1 2017-09-13 Martin Liska <mliska@suse.cz>
4 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5 CASE_HIGH is NULL_TREE.
7 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8 Alan Hayward <alan.hayward@arm.com>
9 David Sherwood <david.sherwood@arm.com>
11 * target.def (secondary_memory_needed): New hook.
12 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
13 instead of SECONDARY_MEMORY_NEEDED.
14 (secondary_memory_needed_mode): Likewise.
15 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
16 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
17 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
18 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
19 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
20 * doc/tm.texi: Regenerate.
21 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
22 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
23 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
24 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
25 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
26 * config/i386/i386.c (inline_secondary_memory_needed): Put the
27 mode argument first and change the reg_class arguments to reg_class_t.
28 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
29 Make static. Update the call to inline_secondary_memory_needed.
30 (ix86_register_move_cost): Update the call to
31 inline_secondary_memory_needed.
32 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
33 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
35 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
36 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
38 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
39 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
40 * config/mips/mips.c (mips_secondary_memory_needed): Make static
41 and match hook interface. Add comment from mips.h.
42 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
43 * config/mmix/mmix.md (truncdfsf2): Refer to
44 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
46 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
47 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
48 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
49 (pa_secondary_memory_needed): New function.
50 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
51 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
52 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
53 (pdp11_secondary_memory_needed): Make static and match hook interface.
54 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
55 * config/powerpcspe/powerpcspe-protos.h
56 (rs6000_secondary_memory_needed_ptr): Delete.
57 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
59 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
60 (rs6000_option_override_internal): Assign to
61 targetm.secondary_memory_needed rather than
62 rs6000_secondary_memory_needed_ptr.
63 (rs6000_secondary_memory_needed): Match hook interface.
64 (rs6000_debug_secondary_memory_needed): Likewise.
65 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
66 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
67 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
68 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
69 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
70 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
72 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
73 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
74 (rs6000_option_override_internal): Assign to
75 targetm.secondary_memory_needed rather than
76 rs6000_secondary_memory_needed_ptr.
77 (rs6000_secondary_memory_needed): Match hook interface.
78 (rs6000_debug_secondary_memory_needed): Likewise.
79 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
80 * config/s390/s390.c (s390_secondary_memory_needed): New function.
81 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
82 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
83 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
84 (sparc_secondary_memory_needed): New function.
85 * lra-constraints.c (check_and_process_move): Refer to
86 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
88 (curr_insn_transform): Likewise.
89 (process_alt_operands): Use targetm.secondary_memory_needed
90 instead of TARGET_SECONDARY_MEMORY_NEEDED.
91 (check_secondary_memory_needed_p): Likewise.
92 (choose_split_class): Likewise.
93 * reload.c: Unconditionally include code that was previously
94 conditional on SECONDARY_MEMORY_NEEDED.
95 (push_secondary_reload): Use targetm.secondary_memory_needed
96 instead of TARGET_SECONDARY_MEMORY_NEEDED.
97 (push_reload): Likewise.
98 * reload1.c: Unconditionally include code that was previously
99 conditional on SECONDARY_MEMORY_NEEDED.
100 (choose_reload_regs): Use targetm.secondary_memory_needed
101 instead of TARGET_SECONDARY_MEMORY_NEEDED.
102 (gen_reload): Likewise.
103 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
105 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
106 Alan Hayward <alan.hayward@arm.com>
107 David Sherwood <david.sherwood@arm.com>
109 * target.def (secondary_memory_needed_mode): New hook:
110 * targhooks.c (default_secondary_memory_needed_mode): Declare.
111 * targhooks.h (default_secondary_memory_needed_mode): New function.
112 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
113 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
114 * doc/tm.texi: Regenerate.
115 * lra-constraints.c (check_and_process_move): Use
116 targetm.secondary_memory_needed_mode instead of
117 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
118 (curr_insn_transform): Likewise.
119 * reload.c (get_secondary_mem): Likewise.
120 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
121 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
123 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
124 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
125 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
126 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
127 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
129 * config/powerpcspe/powerpcspe-protos.h
130 (rs6000_secondary_memory_needed_mode): Delete.
131 * config/powerpcspe/powerpcspe.c
132 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
133 (rs6000_secondary_memory_needed_mode): Make static.
134 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
135 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
137 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
139 (rs6000_secondary_memory_needed_mode): Make static.
140 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
141 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
142 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
143 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
144 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
146 (sparc_secondary_memory_needed_mode): New function.
147 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
149 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
151 * config/aarch64/constraints.md (Umq): New constraint.
152 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
154 (mov<mode>): Update condition.
156 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
158 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
159 when bitposition is the same.
161 2017-09-13 Richard Biener <rguenther@suse.de>
163 * dwarf2out.c (output_die_symbol): Remove.
164 (output_die): Do not output a DIEs symbol.
166 2017-09-13 Richard Biener <rguenther@suse.de>
169 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
170 default-def to avoid breaking iterator update with the weird
171 interaction with cgraph_update_edges_for_call_stmt_node.
173 2017-09-13 Richard Biener <rguenther@suse.de>
175 * tree-cfg.c (verify_gimple_assign_binary): Add verification
176 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
177 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
178 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
180 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
182 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
183 Disable pc relative literal load irrespective of
184 TARGET_FIX_ERR_A53_84341 for default.
186 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
188 * config/sparc/sparc.c (output_return): Output the source location of
189 the insn in the delay slot, if any.
190 (output_sibcall): Likewise.
192 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
195 * common.opt (-static-pie): New alias.
196 (shared): Negate static-pie.
197 (-no-pie): Update help text.
199 (static-pie): New option.
200 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
202 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
203 (LINK_EH_SPEC): Likewise.
204 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
205 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
206 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
207 * gcc.c (LINK_COMMAND_SPEC): Likewise.
208 (init_gcc_specs): Likewise.
209 (init_spec): Likewise.
210 (display_help): Update help message for -pie.
211 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
214 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
216 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
217 (movdi_aarch64): Likewise.
218 (movti_aarch64): Likewise.
220 2017-09-12 Simon Wright <simon@pushface.org>
223 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
224 calculation of the minor version, always output as 0.
226 2017-09-12 Jakub Jelinek <jakub@redhat.com>
229 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
230 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
231 on it early, rather than manual conversion late. For
232 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
233 instead of performing manual conversion.
235 2017-09-12 Carl Love <cel@us.ibm.com>
237 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
238 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
240 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
241 VMULOSW): Add definitions.
242 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
243 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
244 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
245 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
246 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
248 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
250 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
252 (movti_aarch64): Likewise.
253 (movdf_aarch64): Likewise.
254 (movtf_aarch64): Likewise.
255 (load_pairdi): Likewise.
256 (store_pairdi): Likewise.
257 (load_pairdf): Likewise.
258 (store_pairdf): Likewise.
259 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
260 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
261 (ldr_got_small_<mode>): Likewise.
262 (ldr_got_small_28k_<mode>): Likewise.
263 (ldr_got_tiny): Likewise.
264 * config/aarch64/iterators.md (ldst_sz): New.
265 (ldpstp_sz): Likewise.
266 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
268 (thunderx_load): Split load_8 to load_16.
269 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
271 (thunderx2t99_storepair_basic): Split store_8 to store_16.
272 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
273 (xgene1_store_pair): Split store_8 to store_16.
274 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
275 (falkor_st_0_st_sd): Split store_8 to store_16.
277 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
279 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
280 and store1/2/3/4 to store_4/8/12/16.
281 * config/aarch64/aarch64.md: Update for rename.
282 * config/arm/arm.md: Likewise.: Likewise.
283 * config/arm/arm.c: Likewise.
284 * config/arm/thumb1.md: Likewise.
285 * config/arm/thumb2.md: Likewise.
286 * config/arm/vfp.md: Likewise.
287 * config/arm/arm-generic.md: Likewise.
288 * config/arm/arm1020e.md: Likewise.
289 * config/arm/arm1026ejs.md: Likewise.
290 * config/arm/arm1136jfs.md: Likewise.
291 * config/arm/arm926ejs.md: Likewise.
292 * config/arm/cortex-a15.md: Likewise.
293 * config/arm/cortex-a17.md: Likewise.
294 * config/arm/cortex-a5.md: Likewise.
295 * config/arm/cortex-a53.md: Likewise.
296 * config/arm/cortex-a57.md: Likewise.
297 * config/arm/cortex-a7.md: Likewise.
298 * config/arm/cortex-a8.md: Likewise.
299 * config/arm/cortex-a9.md: Likewise.
300 * config/arm/cortex-m4.md: Likewise.
301 * config/arm/cortex-m7.md: Likewise.
302 * config/arm/cortex-r4.md: Likewise.
303 * config/arm/exynos-m1.md: Likewise.
304 * config/arm/fa526.md: Likewise.
305 * config/arm/fa606te.md: Likewise.
306 * config/arm/fa626te.md: Likewise.
307 * config/arm/fa726te.md: Likewise.
308 * config/arm/fmp626.md: Likewise.
309 * config/arm/iwmmxt.md: Likewise.
310 * config/arm/ldmstm.md: Likewise.
311 * config/arm/marvell-pj4.md: Likewise.
312 * config/arm/xgene1.md: Likewise.
313 * config/aarch64/thunderx.md: Likewise.
314 * config/aarch64/thunderx2t99.md: Likewise.
315 * config/aarch64/falkor.md: Likewise.
317 2017-09-12 Martin Liska <mliska@suse.cz>
319 * attribs.c (private_lookup_attribute): New function.
320 * attribs.h (private_lookup_attribute): Declared here.
321 (lookup_attribute): Called from this place.
323 2017-09-12 Richard Biener <rguenther@suse.de>
325 PR tree-optimization/82157
326 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
327 stmts with side-effects.
329 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
330 Alan Hayward <alan.hayward@arm.com>
331 David Sherwood <david.sherwood@arm.com>
333 * target.def (hard_regno_nregs): New hook.
334 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
335 * targhooks.h (default_hard_regno_nregs): Declare.
336 * targhooks.c (default_hard_regno_nregs): New function.
337 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
338 (TARGET_HARD_REGNO_NREGS): ...this hook.
339 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
340 (CLASS_MAX_NREGS): Likewise.
341 * doc/tm.texi: Regenerate.
342 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
343 instead of HARD_REGNO_NREGS.
344 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
345 HARD_REGNO_NREGS in the comment.
346 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
347 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
348 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
349 Return an unsigned int.
350 (TARGET_HARD_REGNO_NREGS): Redefine.
351 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
352 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
353 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
354 (arc_hard_regno_nregs): New function.
355 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
356 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
357 (arm_hard_regno_nregs): New function.
358 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
359 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
360 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
361 (TARGET_HARD_REGNO_NREGS): Redefine.
362 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
363 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
364 (HARD_REGNO_NREGS): Delete.
365 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
366 (cr16_hard_regno_nregs): New function.
367 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
368 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
369 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
370 (cris_hard_regno_nregs): New function.
371 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
372 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
373 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
374 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
375 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
376 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
377 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
378 (frv_hard_regno_nregs): Make static. Take and return an
380 (frv_class_max_nregs): Remove outdated copy of documentation.
381 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
382 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
383 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
384 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
385 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
386 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
387 (TARGET_HARD_REGNO_NREGS): Redefine.
388 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
389 (CLASS_MAX_NREGS): Update comment.
390 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
391 (ia64_hard_regno_nregs): New function.
392 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
393 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
394 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
395 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
396 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
398 (m32c_hard_regno_nregs): Likewise. Make static.
399 (TARGET_HARD_REGNO_NREGS): Redefine.
400 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
401 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
402 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
403 (m68k_hard_regno_nregs): New function.
404 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
405 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
406 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
407 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
408 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
409 Take and return an unsigned int.
410 (TARGET_HARD_REGNO_NREGS): Redefine.
411 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
412 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
413 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
414 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
415 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
416 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
417 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
418 (msp430_hard_regno_nregs): Make static. Take and return an
420 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
421 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
422 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
423 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
424 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
425 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
426 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
427 (TARGET_HARD_REGNO_NREGS): Redefine.
428 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
429 (PA_HARD_REGNO_NREGS): ...this.
430 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
431 (PA_HARD_REGNO_NREGS): ...this.
432 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
433 (pa_hard_regno_nregs): New function.
434 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
435 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
436 (pdp11_hard_regno_nregs): New function.
437 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
438 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
439 (rs6000_hard_regno_nregs_hook): New function.
440 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
441 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
442 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
443 Take and return an unsigned int. Move earlier in file.
444 (TARGET_HARD_REGNO_NREGS): Redefine.
445 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
446 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
447 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
448 (rl78_hard_regno_nregs): Make static. Take and return an
450 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
451 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
452 (rs6000_hard_regno_nregs_hook): New function.
453 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
454 * config/rx/rx.c (rx_hard_regno_nregs): New function.
455 (TARGET_HARD_REGNO_NREGS): Redefine.
456 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
457 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
458 instead of HARD_REGNO_NREGS.
459 (s390_hard_regno_nregs): New function.
460 (s390_hard_regno_mode_ok): Add comment from s390.h.
461 (TARGET_HARD_REGNO_NREGS): Redefine.
462 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
463 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
464 (sh_hard_regno_nregs): New function.
465 (sh_pass_in_reg_p): Use it.
466 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
467 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
468 (sparc_hard_regno_nregs): New function.
469 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
470 * config/spu/spu.c (spu_hard_regno_nregs): New function.
471 (spu_function_arg_advance): Use it, supplying a valid register number.
472 (TARGET_HARD_REGNO_NREGS): Redefine.
473 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
474 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
475 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
476 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
477 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
478 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
479 (CLASS_MAX_NREGS): Remove copy of old documentation.
480 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
481 (visium_hard_regno_nregs): New function.
482 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
483 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
484 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
485 xtensa_hard_regno_nregs): New function.
486 * system.h (HARD_REGNO_NREGS): Poison.
488 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
490 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
491 hard_regno_nregs instead of HARD_REGNO_NREGS.
492 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
493 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
494 (c6x_expand_epilogue): Likewise.
495 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
496 (frv_read_iacc_argument): Likewise.
497 * config/sh/sh.c: Include regs.h.
498 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
499 (regs_used): Likewise.
500 (output_stack_adjust): Likewise.
501 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
502 * expmed.c: Include regs.h.
503 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
504 * ree.c: Include regs.h.
505 (combine_reaching_defs): Use hard_regno_nregs instead of
507 (add_removable_extension): Likewise.
509 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
511 * regs.h (hard_regno_nregs): Turn into a function.
512 (end_hard_regno): Update accordingly.
513 * caller-save.c (setup_save_areas): Likewise.
514 (save_call_clobbered_regs): Likewise.
515 (replace_reg_with_saved_mem): Likewise.
516 (insert_restore): Likewise.
517 (insert_save): Likewise.
518 * combine.c (can_change_dest_mode): Likewise.
519 (move_deaths): Likewise.
520 (distribute_notes): Likewise.
521 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
522 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
523 (rs6000_split_multireg_move): Likewise.
524 (rs6000_register_move_cost): Likewise.
525 (rs6000_memory_move_cost): Likewise.
526 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
527 (rs6000_split_multireg_move): Likewise.
528 (rs6000_register_move_cost): Likewise.
529 (rs6000_memory_move_cost): Likewise.
530 * cselib.c (cselib_reset_table): Likewise.
531 (cselib_lookup_1): Likewise.
532 * emit-rtl.c (set_mode_and_regno): Likewise.
533 * function.c (aggregate_value_p): Likewise.
534 * ira-color.c (setup_profitable_hard_regs): Likewise.
535 (check_hard_reg_p): Likewise.
536 (calculate_saved_nregs): Likewise.
537 (assign_hard_reg): Likewise.
538 (improve_allocation): Likewise.
539 (calculate_spill_cost): Likewise.
540 * ira-emit.c (modify_move_list): Likewise.
541 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
542 (ira_hard_reg_in_set_p): Likewise.
543 * ira.c (setup_reg_mode_hard_regset): Likewise.
544 (clarify_prohibited_class_mode_regs): Likewise.
545 (check_allocation): Likewise.
546 * lra-assigns.c (find_hard_regno_for_1): Likewise.
547 (lra_setup_reg_renumber): Likewise.
548 (setup_try_hard_regno_pseudos): Likewise.
549 (spill_for): Likewise.
550 (assign_hard_regno): Likewise.
551 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
552 * lra-constraints.c (in_class_p): Likewise.
553 (lra_constraint_offset): Likewise.
554 (simplify_operand_subreg): Likewise.
555 (lra_constraints): Likewise.
556 (split_reg): Likewise.
557 (split_if_necessary): Likewise.
558 (invariant_p): Likewise.
559 (inherit_in_ebb): Likewise.
560 * lra-lives.c (process_bb_lives): Likewise.
561 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
562 (get_hard_regs): Likewise.
563 (do_remat): Likewise.
564 * lra-spills.c (assign_spill_hard_regs): Likewise.
565 * mode-switching.c (create_pre_exit): Likewise.
566 * postreload.c (reload_combine_recognize_pattern): Likewise.
567 * recog.c (peep2_find_free_register): Likewise.
568 * regcprop.c (kill_value_regno): Likewise.
569 (set_value_regno): Likewise.
570 (copy_value): Likewise.
571 (maybe_mode_change): Likewise.
572 (find_oldest_value_reg): Likewise.
573 (copyprop_hardreg_forward_1): Likewise.
574 * regrename.c (check_new_reg_p): Likewise.
575 (regrename_do_replace): Likewise.
576 * reload.c (push_reload): Likewise.
577 (combine_reloads): Likewise.
578 (find_dummy_reload): Likewise.
579 (operands_match_p): Likewise.
580 (find_reloads): Likewise.
581 (find_equiv_reg): Likewise.
582 (reload_adjust_reg_for_mode): Likewise.
583 * reload1.c (count_pseudo): Likewise.
584 (count_spilled_pseudo): Likewise.
585 (find_reg): Likewise.
586 (clear_reload_reg_in_use): Likewise.
587 (free_for_value_p): Likewise.
588 (allocate_reload_reg): Likewise.
589 (choose_reload_regs): Likewise.
590 (reload_adjust_reg_for_temp): Likewise.
591 (emit_reload_insns): Likewise.
592 (delete_output_reload): Likewise.
593 * rtlanal.c (subreg_get_info): Likewise.
594 * sched-deps.c (sched_analyze_reg): Likewise.
595 * sel-sched.c (init_regs_for_mode): Likewise.
596 (mark_unavailable_hard_regs): Likewise.
597 (choose_best_reg_1): Likewise.
598 (verify_target_availability): Likewise.
599 * valtrack.c (dead_debug_insert_temp): Likewise.
600 * var-tracking.c (track_loc_p): Likewise.
601 (emit_note_insn_var_location): Likewise.
602 * varasm.c (make_decl_rtl): Likewise.
603 * reginfo.c (choose_hard_reg_mode): Likewise.
604 (init_reg_modes_target): Refer directly to
605 this_target_regs->x_hard_regno_nregs.
607 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
609 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
610 instead of hard_regno_nregs.
612 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
614 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
615 end_hard_regno instead of hard_regno_nregs.
616 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
617 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
618 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
619 * ira-color.c (improve_allocation): Likewise.
620 * lra-assigns.c (find_hard_regno_for_1): Likewise.
621 * lra-lives.c (mark_regno_live): Likewise.
622 (mark_regno_dead): Likewise.
623 * lra-remat.c (operand_to_remat): Likewise.
624 * lra.c (collect_non_operand_hard_regs): Likewise.
625 * postreload.c (reload_combine_note_store): Likewise.
626 (move2add_valid_value_p): Likewise.
627 * reload.c (regno_clobbered_p): Likewise.
629 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
631 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
633 * config/v850/v850.c (v850_reorg): Likewise.
634 * reload.c (refers_to_regno_for_reload_p): Likewise.
635 (find_equiv_reg): Likewise.
636 * reload1.c (reload_reg_reaches_end_p): Likewise.
638 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
640 * caller-save.c (add_used_regs): Use REG_NREGS instead of
642 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
643 * config/arm/arm.c (output_move_neon): Likewise.
