling fix.
24799 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24802 * doc/invoke.texi (fabi-version): Mention mangling fix for
24805 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24808 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24810 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24813 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24814 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24816 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24818 (stack_protect_test): Ditto.
24819 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24820 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24822 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24823 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24824 -mstack-protector-guard-offset=.
24825 (RS/6000 and PowerPC Options): Ditto.
24827 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24829 * config/i386/i386.h (MASK_CLASS_P): New define.
24830 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24831 there are no registers from different register sets also when
24832 mask registers are used. Update function comment.
24833 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24834 to (*k/*r) and (*k/*km) alternatives.
24836 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24838 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24839 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24840 (EH_RETURN_HANDLER_RTX): New define.
24841 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24842 Force frame pointer in EH return functions.
24843 (aarch64_expand_epilogue): Add barrier for eh_return.
24844 (aarch64_final_eh_return_addr): Remove.
24845 (aarch64_eh_return_handler_rtx): New function.
24846 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24848 (aarch64_eh_return_handler_rtx): New prototype.
24850 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24852 * config/rs6000/altivec.h (vec_rlmi): New #define.
24853 (vec_vrlnm): Likewise.
24854 (vec_rlnm): Likewise.
24855 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24856 (UNSPEC_VRLNM): Likewise.
24857 (VIlong): New mode iterator.
24858 (altivec_vrl<VI_char>mi): New define_insn.
24859 (altivec_vrl<VI_char>nm): Likewise.
24860 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24862 (VRLDNM): Likewise.
24863 (RLNM): New polymorphic function entry.
24864 (VRLWMI): New monomorphic function entry.
24865 (VRLDMI): Likewise.
24866 (RLMI): New polymorphic function entry.
24867 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24868 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24869 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24872 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24875 * dwarf2out.c (field_byte_offset): Restore the
24876 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24877 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24878 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24881 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24884 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24886 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24889 * dwarf2out.c (add_data_member_location_attribute): For constant
24890 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24891 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24892 DW_AT_byte_size attributes.
24894 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24896 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24897 after forcing to constant memory when the code model is medium.
24899 2017-01-17 Julia Koval <julia.koval@intel.com>
24902 * config/i386/avx512fintrin.h
24903 (_mm512_i32gather_ps): Change __addr type to void const*.
24904 (_mm512_mask_i32gather_ps): Ditto.
24905 (_mm512_i32gather_pd): Ditto.
24906 (_mm512_mask_i32gather_pd): Ditto.
24907 (_mm512_i64gather_ps): Ditto.
24908 (_mm512_mask_i64gather_ps): Ditto.
24909 (_mm512_i64gather_pd): Ditto.
24910 (_mm512_mask_i64gather_pd): Ditto.
24911 (_mm512_i32gather_epi32): Ditto.
24912 (_mm512_mask_i32gather_epi32): Ditto.
24913 (_mm512_i32gather_epi64): Ditto.
24914 (_mm512_mask_i32gather_epi64): Ditto.
24915 (_mm512_i64gather_epi32): Ditto.
24916 (_mm512_mask_i64gather_epi32): Ditto.
24917 (_mm512_i64gather_epi64): Ditto.
24918 (_mm512_mask_i64gather_epi64): Ditto.
24919 (_mm512_i32scatter_ps): Change __addr type to void*.
24920 (_mm512_mask_i32scatter_ps): Ditto.
24921 (_mm512_i32scatter_pd): Ditto.
24922 (_mm512_mask_i32scatter_pd): Ditto.
24923 (_mm512_i64scatter_ps): Ditto.
24924 (_mm512_mask_i64scatter_ps): Ditto.
24925 (_mm512_i64scatter_pd): Ditto.
24926 (_mm512_mask_i64scatter_pd): Ditto.
24927 (_mm512_i32scatter_epi32): Ditto.
24928 (_mm512_mask_i32scatter_epi32): Ditto.
24929 (_mm512_i32scatter_epi64): Ditto.
24930 (_mm512_mask_i32scatter_epi64): Ditto.
24931 (_mm512_i64scatter_epi32): Ditto.
24932 (_mm512_mask_i64scatter_epi32): Ditto.
24933 (_mm512_i64scatter_epi64): Ditto.
24934 (_mm512_mask_i64scatter_epi64): Ditto.
24935 * config/i386/avx512pfintrin.h
24936 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24937 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24938 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24939 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24940 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24941 (_mm512_prefetch_i32scatter_ps): Ditto.
24942 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24943 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24944 (_mm512_prefetch_i64scatter_pd): Ditto.
24945 (_mm512_prefetch_i64scatter_ps): Ditto.
24946 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24947 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24948 * config/i386/avx512vlintrin.h
24949 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24950 (_mm_mmask_i32gather_ps): Ditto.
24951 (_mm256_mmask_i32gather_pd): Ditto.
24952 (_mm_mmask_i32gather_pd): Ditto.
24953 (_mm256_mmask_i64gather_ps): Ditto.
24954 (_mm_mmask_i64gather_ps): Ditto.
24955 (_mm256_mmask_i64gather_pd): Ditto.
24956 (_mm_mmask_i64gather_pd): Ditto.
24957 (_mm256_mmask_i32gather_epi32): Ditto.
24958 (_mm_mmask_i32gather_epi32): Ditto.
24959 (_mm256_mmask_i32gather_epi64): Ditto.
24960 (_mm_mmask_i32gather_epi64): Ditto.
24961 (_mm256_mmask_i64gather_epi32): Ditto.
24962 (_mm_mmask_i64gather_epi32): Ditto.
24963 (_mm256_mmask_i64gather_epi64): Ditto.
24964 (_mm_mmask_i64gather_epi64): Ditto.
24965 (_mm256_i32scatter_ps): Change __addr type to void*.
24966 (_mm256_mask_i32scatter_ps): Ditto.
24967 (_mm_i32scatter_ps): Ditto.
24968 (_mm_mask_i32scatter_ps): Ditto.
24969 (_mm256_i32scatter_pd): Ditto.
24970 (_mm256_mask_i32scatter_pd): Ditto.
24971 (_mm_i32scatter_pd): Ditto.
24972 (_mm_mask_i32scatter_pd): Ditto.
24973 (_mm256_i64scatter_ps): Ditto.
24974 (_mm256_mask_i64scatter_ps): Ditto.
24975 (_mm_i64scatter_ps): Ditto.
24976 (_mm_mask_i64scatter_ps): Ditto.
24977 (_mm256_i64scatter_pd): Ditto.
24978 (_mm256_mask_i64scatter_pd): Ditto.
24979 (_mm_i64scatter_pd): Ditto.
24980 (_mm_mask_i64scatter_pd): Ditto.
24981 (_mm256_i32scatter_epi32): Ditto.
24982 (_mm256_mask_i32scatter_epi32): Ditto.
24983 (_mm_i32scatter_epi32): Ditto.
24984 (_mm_mask_i32scatter_epi32): Ditto.
24985 (_mm256_i32scatter_epi64): Ditto.
24986 (_mm256_mask_i32scatter_epi64): Ditto.
24987 (_mm_i32scatter_epi64): Ditto.
24988 (_mm_mask_i32scatter_epi64): Ditto.
24989 (_mm256_i64scatter_epi32): Ditto.
24990 (_mm256_mask_i64scatter_epi32): Ditto.
24991 (_mm_i64scatter_epi32): Ditto.
24992 (_mm_mask_i64scatter_epi32): Ditto.
24993 (_mm256_i64scatter_epi64): Ditto.
24994 (_mm256_mask_i64scatter_epi64): Ditto.
24995 (_mm_i64scatter_epi64): Ditto.
24996 (_mm_mask_i64scatter_epi64): Ditto.
24997 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24998 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24999 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25000 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25001 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25002 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25003 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25004 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25005 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25006 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25007 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25008 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25009 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25010 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25011 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25012 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25013 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25014 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25015 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25016 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25017 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25018 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25019 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25020 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25021 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25022 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25023 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25024 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25025 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25026 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25027 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25028 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25029 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25030 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25031 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25032 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25033 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25034 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25035 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25036 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25037 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25038 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25039 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25040 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25041 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25042 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25043 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25044 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25045 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25046 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25047 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25048 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25049 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25050 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25051 definitions accordingly.
25053 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25054 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25057 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25060 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25063 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25064 hard regs for inner regno.
25066 2017-01-17 Martin Liska <mliska@suse.cz>
25069 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25070 assumption and add comment.
25072 2017-01-17 Nathan Sidwell <nathan@acm.org>
25074 * ipa-visibility.c (localize_node): New function, broken out of ...
25075 (function_and_variable_visibility): ... here. Call it.
25077 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25079 PR middle-end/77445
25080 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25081 correctly set frequency of oudgoing edge.
25082 (duplicate_thread_path): Fix profile updating.
25084 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25087 * configure.ac: Add GCC_BASE_VER.
25088 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25089 version from BASE-VER file.
25090 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25091 (gcc.o): Depend on $(BASEVER).
25092 * common.opt (dumpfullversion): New option.
25093 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25094 * doc/invoke.texi: Document -dumpfullversion.
25095 * doc/install.texi: Document --with-gcc-major-version-only.
25096 * configure: Regenerated.
25098 2017-01-17 Richard Biener <rguenther@suse.de>
25100 PR tree-optimization/71433
25101 * tree-vrp.c (register_new_assert_for): Merge same asserts
25102 on all incoming edges.
25103 (process_assert_insertions_for): Handle insertions at the
25106 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25108 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25109 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25111 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25114 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25117 2017-01-17 Alan Modra <amodra@gmail.com>
25120 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25121 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25122 symbolic stack limit when pic.
25124 2017-01-16 Martin Sebor <msebor@redhat.com>
25126 PR tree-optimization/78608
25127 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25129 2017-01-16 Jeff Law <law@redhat.com>
25132 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25133 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25134 for several include directories that may be relative to sysroot.
25135 * config/i386/x-mingw32 (gplus_includedir): Define.
25136 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25137 (native_system_includedir): Likewise.
25138 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25139 override if TARGET_SYSTEM_ROOT is defined.
25140 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25142 PR tree-optimization/79090
25143 PR tree-optimization/33562
25144 PR tree-optimization/61912
25145 PR tree-optimization/77485
25146 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25147 and computed trims into the dump file.
25149 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25151 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25153 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25156 * gimplify.c (gimplify_init_constructor): If want_value and
25157 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25161 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25162 sequence. Formatting fixes.
25163 (doloop_optimize): Formatting fixes.
25166 * gcc.c (debug_level_greater_than_spec_func): New function.
25167 (static_spec_functions): Add debug-level-gt spec function.
25168 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25170 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25171 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25172 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25173 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25174 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25175 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25177 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25179 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25180 QImode fixups to general and mask registers only.
25182 2017-01-16 Carl Love <cel@us.ibm.com>
25184 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25185 for built-in functions
25186 vector signed char vec_nabs (vector signed char)
25187 vector signed short vec_nabs (vector signed short)
25188 vector signed int vec_nabs (vector signed int)
25189 vector signed long long vec_nabs (vector signed long long)
25190 vector float vec_nabs (vector float)
25191 vector double vec_nabs (vector double)
25192 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25194 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25195 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25196 * doc/extend.texi: Update the documentation file for the new built-in
25199 2017-01-16 Martin Sebor <msebor@redhat.com>
25201 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25204 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25206 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25207 UNSPEC_VSX__XXSPLTD to require special splat handling.
25209 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25212 * system.h: Poison strndup.
25214 2017-01-16 Alan Modra <amodra@gmail.com>
25217 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25220 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25222 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25224 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25226 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25227 call recog here. Assert that INSN_CODE (insn) is non-negative.
25229 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25232 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25234 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25235 in the currently scheduled RTL fragment.
25237 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25239 PR rtl-optimization/78751
25240 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25243 2017-01-14 Jeff Law <law@redhat.com>
25245 PR tree-optimization/79090
25246 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25247 variable length stores.
25248 (compute_trims): Delete dead assignment to *trim_tail.
25249 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25252 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25254 PR rtl-optimization/78626
25255 PR rtl-optimization/78727
25256 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25257 of a block, and split such blocks after everything else is finished.
25259 2017-01-14 Alan Modra <amodra@gmail.com>
25262 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25263 target legitimate_combined_insn.
25264 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25265 (rs6000_legitimate_combined_insn): New function.
25266 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25268 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25269 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25270 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25272 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25274 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25276 2017-01-13 Jeff Law <law@redhat.com>
25278 PR tree-optimization/33562
25279 PR tree-optimization/61912
25280 PR tree-optimization/77485
25281 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25283 (delete_dead_assignment): Likewise.
25284 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25285 statement to delete_dead_call and delete_dead_assignment.