644 (arm_attr_length_move_neon): Likewise.
645 (neon_split_vcombine): Likewise.
646 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
647 (c6x_mark_reg_written): Likewise.
648 (c6x_dwarf_register_span): Likewise.
649 * config/i386/i386.c (ix86_save_reg): Likewise.
650 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
651 (rws_access_reg): Likewise.
652 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
653 * mode-switching.c (create_pre_exit): Likewise.
654 * ree.c (combine_reaching_defs): Likewise.
655 (add_removable_extension): Likewise.
656 * regcprop.c (find_oldest_value_reg): Likewise.
657 (copyprop_hardreg_forward_1): Likewise.
658 * reload.c (reload_inner_reg_of_subreg): Likewise.
659 (push_reload): Likewise.
660 (combine_reloads): Likewise.
661 (find_dummy_reload): Likewise.
662 (reload_adjust_reg_for_mode): Likewise.
663 * reload1.c (find_reload_regs): Likewise.
664 (forget_old_reloads_1): Likewise.
665 (reload_reg_free_for_value_p): Likewise.
666 (reload_adjust_reg_for_temp): Likewise.
667 (emit_reload_insns): Likewise.
668 (delete_output_reload): Likewise.
669 * sel-sched.c (choose_best_reg_1): Likewise.
670 (choose_best_pseudo_reg): Likewise.
672 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
673 Alan Hayward <alan.hayward@arm.com>
674 David Sherwood <david.sherwood@arm.com>
676 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
677 * target.def (slow_unaligned_access): New hook.
678 * targhooks.h (default_slow_unaligned_access): Declare.
679 * targhooks.c (default_slow_unaligned_access): New function.
680 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
681 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
682 * doc/tm.texi: Regenerate.
683 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
684 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
685 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
687 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
688 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
690 (rs6000_slow_unaligned_access): New function.
691 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
692 (expand_block_compare): Likewise.
693 (expand_strn_compare): Likewise.
694 (rs6000_rtx_costs): Likewise.
695 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
696 (riscv_slow_unaligned_access): Likewise.
697 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
698 (riscv_slow_unaligned_access_p): ...this and make static.
699 (riscv_option_override): Update accordingly.
700 (riscv_slow_unaligned_access): New function.
701 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
702 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
703 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
704 (rs6000_slow_unaligned_access): New function.
705 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
706 (rs6000_rtx_costs): Likewise.
707 * config/rs6000/rs6000-string.c (expand_block_compare)
708 (expand_strn_compare): Use targetm.slow_unaligned_access instead
709 of SLOW_UNALIGNED_ACCESS.
710 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
711 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
712 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
713 of SLOW_UNALIGNED_ACCESS.
714 * expmed.c (simple_mem_bitfield_p): Likewise.
715 * expr.c (alignment_for_piecewise_move): Likewise.
716 (emit_group_load_1): Likewise.
717 (emit_group_store): Likewise.
718 (copy_blkmode_from_reg): Likewise.
719 (emit_push_insn): Likewise.
720 (expand_assignment): Likewise.
721 (store_field): Likewise.
722 (expand_expr_real_1): Likewise.
723 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
724 * lra-constraints.c (simplify_operand_subreg): Likewise.
725 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
726 * gimple-ssa-store-merging.c: Likewise in block comment at start
728 * tree-ssa-strlen.c: Include target.h.
729 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
730 of SLOW_UNALIGNED_ACCESS.
731 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
733 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
735 PR rtl-optimization/82185
736 * expmed.c (emit_store_flag_int): Only test tem if it has been
739 2017-09-12 Richard Biener <rguenther@suse.de>
742 * match.pd ((FTYPE) N CMP CST): Fix typo.
744 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
746 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
748 (mips_near_type_p): Add 'short_call' attribute as a synonym
750 * doc/extend.texi (short_call): Document new function attribute.
752 2017-09-12 Jakub Jelinek <jakub@redhat.com>
755 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
756 assertion check that in the condition.
757 (get_atomic_generic_size): Likewise. Before testing if parameter
758 has pointer type, if it has array type, call for C++
759 default_conversion to perform array-to-pointer conversion.
761 2017-09-12 Richard Biener <rguenther@suse.de>
763 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
764 for operations we cannot scalarize.
766 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
768 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
769 vectors heap vectors. Clean up comments.
770 Make visited_bbs a reference.
771 (profitable_jump_thread_path): Make GC
772 vectors heap vectors. Clean up comments.
774 (convert_and_register_jump_thread_path): Make GC vectors heap
776 (check_subpath_and_update_thread_path): Same. Clean up comments.
777 Make visited_bbs a reference.
778 (handle_phi): Abstract common code to to
779 register_jump_thread_path_if_profitable.
780 Rename VAR_BB to DEF_BB.
782 Make GC vectors heap vectors.
783 Make visited_bbs a reference.
784 (handle_assignment): Same.
785 (register_jump_thread_path_if_profitable): New.
786 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
788 Make GC vectors heap vectors. Clean up comments.
789 Make visited_bbs a reference.
790 (find_jump_threads_backwards): Make visited_bbs live in the stack.
791 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
794 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
797 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
798 words of E_DImode object are reachable by xtensa_uimm8x4 access.
800 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
802 Revert r251800 and r251799.
804 2017-09-11 Martin Jambor <mjambor@suse.cz>
807 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
808 arguments in advance.
809 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
810 use it to find predecessor edges.
811 (naive_outof_ssa): Collect vector of predecessors.
813 2017-09-08 Jason Merrill <jason@redhat.com>
815 PR c++/70029 - ICE with ref-qualifier and -flto
816 * langhooks.h (struct lang_hooks_for_types): Add
817 copy_lang_qualifiers.
818 * attribs.c (build_type_attribute_qual_variant): Use it.
819 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
821 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
822 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
824 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
827 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
828 (*mulsi3_sp64): New instruction.
829 (mulsi3): New expander.
831 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
833 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
835 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
837 * sancov.c: Include memmodel.h.
839 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
842 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
845 2017-09-07 Carl Love <cel@us.ibm.com>
847 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
848 the sldi instruction.
850 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
852 * sancov.c: Include tm_p.h.
854 2017-09-07 Jakub Jelinek <jakub@redhat.com>
857 * output.h (switch_to_other_text_partition): New declaration.
858 * varasm.c (switch_to_other_text_partition): New function.
859 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
860 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
861 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
862 to the other text partition before emitting LCL label and switch back
863 after emitting the word after it.
865 2017-09-07 Richard Biener <rguenther@suse.de>
867 * passes.def (pass_split_crit_edges): Remove instance before PRE.
868 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
869 critical edges here, after loop init.
870 (pass_data_pre): Remove PROP_no_crit_edges flags.
871 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
872 for valueization of call args to avoid leaking VN_TOP.
873 (visit_use): Assert we do not visit default defs.
874 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
875 Use error_mark_node to more easily detect leaking VN_TOP.
876 All default-defs are varying, not VN_TOP. Mark them visited.
877 (run_scc_vn): Make code match comment.
879 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
881 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
882 OPTION_MASK_FLOAT128_KEYWORD.
883 (POWERPC_MASKS): Likewise.
884 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
885 support for the -mfloat128-type option, and make -mfloat128
886 default on PowerPC Linux systems. Define or undefine
887 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
888 Define __float128 to be __ieee128 if IEEE 128-bit support is
889 enabled, or undefine it.
890 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
891 Delete defining __FLOAT128_TYPE__.
892 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
893 -mfloat128-type option and make -mfloat128 default on PowerPC
895 (TARGET_FLOAT128_TYPE): Likewise.
896 (-mfloat128-type): Likewise.
897 * config/rs6000/rs6000.c (rs6000_option_override_internal):
898 Delete the -mfloat128-type option and make -mfloat128 default on
899 PowerPC Linux systems. Always use __ieee128 to be the keyword for
900 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
901 128-bit floating point is enabled. Change tests from using
902 -mfloat128-type to -mfloat128.
903 (rs6000_mangle_type): Use the correct mangling for the __float128
904 type even if normal long double is restricted to 64-bits.
905 (floatn_mode): Enable the _Float128 type by default on VSX Linux
907 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
908 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
909 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
911 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
912 documentation for -mfloat128.
914 2017-09-06 Olivier Hainque <hainque@adacore.com>
916 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
918 2017-09-06 Wish Wu <wishwu007@gmail.com>
919 Jakub Jelinek <jakub@redhat.com>
921 * asan.c (initialize_sanitizer_builtins): Add
922 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
923 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
924 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
925 BT_FN_VOID_UINT64_PTR variables.
926 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
927 (BT_FN_VOID_UINT16_UINT16): Likewise.
928 (BT_FN_VOID_UINT32_UINT32): Likewise.
929 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
930 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
931 (BT_FN_VOID_UINT64_PTR): Likewise.
932 * common.opt (flag_sanitize_coverage): New variable.
933 (fsanitize-coverage=trace-pc): Remove.
934 (fsanitize-coverage=): Add.
935 * flag-types.h (enum sanitize_coverage_code): New enum.
936 * fold-const.c (fold_range_test): Disable non-short-circuit
937 optimization if flag_sanitize_coverage.
938 (fold_truth_andor): Likewise.
939 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
940 * opts.c (COVERAGE_SANITIZER_OPT): Define.
941 (coverage_sanitizer_opts): New array.
942 (get_closest_sanitizer_option): Add OPTS argument, handle also
943 OPT_fsanitize_coverage_.
944 (parse_sanitizer_options): Adjusted to also handle
945 OPT_fsanitize_coverage_.
946 (common_handle_option): Add OPT_fsanitize_coverage_.
947 * sancov.c (instrument_comparison, instrument_switch): New function.
948 (sancov_pass): Add trace-cmp support.
949 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
950 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
951 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
952 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
953 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
954 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
955 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
956 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
957 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
958 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
960 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
962 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
963 error. Only quit immediately if parsing is complete.
964 (BEGIN): Initialize fatal_err and parse_done.
965 (begin fpu, end fpu): Check number of arguments.
966 (begin arch, end arch): Likewise.
967 (begin cpu, end cpu): Likewise.
968 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
969 (optalias): Likewise.
971 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
973 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
974 * config/arm/arm-isa.h: Delete. Move definitions to ...
975 * arm-cpus.in: ... here. Use new feature and fgroup values.
976 * config/arm/arm.c (arm_option_override): Use lower case for feature
978 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
979 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
980 * config/arm/parsecpu.awk (END): Add new command 'isa'.
982 (print_isa_bits_for): New function.
983 (gen_isa): New function.
984 (gen_comm_data): Use print_isa_bits_for.
985 (define feature): New keyword.
986 (define fgroup): New keyword.
987 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
988 (arm-isa.h): Add rule to generate file.
989 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
990 case for feature bit names.
992 2017-09-06 Richard Biener <rguenther@suse.de>
994 * tree-ssa-pre.c (NECESSARY): Remove.
995 (create_expression_by_pieces): Do not touch pass-local flags.
996 (insert_into_preds_of_block): Likewise.
997 (do_pre_regular_insertion): Likewise.
998 (eliminate_insert): Likewise.
999 (eliminate_dom_walker::before_dom_children): Likewise.
1000 (fini_eliminate): Do not look at inserted_exprs.
1001 (mark_operand_necessary): Remove.
1002 (remove_dead_inserted_code): Replace with simple work-list
1003 algorithm based on inserted_exprs and SSA uses.
1004 (pass_pre::execute): Re-order fini_eliminate and
1005 remove_dead_inserted_code.
1007 2017-09-06 Olivier Hainque <hainque@adacore.com>
1009 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1010 for VxWorks 7. Adjust surrounding comments.
1012 2017-09-06 Richard Biener <rguenther@suse.de>
1014 * gimple-ssa-strength-reduction.c
1015 (find_candidates_dom_walker::before_dom_children): Also allow
1018 2017-09-06 Richard Biener <rguenther@suse.de>
1020 PR tree-optimization/82108
1021 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1022 for gap in the non-permutation SLP case.
1024 2017-09-06 Martin Jambor <mjambor@suse.cz>
1026 PR tree-optimization/82078
1027 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1028 add_access_to_work_queue...
1029 (build_accesses_from_assign): ...here.
1030 (propagate_all_subaccesses): Make sure racc is the group
1031 representative, if there is one.
1033 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1036 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1039 2017-09-06 Richard Biener <rguenther@suse.de>
1041 * gimple-ssa-strength-reduction.c
1042 (find_candidates_dom_walker::before_doom_children): Use a
1043 type and not a mode check.
1045 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1048 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1050 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1052 2017-09-05 Jeff Law <law@redhat.com>
1054 PR tree-optimization/64910
1055 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1056 cases where we have 3 or more operands.
1058 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1061 * omp-low.c (lower_omp_for): Recompute tree invariant if
1062 gimple_omp_for_initial/final is ADDR_EXPR.
1065 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1066 into gimple val before gimplification fo the COND_EXPR.
1068 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1070 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1071 REGION_COPY argument.
1072 (thread_through_all_blocks): Remove unused argument to
1073 duplicate_thread_path.
1075 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1076 Alan Hayward <alan.hayward@arm.com>
1077 David Sherwood <david.sherwood@arm.com>
1079 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1080 Take a scalar_mode rather than a machine_mode.
1081 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1082 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1083 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1084 (aarch64_gen_adjusted_ldpstp): Likewise.
1085 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1087 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1088 Alan Hayward <alan.hayward@arm.com>
1089 David Sherwood <david.sherwood@arm.com>
1091 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1092 Take a scalar_int_mode instead of a machine_mode.
1093 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1094 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1095 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1096 (aarch64_simd_attr_length_rglist): Delete.
1097 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1098 a scalar_int_mode instead of a machine_mode.
1099 (aarch64_add_offset): Likewise.
1100 (aarch64_internal_mov_immediate): Likewise
1101 (aarch64_add_constant_internal): Likewise.
1102 (aarch64_add_constant): Likewise.
1103 (aarch64_movw_imm): Likewise.
1104 (aarch64_rtx_arith_op_extract_p): Likewise.
1105 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1106 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1107 Remove assert that the mode isn't a vector.
1108 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1109 (aarch64_expand_mov_immediate): Update calls after above changes.
1110 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1111 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1112 (aarch64_move_imm): Likewise.
1113 (aarch64_can_const_movi_rtx_p): Likewise.
1114 (aarch64_strip_extend): Likewise.
1115 (aarch64_extr_rtx_p): Likewise.
1116 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1117 a CONST_INT when the mode parameter is VOIDmode.
1118 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1120 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1122 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1123 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1124 (bitwise_type_for_mode): Update accordingly.
1126 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1128 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1129 * stor-layout.c (mode_for_size_tree): Likewise.
1130 (mode_for_array): Update accordingly.
1131 (layout_decl): Likewise.
1132 (compute_record_mode): Likewise. Only set the mode once.
1134 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1136 * target.def (get_mask_mode): Change return type to opt_mode.
1138 * doc/tm.texi: Regenerate.
1139 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1140 * targhooks.c (default_get_mask_mode): Likewise.
1141 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1142 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1143 targetm.get_mask_mode.
1144 * tree.c (build_truth_vector_type): Likewise.
1146 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1148 * machmode.h (mode_for_vector): Return an opt_mode.
1149 * stor-layout.c (mode_for_vector): Likewise.
1150 (mode_for_int_vector): Update accordingly.
1151 (layout_type): Likewise.
1152 * config/i386/i386.c (emit_memmov): Likewise.
1153 (ix86_expand_set_or_movmem): Likewise.
1154 (ix86_expand_vector_init): Likewise.
1155 (ix86_get_mask_mode): Likewise.
1156 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1158 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1159 * expmed.c (extract_bit_field_1): Likewise.
1160 * expr.c (expand_expr_real_2): Likewise.
1161 * optabs-query.c (can_vec_perm_p): Likewise.
1162 (can_vec_mask_load_store_p): Likewise.
1163 * optabs.c (expand_vec_perm): Likewise.
1164 * targhooks.c (default_get_mask_mode): Likewise.
1165 * tree-vect-stmts.c (vectorizable_store): Likewise.
1166 (vectorizable_load): Likewise.
1167 (get_vectype_for_scalar_type_and_size): Likewise.
1169 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1171 * machmode.h (mode_for_int_vector): New function.
1172 * stor-layout.c (mode_for_int_vector): Likewise.
1173 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1174 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1175 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1176 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1177 (s390_expand_vcond): Likewise.
1179 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1181 * machmode.h (opt_machine_mode): New type.
1182 (opt_mode<T>): Allow construction from anything that can be
1184 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1185 (mode_for_size): Return an opt_machine_mode.
1186 * stor-layout.c (mode_for_size): Likewise.
1187 (mode_for_size_tree): Update call accordingly.
1188 (bitwise_mode_for_mode): Likewise.
1189 (make_fract_type): Likewise.
1190 (make_accum_type): Likewise.
1191 * caller-save.c (replace_reg_with_saved_mem): Update call
1193 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1194 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1195 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1196 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1197 * expmed.c (extract_bit_field_1): Likewise.
1198 * reload.c (get_secondary_mem): Likewise.
1199 * varasm.c (assemble_integer): Likewise.
1200 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1203 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1205 * machmode.h (decimal_float_mode_for_size): New function.
1206 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1207 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1208 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1209 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1210 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1211 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1213 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1215 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1216 (get_builtin_sync_mode): Likewise.
1217 (expand_ifn_atomic_compare_exchange): Likewise.
1218 (expand_builtin_atomic_clear): Likewise.
1219 (expand_builtin_atomic_test_and_set): Likewise.
1220 (fold_builtin_atomic_always_lock_free): Likewise.
1221 * calls.c (compute_argument_addresses): Likewise.
1222 (emit_library_call_value_1): Likewise.
1223 (store_one_arg): Likewise.
1224 * combine.c (combine_instructions): Likewise.
1225 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1226 * config/arm/arm.c (arm_function_value): Likewise.
1227 (aapcs_allocate_return_reg): Likewise.
1228 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1229 * config/i386/i386.c (construct_container): Likewise.
1230 (ix86_gimplify_va_arg): Likewise.
1231 (ix86_expand_sse_cmp): Likewise.
1232 (emit_memmov): Likewise.
1233 (emit_memset): Likewise.
1234 (expand_small_movmem_or_setmem): Likewise.
1235 (ix86_expand_pextr): Likewise.
1236 (ix86_expand_pinsr): Likewise.
1237 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1238 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1240 * config/mips/mips.c (mips_function_value_1) Likewise.
1241 (mips_block_move_straight): Likewise.
1242 (mips_expand_ins_as_unaligned_store): Likewise.
1243 * config/powerpcspe/powerpcspe.c
1244 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1245 (rs6000_darwin64_record_arg_flush): Likewise.
1246 * config/rs6000/rs6000.c
1247 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1248 (rs6000_darwin64_record_arg_flush): Likewise.
1249 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1250 (sparc_function_value_1): Likewise.
1251 * config/spu/spu.c (adjust_operand): Likewise.
1252 (spu_emit_branch_or_set): Likewise.
1253 (arith_immediate_p): Likewise.
1254 * emit-rtl.c (gen_lowpart_common): Likewise.
1255 * expr.c (expand_expr_real_1): Likewise.
1256 * function.c (assign_parm_setup_block): Likewise.
1257 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1258 * reload1.c (alter_reg): Likewise.
1259 * stor-layout.c (mode_for_vector): Likewise.
1260 (layout_type): Likewise.
1262 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1264 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1265 (spu_convert_move): Likewise.
1266 * lower-subreg.c (resolve_simple_move): Likewise.
1268 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1271 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1272 define_insn to a define_expand.
1273 (altivec_vsum2sws_direct): New define_insn.
1274 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1276 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1278 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1281 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1286 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1287 realign stack if stack alignment needed is less than incoming
1290 2017-09-05 Marek Polacek <polacek@redhat.com>
1293 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1296 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1298 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1300 2017-09-05 Richard Biener <rguenther@suse.de>
1302 PR tree-optimization/82084
1303 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1305 2017-09-05 Richard Biener <rguenther@suse.de>
1307 PR tree-optimization/82102
1308 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1310 2017-09-05 Martin Liska <mliska@suse.cz>
1312 PR tree-optimization/82032
1313 * tree-cfg.c (generate_range_test): New function.
1314 * tree-cfg.h (generate_range_test): Declared here.
1315 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1316 (cleanup_control_expr_graph): Use it.
1317 * tree-switch-conversion.c (try_switch_expansion): Remove
1319 (emit_case_nodes): Use generate_range_test.
1321 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1324 * config/i386/i386.md (*<btsc><mode>_mask): Add
1325 TARGET_USE_BT to insn constraint.
1326 (*btr<mode>_mask): Ditto.
1328 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1330 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1331 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1333 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1336 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1337 TARGET_NEON and TARGET_IWMMXT.
1338 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1339 TARGET_NEON and TARGET_IWMMXT.
1340 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1342 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1344 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1345 (ix86_rewrite_tls_address): Ditto.
1346 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1347 (ix86_rewrite_tls_address_1): Ditto.
1348 (ix86_rewrite_tls_address): Ditto.
1349 * config/i386/predicates.md (tls_address_pattern): New predicate.
1350 * config/i386/i386.md (TLS address splitter): New splitter.
1352 2017-09-04 Richard Biener <rguenther@suse.de>
1354 PR tree-optimization/82084
1355 * fold-const.h (can_native_encode_string_p): Declare.
1356 * fold-const.c (can_native_encode_string_p): Factor out from ...
1357 (native_encode_string): ... here.
1358 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1359 vectorizing stores from constants we later cannot handle.
1361 2017-09-04 Marek Polacek <polacek@redhat.com>
1364 * doc/invoke.texi: Update -Wtautological-compare documentation.
1366 2017-09-04 Jeff Law <law@redhat.com>
1368 PR tree-optimization/64910
1369 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1370 swap the first and last operand if the last is a constant.
1372 2017-09-04 Marek Polacek <polacek@redhat.com>
1375 * convert.c (do_narrow): When sanitizing signed integer overflows,
1376 bail out for signed types.
1377 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1379 2017-09-04 Richard Biener <rguenther@suse.de>
1381 PR tree-optimization/82060
1382 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1383 Move devirtualization after stmt folding and before EH/AB/noreturn
1384 cleanup to get the stmt refs canonicalized. Use a bool instead
1385 of gimple_modified_p since that doesn't work for NOPs. Schedule
1386 NOPs generated by folding for removal.
1388 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1389 Alan Hayward <alan.hayward@arm.com>
1390 David Sherwood <david.sherwood@arm.com>
1392 * coretypes.h (pad_direction): New enum.
1393 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1394 (FUNCTION_ARG_PADDING): Likewise.
1395 * target.def (function_arg_padding): New hook.
1396 * targhooks.h (default_function_arg_padding): Declare.
1397 * targhooks.c (default_function_arg_padding): New function.
1398 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1399 (TARGET_FUNCTION_ARG_PADDING): ...this.
1400 * doc/tm.texi: Regenerate.
1401 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1402 instead of direction.
1403 (compute_argument_addresses): Likewise.
1404 (load_register_parameters): Likewise.
1405 (emit_library_call_value_1): Likewise.
1406 (store_one_arg): Use targetm.calls.function_arg_padding instead
1407 of FUNCTION_ARG_PADDING.
1408 (must_pass_in_stack_var_size_or_pad): Likewise.
1409 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1410 (emit_group_store): Likewise.
1411 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1412 instead of FUNCTION_ARG_PADDING.
1413 (emit_push_insn): Likewise, and propagate enum change throughout
1415 * function.h (direction): Delete.
1416 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1418 * function.c (assign_parm_find_stack_rtl): Likewise.
1419 (assign_parm_setup_block_p): Likewise.
1420 (assign_parm_setup_block): Likewise.
1421 (gimplify_parameters): Likewise.
1422 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1423 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1425 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1426 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1427 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1428 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1429 (aarch64_function_arg_padding): ...this new function.
1430 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1431 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1432 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1433 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1434 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1435 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1436 (arm_pad_arg_upward): Replace with...
1437 (arm_function_arg_padding): ...this new function.
1438 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1440 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1441 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1442 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1443 (ia64_hpux_function_arg_padding): Replace with...
1444 (ia64_function_arg_padding): ...this new function. Use pad_direction
1445 instead of direction. Check for TARGET_HPUX.
1446 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1447 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1448 (iq2000_function_arg_padding): New function.
1449 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1450 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1451 (mips_function_arg_padding): ...this new function.
1452 (mips_pad_reg_upward): Update accordingly.
1453 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1454 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1455 targetm.calls.function_arg_padding.
1456 (FUNCTION_ARG_PADDING): Delete.
1457 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1458 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1459 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1460 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1461 (nios2_block_reg_padding): Return pad_direction instead of direction.
1462 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1463 instead of direction.
1464 (nios2_function_arg_padding): Likewise. Make static.
1465 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1466 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1467 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1468 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1469 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1470 (pa_function_arg_padding): Make static. Return pad_direction instead
1472 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1473 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1474 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1475 instead of direction. Use targetm.calls.function_arg_padding.
1476 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1477 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1478 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1479 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1480 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1482 (function_arg_padding): Rename to...
1483 (rs6000_function_arg_padding): ...this. Make static. Return
1484 pad_direction instead of direction.
1485 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1486 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1487 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1488 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1489 instead of direction. Use targetm.calls.function_arg_padding.
1490 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1491 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1492 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1493 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1494 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1495 (function_arg_padding): Rename to...
1496 (rs6000_function_arg_padding): ...this. Make static. Return
1497 pad_direction instead of direction.
1498 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1499 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1500 * config/s390/s390.c (s390_function_arg_padding): New function.
1501 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1502 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1503 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1504 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1505 (function_arg_padding): Rename to...
1506 (sparc_function_arg_padding): ...this. Make static. Return
1507 pad_direction instead of direction.
1508 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1509 * config/spu/spu.c (spu_function_arg_padding): New function.
1510 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1511 * system.h (FUNCTION_ARG_PADDING): Poison.
1513 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1514 Alan Hayward <alan.hayward@arm.com>
1515 David Sherwood <david.sherwood@arm.com>
1517 * target.def (modes_tieable_p): New hook.
1518 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1519 (TARGET_MODES_TIEABLE_P): ...this.
1520 * doc/tm.texi.in: Regenerate.
1521 * hooks.h (hook_bool_mode_mode_true): Declare.
1522 * hooks.c (hook_bool_mode_mode_true): New function.
1523 * combine.c (subst): Use targetm.modes_tieable_p instead of
1525 * dse.c (find_shift_sequence): Likewise.
1526 * expmed.c (extract_low_bits): Likewise.
1527 * lower-subreg.c: Include target.h.
1528 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1530 * rtlanal.c (rtx_cost): Likewise.
1531 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1532 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1533 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1534 (TARGET_MODES_TIEABLE_P): Redefine.
1535 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1536 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1537 (TARGET_MODES_TIEABLE_P): Redefine.
1538 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1539 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1540 (arc_modes_tieable_p): New function.
1541 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1542 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1543 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1544 (arm_modes_tieable_p): Make static.
1545 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1546 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1547 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1548 (TARGET_MODES_TIEABLE_P): Redefine.
1549 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1550 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1551 (TARGET_MODES_TIEABLE_P): Redefine.
1552 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1553 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1554 (cr16_modes_tieable_p): New function.
1555 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1556 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1557 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1558 (TRULY_NOOP_TRUNCATION): Update comment.
1559 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1560 (TRULY_NOOP_TRUNCATION): Update comment.
1561 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1562 (frv_modes_tieable_p): New function.
1563 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1564 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1565 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1566 (TARGET_MODES_TIEABLE_P): Redefine.
1567 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1568 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1569 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1570 (TARGET_MODES_TIEABLE_P): Redefine.
1571 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1572 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1573 (ia64_modes_tieable_p): New function.
1574 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1575 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1576 (iq2000_modes_tieable_p): New function.
1577 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1578 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1579 (lm32_modes_tieable_p): New function.
1580 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1581 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1582 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1583 (TARGET_MODES_TIEABLE_P): Redefine.
1584 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1585 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1586 (m32r_modes_tieable_p): New function.
1587 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1588 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1589 (m68k_modes_tieable_p): New function.
1590 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1591 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1592 (mcore_modes_tieable_p): New function.
1593 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1594 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1596 (TARGET_MODES_TIEABLE_P): Redefine.
1597 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1598 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1599 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1600 (TARGET_MODES_TIEABLE_P): Redefine.
1601 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1602 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1603 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1604 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1605 (mn10300_modes_tieable_p): ...this and make static.
1606 (TARGET_MODES_TIEABLE_P): Redefine.
1607 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1608 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1609 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1610 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1611 (msp430_modes_tieable_p): Make static.
1612 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1613 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1614 (TARGET_MODES_TIEABLE_P): Redefine.
1615 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1616 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1617 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1618 (TARGET_MODES_TIEABLE_P): Redefine.
1619 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1620 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1621 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1622 (TARGET_MODES_TIEABLE_P): Redefine.
1623 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1624 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1625 (pdp11_modes_tieable_p): New function.
1626 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1627 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1628 (rs6000_modes_tieable_p): New function.
1629 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1630 * config/powerpcspe/powerpcspe.md: Update comment.
1631 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1632 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1633 (TARGET_MODES_TIEABLE_P): Redefine.
1634 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1635 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1636 (rl78_modes_tieable_p): New function.
1637 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1638 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1639 (rs6000_modes_tieable_p): New function.
1640 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1641 * config/rs6000/rs6000.md: Update comment.
1642 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1643 * config/rx/rx.c (rx_modes_tieable_p): New function.
1644 (TARGET_MODES_TIEABLE_P): Redefine.
1645 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1646 * config/s390/s390.c (s390_modes_tieable_p): New function.
1647 (TARGET_MODES_TIEABLE_P): Redefine.
1648 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1649 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1650 (sh_modes_tieable_p): New function.
1651 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1652 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1653 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1654 (sparc_modes_tieable_p): Make static.
1655 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1656 * config/spu/spu.c (spu_modes_tieable_p): New function.
1657 (TARGET_MODES_TIEABLE_P): Redefine.
1658 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1659 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1660 (TARGET_MODES_TIEABLE_P): Redefine.
1661 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1662 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1663 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1664 * config/v850/v850.c (v850_modes_tieable_p): New function.
1665 (TARGET_MODES_TIEABLE_P): Redefine.
1666 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1667 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1668 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1669 (visium_modes_tieable_p): New function.
1670 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1671 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1672 (xtensa_modes_tieable_p): New function.
1673 * system.h (MODES_TIEABLE_P): Poison.
1675 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1676 Alan Hayward <alan.hayward@arm.com>
1677 David Sherwood <david.sherwood@arm.com>
1679 * target.def (hard_regno_mode_ok): New hook.
1680 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1681 (TARGET_HARD_REGNO_MODE_OK): ...this.
1682 * doc/tm.texi.in: Regenerate.
1683 * hooks.h (hook_bool_uint_mode_true): Declare.
1684 * hooks.c (hook_bool_uint_mode_true): New function.
1685 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1687 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1688 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1689 instead of HARD_REGNO_MODE_OK.
1690 * caller-save.c: Include target.h.
1691 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1693 * combine.c (can_combine_p): Likewise.
1694 (combinable_i3pat): Likewise.
1695 (can_change_dest_mode): Likewise.
1696 * expr.c (init_expr_target): Likewise.
1697 (convert_move): Likewise.
1698 (convert_modes): Likewise.
1699 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1700 (setup_prohibited_mode_move_regs): Likewise.
1701 * ira.h (target_ira): Likewise.
1702 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1703 * lra-constraints.c (process_alt_operands): Likewise.
1704 (split_reg): Likewise.
1705 * recog.c (peep2_find_free_register): Likewise.
1706 * ree.c (combine_reaching_defs): Likewise.
1707 * regcprop.c (maybe_mode_change): Likewise.
1708 * reginfo.c (init_reg_sets_1): Likewise.
1709 (choose_hard_reg_mode): Likewise.
1710 (simplifiable_subregs): Likewise.
1711 * regrename.c (check_new_reg_p): Likewise.
1712 * reload.c (find_valid_class): Likewise.
1713 (find_valid_class_1): Likewise.
1714 (reload_inner_reg_of_subreg): Likewise.
1715 (push_reload): Likewise.
1716 (combine_reloads): Likewise.
1717 (find_dummy_reload): Likewise.
1718 (find_reloads): Likewise.
1719 * reload1.c (find_reg): Likewise.
1720 (set_reload_reg): Likewise.
1721 (allocate_reload_reg): Likewise.
1722 (choose_reload_regs): Likewise.
1723 (reload_adjust_reg_for_temp): Likewise.
1724 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1725 (simplify_subreg_regno): Likewise.
1726 * sel-sched.c (init_regs_for_mode): Likewise.
1727 * varasm.c (make_decl_rtl): Likewise.
1728 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1729 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1731 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1732 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1734 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1735 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1736 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1737 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1738 (arc_mode_class): Delete.
1739 (HARD_REGNO_MODE_OK): Delete.
1740 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1741 (arc_hard_regno_mode_ok): Rename old array to...
1742 (arc_hard_regno_mode_ok_modes): ...this.
1743 (arc_conditional_register_usage): Update accordingly.
1744 (arc_mode_class): Make static.
1745 (arc_hard_regno_mode_ok): New function.
1746 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1747 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1748 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1749 (arm_hard_regno_mode_ok): Make static.
1750 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1752 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1753 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1754 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1756 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1757 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1758 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1759 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1760 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1761 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1762 * config/bfin/predicates.md (valid_reg_operand): Use
1763 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1764 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1765 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1766 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1767 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1768 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1769 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1770 (cr16_hard_regno_mode_ok): Make static and return a bool.
1771 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1772 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1773 (cris_hard_regno_mode_ok): New function.
1774 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1775 (epiphany_mode_class): Delete.
1776 (HARD_REGNO_MODE_OK): Delete.
1777 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1778 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1779 (hard_regno_mode_ok): Rename to...
1780 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1781 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1782 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1784 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1785 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1786 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1787 (frv_hard_regno_mode_ok): Make static and return a bool.
1788 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1790 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1791 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1792 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1793 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1795 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1796 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1797 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1798 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1801 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1802 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1803 (ia64_hard_regno_mode_ok): New function.
1804 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1805 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1806 (iq2000_hard_regno_mode_ok): New function.
1807 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1808 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1809 (lm32_hard_regno_mode_ok): New function.
1810 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1811 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1812 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1813 instead of HARD_REGNO_MODE_OK.
1814 (m32c_hard_regno_ok): Rename to...
1815 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1816 (m32c_cannot_change_mode_class): Update accordingly.
1817 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1818 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1819 (m32r_mode_class): Delete.
1820 (HARD_REGNO_MODE_OK): Delete.
1821 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1822 (m32r_hard_regno_mode_ok): Rename to...
1823 (m32r_hard_regno_modes): ...this.
1824 (m32r_mode_class): Make static.
1825 (m32r_hard_regno_mode_ok): New function.
1826 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1827 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1828 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1829 (m68k_hard_regno_mode_ok): Make static.
1830 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1831 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1832 (mcore_hard_regno_mode_ok): New function.
1833 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1834 (HARD_REGNO_MODE_OK): Delete.
1835 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1837 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1838 (microblaze_hard_regno_mode_ok): New function.
1839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1840 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1841 (mips_hard_regno_mode_ok): Delete.
1842 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1843 (mips_hard_regno_mode_ok_p): ...this and make static.
1844 (mips_hard_regno_mode_ok_p): Rename to...
1845 (mips_hard_regno_mode_ok_uncached): ...this.
1846 (mips_hard_regno_mode_ok): New function.
1847 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1848 of HARD_REGNO_MODE_OK.
1849 (mips_option_override): Update after above name changes.
1850 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1851 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1852 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1853 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1854 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1855 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1856 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1857 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1858 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1859 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1860 (msp430_hard_regno_mode_ok): Make static and return a bool.
1861 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1862 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1863 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1866 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1867 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1868 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1869 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1870 (PA_HARD_REGNO_MODE_OK): ...this
1871 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1872 (PA_HARD_REGNO_MODE_OK): ...this.
1873 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1874 (pa_hard_regno_mode_ok): New function.
1875 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1876 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1877 (pdp11_hard_regno_mode_ok): New function.
1878 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1879 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1881 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1884 (rs6000_hard_regno_mode_ok): Rename to...
1885 (rs6000_hard_regno_mode_ok_uncached): ...this.
1886 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1887 (rs6000_hard_regno_mode_ok): New function.
1888 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1889 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1890 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1891 (riscv_hard_regno_mode_ok): ...this and make static.
1892 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1893 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1894 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1895 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1896 (rl78_hard_regno_mode_ok): Make static and return bool.
1897 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1898 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1900 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1901 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1902 (rs6000_hard_regno_mode_ok): Rename to...
1903 (rs6000_hard_regno_mode_ok_uncached): ...this.
1904 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1905 (rs6000_hard_regno_mode_ok): New function.
1906 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1907 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1908 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1909 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1910 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1911 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1912 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1913 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1914 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1915 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1916 (sh_hard_regno_mode_ok): Make static.
1917 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1918 instead of HARD_REGNO_MODE_OK.
1919 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1920 (sparc_mode_class): Delete.
1921 (HARD_REGNO_MODE_OK): Delete.
1922 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1923 (hard_regno_mode_classes): Make static.
1924 (sparc_mode_class): Likewise.
1925 (sparc_hard_regno_mode_ok): New function.
1926 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1927 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1928 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1930 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1931 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1932 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1933 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1934 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1935 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1936 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1937 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1938 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1939 (visium_hard_regno_mode_ok): New function.
1940 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1941 instead of HARD_REGNO_MODE_OK.
1942 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1943 (HARD_REGNO_MODE_OK): Delete.
1944 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1945 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1946 (xtensa_option_override): Update accordingly.
1947 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1948 (xtensa_hard_regno_mode_ok): New function.
1949 * system.h (HARD_REGNO_MODE_OK): Poison.
1951 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1952 Alan Hayward <alan.hayward@arm.com>
1953 David Sherwood <david.sherwood@arm.com>
1955 * target.def (hard_regno_call_part_clobbered): New hook.
1956 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1957 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1958 * doc/tm.texi: Regenerate.
1959 * hooks.h (hook_bool_uint_mode_false): Declare.
1960 * hooks.c (hook_bool_uint_mode_false): New function.
1961 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1962 * cselib.c (cselib_process_insn): Use
1963 targetm.hard_regno_call_part_clobbered instead of
1964 HARD_REGNO_CALL_PART_CLOBBERED.
1965 * ira-conflicts.c (ira_build_conflicts): Likewise.
1966 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1967 * lra-constraints.c (need_for_call_save_p): Likewise.
1968 * lra-lives.c: Include target.h.
1969 (check_pseudos_live_through_calls): Use
1970 targetm.hard_regno_call_part_clobbered instead of
1971 HARD_REGNO_CALL_PART_CLOBBERED.
1972 * regcprop.c: Include target.h.
1973 (copyprop_hardreg_forward_1): Use
1974 targetm.hard_regno_call_part_clobbered instead of
1975 HARD_REGNO_CALL_PART_CLOBBERED.
1976 * reginfo.c (choose_hard_reg_mode): Likewise.
1977 * regrename.c (check_new_reg_p): Likewise.
1978 * reload.c (find_equiv_reg): Likewise.
1979 * reload1.c (emit_reload_insns): Likewise.
1980 * sched-deps.c (deps_analyze_insn): Likewise.
1981 * sel-sched.c (init_regs_for_mode): Likewise.
1982 (mark_unavailable_hard_regs): Likewise.
1983 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1984 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1985 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1987 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1988 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1989 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1991 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1993 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1994 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1995 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1997 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1998 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1999 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2001 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2002 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2004 * config/powerpcspe/powerpcspe.c
2005 (rs6000_hard_regno_call_part_clobbered): New function.