25287 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25290 * substring-locations.c (format_warning_va): Strengthen case 1 so
25291 that both endpoints of the substring must be within the format
25292 range for just the substring to be printed.
25294 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25296 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25297 * config/i386/i386.c (ix86_target_string): Add missing options
25298 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25299 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25300 flags_other and ix86_target_other to flags2_other. Display unknown
25302 (ix86_valid_target_attribute_inner_p): Add missing options and
25303 reorder options by implied ISAs, as in ix86_target_string.
25305 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25307 * hash-table.h (hash_table::too_empty_p): New function.
25308 (hash_table::expand): Use it.
25309 (hash_table::traverse): Likewise.
25310 (hash_table::empty_slot): Use sizeof (value_type) instead of
25311 sizeof (PTR) to convert bytes to elements. Shrink the table
25312 if the current size is excessive for the current number of
25315 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25317 * ira-costs.c (record_reg_classes): Break from the inner loop
25318 early once alt_fail is known to be true. Update outer loop
25319 handling accordingly.
25321 2017-01-13 Jeff Law <law@redhat.com>
25323 * tree-ssa-dse.c (decrement_count): New function.
25324 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25325 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25326 when we know the partially dead statement is a mem* function.
25328 PR tree-optimization/61912
25329 PR tree-optimization/77485
25330 * tree-ssa-dse.c: Include expr.h.
25331 (maybe_trim_constructor_store): New function.
25332 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25334 PR tree-optimization/33562
25335 PR tree-optimization/61912
25336 PR tree-optimization/77485
25337 * doc/invoke.texi: Document new dse-max-object-size param.
25338 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25339 * tree-ssa-dse.c: Include params.h.
25340 (dse_store_status): New enum.
25341 (initialize_ao_ref_for_dse): New, partially extracted from
25343 (valid_ao_ref_for_dse, normalize_ref): New.
25344 (setup_live_bytes_from_ref, compute_trims): Likewise.
25345 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25346 (maybe_trim_partially_dead_store): Likewise.
25347 (maybe_trim_complex_store): Likewise.
25348 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25349 Track what bytes live from the original store. Return tri-state
25350 for dead, partially dead or live.
25351 (dse_dom_walker): Add constructor, destructor and new private members.
25352 (delete_dead_call, delete_dead_assignment): New extracted from
25354 (dse_optimize_stmt): Make a member of dse_dom_walker.
25355 Use initialize_ao_ref_for_dse.
25357 PR tree-optimization/33562
25358 PR tree-optimization/61912
25359 PR tree-optimization/77485
25360 * sbitmap.h (bitmap_count_bits): Prototype.
25361 (bitmap_clear_range, bitmap_set_range): Likewise.
25362 * sbitmap.c (bitmap_clear_range): New function.
25363 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25365 2017-01-13 Martin Liska <mliska@suse.cz>
25368 * function.c (set_cfun): Add new argument force.
25369 * function.h (set_cfun): Likewise.
25370 * ipa-inline-transform.c (inline_call): Use the function when
25371 strict alising from is dropped for function we inline to.
25373 2017-01-13 Richard Biener <rguenther@suse.de>
25375 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25376 for dumping GIMPLE INTEGER_CSTs.
25378 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25380 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25381 to 201112L since C++17.
25383 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25386 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25387 if -fsanitize=kernel-address is present.
25389 2017-01-13 Richard Biener <rguenther@suse.de>
25391 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25392 as _Literal ( type ) number in case usual suffixes do not
25393 preserve all information.
25395 2017-01-13 Richard Biener <rguenther@suse.de>
25397 PR tree-optimization/77283
25398 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25399 and ssa-iterators.h.
25400 (is_feasible_trace): Implement a cost model based on joiner
25403 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25406 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25407 char or short to __float128/_Float128 directly.
25409 2017-01-12 Martin Sebor <msebor@redhat.com>
25411 to -Wformat-overflow.
25412 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25413 (min_bytes_remaining): Same.
25414 (get_string_length): Same.
25415 (format_string): Same.
25416 (format_directive): Same.
25418 (pass_sprintf_length::handle_gimple_call): Same.
25420 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25422 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25423 info.nowrite calls with no lhs that can't throw. Return bool
25424 whether gsi_remove has been called or not.
25425 (pass_sprintf_length::handle_gimple_call): Return bool whether
25426 try_substitute_return_value called gsi_remove. Formatting fix.
25427 (pass_sprintf_length::execute): Don't use gsi_remove if
25428 handle_gimple_call returned true.
25431 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25432 be removed due to side-effects, don't remove following barrier nor
25433 turn the successor edge into fallthru edge.
25435 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25438 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25439 element-reversing loads and stores as not swappable.
25441 2017-01-12 Nathan Sidwell <nathan@acm.org>
25442 Nicolai Stange <nicstange@gmail.com>
25444 * combine.c (try_combine): Don't ignore result of overlap checking
25445 loop. Combine overlap & asm check into single loop.
25447 2017-01-12 Richard Biener <rguenther@suse.de>
25449 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25450 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25452 2017-01-12 Richard Biener <rguenther@suse.de>
25454 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25455 and TS_TARGET_OPTION directly derive from TS_BASE.
25456 * tree-core.h (tree_optimization_option): Derive from tree_base.
25457 (tree_target_option): Likewise.
25459 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25461 * config/i386/i386.c (memory_address_length): Increase len
25462 only when rip_relative_addr_p returns false.
25464 2017-01-11 Julia Koval <julia.koval@intel.com>
25466 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25467 (OPTION_MASK_ISA_SGX_SET): New.
25468 (ix86_handle_option): Handle OPT_msgx.
25469 * config.gcc: Added sgxintrin.h.
25470 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25471 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25472 * config/i386/i386.c (ix86_target_string): Add -msgx.
25474 (ix86_option_override_internal): Handle new options.
25475 (ix86_valid_target_attribute_inner_p): Add sgx.
25476 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25477 * config/i386/i386.opt: Add msgx.
25478 * config/i386/sgxintrin.h: New file.
25479 * config/i386/x86intrin.h: Add sgxintrin.h.
25481 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25484 * fold-const.c (maybe_nonzero_address): Return 1 for function
25486 (tree_single_nonzero_warnv_p): Don't handle function local objects
25490 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25493 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25496 * opts.c: Include "spellcheck.h"
25497 (struct string_fragment): New struct.
25498 (struct edit_distance_traits<const string_fragment &>): New
25500 (get_closest_sanitizer_option): New function.