2006 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2007 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2008 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2010 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2011 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2012 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2014 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2015 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2016 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2018 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2019 Alan Hayward <alan.hayward@arm.com>
2020 David Sherwood <david.sherwood@arm.com>
2022 * rtl.h (subreg_memory_offset): Declare.
2023 * emit-rtl.c (subreg_memory_offset): New function.
2024 * expmed.c (store_bit_field_1): Use it.
2025 * expr.c (undefined_operand_subword_p): Likewise.
2026 * simplify-rtx.c (simplify_subreg): Likewise.
2028 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2030 PR rtl-optimization/57448
2033 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2034 using atomic_load pattern.
2035 (expand_atomic_store): Likewise.
2037 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2040 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2041 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2042 handling. Use replace_call_with_value with NULL instead of
2043 gsi_replace, unlink_stmt_vdef and release_defs.
2045 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2048 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2050 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2053 * rtl.h (emit_library_call_value_1): Declare.
2054 (emit_library_call): Replace declaration with a series of overloads.
2055 Remove the parameter count argument.
2056 (emit_library_call_value): Likewise.
2057 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2058 with an "rtx_mode_t *".
2059 (emit_library_call_value): Delete.
2060 (emit_library_call): Likewise.
2061 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2062 (asan_emit_allocas_unpoison): Likewise.
2063 * builtins.c (expand_builtin_powi): Likewise.
2064 (expand_asan_emit_allocas_unpoison): Likewise.
2065 * cfgexpand.c (expand_main_function): Likewise.
2066 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2067 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2068 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2069 * config/arm/arm.c (arm_trampoline_init): Likewise.
2070 (arm_call_tls_get_addr): Likewise.
2071 (arm_expand_divmod_libfunc): Likewise.
2072 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2073 (smulsi3_highpart): Likewise.
2074 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2075 (c6x_expand_compare): Likewise.
2076 (c6x_expand_movmem): Likewise.
2077 * config/frv/frv.c (frv_trampoline_init): Likewise.
2078 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2079 (ix86_expand_divmod_libfunc): Likewise.
2080 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2081 (ia64_expand_compare): Likewise.
2082 (ia64_profile_hook): Likewise.
2083 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2084 (nonlocal_goto): Likewise.
2085 (restore_stack_nonlocal): Likewise.
2086 * config/m32r/m32r.c (block_move_call): Likewise.
2087 (m32r_trampoline_init): Likewise.
2088 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2089 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2090 (m68k_call_m68k_read_tp): Likewise.
2091 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2092 (microblaze_expand_divide): Likewise.
2093 * config/mips/mips.h (mips_args): Likewise.
2094 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2095 (MIPS_ICACHE_SYNC): Likewise.
2096 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2097 (nios2_trampoline_init): Likewise.
2098 * config/pa/pa.c (hppa_tls_call): Likewise.
2099 (pa_trampoline_init): Likewise.
2100 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2101 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2102 (expand_strn_compare): Likewise.
2103 (rs6000_generate_compare): Likewise.
2104 (rs6000_expand_float128_convert): Likewise.
2105 (output_profile_hook): Likewise.
2106 (rs6000_trampoline_init): Likewise.
2107 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2108 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2109 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2110 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2111 (rs6000_generate_compare): Likewise.
2112 (rs6000_expand_float128_convert): Likewise.
2113 (output_profile_hook): Likewise.
2114 (rs6000_trampoline_init): Likewise.
2115 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2116 * config/sh/sh.c (sh_trampoline_init): Likewise.
2117 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2118 (sparc_emit_float_lib_cmp): Likewise.
2119 (sparc32_initialize_trampoline): Likewise.
2120 (sparc64_initialize_trampoline): Likewise.
2121 (sparc_profile_hook): Likewise.
2122 * config/spu/spu.c (ea_load_store): Likewise.
2123 * config/spu/spu.md (floatunssidf2): Likewise.
2124 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2125 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2126 * config/visium/visium.c (expand_block_move_4): Likewise.
2127 (expand_block_move_2): Likewise.
2128 (expand_block_move_1): Likewise.
2129 (expand_block_set_4): Likewise.
2130 (expand_block_set_2): Likewise.
2131 (expand_block_set_1): Likewise.
2132 (visium_trampoline_init): Likewise.
2133 (visium_profile_hook): Likewise.
2134 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2135 (xtensa_setup_frame_addresses): Likewise.
2136 (xtensa_trampoline_init): Likewise.
2137 * except.c (sjlj_emit_function_enter): Likewise.
2138 (sjlj_emit_function_exit): Likewise.
2139 * explow.c (allocate_dynamic_stack_space): Likewise.
2140 (probe_stack_range): Likewise.
2141 * expr.c (convert_mode_scalar): Likewise.
2142 * optabs.c (expand_binop): Likewise.
2143 (expand_twoval_binop_libfunc): Likewise.
2144 (expand_unop): Likewise.
2145 (prepare_cmp_insn): Likewise.
2146 (prepare_float_lib_cmp): Likewise.
2147 (expand_float): Likewise.
2148 (expand_fix): Likewise.
2149 (expand_fixed_convert): Likewise.
2150 (maybe_emit_sync_lock_test_and_set): Likewise.
2151 (expand_atomic_compare_and_swap): Likewise.
2152 (expand_mem_thread_fence): Likewise.
2153 (expand_atomic_fetch_op): Likewise.
2155 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2157 * doc/generic.texi (OpenACC): Adjust URL.
2158 * doc/invoke.texi (C Dialect Options): Ditto.
2160 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2162 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2163 predicate for operand 1. Add (m,<S>) constraint.
2164 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2165 Prevent memory operand 1 with register operand 2.
2167 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2169 PR rtl-optimization/82024
2170 * combine.c (try_combine): If the combination result is a PARALLEL,
2171 and we only need to retain the SET in there that would be placed
2172 at I2, check that we can place that at I3 instead, before doing so.
2174 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2177 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2179 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2180 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2183 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2184 Jeff Law <law@redhat.com>
2186 * varasm.c (bss_initializer_p): Do not put constants into .bss
2187 (categorize_decl_for_section): Handle bss_initializer_p returning
2188 false when DECL_INITIAL is NULL.
2190 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2193 * config/s390/s390.c (s390_can_inline_p): New function.
2195 2017-09-01 Jeff Law <law@redhat.com>
2197 PR tree-optimization/82052
2198 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2199 Always initialize the returned slot after a hash table miss
2200 when INSERT is true.
2202 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2204 * config/s390/s390.md (mem_signal_fence): Remove.
2205 * doc/md.texi (mem_signal_fence): Remove.
2206 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2208 * target-insns.def (mem_signal_fence): Remove.
2210 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2213 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2216 * asan.c (create_odr_indicator): Strip name encoding from assembler
2217 name before appending it after __odr_asan_.
2219 2017-09-01 Martin Liska <mliska@suse.cz>
2221 PR tree-optimization/82059
2222 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2223 frequency only when an edge is redirected.
2225 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2227 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2228 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2229 (arc_conditional_register_usage): Remove ARC600 lp_count
2231 (arc_file_start): Emit Tag_ARC_CPU_variation.
2232 (arc_can_use_doloop_p): New conditions to use ZOLs.
2233 (hwloop_fail): New function.
2234 (hwloop_optimize): Likewise.
2235 (hwloop_pattern_reg): Likewise.
2236 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2237 (arc_reorg_loops): New function, calls reorg_loops.
2238 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2239 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2241 (arc_loop_hazard): Remove function, functionality moved into
2243 (arc_hazard): Remove arc_loop_hazard call.
2244 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2245 into hwloop_optimize.
2246 (arc_label_align): Remove ZOL handling.
2247 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2248 * config/arc/arc.md (doloop_begin): Remove pattern.
2249 (doloop_begin_i): Likewise.
2250 (doloop_end_i): Likewise.
2251 (doloop_fallback): Likewise.
2252 (doloop_fallback_m): Likewise.
2253 (doloop_end): Reimplement expand.
2254 (arc_lp): New pattern for LP instruction.
2255 (loop_end): New pattern.
2256 (loop_fail): Likewise.
2257 (decrement_and_branch_until_zero): Likewise.
2258 * config/arc/arc.opt (mlpc-width): New option.
2259 * doc/invoke.texi (mlpc-width): Document option.
2261 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2263 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2264 (arc_ccfsm_advance): Fix checking for delay slots.
2265 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2267 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2269 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2271 (movhi_insn): Update store instruction constraint which are saving
2272 6-bit short immediates.
2273 (movsi_insn): Consider also short scaled load operations.
2274 (zero_extendhisi2_i): Use Usd constraint instead of T.
2275 (extendhisi2_i): Add q constraint.
2276 (arc_clzsi2): Add type and length attributes.
2277 (arc_ctzsi2): Likewise.
2278 * config/arc/constraints.md (Usc): Update constraint, the
2279 assembler can parse two relocations for a single instruction.
2281 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2283 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2284 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2286 2017-08-31 Olivier Hainque <hainque@adacore.com>
2288 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2289 match as powerpc-wrs-vxworks*.
2291 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2293 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2294 register constraint for by-element operand.
2295 (aarch64_mls_elt_merge<mode>): Likewise.
2297 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2299 * config/arc/arc.c (arc_can_follow_jump): Check for short
2302 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2304 * config.gcc: Use g.opt for arc.
2305 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2306 functionality moved to ...
2307 (legitimate_scaled_address_p): New function, ...here.
2308 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2309 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2310 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2312 (arc_override_options): Handle G option.
2313 (arc_output_pic_addr_const): Correct function definition.
2314 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2315 (arc_decl_anon_ns_mem_p): Delete.
2316 (arc_in_small_data_p): Overhaul this function to take into
2317 consideration the value given via G option.
2318 (arc_rewrite_small_data_1): Renamed and corrected old
2319 arc_rewrite_small_data function.
2320 (arc_rewrite_small_data): New function.
2321 (small_data_pattern): Don't use pic_offset_table_rtx.
2322 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2323 * config/arc/simdext.md (movmisalignv2hi): Use
2324 prepare_move_operands function.
2326 (movmisalign*): Likewise.
2327 * doc/invoke.texi (ARC options): Document -G option.
2329 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2331 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2333 * config/arc/arc.c (arc_print_operand): Output scalled address for
2334 sdata whenever is possible.
2335 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2336 load/stores are available.
2337 (compact_sda_memory_operand): Check for the alignment required by
2338 code density instructions.
2339 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2341 * config/arc/constraints.md (Usd): Update constraint.
2342 (Us0): New constraint.
2343 (Usc): Update constraint.
2345 2017-08-31 Richard Biener <rguenther@suse.de>
2348 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2351 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2353 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2354 Resize type_signature.
2356 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2357 Alan Hayward <alan.hayward@arm.com>
2358 David Sherwood <david.sherwood@arm.com>
2360 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2361 subregs whose inner modes can be stored in GPRs.
2362 (aarch64_classify_index): Likewise.
2364 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2365 Alan Hayward <alan.hayward@arm.com>
2366 David Sherwood <david.sherwood@arm.com>
2368 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2369 (V_INT_EQUIV): ...this.
2370 (v_cmp_result): Rename to...
2371 (v_int_equiv): ...this.
2372 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2373 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2374 (copysign<mode>3): Likewise.
2375 (aarch64_simd_bsl<mode>_internal): Likewise.
2376 (aarch64_simd_bsl<mode>): Likewise.
2377 (vec_cmp<mode><mode>): Likewise.
2378 (vcond<mode><mode>): Likewise.
2379 (vcond<v_cmp_mixed><mode>): Likewise.
2380 (vcondu<mode><v_cmp_mixed>): Likewise.
2381 (aarch64_cm<optab><mode>): Likewise.
2382 (aarch64_cmtst<mode>): Likewise.
2383 (aarch64_fac<optab><mode>): Likewise.
2384 (vec_perm_const<mode>): Likewise.
2385 (vcond_mask_<mode><v_cmp_result>): Rename to...
2386 (vcond_mask_<mode><v_int_equiv>): ...this.
2387 (vec_cmp<mode><v_cmp_result>): Rename to...
2388 (vec_cmp<mode><v_int_equiv>): ...this.
2390 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2391 Alan Hayward <alan.hayward@arm.com>
2392 David Sherwood <david.sherwood@arm.com>
2394 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2396 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2397 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2398 (UNSPEC_LD4_DREG): New unspecs.
2399 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2400 (aarch64_ld2<mode>_dreg_be): Replace with...
2401 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2403 (aarch64_ld3<mode>_dreg_le)
2404 (aarch64_ld3<mode>_dreg_be): Replace with...
2405 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2407 (aarch64_ld4<mode>_dreg_le)
2408 (aarch64_ld4<mode>_dreg_be): Replace with...
2409 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2412 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2414 PR tree-optimization/81987
2415 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2416 insert an initializer in a location not dominated by the stride
2419 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2421 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2422 on the entire header of the finally block in the fallthru case.
2424 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2426 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2428 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2430 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2431 rs6000_emit_move_from_cr and call renamed function.
2432 (rs6000_emit_prologue): Call renamed functions.
2433 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2434 movesi_from_cr, remove volatile CRs.
2436 2017-08-30 Jon Beniston <jon@beniston.com>
2437 Richard Biener <rguenther@suse.de>
2439 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2440 of VECTOR_MODE_P check.
2441 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2442 element vector types.
2444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2446 * df.h (df_read_modify_subreg_p): Remove in favor of...
2447 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2448 const_rtx instead of an rtx.
2449 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2450 * df-problems.c (df_word_lr_mark_ref): Likewise.
2451 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2452 (mark_pseudo_reg_dead): Likewise.
2453 (mark_ref_dead): Likewise.
2454 * reginfo.c (init_subregs_of_mode): Likewise.
2455 * sched-deps.c (sched_analyze_1): Likewise.
2456 * df-scan.c (df_def_record_1): Likewise.
2457 (df_uses_record): Likewise.
2458 (df_read_modify_subreg_p): Remove in favor of...
2459 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2460 const_rtx instead of an rtx.
2462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2463 Alan Hayward <alan.hayward@arm.com>
2464 David Sherwood <david.sherwood@arm.com>
2466 * rtl.h (partial_subreg_p): New function.
2467 * caller-save.c (save_call_clobbered_regs): Use it.
2468 * calls.c (expand_call): Likewise.
2469 * combine.c (combinable_i3pat): Likewise.
2470 (simplify_set): Likewise.
2471 (make_extraction): Likewise.
2472 (make_compound_operation_int): Likewise.
2473 (gen_lowpart_or_truncate): Likewise.
2474 (force_to_mode): Likewise.
2475 (make_field_assignment): Likewise.
2476 (reg_truncated_to_mode): Likewise.
2477 (record_truncated_value): Likewise.
2478 (move_deaths): Likewise.
2479 * cse.c (record_jump_cond): Likewise.
2480 (cse_insn): Likewise.
2481 * cselib.c (cselib_lookup_1): Likewise.
2482 * expmed.c (extract_bit_field_using_extv): Likewise.
2483 * function.c (assign_parm_setup_reg): Likewise.
2484 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2485 * ira-build.c (create_insn_allocnos): Likewise.
2486 * lra-coalesce.c (merge_pseudos): Likewise.
2487 * lra-constraints.c (match_reload): Likewise.
2488 (simplify_operand_subreg): Likewise.
2489 (curr_insn_transform): Likewise.
2490 * lra-lives.c (process_bb_lives): Likewise.
2491 * lra.c (new_insn_reg): Likewise.
2492 (lra_substitute_pseudo): Likewise.
2493 * regcprop.c (mode_change_ok): Likewise.
2494 (maybe_mode_change): Likewise.
2495 (copyprop_hardreg_forward_1): Likewise.
2496 * reload.c (push_reload): Likewise.
2497 (find_reloads): Likewise.
2498 (find_reloads_subreg_address): Likewise.
2499 * reload1.c (alter_reg): Likewise.
2500 (eliminate_regs_1): Likewise.
2501 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2503 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2505 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2506 back to if statements, including unpack.
2508 2017-08-30 Martin Liska <mliska@suse.cz>
2511 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2513 (func_checker::compare_asm_inputs_outputs): ... this function.
2514 (func_checker::compare_gimple_asm): Use the function to compare
2515 also ASM constrains.
2516 * ipa-icf-gimple.h: Rename the function.
2518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2519 Alan Hayward <alan.hayward@arm.com>
2520 David Sherwood <david.sherwood@arm.com>
2522 * coretypes.h (complex_mode): New type.
2523 * gdbhooks.py (build_pretty_printer): Handle it.
2524 * machmode.h (complex_mode): New class.
2525 (complex_mode::includes_p): New function.
2526 (is_complex_int_mode): Likewise.
2527 (is_complex_float_mode): Likewise.
2528 * genmodes.c (get_mode_class): Handle complex mode classes.
2529 * function.c (expand_function_end): Use is_complex_int_mode.
2531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2532 Alan Hayward <alan.hayward@arm.com>
2533 David Sherwood <david.sherwood@arm.com>
2535 * coretypes.h (scalar_mode_pod): New typedef.
2536 * gdbhooks.py (build_pretty_printer): Handle it.
2537 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2538 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2539 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2540 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2544 Alan Hayward <alan.hayward@arm.com>
2545 David Sherwood <david.sherwood@arm.com>
2547 * machmode.h (mode_for_vector): Take a scalar_mode instead
2549 * stor-layout.c (mode_for_vector): Likewise.
2550 * explow.c (promote_mode): Use as_a <scalar_mode>.
2551 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2554 Alan Hayward <alan.hayward@arm.com>
2555 David Sherwood <david.sherwood@arm.com>
2557 * target.def (preferred_simd_mode): Take a scalar_mode
2558 instead of a machine_mode.
2559 * targhooks.h (default_preferred_simd_mode): Likewise.
2560 * targhooks.c (default_preferred_simd_mode): Likewise.
2561 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2562 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2563 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2564 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2565 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2566 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2567 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2568 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2570 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2571 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2572 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2573 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2574 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2575 * doc/tm.texi: Regenerate.
2576 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2580 Alan Hayward <alan.hayward@arm.com>
2581 David Sherwood <david.sherwood@arm.com>
2583 * target.def (scalar_mode_supported_p): Take a scalar_mode
2584 instead of a machine_mode.
2585 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2586 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2587 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2588 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2589 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2590 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2591 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2592 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2593 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2594 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2595 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2596 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2597 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2598 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2600 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2601 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2602 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2603 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2604 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2606 * doc/tm.texi: Regenerate.
2608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2609 Alan Hayward <alan.hayward@arm.com>
2610 David Sherwood <david.sherwood@arm.com>
2612 * coretypes.h (opt_scalar_mode): New typedef.
2613 * gdbhooks.py (build_pretty_printers): Handle it.
2614 * machmode.h (mode_iterator::get_2xwider): Add overload for
2616 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2618 * expr.c (convert_mode_scalar): Likewise.
2619 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2620 * optabs.c (expand_float): Likewise.
2621 (expand_fix): Likewise.
2622 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
2629 using a branching expansion.
2630 (expand_fix): Likewise.
2632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2633 Alan Hayward <alan.hayward@arm.com>
2634 David Sherwood <david.sherwood@arm.com>
2636 * expr.c (convert_mode): Split scalar handling out into...
2637 (convert_mode_scalar): ...this new function. Treat the modes
2640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2641 Alan Hayward <alan.hayward@arm.com>
2642 David Sherwood <david.sherwood@arm.com>
2644 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2646 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2649 Alan Hayward <alan.hayward@arm.com>
2650 David Sherwood <david.sherwood@arm.com>
2652 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2653 rather than a machine_mode.
2654 (fixed_from_string): Likewise.
2655 (fixed_convert): Likewise.
2656 (fixed_convert_from_int): Likewise.
2657 (fixed_convert_from_real): Likewise.
2658 (real_convert_from_fixed): Likewise.
2659 * fixed-value.c (fixed_from_double_int): Likewise.
2660 (fixed_from_string): Likewise.
2661 (fixed_convert): Likewise.
2662 (fixed_convert_from_int): Likewise.
2663 (fixed_convert_from_real): Likewise.
2664 (real_convert_from_fixed): Likewise.
2665 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2668 Alan Hayward <alan.hayward@arm.com>
2669 David Sherwood <david.sherwood@arm.com>
2671 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2672 of separate mode class checks. Do not allow vector modes here.
2673 (immed_wide_int_const): Use as_a <scalar_mode>.
2674 * explow.c (trunc_int_for_mode): Likewise.
2675 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2676 (wi::shwi): Likewise.
2677 (wi::min_value): Likewise.
2678 (wi::max_value): Likewise.
2679 * dwarf2out.c (loc_descriptor): Likewise.
2680 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2684 Alan Hayward <alan.hayward@arm.com>
2685 David Sherwood <david.sherwood@arm.com>
2687 * tree.h (SCALAR_TYPE_MODE): New macro.
2688 * expr.c (expand_expr_addr_expr_1): Use it.
2689 (expand_expr_real_2): Likewise.
2690 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2691 (fold_convert_const_fixed_from_int): Likewise.
2692 (fold_convert_const_fixed_from_real): Likewise.
2693 (native_encode_fixed): Likewise
2694 (native_encode_complex): Likewise
2695 (native_encode_vector): Likewise.
2696 (native_interpret_fixed): Likewise.
2697 (native_interpret_real): Likewise.
2698 (native_interpret_complex): Likewise.
2699 (native_interpret_vector): Likewise.
2700 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2701 (simd_clone_adjust_argument_types): Likewise.
2702 (simd_clone_init_simd_arrays): Likewise.
2703 (simd_clone_adjust): Likewise.
2704 * stor-layout.c (layout_type): Likewise.
2705 * tree.c (build_minus_one_cst): Likewise.
2706 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2707 * tree-inline.c (estimate_move_cost): Likewise.
2708 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2709 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2710 (vectorizable_reduction): Likewise.
2711 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2712 (vect_recog_mixed_size_cond_pattern): Likewise.
2713 (check_bool_pattern): Likewise.
2714 (adjust_bool_pattern): Likewise.
2715 (search_type_for_mask_1): Likewise.
2716 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2717 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2718 (vectorizable_load): Likewise.
2719 (vectorizable_store): Likewise.
2720 * ubsan.c (ubsan_encode_value): Likewise.
2721 * varasm.c (output_constant): Likewise.
2723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2724 Alan Hayward <alan.hayward@arm.com>
2725 David Sherwood <david.sherwood@arm.com>
2727 * coretypes.h (scalar_mode): New class.
2728 * machmode.h (scalar_mode): Likewise.
2729 (scalar_mode::includes_p): New function.
2730 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2731 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2732 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2733 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2734 * expmed.c (store_bit_field_1): Likewise.
2735 (extract_bit_field_1): Likewise.
2736 * expr.c (write_complex_part): Likewise.
2737 (read_complex_part): Likewise.
2738 (emit_move_complex_push): Likewise.
2739 (expand_expr_real_2): Likewise.
2740 * function.c (assign_parm_setup_reg): Likewise.
2741 (assign_parms_unsplit_complex): Likewise.
2742 * optabs.c (expand_binop): Likewise.
2743 * rtlanal.c (subreg_get_info): Likewise.
2744 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2745 * varasm.c (output_constant_pool_2): Likewise.
2747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2753 (expmed_mult_highpart_optab): Likewise.
2754 (expmed_mult_highpart): Likewise.
2756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2757 Alan Hayward <alan.hayward@arm.com>
2758 David Sherwood <david.sherwood@arm.com>
2760 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2761 instead of a machine_mode.
2762 (builtin_memset_read_str): Likewise.
2763 * builtins.c (c_readstr): Likewise.
2764 (builtin_memcpy_read_str): Likewise.
2765 (builtin_strncpy_read_str): Likewise.
2766 (builtin_memset_read_str): Likewise.
2767 (builtin_memset_gen_str): Likewise.
2768 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2769 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2770 instead of a machine_mode.
2771 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2773 (make_extraction): Likewise.
2774 (try_widen_shift_mode): Take and return scalar_int_modes instead
2776 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2777 a scalar_int_mode instead of a machine_mode.
2778 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2779 (avr_addr_space_pointer_mode): Likewise.
2780 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2781 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2782 (msp430_unwind_word_mode): Likewise.
2783 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2784 (spu_addr_space_pointer_mode): Likewise.
2785 (spu_addr_space_address_mode): Likewise.
2786 (spu_libgcc_cmp_return_mode): Likewise.
2787 (spu_libgcc_shift_count_mode): Likewise.
2788 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2789 (rl78_addr_space_pointer_mode): Likewise.
2790 (fl78_unwind_word_mode): Likewise.
2791 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2793 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2794 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2795 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2796 (mips_valid_pointer_mode): Likewise.
2797 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2798 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2799 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2801 (ft32_addr_space_address_mode): Likewise.
2802 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2803 scalar_int_mode instead of a machine_mode.
2804 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2806 (m32c_addr_space_address_mode): Likewise.
2807 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2808 (rs6000_eh_return_filter_mode): Likewise.
2809 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2810 (rs6000_eh_return_filter_mode): Likewise.
2811 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2812 (s390_libgcc_shift_count_mode): Likewise.
2813 (s390_unwind_word_mode): Likewise.
2814 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2816 * target.def (mode_rep_extended): Likewise.
2817 (valid_pointer_mode): Likewise.
2818 (addr_space.valid_pointer_mode): Likewise.
2819 (eh_return_filter_mode): Return a scalar_int_mode rather than
2821 (libgcc_cmp_return_mode): Likewise.
2822 (libgcc_shift_count_mode): Likewise.
2823 (unwind_word_mode): Likewise.
2824 (addr_space.pointer_mode): Likewise.
2825 (addr_space.address_mode): Likewise.
2826 * doc/tm.texi: Regenerate.
2827 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2829 (do_jump): Use scalar_int_mode for local variables.
2830 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2831 rather than a machine_mode.
2832 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2833 (scompare_loc_descriptor_wide): Likewise.
2834 (scompare_loc_descriptor_narrow): Likewise.
2835 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2837 * except.c (sjlj_emit_dispatch_table): Likewise.
2838 (expand_builtin_eh_copy_values): Likewise.
2839 * explow.c (convert_memory_address_addr_space_1): Likewise.
2840 Take a scalar_int_mode rather than a machine_mode.
2841 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2842 than a machine_mode.
2843 (memory_address_addr_space): Use scalar_int_mode for local variables.
2844 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2845 rather than a machine_mode.
2846 * expmed.c (mask_rtx): Likewise.
2847 (init_expmed_one_conv): Likewise.
2848 (expand_mult_highpart_adjust): Likewise.
2849 (extract_high_half): Likewise.
2850 (expmed_mult_highpart_optab): Likewise.
2851 (expmed_mult_highpart): Likewise.
2852 (expand_smod_pow2): Likewise.
2853 (expand_sdiv_pow2): Likewise.
2854 (emit_store_flag_int): Likewise.
2855 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2857 (extract_low_bits): Likewise.
2858 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2860 * expr.c (pieces_addr::adjust): Likewise.
2861 (can_store_by_pieces): Likewise.
2862 (store_by_pieces): Likewise.
2863 (clear_by_pieces_1): Likewise.
2864 (expand_expr_addr_expr_1): Likewise.
2865 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2866 (expand_expr_real_1): Likewise.
2867 (try_casesi): Likewise.
2868 * final.c (shorten_branches): Likewise.
2869 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2870 type of "mode" to machine_mode.
2871 * internal-fn.c (expand_arith_overflow_result_store): Take a
2872 scalar_int_mode rather than a machine_mode.
2873 (expand_mul_overflow): Use scalar_int_mode for local variables.
2874 * loop-doloop.c (doloop_modify): Likewise.
2875 (doloop_optimize): Likewise.
2876 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2877 than a machine_mode.
2878 (expand_doubleword_shift_condmove): Likewise.
2879 (expand_doubleword_shift): Likewise.
2880 (expand_doubleword_clz): Likewise.
2881 (expand_doubleword_popcount): Likewise.
2882 (expand_doubleword_parity): Likewise.
2883 (expand_absneg_bit): Use scalar_int_mode for local variables.
2884 (prepare_float_lib_cmp): Likewise.
2885 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2886 rather than a machine_mode.
2887 (convert_memory_address_addr_space): Likewise.
2888 (get_mode_bounds): Likewise.
2889 (get_address_mode): Return a scalar_int_mode rather than a
2891 * rtlanal.c (get_address_mode): Likewise.
2892 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2893 than a machine_mode.
2894 * targhooks.c (default_mode_rep_extended): Likewise.
2895 (default_valid_pointer_mode): Likewise.
2896 (default_addr_space_valid_pointer_mode): Likewise.
2897 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2898 than a machine_mode.
2899 (default_libgcc_cmp_return_mode): Likewise.
2900 (default_libgcc_shift_count_mode): Likewise.
2901 (default_unwind_word_mode): Likewise.
2902 (default_addr_space_pointer_mode): Likewise.
2903 (default_addr_space_address_mode): Likewise.
2904 * targhooks.h (default_eh_return_filter_mode): Likewise.
2905 (default_libgcc_cmp_return_mode): Likewise.
2906 (default_libgcc_shift_count_mode): Likewise.
2907 (default_unwind_word_mode): Likewise.
2908 (default_addr_space_pointer_mode): Likewise.
2909 (default_addr_space_address_mode): Likewise.
2910 (default_mode_rep_extended): Take a scalar_int_mode rather than
2912 (default_valid_pointer_mode): Likewise.
2913 (default_addr_space_valid_pointer_mode): Likewise.
2914 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2916 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2917 rather than a machine_mode.
2918 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2919 for local variables.
2920 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2921 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2922 than a machine_mode.
2924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2925 Alan Hayward <alan.hayward@arm.com>
2926 David Sherwood <david.sherwood@arm.com>
2928 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2929 the mode argument to scalar_int_mode.
2930 (do_jump_by_parts_zero_rtx): Likewise.
2931 (do_jump_by_parts_equality_rtx): Likewise.
2932 (do_jump_by_parts_greater): Take a mode argument.
2933 (do_jump_by_parts_equality): Likewise.
2934 (do_jump_1): Update calls accordingly.
2936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2937 Alan Hayward <alan.hayward@arm.com>
2938 David Sherwood <david.sherwood@arm.com>
2940 * is-a.h (safe_dyn_cast): New function.
2941 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2942 (jump_table_for_label): Likewise.
2943 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2944 instead of an rtx_insn *.
2945 (shorten_branches): Use dyn_cast instead of LABEL_P and
2946 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2947 rtx_jump_table_data::get_data_mode.
2948 (final_scan_insn): Likewise.
2950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2951 Alan Hayward <alan.hayward@arm.com>
2952 David Sherwood <david.sherwood@arm.com>
2954 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2955 trying to combine a full-register integer set with a subreg
2958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2959 Alan Hayward <alan.hayward@arm.com>
2960 David Sherwood <david.sherwood@arm.com>
2962 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2963 that is always either address_mode or pointer_mode.
2965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2966 Alan Hayward <alan.hayward@arm.com>
2967 David Sherwood <david.sherwood@arm.com>
2969 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2970 when the two are known to be equal.
2972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2973 Alan Hayward <alan.hayward@arm.com>
2974 David Sherwood <david.sherwood@arm.com>
2976 * simplify-rtx.c (simplify_const_unary_operation): Use
2977 is_a <scalar_int_mode> instead of checking for a nonzero
2978 precision. Forcibly convert op_mode to a scalar_int_mode
2979 in that case. More clearly differentiate the operand and
2980 result modes and use the former when deciding what the value
2981 of a count-bits operation should be. Use is_int_mode instead
2982 of checking for a MODE_INT. Remove redundant check for whether
2983 this mode has a zero precision.
2985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2986 Alan Hayward <alan.hayward@arm.com>
2987 David Sherwood <david.sherwood@arm.com>
2989 * optabs.c (widen_leading): Change the type of the mode argument
2990 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2991 (widen_bswap): Likewise.
2992 (expand_parity): Likewise.
2993 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2994 (expand_ffs): Likewise.
2995 (epand_unop): Check for scalar integer modes before calling the
2998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2999 Alan Hayward <alan.hayward@arm.com>
3000 David Sherwood <david.sherwood@arm.com>
3002 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3004 (expand_expr_real_1): Update call accordingly.
3006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3007 Alan Hayward <alan.hayward@arm.com>
3008 David Sherwood <david.sherwood@arm.com>
3010 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3011 value_mode arguments. Use scalar_int_mode internally.
3012 (store_bit_field_1): Rename the new integer mode from imode
3013 to op0_mode and use it instead of GET_MODE (op0). Update calls
3014 to store_split_bit_field, store_bit_field_using_insv and
3015 store_fixed_bit_field.
3016 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3017 Use scalar_int_mode internally. Use a bit count rather than a mode
3018 when calculating the largest bit size for get_best_mode.
3019 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3020 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3021 Remove assertion that OP0 has a scalar integer mode.
3022 (store_split_bit_field): Add op0_mode and value_mode arguments.
3023 Update calls to extract_fixed_bit_field.
3024 (extract_bit_field_using_extv): Add an op0_mode argument.
3025 Use scalar_int_mode internally.
3026 (extract_bit_field_1): Rename the new integer mode from imode to
3027 op0_mode and use it instead of GET_MODE (op0). Update calls to
3028 extract_split_bit_field, extract_bit_field_using_extv and
3029 extract_fixed_bit_field.
3030 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3031 to extract_split_bit_field and extract_fixed_bit_field_1.
3032 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3033 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3035 (extract_split_bit_field): Add an op0_mode argument. Update call
3036 to extract_fixed_bit_field.
3038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3039 Alan Hayward <alan.hayward@arm.com>
3040 David Sherwood <david.sherwood@arm.com>
3042 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3043 * explow.c (hard_function_value): Likewise.
3044 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3045 convert_to_mode call outside the loop.
3046 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3047 for the mode iterator. Require the mode specified by max_pieces
3049 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3051 (copy_blkmode_to_reg): Likewise.
3052 (set_storage_via_setmem): Likewise.
3053 * optabs.c (prepare_cmp_insn): Likewise.
3054 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3055 * stor-layout.c (finish_bitfield_representative): Likewise.
3057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3058 Alan Hayward <alan.hayward@arm.com>
3059 David Sherwood <david.sherwood@arm.com>
3061 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3062 * expr.c (convert_move): Use them.
3063 (convert_modes): Likewise.
3064 (store_expr_with_bounds): Likewise.
3066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3067 Alan Hayward <alan.hayward@arm.com>
3068 David Sherwood <david.sherwood@arm.com>
3070 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3071 parameter for the mode of "x". Remove the "known_x", "known_mode"
3072 and "known_ret" arguments. Change the type of the mode argument
3074 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3075 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3076 (reg_num_sign_bit_copies_for_combine): Likewise.
3077 * rtlanal.c (nonzero_bits1): Likewise.
3078 (num_sign_bit_copies1): Likewise.
3079 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3080 (reg_num_sign_bit_copies_general): Likewise.
3081 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3082 (reg_nonzero_bits_general): Likewise.
3084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3085 Alan Hayward <alan.hayward@arm.com>
3086 David Sherwood <david.sherwood@arm.com>
3088 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3089 than in subroutines. Return 1 for non-integer modes.
3090 (cached_num_sign_bit_copies): Change the type of the mode parameter
3092 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3093 classes. Handle CONST_INT_P first and then check whether X also
3094 has a scalar integer mode. Check the same thing for inner registers
3095 of a SUBREG and for values that are being extended or truncated.
3097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3098 Alan Hayward <alan.hayward@arm.com>
3099 David Sherwood <david.sherwood@arm.com>
3101 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3102 in subroutines. Return the mode mask for non-integer modes.
3103 (cached_nonzero_bits): Change the type of the mode parameter
3105 (nonzero_bits1): Likewise. Remove early exit for other mode
3106 classes. Handle CONST_INT_P first and then check whether X
3107 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3114 Return a scalar_int_mode and assert that the size is greater than
3116 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3117 (op_by_pieces_d::op_by_pieces_d): Likewise.
3118 (op_by_pieces_d::run): Likewise.
3119 (can_store_by_pieces): Likewise.
3121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3122 Alan Hayward <alan.hayward@arm.com>
3123 David Sherwood <david.sherwood@arm.com>
3125 * combine.c (extract_left_shift): Add a mode argument and update
3127 (make_compound_operation_int): Change the type of the mode parameter
3128 to scalar_int_mode and update the call to extract_left_shift.
3130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3131 Alan Hayward <alan.hayward@arm.com>
3132 David Sherwood <david.sherwood@arm.com>
3134 * combine.c (simplify_and_const_int): Change the type of the mode
3135 parameter to scalar_int_mode.
3136 (simplify_and_const_int_1): Likewise. Update recursive call.
3138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3139 Alan Hayward <alan.hayward@arm.com>
3140 David Sherwood <david.sherwood@arm.com>
3142 * combine.c (simplify_compare_const): Check that the mode is a
3143 scalar_int_mode (rather than VOIDmode) before testing its
3145 (simplify_comparison): Move COMPARISON_P handling out of the
3146 loop and restrict the latter part of the loop to scalar_int_modes.
3147 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3148 and when considering SUBREG_REGs. Use is_int_mode instead of
3149 checking GET_MODE_CLASS against MODE_INT.
3151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3152 Alan Hayward <alan.hayward@arm.com>
3153 David Sherwood <david.sherwood@arm.com>
3155 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3156 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3157 shift_unit_mode and for modes involved in scalar shifts.
3159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3160 Alan Hayward <alan.hayward@arm.com>
3161 David Sherwood <david.sherwood@arm.com>
3163 * combine.c (force_int_to_mode): New function, split out from...
3164 (force_to_mode): ...here. Keep xmode up-to-date and use it
3165 instead of GET_MODE (x).
3167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3168 Alan Hayward <alan.hayward@arm.com>
3169 David Sherwood <david.sherwood@arm.com>
3171 * optabs-query.h (extraction_insn::struct_mode): Change type to
3172 opt_scalar_int_mode and update comment.
3173 (extraction_insn::field_mode): Change type to scalar_int_mode.
3174 (extraction_insn::pos_mode): Likewise.
3175 * combine.c (make_extraction): Update accordingly.
3176 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3177 (get_optab_extraction_insn): Likewise.
3178 * recog.c (simplify_while_replacing): Likewise.
3179 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3180 parameter to opt_scalar_int_mode.
3182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3183 Alan Hayward <alan.hayward@arm.com>
3184 David Sherwood <david.sherwood@arm.com>
3186 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3187 to a scalar_int_mode instead of a machine_mode.
3188 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3189 (get_best_mode): Return a boolean and use a pointer argument to store
3190 the selected mode. Replace the limit mode parameter with a bit limit.
3191 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3192 for the values returned by bit_field_mode_iterator::next_mode.
3193 (store_bit_field): Update call to get_best_mode.
3194 (store_fixed_bit_field): Likewise.
3195 (extract_fixed_bit_field): Likewise.
3196 * expr.c (optimize_bitfield_assignment_op): Likewise.
3197 * fold-const.c (optimize_bit_field_compare): Likewise.
3198 (fold_truth_andor_1): Likewise.
3199 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3200 Update for new type of m_mode.
3201 (get_best_mode): As above.
3203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3204 Alan Hayward <alan.hayward@arm.com>
3205 David Sherwood <david.sherwood@arm.com>
3207 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3208 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3209 (store_bit_field): Check is_a <scalar_int_mode> before calling
3210 strict_volatile_bitfield_p.
3211 (extract_bit_field): Likewise.
3213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3214 Alan Hayward <alan.hayward@arm.com>
3215 David Sherwood <david.sherwood@arm.com>
3217 * target.def (cstore_mode): Return a scalar_int_mode.
3218 * doc/tm.texi: Regenerate.
3219 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3220 * targhooks.h (default_cstore_mode): Likewise.