25501 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25503 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25505 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25507 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25508 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25509 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25510 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25511 for initial die_offset if dwarf_split_debug_info.
25512 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25513 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25515 (output_skeleton_debug_sections): Formatting fix. Use
25516 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25517 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25519 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25521 * config/arm/cortex-a53.md: Add bypasses for
25522 cortex_a53_r2f_cvt.
25523 (cortex_a53_r2f): Only use for transfers.
25524 (cortex_a53_f2r): Likewise.
25525 (cortex_a53_r2f_cvt): Add reservation for conversions.
25526 (cortex_a53_f2r_cvt): Likewise.
25528 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25530 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25531 to all inlined functions, change static to extern.
25533 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25536 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25538 (arm_assemble_integer): Likewise.
25540 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25542 * config.gcc: Use new awk script to check CPU, FPU and architecture
25543 parameters for --with-... options.
25544 * config/arm/parsecpu.awk: New file
25545 * config/arm/arm-cpus.in: New file.
25546 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25548 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25550 * config/arm/t-arm: Update dependency rules.
25551 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25552 of processing .def files.
25553 * config/arm/genopt.sh: Deleted.
25554 * config/arm/gentune.sh: Deleted.
25555 * config/arm/arm-cores.def: Deleted.
25556 * config/arm/arm-arches.def: Deleted.
25557 * config/arm/arm-fpus.def: Deleted.
25558 * config/arm/arm-tune.md: Regenerated.
25559 * config/arm/arm-tables.opt: Regenerated.
25560 * config/arm/arm-cpu.h: New generated file.
25561 * config/arm/arm-cpu-data.h: New generated file.
25562 * config/arm/arm-cpu-cdata.h: New generated file.
25564 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25567 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25569 (input_varpool_node): Unpack dynamically_initialized bit.
25571 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25573 PR rtl-optimization/79032
25574 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25575 the alignment of the adjusted memory reference against that of MODE,
25576 instead of the alignment of the original memory reference.
25578 2017-01-11 Martin Jambor <mjambor@suse.cz>
25580 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25582 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25583 decorated functions.
25585 2017-01-11 Richard Biener <rguenther@suse.de>
25587 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25588 set range/nonnull info for PHI results. Do not set it on
25589 stmts marked for removal.
25591 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25593 * expr.c (store_field): In the bitfield case, fetch the return value
25594 from the registers before applying a single big-endian adjustment.
25595 Always do a final load for a BLKmode value not larger than a word.
25597 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25600 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25601 that we correctly handle column numbers greater than
25602 LINE_MAP_MAX_COLUMN_NUMBER.
25604 2017-01-10 Martin Sebor <msebor@redhat.com>
25606 PR middle-end/78245
25607 * gimple-ssa-sprintf.c (get_destination_size): Call
25608 {init,fini}object_sizes.
25609 * tree-object-size.c (addr_object_size): Adjust.
25610 (pass_through_call): Adjust.
25611 (pass_object_sizes::execute): Adjust.
25612 * tree-object-size.h (fini_object_sizes): Declare.
25614 2017-01-10 Martin Sebor <msebor@redhat.com>
25616 PR tree-optimization/78775
25617 * builtins.c (get_size_range): Move...
25618 * calls.c: ...to here.
25619 (alloc_max_size): Accept zero argument.
25620 (operand_signed_p): Remove.
25621 (maybe_warn_alloc_args_overflow): Call get_size_range.
25622 * calls.h (get_size_range): Declare.
25624 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25626 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25627 from TI's devices.csv file as of September 2016.
25628 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25630 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25632 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25633 * doc/invoke.texi: Likewise.
25634 * doc/md.texi: Likewise.
25635 * doc/objc.texi: Likewise.
25637 2017-01-10 Joshua Conner <joshconner@google.com>
25639 * config/arm/fuchsia-elf.h: New file.
25640 * config/fuchsia.h: New file.
25641 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25642 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25644 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25646 2016-01-10 Richard Biener <rguenther@suse.de>
25648 PR tree-optimization/79034
25649 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25650 Propagate out degenerate PHIs in the joiner.
25652 2017-01-10 Martin Liska <mliska@suse.cz>
25654 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25655 (sort_congruence_classes_by_decl_uid): Likewise.
25656 (sort_congruence_class_groups_by_decl_uid): Likewise.
25657 (sem_item_optimizer::merge_classes): Sort class, groups in these
25658 classes and members in the groups by DECL_UID of declarations.
25659 This would make merge operations stable.
25661 2017-01-10 Martin Liska <mliska@suse.cz>
25663 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25664 usage of m_classes_vec.
25665 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25666 (sem_item_optimizer::get_group_by_hash): Likewise.
25667 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25668 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25669 (sem_item_optimizer::verify_classes): Likewise.
25670 (sem_item_optimizer::process_cong_reduction): Likewise.
25671 (sem_item_optimizer::dump_cong_classes): Likewise.
25672 (sem_item_optimizer::merge_classes): Likewise.
25673 * ipa-icf.h (congruence_class_hash): Rename from
25674 congruence_class_group_hash. Remove declaration of m_classes_vec.
25676 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25678 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25679 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25680 * config.gcc: Add avx512vpopcntdqintrin.h.
25681 * config/i386/avx512vpopcntdqintrin.h: New.
25682 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25683 * config/i386/i386-builtin-types.def: Add new types.
25684 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25685 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25686 __builtin_ia32_vpopcountq_v8di_mask): New.
25687 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25688 __AVX512VPOPCNTDQ__.
25689 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25690 (PTA_AVX512VPOPCNTDQ): Define.
25691 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25692 TARGET_AVX512VPOPCNTDQ_P): Define.
25693 * config/i386/i386.opt: Add mavx512vpopcntdq.
25694 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25695 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25697 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25699 PR middle-end/77484
25700 * predict.def (PRED_CALL): Set to 67.
25702 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25704 * expr.c (store_field): In the bitfield case, if the value comes from
25705 a function call and is of an aggregate type returned in registers, do
25706 not modify the field mode; extract the value in all cases if the mode
25707 is BLKmode and the size is not larger than a word.
25709 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25712 * config/i386/cpuid.h: Fix undefined behavior.
25714 2017-01-04 Jeff Law <law@redhat.com>
25716 PR tree-optimization/79007
25717 PR tree-optimization/67955
25718 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25719 conservative for pt.null when flag_non_call_exceptions is on.
25721 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25723 PR translation/79019
25724 PR translation/79020
25725 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25726 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25727 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25729 * config/avr/avr.opt (maccumulate-args): Likewise.