3221 * targhooks.c (default_cstore_mode): Likewise, using a forced
3223 * expmed.c (emit_cstore): Expect the target of the cstore to be
3226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3227 Alan Hayward <alan.hayward@arm.com>
3228 David Sherwood <david.sherwood@arm.com>
3230 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3232 (niter_desc): Likewise mode.
3233 (iv_analyze): Add a mode parameter.
3235 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3236 and change its type to scalar_int_mode.
3237 * loop-iv.c: Update commentary at head of file.
3238 (iv_constant): Pass the mode paraeter before the rtx it describes
3239 and change its type to scalar_int_mode. Remove VOIDmode handling.
3240 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3241 (iv_extend): Likewise.
3242 (shorten_into_mode): Likewise.
3243 (iv_add): Use scalar_int_mode.
3244 (iv_mult): Likewise.
3245 (iv_shift): Likewise.
3246 (canonicalize_iv_subregs): Likewise.
3247 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3248 it describes and change its mode to scalar_int_mode. Also change
3249 the type of the returned inner_mode to scalar_int_mode.
3250 (get_biv_step): Likewise, turning outer_mode from a pointer
3251 into a direct parameter. Update call to get_biv_step_1.
3252 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3253 iv_constant and get_biv_step.
3254 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3255 and change its type to scalar_int_mode. Don't initialise iv->mode
3256 to VOIDmode and remove later checks for its still being VOIDmode.
3257 Update calls to iv_analyze_op and iv_analyze_expr. Check
3258 is_a <scalar_int_mode> when changing the mode under consideration.
3259 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3260 Update call to iv_analyze_expr.
3261 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3262 inner register is not also a scalar_int_mode. Update call to
3264 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3265 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3266 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3267 separate mode class checks. Update calls to iv_analyze. Remove
3268 fix-up of VOIDmodes after iv_analyze_biv.
3269 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3270 don't have a scalar_int_mode. Update call to biv_p.
3272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3273 Alan Hayward <alan.hayward@arm.com>
3274 David Sherwood <david.sherwood@arm.com>
3276 * cfgexpand.c (convert_debug_memory_address): Use
3277 as_a <scalar_int_mode>.
3278 * combine.c (expand_compound_operation): Likewise.
3279 (make_extraction): Likewise.
3280 (change_zero_ext): Likewise.
3281 (simplify_comparison): Likewise.
3282 * cse.c (cse_insn): Likewise.
3283 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3284 (mem_loc_descriptor): Likewise.
3285 (loc_descriptor): Likewise.
3286 * expmed.c (init_expmed_one_mode): Likewise.
3287 (synth_mult): Likewise.
3288 (emit_store_flag_1): Likewise.
3289 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3290 of a comparison with size.
3291 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3292 (reduce_to_bit_field_precision): Likewise.
3293 * function.c (expand_function_end): Likewise.
3294 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3295 * loop-doloop.c (doloop_modify): Likewise.
3296 * optabs.c (expand_binop): Likewise.
3297 (expand_unop): Likewise.
3298 (expand_copysign_absneg): Likewise.
3299 (prepare_cmp_insn): Likewise.
3300 (maybe_legitimize_operand): Likewise.
3301 * recog.c (const_scalar_int_operand): Likewise.
3302 * rtlanal.c (get_address_mode): Likewise.
3303 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3304 (simplify_cond_clz_ctz): Likewise.
3305 * tree-nested.c (get_nl_goto_field): Likewise.
3306 * tree.c (build_vector_type_for_mode): Likewise.
3307 * var-tracking.c (use_narrower_mode): Likewise.
3309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3310 Alan Hayward <alan.hayward@arm.com>
3311 David Sherwood <david.sherwood@arm.com>
3313 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3314 * builtins.c (expand_builtin_signbit): Use it.
3315 * cfgexpand.c (expand_debug_expr): Likewise.
3316 * dojump.c (do_jump): Likewise.
3317 (do_compare_and_jump): Likewise.
3318 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3319 * expmed.c (make_tree): Likewise.
3320 * expr.c (expand_expr_real_2): Likewise.
3321 (expand_expr_real_1): Likewise.
3322 (try_casesi): Likewise.
3323 * fold-const-call.c (fold_const_call_ss): Likewise.
3324 * fold-const.c (unextend): Likewise.
3325 (extract_muldiv_1): Likewise.
3326 (fold_single_bit_test): Likewise.
3327 (native_encode_int): Likewise.
3328 (native_encode_string): Likewise.
3329 (native_interpret_int): Likewise.
3330 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3331 * internal-fn.c (expand_addsub_overflow): Likewise.
3332 (expand_neg_overflow): Likewise.
3333 (expand_mul_overflow): Likewise.
3334 (expand_arith_overflow): Likewise.
3335 * match.pd: Likewise.
3336 * stor-layout.c (layout_type): Likewise.
3337 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3338 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3339 * tree-ssanames.c (get_range_info): Likewise.
3340 * tree-switch-conversion.c (array_value_type) Likewise.
3341 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3342 (vect_recog_divmod_pattern): Likewise.
3343 (vect_recog_mixed_size_cond_pattern): Likewise.
3344 * tree-vrp.c (extract_range_basic): Likewise.
3345 (simplify_float_conversion_using_ranges): Likewise.
3346 * tree.c (int_fits_type_p): Likewise.
3347 * ubsan.c (instrument_bool_enum_load): Likewise.
3348 * varasm.c (mergeable_string_section): Likewise.
3349 (narrowing_initializer_constant_valid_p): Likewise.
3350 (output_constant): Likewise.
3352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3353 Alan Hayward <alan.hayward@arm.com>
3354 David Sherwood <david.sherwood@arm.com>
3356 * machmode.h (NARROWEST_INT_MODE): New macro.
3357 * expr.c (alignment_for_piecewise_move): Use it instead of
3358 GET_CLASS_NARROWEST_MODE (MODE_INT).
3359 (push_block): Likewise.
3360 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3362 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3365 Alan Hayward <alan.hayward@arm.com>
3366 David Sherwood <david.sherwood@arm.com>
3368 * postreload.c (move2add_valid_value_p): Change the type of the
3369 mode parameter to scalar_int_mode.
3370 (move2add_use_add2_insn): Add a mode parameter and use it instead
3372 (move2add_use_add3_insn): Likewise.
3373 (reload_cse_move2add): Update accordingly.
3375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3376 Alan Hayward <alan.hayward@arm.com>
3377 David Sherwood <david.sherwood@arm.com>
3379 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3381 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3382 * optabs.c (expand_unop): Likewise.
3384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3385 Alan Hayward <alan.hayward@arm.com>
3386 David Sherwood <david.sherwood@arm.com>
3388 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3389 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3390 (popcount_loc_descriptor): Likewise.
3391 (bswap_loc_descriptor): Likewise.
3392 (rotate_loc_descriptor): Likewise.
3393 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3394 calling the functions above.
3396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3397 Alan Hayward <alan.hayward@arm.com>
3398 David Sherwood <david.sherwood@arm.com>
3400 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3402 (try_combine): Likewise.
3403 (simplify_if_then_else): Likewise.
3404 * cse.c (cse_insn): Likewise.
3405 * dwarf2out.c (mem_loc_descriptor): Likewise.
3406 * emit-rtl.c (gen_lowpart_common): Likewise.
3407 * simplify-rtx.c (simplify_truncation): Likewise.
3408 (simplify_binary_operation_1): Likewise.
3409 (simplify_const_relational_operation): Likewise.
3410 (simplify_ternary_operation): Likewise.
3411 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3414 Alan Hayward <alan.hayward@arm.com>
3415 David Sherwood <david.sherwood@arm.com>
3417 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3418 * reload.c (push_reload): Likewise.
3419 (find_reloads): Likewise.
3421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3422 Alan Hayward <alan.hayward@arm.com>
3423 David Sherwood <david.sherwood@arm.com>
3425 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3426 (make_compound_operation_int): Likewise.
3427 (change_zero_ext): Likewise.
3428 * expr.c (convert_move): Likewise.
3429 (convert_modes): Likewise.
3430 * fwprop.c (forward_propagate_subreg): Likewise.
3431 * loop-iv.c (get_biv_step_1): Likewise.
3432 * optabs.c (widen_operand): Likewise.
3433 * postreload.c (move2add_valid_value_p): Likewise.
3434 * recog.c (simplify_while_replacing): Likewise.
3435 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3436 (simplify_binary_operation_1): Likewise. Remove redundant
3437 mode equality check.
3439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3440 Alan Hayward <alan.hayward@arm.com>
3441 David Sherwood <david.sherwood@arm.com>
3443 * combine.c (combine_simplify_rtx): Add checks for
3444 is_a <scalar_int_mode>.
3445 (simplify_if_then_else): Likewise.
3446 (make_field_assignment): Likewise.
3447 (simplify_comparison): Likewise.
3448 * ifcvt.c (noce_try_bitop): Likewise.
3449 * loop-invariant.c (canonicalize_address_mult): Likewise.
3450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3453 Alan Hayward <alan.hayward@arm.com>
3454 David Sherwood <david.sherwood@arm.com>
3456 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3457 is_a <scalar_int_mode> instead of != BLKmode.
3459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3460 Alan Hayward <alan.hayward@arm.com>
3461 David Sherwood <david.sherwood@arm.com>
3463 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3464 instead of != VOIDmode.
3465 * combine.c (if_then_else_cond): Likewise.
3466 (change_zero_ext): Likewise.
3467 * dwarf2out.c (mem_loc_descriptor): Likewise.
3468 (loc_descriptor): Likewise.
3469 * rtlanal.c (canonicalize_condition): Likewise.
3470 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3473 Alan Hayward <alan.hayward@arm.com>
3474 David Sherwood <david.sherwood@arm.com>
3476 * simplify-rtx.c (simplify_binary_operation_1): Use
3477 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3480 Alan Hayward <alan.hayward@arm.com>
3481 David Sherwood <david.sherwood@arm.com>
3483 * wide-int.h (int_traits<unsigned char>) New class.
3484 (int_traits<unsigned short>) Likewise.
3485 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3486 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3488 * combine.c (set_nonzero_bits_and_sign_copies): Use
3489 is_a <scalar_int_mode>.
3490 (find_split_point): Likewise.
3491 (combine_simplify_rtx): Likewise.
3492 (simplify_logical): Likewise.
3493 (expand_compound_operation): Likewise.
3494 (expand_field_assignment): Likewise.
3495 (make_compound_operation): Likewise.
3496 (extended_count): Likewise.
3497 (change_zero_ext): Likewise.
3498 (simplify_comparison): Likewise.
3499 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3500 (ucompare_loc_descriptor): Likewise.
3501 (minmax_loc_descriptor): Likewise.
3502 (mem_loc_descriptor): Likewise.
3503 (loc_descriptor): Likewise.
3504 * expmed.c (init_expmed_one_mode): Likewise.
3505 * lra-constraints.c (lra_constraint_offset): Likewise.
3506 * optabs.c (prepare_libcall_arg): Likewise.
3507 * postreload.c (move2add_note_store): Likewise.
3508 * reload.c (operands_match_p): Likewise.
3509 * rtl.h (load_extend_op): Likewise.
3510 * rtlhooks.c (gen_lowpart_general): Likewise.
3511 * simplify-rtx.c (simplify_truncation): Likewise.
3512 (simplify_unary_operation_1): Likewise.
3513 (simplify_binary_operation_1): Likewise.
3514 (simplify_const_binary_operation): Likewise.
3515 (simplify_const_relational_operation): Likewise.
3516 (simplify_subreg): Likewise.
3517 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3518 * var-tracking.c (adjust_mems): Likewise.
3519 (prepare_call_arguments): Likewise.
3521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3522 Alan Hayward <alan.hayward@arm.com>
3523 David Sherwood <david.sherwood@arm.com>
3525 * machmode.h (is_int_mode): New fuction.
3526 * combine.c (find_split_point): Use it.
3527 (combine_simplify_rtx): Likewise.
3528 (simplify_if_then_else): Likewise.
3529 (simplify_set): Likewise.
3530 (simplify_shift_const_1): Likewise.
3531 (simplify_comparison): Likewise.
3532 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3533 * cse.c (notreg_cost): Likewise.
3534 (cse_insn): Likewise.
3535 * cselib.c (cselib_lookup_1): Likewise.
3536 * dojump.c (do_jump_1): Likewise.
3537 (do_compare_rtx_and_jump): Likewise.
3538 * dse.c (get_call_args): Likewise.
3539 * dwarf2out.c (rtl_for_decl_init): Likewise.
3540 (native_encode_initializer): Likewise.
3541 * expmed.c (emit_store_flag_1): Likewise.
3542 (emit_store_flag): Likewise.
3543 * expr.c (convert_modes): Likewise.
3544 (store_field): Likewise.
3545 (expand_expr_real_1): Likewise.
3546 * fold-const.c (fold_read_from_constant_string): Likewise.
3547 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3548 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3549 * optabs.c (expand_binop): Likewise.
3550 (expand_unop): Likewise.
3551 (expand_abs_nojump): Likewise.
3552 (expand_one_cmpl_abs_nojump): Likewise.
3553 * simplify-rtx.c (mode_signbit_p): Likewise.
3554 (val_signbit_p): Likewise.
3555 (val_signbit_known_set_p): Likewise.
3556 (val_signbit_known_clear_p): Likewise.
3557 (simplify_relational_operation_1): Likewise.
3558 * tree.c (vector_type_mode): Likewise.
3560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3561 Alan Hayward <alan.hayward@arm.com>
3562 David Sherwood <david.sherwood@arm.com>
3564 * machmode.h (smallest_mode_for_size): Fix formatting.
3565 (smallest_int_mode_for_size): New function.
3566 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3567 instead of smallest_mode_for_size.
3568 * combine.c (make_extraction): Likewise.
3569 * config/arc/arc.c (arc_expand_movmem): Likewise.
3570 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3571 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3572 * config/s390/s390.c (s390_expand_insv): Likewise.
3573 * config/sparc/sparc.c (assign_int_registers): Likewise.
3574 * config/spu/spu.c (spu_function_value): Likewise.
3575 (spu_function_arg): Likewise.
3576 * coverage.c (get_gcov_type): Likewise.
3577 (get_gcov_unsigned_t): Likewise.
3578 * dse.c (find_shift_sequence): Likewise.
3579 * expmed.c (store_bit_field_1): Likewise.
3580 * expr.c (convert_move): Likewise.
3581 (store_field): Likewise.
3582 * internal-fn.c (expand_arith_overflow): Likewise.
3583 * optabs-query.c (get_best_extraction_insn): Likewise.
3584 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3585 * stor-layout.c (layout_type): Likewise.
3586 (initialize_sizetypes): Likewise.
3587 * targhooks.c (default_get_mask_mode): Likewise.
3588 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3591 Alan Hayward <alan.hayward@arm.com>
3592 David Sherwood <david.sherwood@arm.com>
3594 * machmode.h (opt_mode::else_blk): New function.
3595 (int_mode_for_mode): Declare.
3596 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3597 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3599 * cfgexpand.c (expand_debug_expr): Likewise.
3600 * combine.c (gen_lowpart_or_truncate): Likewise.
3601 (gen_lowpart_for_combine): Likewise.
3602 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3603 * config/avr/avr.c (avr_to_int_mode): Likewise.
3604 (avr_out_plus_1): Likewise.
3605 (avr_out_plus): Likewise.
3606 (avr_out_round): Likewise.
3607 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3608 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3609 (s390_expand_vcond): Likewise.
3610 * config/spu/spu.c (spu_split_immediate): Likewise.
3611 (spu_expand_mov): Likewise.
3612 * dse.c (get_stored_val): Likewise.
3613 * expmed.c (store_bit_field_1): Likewise.
3614 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3616 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3617 (extract_low_bits): Likewise.
3618 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3619 handling rather than repeating the check.
3620 (emit_group_store): Likewise.
3621 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3622 * optabs.c (expand_absneg_bit): Likewise.
3623 (expand_copysign_absneg): Likewise.
3624 (expand_copysign_bit): Likewise.
3625 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3626 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3627 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3628 * var-tracking.c (prepare_call_arguments): Likewise.
3629 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3630 int_mode_for_mode instead of mode_for_size.
3631 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3634 Alan Hayward <alan.hayward@arm.com>
3635 David Sherwood <david.sherwood@arm.com>
3637 * machmode.h (int_mode_for_size): New function.
3638 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3639 instead of mode_for_size.
3640 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3642 * combine.c (expand_field_assignment): Use int_mode_for_size
3643 instead of mode_for_size.
3644 (make_extraction): Likewise.
3645 (simplify_shift_const_1): Likewise.
3646 (simplify_comparison): Likewise.
3647 * dojump.c (do_jump): Likewise.
3648 * dwarf2out.c (mem_loc_descriptor): Likewise.
3649 * emit-rtl.c (init_derived_machine_modes): Likewise.
3650 * expmed.c (flip_storage_order): Likewise.
3651 (convert_extracted_bit_field): Likewise.
3652 * expr.c (copy_blkmode_from_reg): Likewise.
3653 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3654 * internal-fn.c (expand_mul_overflow): Likewise.
3655 * lower-subreg.c (simple_move): Likewise.
3656 * optabs-libfuncs.c (init_optabs): Likewise.
3657 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3658 * tree.c (vector_type_mode): Likewise.
3659 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3660 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3661 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3662 * tree-vect-stmts.c (vectorizable_load): Likewise.
3663 (vectorizable_store): Likewise.
3665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3666 Alan Hayward <alan.hayward@arm.com>
3667 David Sherwood <david.sherwood@arm.com>
3669 * coretypes.h (pod_mode): New type.
3670 (scalar_int_mode_pod): New typedef.
3671 * machmode.h (pod_mode): New class.
3672 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3673 * genmodes.c (emit_mode_int_n): Update accordingly.
3674 * lower-subreg.h (target_lower_subreg): Change type to
3675 scalar_int_mode_pod.
3676 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3677 scalar_int_mode_pod.
3679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3680 Alan Hayward <alan.hayward@arm.com>
3681 David Sherwood <david.sherwood@arm.com>
3683 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3684 machine_mode to scalar_int_mode.
3685 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3686 (rs6000_option_override_internal): Remove cast to int.
3687 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3688 machine_mode to scalar_int_mode.
3689 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3690 (rs6000_option_override_internal): Remove cast to int.
3691 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3692 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3694 * config/s390/s390.c (s390_expand_builtin): Likewise.
3695 * coretypes.h (scalar_int_mode): New type.
3696 (opt_scalar_int_mode): New typedef.
3697 * machmode.h (scalar_int_mode): New class.
3698 (scalar_int_mode::includes_p): New function.
3699 (byte_mode): Change type to scalar_int_mode.
3700 (word_mode): Likewise.
3701 (ptr_mode): Likewise.
3702 * emit-rtl.c (byte_mode): Likewise.
3703 (word_mode): Likewise.
3704 (ptr_mode): Likewise.
3705 (init_derived_machine_modes): Update accordingly.
3706 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3707 and MODE_PARTIAL_INT.
3708 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3709 opt_scalar_int_mode.
3711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3712 Alan Hayward <alan.hayward@arm.com>
3713 David Sherwood <david.sherwood@arm.com>
3715 * target.def (libgcc_floating_mode_supported_p): Take a
3717 * doc/tm.texi: Regenerate.
3718 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3720 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3721 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3725 Alan Hayward <alan.hayward@arm.com>
3726 David Sherwood <david.sherwood@arm.com>
3728 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3729 * doc/tm.texi: Regenerate.
3730 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3731 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3732 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3733 * targhooks.h (default_floatn_mode): Likewise.
3734 * targhooks.c (default_floatn_mode): Likewise.
3735 * tree.c (build_common_tree_nodes): Update accordingly.
3737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3738 Alan Hayward <alan.hayward@arm.com>
3739 David Sherwood <david.sherwood@arm.com>
3741 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3742 (mode_iterator::iterate_p): Likewise.
3743 (mode_iterator::get_wider): Likewise.