25730 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25731 * common.opt (freport-bug): Likewise.
25732 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25733 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25734 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25735 translatable string.
25736 * config/i386/i386.c (function_value_32): Likewise.
25737 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25738 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25740 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25741 * common/config/msp430/msp430-common.c (msp430_handle_option):
25743 * symtab.c (symtab_node::verify_base): Likewise.
25744 * opts.c (set_debug_level): Likewise.
25745 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25746 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25747 missing whitespace to translatable strings.
25748 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25749 * config/sh/superh.h: Likewise.
25750 * config/i386/xopintrin.h: Likewise.
25751 * config/i386/znver1.md: Likewise.
25752 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25753 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25754 * double-int.h (struct double_int): Likewise.
25755 * double-int.c (div_and_round_double): Likewise.
25756 * wide-int.cc: Likewise.
25757 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25758 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25759 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25760 (crossjumps_occurred): ... this.
25761 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25764 PR tree-optimization/78899
25765 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25766 returning bool return struct loop *, NULL for failure and the new
25768 (versionable_outer_loop_p): Don't version outer loop if it has
25769 dont_vectorized bit set.
25770 (tree_if_conversion): When versioning outer loop, ensure
25771 tree_if_conversion is performed also on the inner loop of the
25772 non-vectorizable outer loop copy.
25773 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25774 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25775 prevent vectorization of it.
25776 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25777 the outer loop vectorization of the non-scalar version is attempted
25778 before vectorization of the inner loop in scalar version. If
25779 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25780 vectorization of its inner loop.
25781 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25782 has 2 inner loops, rename also on edges from bb whose single pred
25783 is outer_loop->header. Fix typo in function comment.
25785 2017-01-09 Martin Sebor <msebor@redhat.com>
25788 * asan.c (asan_emit_stack_protection): Increase local buffer size
25789 to avoid snprintf truncation warning.
25791 2017-01-09 Andrew Pinski <apinski@cavium.com>
25793 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25794 to reference thunderx2t99 for the tuning structure
25795 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25797 (thunderx2t99_extra_costs): This.
25798 * config/aarch64/aarch64-tune.md: Regenerate.
25799 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25800 (vulcan_addrcost_table): This.
25801 (vulcan_regmove_cost): Rename to ...
25802 (thunderx2t99_regmove_cost): This.
25803 (vulcan_vector_cost): Rename to ...
25804 (thunderx2t99_vector_cost): this.
25805 (vulcan_branch_cost): Rename to ...
25806 (thunderx2t99_branch_cost): This.
25807 (vulcan_tunings): Rename to ...
25808 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25809 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25811 2017-01-09 Martin Jambor <mjambor@suse.cz>
25815 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25816 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25817 (propagate_vr_accross_jump_function): Use the above function for all
25818 value range computations for pass-through jump functions and type
25819 converasion from explicit value range values.
25820 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25821 parameters from TYPE_ARG_TYPES.
25822 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25823 (ipa_write_node_info): Stream type of the actual argument.
25824 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25826 2017-01-09 Martin Liska <mliska@suse.cz>
25829 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25830 (lookup_compiler): Do not show error message with have_E.
25832 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25834 PR tree-optimization/78938
25835 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25836 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25837 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25838 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25841 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25843 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25846 2017-01-09 Richard Biener <rguenther@suse.de>
25848 PR tree-optimization/78997
25849 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25850 name condition properly.
25852 2017-01-09 Richard Biener <rguenther@suse.de>
25855 * dwarf2out.c (is_cxx): New overload with context.
25856 (is_naming_typedef_decl): Use it.
25858 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25860 * invoke.texi (Option Summary): Correct spacing in option lists
25861 and add line breaks to fix over-long lines.
25863 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25865 PR middle-end/17660
25867 * extend.texi (Common Variable Attributes): Add xref to GCC
25868 Internals manual to explain mode attribute keywords.
25870 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25873 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25874 and Preprocessor Options.
25875 (Options for Linking): Document -pthread here....
25876 (RS/6000 and PowerPC Options): ...not here.
25877 (Solaris 2 Options): ...or here.
25878 * doc/cppopts.texi: Document -pthread.
25880 2017-01-08 Martin Sebor <msebor@redhat.com>
25882 PR middle-end/77708
25883 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25884 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25885 New member functions.
25886 (format_directive): Used them.
25888 (pass_sprintf_length::handle_gimple_call): Same.
25889 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25890 to avoid truncation for any argument.
25891 (extract_affine_mul): Same.
25892 * tree.c (get_file_function_name): Same.
25894 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25896 PR middle-end/77484
25897 * predict.def (PRED_INDIR_CALL): Set to 86.
25899 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25901 PR preprocessor/54124
25902 * doc/cppopts.texi: Reformat -d subtable to list the full name
25903 of the options. Add cross-reference to the docs for the general
25904 compiler -d options.
25905 * doc/invoke.texi (Developer Options): Add cross-reference to the
25906 preprocessor-specific -d option documentation.
25908 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25910 PR preprocessor/13498
25911 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25912 redudant material, and reflect new command-line options.
25913 (System Headers): Likewise.
25915 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25917 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25918 -isystem, and -idirafter. Copy-edit.
25919 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25920 default for -ftrack-macro-expansion. Delete obsolete and
25921 badly-formatted implementation details about -fdebug-cpp output.
25922 * doc/cppwarnopts.texi: Copy-edit.
25924 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25927 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25928 that the transition from a max line width >= 1<<10 to narrower
25929 lines works correctly.
25931 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25933 * doc/options.texi (PerFunction): New.
25934 * opt-functions.awk (switch_flags): Map both Optimization and
25935 PerFunction to CL_OPTIMIZATION.
25936 * opth-gen.awk: Test for PerFunction flag along with
25938 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25939 it only when the latter is present. Skip those that don't in
25940 the hash function generator.
25941 * common.opt (fvar-tracking): Mark as PerFunction instead of
25943 (fvar-tracking-assignments): Likewise.
25944 (fvar-tracking-assignments-toggle): Likewise.
25945 (fvar-tracking-uninit): Likewise.
25947 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25949 PR translation/79018
25950 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25953 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25956 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25957 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25958 TARGET_LONG_JUMP_TABLE_OFFSETS.
25959 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25960 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25961 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25962 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25963 * config/m68k/m68k.md (tablejump expander): Likewise.
25964 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25965 TARGET_LONG_JUMP_TABLE_OFFSETS.