3744 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3747 Alan Hayward <alan.hayward@arm.com>
3748 David Sherwood <david.sherwood@arm.com>
3750 * coretypes.h (opt_scalar_float_mode): New typedef.
3751 * machmode.h (float_mode_for_size): New function.
3752 * emit-rtl.c (double_mode): Delete.
3753 (init_emit_once): Use float_mode_for_size.
3754 * stor-layout.c (layout_type): Likewise.
3755 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3758 Alan Hayward <alan.hayward@arm.com>
3759 David Sherwood <david.sherwood@arm.com>
3761 * output.h (assemble_real): Take a scalar_float_mode.
3762 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3763 * config/arm/arm.md (consttable_4): Likewise.
3764 (consttable_8): Likewise.
3765 (consttable_16): Likewise.
3766 * config/mips/mips.md (consttable_float): Likewise.
3767 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3768 * varasm.c (assemble_real): Take a scalar_float_mode.
3769 (output_constant_pool_2): Update accordingly.
3770 (output_constant): Likewise.
3772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3773 Alan Hayward <alan.hayward@arm.com>
3774 David Sherwood <david.sherwood@arm.com>
3776 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3777 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3778 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3779 (native_encode_real): Likewise.
3780 (native_interpret_real): Likewise.
3781 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3782 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3785 Alan Hayward <alan.hayward@arm.com>
3786 David Sherwood <david.sherwood@arm.com>
3788 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3789 <scalar_float_mode>. Simplify.
3790 (gen_extend_conv_libfunc): Likewise.
3792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3793 Alan Hayward <alan.hayward@arm.com>
3794 David Sherwood <david.sherwood@arm.com>
3796 * coretypes.h (scalar_float_mode): New type.
3797 * machmode.h (mode_traits::from_int): Use machine_mode if
3798 USE_ENUM_MODES is defined.
3799 (is_a): New function.
3801 (dyn_cast): Likewise.
3802 (scalar_float_mode): New class.
3803 (scalar_float_mode::includes_p): New function.
3804 (is_float_mode): Likewise.
3805 * gdbhooks.py (MachineModePrinter): New class.
3806 (build_pretty_printer): Use it for scalar_float_mode.
3807 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3808 (format_helper::format_helper): Turn into a template.
3809 * genmodes.c (get_mode_class): New function.
3810 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3811 or machine_mode if none.
3812 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3813 as_a <scalar_float_mode>.
3814 * dwarf2out.c (mem_loc_descriptor): Likewise.
3815 (insert_float): Likewise.
3816 (add_const_value_attribute): Likewise.
3817 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3818 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3819 (expand_unop): Update accordingly.
3820 (expand_abs_nojump): Likewise.
3821 (expand_copysign_absneg): Take a scalar_float_mode.
3822 (expand_copysign_bit): Likewise.
3823 (expand_copysign): Update accordingly.
3825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3826 Alan Hayward <alan.hayward@arm.com>
3827 David Sherwood <david.sherwood@arm.com>
3829 * coretypes.h (opt_mode): New class.
3830 * machmode.h (opt_mode): Likewise.
3831 (opt_mode::else_void): New function.
3832 (opt_mode::require): Likewise.
3833 (opt_mode::exists): Likewise.
3834 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3835 (GET_MODE_2XWIDER_MODE): Likewise.
3836 (mode_iterator::get_wider): Update accordingly.
3837 (mode_iterator::get_2xwider): Likewise.
3838 (mode_iterator::get_known_wider): Likewise, turning into a template.
3839 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3840 forcing a wider mode to exist.
3841 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3842 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3843 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3844 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3845 * lower-subreg.c (init_lower_subreg): Likewise.
3846 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3847 on the final iteration.
3848 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3849 a wider mode exists before asking for a move pattern.
3850 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3851 forcing a wider mode to exist.
3852 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3853 returning false if no such mode exists.
3854 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3855 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3856 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3857 Avoid checking for a MODE_INT if we already know the mode is not a
3859 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3860 forcing a wider mode to exist.
3861 (expmed_mult_highpart_optab): Likewise.
3862 (expmed_mult_highpart): Likewise.
3863 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3865 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3866 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3867 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3868 * internal-fn.c (expand_mul_overflow): Update use of
3869 GET_MODE_2XWIDER_MODE.
3870 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3871 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3872 GET_MODE_WIDER_MODE.
3873 (convert_plusminus_to_widen): Likewise.
3874 * tree-switch-conversion.c (array_value_type): Likewise.
3875 * var-tracking.c (emit_note_insn_var_location): Likewise.
3876 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3877 Return false inside rather than outside the loop if no wider mode
3879 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3880 and GET_MODE_2XWIDER_MODE
3881 (can_compare_p): Use else_void.
3882 * gdbhooks.py (OptMachineModePrinter): New class.
3883 (build_pretty_printer): Use it for opt_mode.
3885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3886 Alan Hayward <alan.hayward@arm.com>
3887 David Sherwood <david.sherwood@arm.com>
3889 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3890 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3893 Alan Hayward <alan.hayward@arm.com>
3894 David Sherwood <david.sherwood@arm.com>
3896 * machmode.h (mode_traits): New structure.
3897 (get_narrowest_mode): New function.
3898 (mode_iterator::start): Likewise.
3899 (mode_iterator::iterate_p): Likewise.
3900 (mode_iterator::get_wider): Likewise.
3901 (mode_iterator::get_known_wider): Likewise.
3902 (mode_iterator::get_2xwider): Likewise.
3903 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3904 (FOR_EACH_MODE): Likewise.
3905 (FOR_EACH_MODE_FROM): Likewise.
3906 (FOR_EACH_MODE_UNTIL): Likewise.
3907 (FOR_EACH_WIDER_MODE): Likewise.
3908 (FOR_EACH_2XWIDER_MODE): Likewise.
3909 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3910 * combine.c (simplify_comparison): Likewise
3911 * config/i386/i386.c (type_natural_mode): Likewise.
3912 * cse.c (cse_insn): Likewise.
3913 * dse.c (find_shift_sequence): Likewise.
3914 * emit-rtl.c (init_derived_machine_modes): Likewise.
3915 (init_emit_once): Likewise.
3916 * explow.c (hard_function_value): Likewise.
3917 * expmed.c (extract_fixed_bit_field_1): Likewise.
3918 (extract_bit_field_1): Likewise.
3919 (expand_divmod): Likewise.
3920 (emit_store_flag_1): Likewise.
3921 * expr.c (init_expr_target): Likewise.
3922 (convert_move): Likewise.
3923 (alignment_for_piecewise_move): Likewise.
3924 (widest_int_mode_for_size): Likewise.
3925 (emit_block_move_via_movmem): Likewise.
3926 (copy_blkmode_to_reg): Likewise.
3927 (set_storage_via_setmem): Likewise.
3928 (compress_float_constant): Likewise.
3929 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3930 * optabs-query.c (get_best_extraction_insn): Likewise.
3931 * optabs.c (expand_binop): Likewise.
3932 (expand_twoval_unop): Likewise.
3933 (expand_twoval_binop): Likewise.
3934 (widen_leading): Likewise.
3935 (widen_bswap): Likewise.
3936 (expand_parity): Likewise.
3937 (expand_unop): Likewise.
3938 (prepare_cmp_insn): Likewise.
3939 (prepare_float_lib_cmp): Likewise.
3940 (expand_float): Likewise.
3941 (expand_fix): Likewise.
3942 (expand_sfix_optab): Likewise.
3943 * postreload.c (move2add_use_add2_insn): Likewise.
3944 * reg-stack.c (reg_to_stack): Likewise.
3945 * reginfo.c (choose_hard_reg_mode): Likewise.
3946 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3947 * stor-layout.c (mode_for_size): Likewise.
3948 (smallest_mode_for_size): Likewise.
3949 (mode_for_vector): Likewise.
3950 (finish_bitfield_representative): Likewise.
3951 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3952 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3953 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3954 * var-tracking.c (prepare_call_arguments): Likewise.
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 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3961 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3962 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3963 * machmode.h (mode_size): Move earlier in file.
3964 (mode_precision): Likewise.
3965 (mode_inner): Likewise.
3966 (mode_nunits): Likewise.
3967 (mode_unit_size): Likewise.
3968 (unit_unit_precision): Likewise.
3969 (mode_wider): Likewise.
3970 (mode_2xwider): Likewise.
3971 (machine_mode): New class.
3972 (mode_to_bytes): New function.
3973 (mode_to_bits): Likewise.
3974 (mode_to_precision): Likewise.
3975 (mode_to_inner): Likewise.
3976 (mode_to_unit_size): Likewise.
3977 (mode_to_unit_precision): Likewise.
3978 (mode_to_nunits): Likewise.
3979 (GET_MODE_SIZE): Use mode_to_bytes.
3980 (GET_MODE_BITSIZE): Use mode_to_bits.
3981 (GET_MODE_PRECISION): Use mode_to_precision.
3982 (GET_MODE_INNER): Use mode_to_inner.
3983 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3984 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3985 (GET_MODE_NUNITS): Use mode_to_nunits.
3986 * system.h (ALWAYS_INLINE): New macro.
3987 * config/powerpcspe/powerpcspe-c.c
3988 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3989 int for arg1_mode and arg2_mode.
3991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3992 Alan Hayward <alan.hayward@arm.com>
3993 David Sherwood <david.sherwood@arm.com>
3995 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3996 Prefix mode names with E_ in case statements.
3997 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3998 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3999 (aarch64_split_simd_move): Likewise.
4000 (aarch64_gen_storewb_pair): Likewise.
4001 (aarch64_gen_loadwb_pair): Likewise.
4002 (aarch64_gen_store_pair): Likewise.
4003 (aarch64_gen_load_pair): Likewise.
4004 (aarch64_get_condition_code_1): Likewise.
4005 (aarch64_constant_pool_reload_icode): Likewise.
4006 (get_rsqrte_type): Likewise.
4007 (get_rsqrts_type): Likewise.
4008 (get_recpe_type): Likewise.
4009 (get_recps_type): Likewise.
4010 (aarch64_gimplify_va_arg_expr): Likewise.
4011 (aarch64_simd_container_mode): Likewise.
4012 (aarch64_emit_load_exclusive): Likewise.
4013 (aarch64_emit_store_exclusive): Likewise.
4014 (aarch64_expand_compare_and_swap): Likewise.
4015 (aarch64_gen_atomic_cas): Likewise.
4016 (aarch64_emit_bic): Likewise.
4017 (aarch64_emit_atomic_swap): Likewise.
4018 (aarch64_emit_atomic_load_op): Likewise.
4019 (aarch64_evpc_trn): Likewise.
4020 (aarch64_evpc_uzp): Likewise.
4021 (aarch64_evpc_zip): Likewise.
4022 (aarch64_evpc_ext): Likewise.
4023 (aarch64_evpc_rev): Likewise.
4024 (aarch64_evpc_dup): Likewise.
4025 (aarch64_gen_ccmp_first): Likewise.
4026 (aarch64_gen_ccmp_next): Likewise.
4027 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4028 (alpha_emit_xfloating_libcall): Likewise.
4029 (emit_insxl): Likewise.
4030 (alpha_arg_type): Likewise.
4031 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4032 (arc_preferred_simd_mode): Likewise.
4033 (arc_secondary_reload): Likewise.
4034 (get_arc_condition_code): Likewise.
4035 (arc_print_operand): Likewise.
4036 (arc_legitimate_constant_p): Likewise.
4037 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4038 * config/arc/arc.md (casesi_load): Likewise.
4039 (casesi_compact_jump): Likewise.
4040 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4041 (cc_use_register): Likewise.
4042 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4043 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4044 (arm_init_iwmmxt_builtins): Likewise.
4045 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4046 (neon_expand_vector_init): Likewise.
4047 (arm_attr_length_move_neon): Likewise.
4048 (maybe_get_arm_condition_code): Likewise.
4049 (arm_emit_vector_const): Likewise.
4050 (arm_preferred_simd_mode): Likewise.
4051 (arm_output_iwmmxt_tinsr): Likewise.
4052 (thumb1_output_casesi): Likewise.
4053 (thumb2_output_casesi): Likewise.
4054 (arm_emit_load_exclusive): Likewise.
4055 (arm_emit_store_exclusive): Likewise.
4056 (arm_expand_compare_and_swap): Likewise.
4057 (arm_evpc_neon_vuzp): Likewise.
4058 (arm_evpc_neon_vzip): Likewise.
4059 (arm_evpc_neon_vrev): Likewise.
4060 (arm_evpc_neon_vtrn): Likewise.
4061 (arm_evpc_neon_vext): Likewise.
4062 (arm_validize_comparison): Likewise.
4063 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4064 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4065 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4066 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4067 (c6x_preferred_simd_mode): Likewise.
4068 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4069 (epiphany_rtx_costs): Likewise.
4070 * config/epiphany/predicates.md (proper_comparison_operator):
4072 * config/frv/frv.c (condexec_memory_operand): Likewise.
4073 (frv_emit_move): Likewise.
4074 (output_move_single): Likewise.
4075 (output_condmove_single): Likewise.
4076 (frv_hard_regno_mode_ok): Likewise.
4077 (frv_matching_accg_mode): Likewise.
4078 * config/h8300/h8300.c (split_adds_subs): Likewise.
4079 (h8300_rtx_costs): Likewise.
4080 (h8300_print_operand): Likewise.
4081 (compute_mov_length): Likewise.
4082 (output_logical_op): Likewise.
4083 (compute_logical_op_length): Likewise.
4084 (compute_logical_op_cc): Likewise.
4085 (h8300_shift_needs_scratch_p): Likewise.
4086 (output_a_shift): Likewise.
4087 (compute_a_shift_length): Likewise.
4088 (compute_a_shift_cc): Likewise.
4089 (expand_a_rotate): Likewise.
4090 (output_a_rotate): Likewise.
4091 * config/i386/i386.c (classify_argument): Likewise.
4092 (function_arg_advance_32): Likewise.
4093 (function_arg_32): Likewise.
4094 (function_arg_64): Likewise.
4095 (function_value_64): Likewise.
4096 (ix86_gimplify_va_arg): Likewise.
4097 (ix86_legitimate_constant_p): Likewise.
4098 (put_condition_code): Likewise.
4099 (split_double_mode): Likewise.
4100 (ix86_avx256_split_vector_move_misalign): Likewise.
4101 (ix86_expand_vector_logical_operator): Likewise.
4102 (ix86_split_idivmod): Likewise.
4103 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4104 (ix86_build_const_vector): Likewise.
4105 (ix86_build_signbit_mask): Likewise.
4106 (ix86_match_ccmode): Likewise.
4107 (ix86_cc_modes_compatible): Likewise.
4108 (ix86_expand_branch): Likewise.
4109 (ix86_expand_sse_cmp): Likewise.
4110 (ix86_expand_sse_movcc): Likewise.
4111 (ix86_expand_int_sse_cmp): Likewise.
4112 (ix86_expand_vec_perm_vpermi2): Likewise.
4113 (ix86_expand_vec_perm): Likewise.
4114 (ix86_expand_sse_unpack): Likewise.
4115 (ix86_expand_int_addcc): Likewise.
4116 (ix86_split_to_parts): Likewise.
4117 (ix86_vectorize_builtin_gather): Likewise.
4118 (ix86_vectorize_builtin_scatter): Likewise.
4119 (avx_vpermilp_parallel): Likewise.
4120 (inline_memory_move_cost): Likewise.
4121 (ix86_tieable_integer_mode_p): Likewise.
4122 (x86_maybe_negate_const_int): Likewise.
4123 (ix86_expand_vector_init_duplicate): Likewise.
4124 (ix86_expand_vector_init_one_nonzero): Likewise.
4125 (ix86_expand_vector_init_one_var): Likewise.
4126 (ix86_expand_vector_init_concat): Likewise.
4127 (ix86_expand_vector_init_interleave): Likewise.
4128 (ix86_expand_vector_init_general): Likewise.
4129 (ix86_expand_vector_set): Likewise.
4130 (ix86_expand_vector_extract): Likewise.
4131 (emit_reduc_half): Likewise.
4132 (ix86_emit_i387_round): Likewise.
4133 (ix86_mangle_type): Likewise.
4134 (ix86_expand_round_sse4): Likewise.
4135 (expand_vec_perm_blend): Likewise.
4136 (canonicalize_vector_int_perm): Likewise.
4137 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4138 (expand_vec_perm_1): Likewise.
4139 (expand_vec_perm_interleave3): Likewise.
4140 (expand_vec_perm_even_odd_pack): Likewise.
4141 (expand_vec_perm_even_odd_1): Likewise.
4142 (expand_vec_perm_broadcast_1): Likewise.
4143 (ix86_vectorize_vec_perm_const_ok): Likewise.
4144 (ix86_expand_vecop_qihi): Likewise.
4145 (ix86_expand_mul_widen_hilo): Likewise.
4146 (ix86_expand_sse2_abs): Likewise.
4147 (ix86_expand_pextr): Likewise.
4148 (ix86_expand_pinsr): Likewise.
4149 (ix86_preferred_simd_mode): Likewise.
4150 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4151 * config/i386/sse.md (*andnot<mode>3): Likewise.
4152 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4153 (*<code><mode>3): Likewise.
4154 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4155 (ia64_expand_atomic_op): Likewise.
4156 (ia64_arg_type): Likewise.
4157 (ia64_mode_to_int): Likewise.
4158 (ia64_scalar_mode_supported_p): Likewise.
4159 (ia64_vector_mode_supported_p): Likewise.
4160 (expand_vec_perm_broadcast): Likewise.
4161 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4162 (iq2000_function_arg_advance): Likewise.
4163 (iq2000_function_arg): Likewise.
4164 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4165 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4166 (m68k_libcall_value): Likewise.
4167 (m68k_function_value): Likewise.
4168 (sched_attr_op_type): Likewise.
4169 * config/mcore/mcore.c (mcore_output_move): Likewise.
4170 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4172 (microblaze_function_arg): Likewise.
4173 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4174 (mips_print_operand): Likewise.
4175 (mips_mode_ok_for_mov_fmt_p): Likewise.
4176 (mips_vector_mode_supported_p): Likewise.
4177 (mips_preferred_simd_mode): Likewise.
4178 (mips_expand_vpc_loongson_even_odd): Likewise.
4179 (mips_expand_vec_unpack): Likewise.
4180 (mips_expand_vi_broadcast): Likewise.
4181 (mips_expand_vector_init): Likewise.
4182 (mips_expand_vec_reduc): Likewise.
4183 (mips_expand_msa_cmp): Likewise.
4184 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4185 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4186 (cc_flags_for_mode): Likewise.
4187 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4188 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4189 (nds32_output_casesi_pc_relative): Likewise.
4190 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4191 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4192 (nvptx_gen_unpack): Likewise.
4193 (nvptx_gen_pack): Likewise.
4194 (nvptx_gen_shuffle): Likewise.
4195 (nvptx_gen_wcast): Likewise.
4196 (nvptx_preferred_simd_mode): Likewise.
4197 * config/pa/pa.c (pa_secondary_reload): Likewise.
4198 * config/pa/predicates.md (base14_operand): Likewise.
4199 * config/powerpcspe/powerpcspe-c.c
4200 (altivec_resolve_overloaded_builtin): Likewise.
4201 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4203 (rs6000_preferred_simd_mode): Likewise.
4204 (output_vec_const_move): Likewise.
4205 (rs6000_expand_vector_extract): Likewise.
4206 (rs6000_split_vec_extract_var): Likewise.
4207 (reg_offset_addressing_ok_p): Likewise.
4208 (rs6000_legitimate_offset_address_p): Likewise.
4209 (rs6000_legitimize_address): Likewise.
4210 (rs6000_emit_set_const): Likewise.
4211 (rs6000_const_vec): Likewise.
4212 (rs6000_emit_move): Likewise.
4213 (spe_build_register_parallel): Likewise.
4214 (rs6000_darwin64_record_arg_recurse): Likewise.
4215 (swap_selector_for_mode): Likewise.
4216 (spe_init_builtins): Likewise.
4217 (paired_init_builtins): Likewise.
4218 (altivec_init_builtins): Likewise.