25966 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25967 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25969 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25970 David Holsgrove <david.holsgrove@xilinx.com>
25972 * common/config/microblaze/microblaze-common.c
25973 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25974 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25976 * config/microblaze/microblaze.c (microblaze_must_save_register)
25977 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25979 (microblaze_eh_return): New function.
25980 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25981 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25982 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25983 * config/microblaze/microblaze.md (eh_return): New pattern.
25985 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25987 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25988 GCC_DIAGNOSTIC_STRINGIFY): Define.
25990 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25992 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25994 * config/arm/arm.md (<mcrr>): New.
25996 * config/arm/arm.c (arm_arch5te): New.
25997 (arm_option_override): Set arm_arch5te.
25998 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26000 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26001 (arm_mcrr_qualifiers): ... this. New.
26002 (MRRC_QUALIFIERS): Define to...
26003 (arm_mrrc_qualifiers): ... this. New.
26004 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26006 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26007 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26008 (MRRCI, mrrc, MRRC): New.
26009 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26010 VUNSPEC_MRRC2): New.
26012 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26014 * config/arm/arm.md (<mcr>): New.
26016 * config/arm/arm.c (arm_coproc_builtin_available): Add
26017 support for mcr, mrc, mcr2 and mrc2.
26018 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26019 (arm_mcr_qualifiers): ... this. New.
26020 (MRC_QUALIFIERS): Define to ...
26021 (arm_mrc_qualifiers): ... this. New.
26022 (MCR_QUALIFIERS): Define to ...
26023 (arm_mcr_qualifiers): ... this. New.
26024 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26026 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26027 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26028 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26029 VUNSPEC_MRC2): New.
26031 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26033 * config/arm/arm.md (*ldc): New.
26037 * config/arm/arm.c (arm_coproc_builtin_available): Add
26038 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26039 (arm_coproc_ldc_stc_legitimate_address): New.
26040 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26041 'qualifier_const_pointer'.
26042 (LDC_QUALIFIERS): Define to...
26043 (arm_ldc_qualifiers): ... this. New.
26044 (STC_QUALIFIERS): Define to...
26045 (arm_stc_qualifiers): ... this. New.
26046 * config/arm/arm-protos.h
26047 (arm_coproc_ldc_stc_legitimate_address): New.
26048 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26049 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26050 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26051 stc2, stcl, stc2l): New.
26052 * config/arm/constraints.md (Uz): New.
26053 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26054 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26055 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26056 VUNSPEC_STC2L): New.
26058 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26060 * config/arm/arm.md (<cdp>): New.
26061 * config/arm/arm.c (neon_const_bounds): Rename this ...
26062 (arm_const_bounds): ... this.
26063 (arm_coproc_builtin_available): New.
26064 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26065 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26066 (CDP_QUALIFIERS): Define to...
26067 (arm_cdp_qualifiers): ... this. New.
26069 (arm_expand_builtin_args): Add case for 6 arguments.
26070 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26071 (arm_const_bounds): ... this.
26072 (arm_coproc_builtin_available): New.
26073 * config/arm/arm_acle.h (__arm_cdp): New.
26075 * config/arm/arm_acle_builtins.def (cdp): New.
26077 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26078 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26079 'arm_const_bounds'.
26080 * config/arm/types.md (coproc): New.
26081 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26082 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26083 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26084 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26086 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26088 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26089 (UBINOP_QUALIFIERS): New.
26091 (acle_builtin_data): New. Change comment.
26092 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26093 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26094 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26095 arm_acle_builtins.def.
26096 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26097 (arm_init_acle_builtins): New.
26098 (CRC32_BUILTIN): Remove.
26099 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26100 crc32cb, crc32ch and crc32cw.
26101 (arm_init_crc32_builtins): Remove.
26102 (arm_init_builtins): Use arm_init_acle_builtins rather
26103 than arm_init_crc32_builtins.
26104 (arm_expand_acle_builtin): New.
26105 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26106 * config/arm/arm_acle_builtins.def: New.
26108 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26110 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26111 (arm_builtin_datum): ... this.
26112 (arm_init_neon_builtin): Rename to ...
26113 (arm_init_builtin): ... this. Add a new parameters PREFIX
26114 and USE_SIG_IN_NAME.
26115 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26116 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26117 'arm_builtin_datum'.
26118 (arm_init_vfp_builtins): Likewise.
26119 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26120 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26121 (arm_expand_neon_args): Rename to ...
26122 (arm_expand_builtin_args): ... this. Rename builtin_arg
26123 enum values and differentiate between ARG_BUILTIN_MEMORY
26124 and ARG_BUILTIN_NEON_MEMORY.
26125 (arm_expand_neon_builtin_1): Rename to ...
26126 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26127 values, arm_expand_builtin_args and add bool parameter NEON.
26128 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26129 (arm_expand_vfp_builtin): Likewise.
26130 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26132 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26134 PR middle-end/77484
26135 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26136 * predict.c (tree_estimate_probability_bb): Reverse direction of
26137 polymorphic call predictor.
26139 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26141 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26142 (determine_pass_name_match): ...this new function and...
26143 (should_skip_pass_p): ...this new function.
26145 2017-01-06 Nathan Sidwell <nathan@acm.org>
26147 * ipa-visibility.c (function_and_variable_visibility): Reformat
26148 comments and long lines. Remove extrneous if.
26149 * symtab.c (symtab_node::make_decl_local): Fix code format.
26150 (symtab_node::set_section_for_node): Fix comment typo.
26152 2017-01-06 Martin Liska <mliska@suse.cz>
26155 * lra-constraints.c: Rename invariant to lra_invariant.
26156 * predict.c (set_even_probabilities): Initialize e to NULL.
26158 2017-01-05 Martin Sebor <msebor@redhat.com>
26160 PR tree-optimization/78910
26161 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26162 (format_integer): Correct off-by-one error in the handling
26163 of precision with negative numbers in signed conversions..
26165 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26167 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26169 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26171 PR tree-optimization/71016
26172 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26173 factor_out_conditional_conversion. Formatting fix.
26174 (factor_out_conditional_conversion): Add cond_stmt argument.
26175 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26176 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26179 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26181 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26182 read-rtl-function.o, and selftest-rtl.o.
26183 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26184 (selftest::aarch64_test_loading_full_dump): New function.
26185 (selftest::aarch64_run_selftests): New function.
26186 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26187 selftest::aarch64_run_selftests.
26188 * config/i386/i386.c
26189 (selftest::ix86_test_loading_dump_fragment_1): New function.
26190 (selftest::ix86_test_loading_call_insn): New function.
26191 (selftest::ix86_test_loading_full_dump): New function.
26192 (selftest::ix86_test_loading_unspec): New function.
26193 (selftest::ix86_run_selftests): Call the new functions.
26194 * emit-rtl.c (maybe_set_max_label_num): New function.
26195 * emit-rtl.h (maybe_set_max_label_num): New decl.
26196 * function.c (instantiate_decls): Guard call to
26197 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26198 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26200 * gensupport.c (gen_reader::gen_reader): Pass "false"
26201 for new "compact" param of rtx_reader.
26202 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26203 rather than an empty string for NULL strings.
26204 * read-md.c: Potentially include config.h rather than bconfig.h.
26205 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26206 (have_error): New global, copied from errors.c.
26207 (md_reader::read_name): Rename to...
26208 (md_reader::read_name_1): ...this, adding "out_loc" param,
26209 and converting "missing name or number" to returning false, rather
26211 (md_reader::read_name): Reimplement in terms of read_name_1.
26212 (md_reader::read_name_or_nil): New function.
26213 (md_reader::read_string): Handle "(nil)" by returning NULL.
26214 (md_reader::md_reader): Add new param "compact".
26215 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26216 (md_reader::read_file): New method.
26217 * read-md.h (md_reader::md_reader): Add new param "compact".
26218 (md_reader::read_file): New method.
26219 (md_reader::is_compact): New accessor.
26220 (md_reader::read_name): Convert return type from void to file_location.
26221 (md_reader::read_name_or_nil): New decl.
26222 (md_reader::read_name_1): New decl.
26223 (md_reader::m_compact): New field.
26224 (noop_reader::noop_reader): Pass "false" for new "compact" param
26226 (rtx_reader::rtx_reader): Add new "compact" param.
26227 (rtx_reader::read_rtx_operand): Make virtual and convert return
26228 type from void to rtx.
26229 (rtx_reader::read_until): New decl.
26230 (rtx_reader::handle_any_trailing_information): New virtual function.
26231 (rtx_reader::postprocess): New virtual function.
26232 (rtx_reader::finalize_string): New virtual function.
26233 (rtx_reader::m_in_call_function_usage): New field.
26234 (rtx_reader::m_reuse_rtx_by_id): New field.
26235 * read-rtl-function.c: New file.
26236 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26237 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26238 (selftest::verify_three_block_rtl_cfg): New decl.
26239 * read-rtl-function.h: New file.
26240 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26241 For host, include function.h, memmodel.h, and emit-rtl.h.
26242 (one_time_initialization): New function.
26243 (struct compact_insn_name): New struct.
26244 (compact_insn_names): New array.
26245 (find_code): Handle insn codes in compact dumps.
26246 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26247 (bind_subst_iter_and_attr): Likewise.
26248 (add_condition_to_string): Likewise.
26249 (add_condition_to_rtx): Likewise.
26250 (apply_attribute_uses): Likewise.
26251 (add_current_iterators): Likewise.
26252 (apply_iterators): Likewise.
26253 (initialize_iterators): Guard usage of apply_subst_iterator with
26254 #ifdef GENERATOR_FILE.
26255 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26256 (md_reader::read_mapping): Likewise.
26257 (add_define_attr_for_define_subst): Likewise.
26258 (add_define_subst_attr): Likewise.
26259 (read_subst_mapping): Likewise.
26260 (check_code_iterator): Likewise.
26261 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26263 (one_time_initialization): New function.
26264 (rtx_reader::read_until): New method.
26265 (read_flags): New function.
26266 (parse_reg_note_name): New function.
26267 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26268 Handle reuse_rtx ids.
26269 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26270 Add parsing support for RTL dumps, mirroring the special-cases in
26271 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26272 values, and calling handle_any_trailing_information.
26273 (rtx_reader::read_rtx_operand): Convert return type from void
26274 to rtx, returning return_rtx. Handle case 'e'. Call
26275 finalize_string on XSTR and XTMPL fields.
26276 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26277 "(nil)" values were omitted. Call the postprocess vfunc on the
26279 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26280 class ctor. Initialize m_in_call_function_usage. Call
26281 one_time_initialization.
26282 * rtl-tests.c (selftest::test_uncond_jump): Call
26283 set_new_first_and_last_insn.
26284 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26285 * selftest-rtl.c: New file.
26286 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26287 (selftest::get_insn_by_uid): New decl.
26288 * selftest-run-tests.c (selftest::run_tests): Call
26289 read_rtl_function_c_tests.
26290 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26291 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26294 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26296 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26297 operands in a special way. Assert that pos+len <= mode precision.
26299 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26301 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26302 3 argument Alias with unlimited for the negative form.
26303 (fno-vect-cost-model): Removed.
26305 2017-01-05 Martin Liska <mliska@suse.cz>
26307 * hsa-gen.c (gen_hsa_divmod): New function.
26308 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26310 2017-01-05 Martin Liska <mliska@suse.cz>
26313 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26316 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26318 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26319 small constant length operands.
26321 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26323 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26324 between loop iterations.
26326 2017-01-05 Martin Liska <mliska@suse.cz>
26329 * gimplify.c (gimplify_decl_expr): Compare to
26330 asan_poisoned_variables instread of checking flags.
26331 (gimplify_target_expr): Likewise.
26332 (gimplify_expr): Likewise.
26333 (gimplify_function_tree): Conditionally initialize
26334 asan_poisoned_variables.
26336 2017-01-04 Jeff Law <law@redhat.com>
26338 PR tree-optimizatin/78812
26339 * rtl.h (contains_mem_rtx_p): Prototype.
26340 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26341 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26342 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26343 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26344 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26346 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26348 * input.c (assert_char_at_range): Default-initialize actual_range.
26350 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26352 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26353 to match the caller.
26355 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26357 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26358 insns after final jump in test to emit dummy move.
26360 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26362 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26363 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26365 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26367 * multiple_target.c (create_dispatcher_calls): Init e_next.
26368 * tree-ssa-loop-split.c (split_loop): Init border.
26369 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26372 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26377 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26378 (altivec_register_operand): Do not return true if the operand
26379 contains a SUBREG mixing SImode and SFmode.
26380 (vsx_register_operand): Likewise.
26381 (vsx_reg_sfsubreg_ok): New predicate.
26382 (vfloat_operand): Do not return true if the operand contains a
26383 SUBREG mixing SImode and SFmode.
26384 (vint_operand): Likewise.
26385 (vlogical_operand): Likewise.
26386 (gpc_reg_operand): Likewise.
26387 (int_reg_operand): Likewise.
26388 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26389 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26390 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26392 (rs6000_emit_move_si_sf_subreg): New helper function.
26393 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26394 fixup SUBREGs involving SImode and SFmode.
26395 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26396 numbers for the new peephole2 optimization.
26397 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26398 the GLIBC math library that do AND/IOR/XOR operations on single
26399 precision floating point.
26400 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26401 target macros to say whether we need to avoid SUBREGs mixing
26403 (TARGET_ALLOW_SF_SUBREG): Likewise.
26404 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26405 (UNSPEC_SI_FROM_SF): Likewise.
26406 (iorxor): Change spacing.
26407 (and_ior_xor): New iterator for AND, IOR, and XOR.
26408 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26409 (movdi_from_sf_zero_ext): Likewise.
26410 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26411 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26412 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26413 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26414 (fms<mode>4): Likewise.
26415 (fnma<mode>4): Likewise.
26416 (fnms<mode>4): Likewise.
26417 (nfma<mode>4): Likewise.
26418 (nfms<mode>4): Likewise.
26420 2017-01-04 Marek Polacek <polacek@redhat.com>
26423 * doc/invoke.texi: Document -Wpointer-compare.
26425 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26427 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26430 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26431 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26434 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26437 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26438 documentation of the powerpc_popcntb_ok attribute.
26439 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26440 code to issue warning messages if a requested CPU configuration is
26441 not supported by the binary (assembler and loader) toolchain.
26442 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26443 made to define a built-in function that has been disabled.
26444 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26445 made to define a built-in function that has been disabled.
26446 (altivec_init_builtins): Add comment explaining why definition
26447 of the DST built-in functions is not preceded by an assertion
26448 check. Add assertions to prevent ICE if attempts are made to
26449 define an altivec predicate or an abs* built-in function that has
26451 (htm_init_builtins): Add comment explaining why definition of the
26452 htm built-in functions is not preceded by an assertion check.
26454 2017-01-04 Jeff Law <law@redhat.com>
26456 PR tree-optimizatin/67955
26457 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26458 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26459 the points-to solution does not include pt_null. Use DECL_PT_UID
26462 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26464 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26465 Use gen_int_mode instead of gen_lopwart for const_int operands.
26467 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26469 PR tree-optimization/71563
26470 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26471 out of range value - has low bits known to be zero.
26473 2017-01-04 Alan Modra <amodra@gmail.com>
26475 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26476 * configure: Regenerate.
26477 * config.in: Regenerate.
26479 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26482 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26483 a substring of the message, but strcmp with the whole message. Ifdef
26484 ENABLE_NLS, translate the message first using dgettext.
26486 2017-01-03 Jeff Law <law@redhat.com>
26488 PR tree-optimizatin/78856
26489 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26490 (mark_threaded_blocks): Remove code to truncate thread paths that
26491 cross multiple loop headers. Instead invalidate the cached loop
26492 iteration information and handle case of a thread path walking
26493 into an irreducible region.
26495 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26498 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26499 assertions. Add support for doing the signbit if the IEEE 128-bit
26500 floating point value is in a GPR.
26501 * config/rs6000/rs6000.md (Fsignbit): Delete.
26502 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26503 Update the length attribute if the value is in a GPR.
26504 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26505 the sign or zero extension instruction, since the value is always 0/1.
26506 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26509 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26510 extracting SImode to a GPR register so that we can generate a
26511 store, limit the vector to be in a traditional Altivec register
26512 for the vextuwrx instruction.
26514 2017-01-03 Ian Lance Taylor <iant@google.com>
26516 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26518 2017-01-03 Martin Sebor <msebor@redhat.com>
26520 PR tree-optimization/78696
26521 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26522 precision. Use MPFR for %f for greater fidelity. Correct handling
26524 (pass_sprintf_length::compute_format_length): Set width and precision
26525 specified by asrerisk to void_node for vararg functions.
26526 (try_substitute_return_value): Adjust dump output.
26528 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26530 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26532 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26534 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26535 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26536 -mlra/-mno-lra was passed to the compiler.
26538 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26540 PR rtl-optimization/65618
26541 * emit-rtl.c (try_split): Move initialization of "before" and
26542 "after" to just before the call to emit_insn_after_setloc.
26544 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26546 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26548 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26550 * dwarf2out.c (gen_enumeration_type_die): When
26551 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26553 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26555 PR tree-optimization/78965
26556 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26557 Change first argument from const call_info & to call_info &. For %n
26558 set info.nowrite to false.
26560 PR middle-end/78901
26561 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26562 possibly throwing calls.
26564 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26565 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26566 and fns handling, rather than in a separate case SSA_NAME.
26568 2017-01-02 Jeff Law <law@redhat.com>
26570 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26571 fixes for first_period and second_period variables.
26573 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26576 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26577 (*insvqi_1): New insn pattern.
26578 (*insvqi_1_mem_rex64): Ditto.
26579 (*insvqi_2): Ditto.
26580 (*insvqi_3): Rename from *insvqi.
26582 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26584 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26586 * doc/cfg.texi (Edges): Remove reference to Java.
26587 (Maintaining the CFG): Ditto.
26589 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26591 PR middle-end/77674
26592 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26593 transparent aliases.
26595 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26597 PR middle-end/77484
26598 * predict.def (PRED_CALL): Update hitrate.
26599 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26600 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26601 into direct/indirect/polymorphic variants.
26603 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26605 Update copyright years.
26607 * gcc.c (process_command): Update copyright notice dates.
26608 * gcov-dump.c (print_version): Ditto.
26609 * gcov.c (print_version): Ditto.
26610 * gcov-tool.c (print_version): Ditto.
26611 * gengtype.c (create_file): Ditto.
26612 * doc/cpp.texi: Bump @copying's copyright year.
26613 * doc/cppinternals.texi: Ditto.
26614 * doc/gcc.texi: Ditto.
26615 * doc/gccint.texi: Ditto.
26616 * doc/gcov.texi: Ditto.
26617 * doc/install.texi: Ditto.
26618 * doc/invoke.texi: Ditto.
26620 Copyright (C) 2017 Free Software Foundation, Inc.
26622 Copying and distribution of this file, with or without modification,
26623 are permitted in any medium without royalty provided the copyright
26624 notice and this notice are preserved.