4219 (do_load_for_compare): Likewise.
4220 (rs6000_generate_compare): Likewise.
4221 (rs6000_expand_float128_convert): Likewise.
4222 (emit_load_locked): Likewise.
4223 (emit_store_conditional): Likewise.
4224 (rs6000_output_function_epilogue): Likewise.
4225 (rs6000_handle_altivec_attribute): Likewise.
4226 (rs6000_function_value): Likewise.
4227 (emit_fusion_gpr_load): Likewise.
4228 (emit_fusion_p9_load): Likewise.
4229 (emit_fusion_p9_store): Likewise.
4230 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4231 (fusion_gpr_mem_load): Likewise.
4232 (fusion_addis_mem_combo_load): Likewise.
4233 (fusion_addis_mem_combo_store): Likewise.
4234 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4235 (fusion_gpr_mem_load): Likewise.
4236 (fusion_addis_mem_combo_load): Likewise.
4237 (fusion_addis_mem_combo_store): Likewise.
4238 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4240 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4241 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4242 (rs6000_preferred_simd_mode): Likewise.
4243 (output_vec_const_move): Likewise.
4244 (rs6000_expand_vector_extract): Likewise.
4245 (rs6000_split_vec_extract_var): Likewise.
4246 (reg_offset_addressing_ok_p): Likewise.
4247 (rs6000_legitimate_offset_address_p): Likewise.
4248 (rs6000_legitimize_address): Likewise.
4249 (rs6000_emit_set_const): Likewise.
4250 (rs6000_const_vec): Likewise.
4251 (rs6000_emit_move): Likewise.
4252 (rs6000_darwin64_record_arg_recurse): Likewise.
4253 (swap_selector_for_mode): Likewise.
4254 (paired_init_builtins): Likewise.
4255 (altivec_init_builtins): Likewise.
4256 (rs6000_expand_float128_convert): Likewise.
4257 (emit_load_locked): Likewise.
4258 (emit_store_conditional): Likewise.
4259 (rs6000_output_function_epilogue): Likewise.
4260 (rs6000_handle_altivec_attribute): Likewise.
4261 (rs6000_function_value): Likewise.
4262 (emit_fusion_gpr_load): Likewise.
4263 (emit_fusion_p9_load): Likewise.
4264 (emit_fusion_p9_store): Likewise.
4265 * config/rx/rx.c (rx_gen_move_template): Likewise.
4266 (flags_from_mode): Likewise.
4267 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4268 (s390_slb_comparison): Likewise.
4269 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4270 (s390_vector_mode_supported_p): Likewise.
4271 (s390_cc_modes_compatible): Likewise.
4272 (s390_match_ccmode_set): Likewise.
4273 (s390_canonicalize_comparison): Likewise.
4274 (s390_emit_compare_and_swap): Likewise.
4275 (s390_branch_condition_mask): Likewise.
4276 (s390_rtx_costs): Likewise.
4277 (s390_secondary_reload): Likewise.
4278 (__SECONDARY_RELOAD_CASE): Likewise.
4279 (s390_expand_cs): Likewise.
4280 (s390_preferred_simd_mode): Likewise.
4281 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4282 * config/sh/sh.c (sh_print_operand): Likewise.
4283 (dump_table): Likewise.
4284 (sh_secondary_reload): Likewise.
4285 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4286 * config/sh/sh.md (casesi_worker_1): Likewise.
4287 (casesi_worker_2): Likewise.
4288 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4289 (fcc_comparison_operator): Likewise.
4290 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4291 (emit_soft_tfmode_cvt): Likewise.
4292 (sparc_preferred_simd_mode): Likewise.
4293 (output_cbranch): Likewise.
4294 (sparc_print_operand): Likewise.
4295 (sparc_expand_vec_perm_bmask): Likewise.
4296 (vector_init_bshuffle): Likewise.
4297 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4298 (spu_vector_mode_supported_p): Likewise.
4299 (spu_expand_insv): Likewise.
4300 (spu_emit_branch_or_set): Likewise.
4301 (spu_handle_vector_attribute): Likewise.
4302 (spu_builtin_splats): Likewise.
4303 (spu_builtin_extract): Likewise.
4304 (spu_builtin_promote): Likewise.
4305 (spu_expand_sign_extend): Likewise.
4306 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4307 (tilegx_simd_int): Likewise.
4308 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4309 (tilepro_simd_int): Likewise.
4310 * config/v850/v850.c (const_double_split): Likewise.
4311 (v850_print_operand): Likewise.
4312 (ep_memory_offset): Likewise.
4313 * config/vax/vax.c (vax_rtx_costs): Likewise.
4314 (vax_output_int_move): Likewise.
4315 (vax_output_int_add): Likewise.
4316 (vax_output_int_subtract): Likewise.
4317 * config/visium/predicates.md (visium_branch_operator): Likewise.
4318 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4319 (visium_print_operand_address): Likewise.
4320 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4321 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4322 (xtensa_expand_conditional_branch): Likewise.
4323 (xtensa_copy_incoming_a7): Likewise.
4324 (xtensa_output_literal): Likewise.
4325 * dfp.c (decimal_real_maxval): Likewise.
4326 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4329 Alan Hayward <alan.hayward@arm.com>
4330 David Sherwood <david.sherwood@arm.com>
4332 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4333 (mode_nunits_inline): Likewise.
4334 (mode_inner_inline): Likewise.
4335 (mode_unit_size_inline): Likewise.
4336 (mode_unit_precision_inline): Likewise.
4337 (emit_insn_modes_h): Likewise. Also emit a #define of the
4339 (emit_mode_wider): Add an E_ prefix to mode names.
4340 (emit_mode_complex): Likewise.
4341 (emit_mode_inner): Likewise.
4342 (emit_mode_adjustments): Likewise.
4343 (emit_mode_int_n): Likewise.
4344 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4345 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4346 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4347 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4348 (CRC32_BUILTIN, ENTRY): Likewise.
4349 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4350 (aarch64_pop_regs): Likewise.
4351 (aarch64_process_components): Likewise.
4352 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4353 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4354 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4355 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4356 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4357 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4359 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4360 (pcmpistr): Likewise.
4361 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4362 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4363 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4365 * config/rl78/rl78.c (mduc_regs): Likewise.
4366 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4367 (htm_expand_builtin): Likewise.
4368 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4369 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4370 * config/xtensa/xtensa.c (print_operand): Likewise.
4371 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4372 (NUM_MODE_VECTOR_INT): Likewise.
4373 * genoutput.c (null_operand): Likewise.
4374 (output_operand_data): Likewise.
4375 * genrecog.c (print_parameter_value): Likewise.
4376 * lra.c (debug_operand_data): Likewise.
4378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4380 * dwarf2out.c (scompare_loc_descriptor_wide)
4381 (scompare_loc_descriptor_narrow): New functions, split out from...
4382 (scompare_loc_descriptor): ...here.
4383 * expmed.c (emit_store_flag_int): New function, split out from...
4384 (emit_store_flag): ...here.
4386 2017-08-30 Richard Biener <rguenther@suse.de>
4388 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4389 (dwarf2out_early_finish): Move setting of AT_pubnames from
4390 early debug output to early finish.
4392 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4394 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4395 and -mdata-region to the assembler.
4397 2017-08-30 Richard Biener <rguenther@suse.de>
4399 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4401 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4402 (dwarf2out_early_global_decl): Only generate a DIE for the
4403 abstract origin if it doesn't already exist or is a declaration DIE.
4404 (resolve_addr): Do not add the linkage name twice when
4405 generating a stub DIE for the DW_TAG_GNU_call_site target.
4407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4410 Use machine_mode rather than int for arg1_mode.
4412 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4415 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4416 that the second argument of the built-in functions to unpack
4417 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4418 switch statement instead a lot of if statements.
4419 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4420 Allow 64-bit values to be in Altivec registers as well as
4421 traditional floating point registers.
4422 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4424 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4426 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4427 MAX_REGS_PER_ADDRESS == 1.
4429 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4431 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4432 * config/i386/i386.c (ix86_option_override_internal): Simplify
4433 setting of opts->x_flag_entry.
4435 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4436 Jakub Jelinek <jakub@redhat.com>
4437 Richard Biener <rguenther@suse.de>
4439 PR tree-optimization/81503
4440 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4441 folded constant fits in the target type; reorder tests for clarity.
4443 2017-08-29 Martin Liska <mliska@suse.cz>
4445 * passes.def: Include pass_lower_switch.
4446 * stmt.c (dump_case_nodes): Remove and move to
4447 tree-switch-conversion.
4448 (case_values_threshold): Likewise.
4449 (expand_switch_as_decision_tree_p): Likewise.
4450 (emit_case_decision_tree): Likewise.
4451 (expand_case): Likewise.
4452 (balance_case_nodes): Likewise.
4453 (node_has_low_bound): Likewise.
4454 (node_has_high_bound): Likewise.
4455 (node_is_bounded): Likewise.
4456 (emit_case_nodes): Likewise.
4457 (struct simple_case_node): New struct.
4458 (add_case_node): Remove.
4459 (emit_case_dispatch_table): Use vector instead of case_list.
4460 (reset_out_edges_aux): Remove.
4461 (compute_cases_per_edge): Likewise.
4462 (expand_case): Build list of simple_case_node.
4463 (expand_sjlj_dispatch_table): Use it.
4464 * tree-switch-conversion.c (struct case_node): Moved from
4465 stmt.c and adjusted.
4466 (emit_case_nodes): Likewise.
4467 (node_has_low_bound): Likewise.
4468 (node_has_high_bound): Likewise.
4469 (node_is_bounded): Likewise.
4470 (case_values_threshold): Likewise.
4471 (reset_out_edges_aux): Likewise.
4472 (compute_cases_per_edge): Likewise.
4473 (add_case_node): Likewise.
4474 (dump_case_nodes): Likewise.
4475 (balance_case_nodes): Likewise.
4476 (expand_switch_as_decision_tree_p): Likewise.
4477 (emit_jump): Likewise.
4478 (emit_case_decision_tree): Likewise.
4479 (try_switch_expansion): Likewise.
4480 (do_jump_if_equal): Likewise.
4481 (emit_cmp_and_jump_insns): Likewise.
4482 (fix_phi_operands_for_edge): New function.
4483 (record_phi_operand_mapping): Likewise.
4484 (class pass_lower_switch): New pass.
4485 (pass_lower_switch::execute): New function.
4486 (make_pass_lower_switch): Likewise.
4487 (conditional_probability):
4488 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4489 * tree-pass.h: Add make_pass_lower_switch.
4491 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4494 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4497 2017-08-29 Richard Biener <rguenther@suse.de>
4499 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4500 we do not add a DW_AT_inline attribute twice.
4501 (gen_subprogram_die): Remove code setting DW_AT_inline on
4502 DECL_ABSTRACT_P nodes.
4504 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4506 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4507 calls to internal functions.
4508 (gimplify_modify_expr): Likewise.
4509 * tree-call-cdce.c (use_internal_fn): Likewise.
4510 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4511 (convert_to_divmod): Set the nothrow flag.
4512 * tree-if-conv.c (predicate_mem_writes): Likewise.
4513 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4514 (vectorizable_call): Likewise.
4515 (vectorizable_store): Likewise.
4516 (vectorizable_load): Likewise.
4517 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4518 (vect_recog_mask_conversion_pattern): Likewise.
4520 2017-08-29 Martin Liska <mliska@suse.cz>
4523 * gcc.c (driver_handle_option): Add new argument.
4524 * opts-common.c (handle_option): Pass
4525 target_option_override_hook.
4526 * opts-global.c (lang_handle_option): Add new option.
4527 (set_default_handlers): Add new argument.
4528 (decode_options): Likewise.
4529 * opts.c (target_handle_option): Likewise.
4530 (common_handle_option): Call target_option_override_hook.
4531 * opts.h (struct cl_option_handler_func): Add hook for
4532 target option override.
4533 (struct cl_option_handlers): Likewise.
4534 (set_default_handlers): Add new argument.
4535 (decode_options): Likewise.
4536 (common_handle_option): Likewise.
4537 (target_handle_option): Likewise.
4538 * toplev.c (toplev::main): Pass targetm.target_option.override
4541 2017-08-29 Richard Biener <rguenther@suse.de>
4542 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4544 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4545 life to the active subtree.
4547 2017-08-28 Jeff Law <law@redhat.com>
4549 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4550 derive_equivalences.
4551 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4553 (edge_info::derive_equivalences): New private member function
4555 * tree-ssa-dom.c (class edge_info): Changed from a struct
4556 to a class. Add ctor/dtor, methods and data members.
4557 (edge_info::edge_info): Renamed from allocate_edge_info.
4558 Initialize additional members.
4559 (edge_info::~edge_info): New.
4560 (free_dom_edge_info): Delete the edge info.
4561 (record_edge_info): Use new class & associated member functions.
4562 Tighten forms for testing for edge equivalences.
4563 (record_temporary_equivalences): Iterate over the simple
4564 equivalences rather than assuming there's only one per edge.
4565 (cprop_into_successor_phis): Iterate over the simple
4566 equivalences rather than assuming there's only one per edge.
4567 (optimize_stmt): Use operand_equal_p rather than pointer
4568 equality for mini-DSE code.
4570 2017-08-28 Nathan Sidwell <nathan@acm.org>
4572 * gcc.c (execute): Fold SIGPIPE handling into switch
4573 statement. Adjust internal error message.
4575 2017-08-28 Richard Biener <rguenther@suse.de>
4578 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4579 Do nothing for removed DIEs.
4581 2017-08-28 Richard Biener <rguenther@suse.de>
4583 PR tree-optimization/81977
4584 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4587 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4590 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4591 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4592 using targetm.gen_mem_thread_fence.
4594 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4597 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4598 predicate to register_operand. Reorder operands.
4599 (*btr<mode>): Ditto.
4600 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4601 (*btr<mode>_mask): Ditto.
4603 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4605 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4606 * config/rs6000/xmmintrin.h: New file.
4607 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4609 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4612 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4613 parameter and_insn and return it.
4614 (recombine_lvx_pattern): Insert a copy to ensure availability of
4615 the base register of the copied masking operation at the point of
4616 the instruction replacement.
4617 (recombine_stvx_pattern): Likewise.
4619 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4621 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4622 undocumented switches.
4623 (-mpower9-dform-vector): Likewise.
4624 (-mpower9-dform): Likewise.
4625 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4626 comments to delete references to -mpower9-dform* switches.
4627 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4628 Delete reference to -mpower9-dform* switches, test for
4629 -mpower9-vector instead.
4630 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4631 (OTHER_P9_VECTOR_MASKS): Likewise.
4632 (POWERPC_MASKS): Likewise.
4633 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4634 tests against -mpower9-dform* to -mpower9-vector. Delete code
4635 that checked for -mpower9-dform* consistancy with other options.
4636 Add test for -mpower9-misc to enable other power9 switches.
4637 (rs6000_init_hard_regno_mode_ok): Likewise.
4638 (rs6000_option_override_internal): Likewise.
4639 (rs6000_emit_prologue): Likewise.
4640 (rs6000_emit_epilogue): Likewise.
4641 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4642 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4643 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4645 (emit_fusion_p9_store): Likewise.
4646 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4647 resetting these macros if the assembler does not support ISA 3.0
4649 (TARGET_P9_DFORM_VECTOR): Likewise.
4650 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4651 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4653 2017-08-25 Alan Modra <amodra@gmail.com>
4655 PR rtl-optimization/81747
4656 * cse.c (cse_extended_basic_block): Don't attempt to record
4657 equivalences for degenerate conditional branches that branch
4658 to their fall-through.
4660 2017-08-24 Martin Sebor <msebor@redhat.com>
4663 * gimple-fold.c (size_must_be_zero_p): New function.
4664 (gimple_fold_builtin_memory_op): Call it.
4666 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4668 * config/rs6000/mm_malloc.h: New file.
4670 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4672 PR tree-optimization/81913
4673 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4674 analysis when either IVs in condition can wrap.
4676 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4678 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4679 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4681 2017-08-24 Richard Biener <rguenther@suse.de>
4684 * targhooks.c (default_target_can_inline_p): Properly
4685 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4686 is present and always compare.
4687 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4688 infer -mfpmath=sse from TARGET_SSE_P.
4689 (ix86_can_inline_p): Properly use target_option_default_node when
4690 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4692 2017-08-24 Richard Biener <rguenther@suse.de>
4695 * dwarf2out.c (output_die): Handle flag_generate_offload like
4697 (output_comp_unit): Likewise.
4698 (gen_array_type_die): Likewise.
4699 (dwarf2out_early_finish): Likewise.
4700 (note_variable_value_in_expr): Likewise.
4701 (dwarf2out_finish): Likewise. Adjust assert.
4702 * cgraphunit.c (symbol_table::compile): Move setting of
4703 flag_generate_offload earlier ...
4704 (symbol_table::finalize_compilation_unit): ... here, before
4705 early debug finalization.
4707 2017-08-24 Richard Biener <rguenther@suse.de>
4709 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4710 and ipa-fnsummary.h.
4711 (ix86_can_inline_p): When ix86_fpmath flags do not match
4712 check whether the callee uses FP math at all.
4714 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4717 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4718 instead of TYPE_PRECISION.
4720 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4721 Alan Hayward <alan.hayward@arm.com>
4722 David Sherwood <david.sherwood@arm.com>
4724 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4726 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4728 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4730 (ix86_frame::stack_realign_allocate): New field.
4731 (struct machine_frame_state): Modify comments.
4732 (machine_frame_state::sp_realigned_fp_end): New field.
4733 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4735 (sp_valid_at): Add assertion to assure no attempt to access invalid
4736 offset of a realigned stack.
4737 (fp_valid_at): Likewise.
4738 (choose_baseaddr): Modify comments.
4739 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4740 ix86_expand_prologue.
4741 (ix86_expand_prologue): Modify stack realignment and allocation.
4742 (ix86_expand_epilogue): Modify comments.
4743 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4744 avx2_runtime, avx512f, and avx512f_runtime.
4746 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4748 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4749 (mstackrealign): Do not init to -1.
4750 * config/i386/i386.c (ix86_option_override_internal):
4751 Check opts_set, not opts when setting default value of
4752 opts->x_ix86_force_align_arg_pointer.
4754 2017-08-23 Richard Biener <rguenther@suse.de>
4756 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4757 lang_hooks.decl_printable_name.
4758 * print-rtl-function.c (print_rtx_function): Likewise.
4759 * tree-pretty-print.c (dump_function_header): Likewise.
4761 2017-08-23 Richard Biener <rguenther@suse.de>
4764 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4765 -g0 at compile-time.
4767 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4770 * doc/sourcebuild.texi (Other hardware attributes):
4773 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4776 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4777 Fix single-use check.
4779 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4781 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4783 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4785 * doc/install.texi: Modify to add more details on running selected
4788 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4790 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4791 is combined with -mabi=ms.
4792 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4795 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4797 PR tree-optimization/81488
4798 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4799 and cached_basis fields.
4800 (MAX_SPREAD): New constant.
4801 (alloc_cand_and_find_basis): Initialize new fields.
4802 (clear_visited): New function.
4803 (create_phi_basis_1): Rename from create_phi_basis, set visited
4804 and cached_basis fields.
4805 (create_phi_basis): New wrapper function.
4806 (phi_add_costs_1): Rename from phi_add_costs, add spread
4807 parameter, set visited field, short-circuit when limits reached.
4808 (phi_add_costs): New wrapper function.
4809 (record_phi_increments_1): Rename from record_phi_increments, set
4811 (record_phi_increments): New wrapper function.
4812 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4813 (phi_incr_cost): New wrapper function.
4814 (all_phi_incrs_profitable_1): Rename from
4815 all_phi_incrs_profitable, set visited field.
4816 (all_phi_incrs_profitable): New wrapper function.
4818 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4819 Alan Hayward <alan.hayward@arm.com>
4820 David Sherwood <david.sherwood@arm.com>
4822 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4823 that takes the outer and inner modes.
4824 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4825 comparison as the canonical test for a paradoxical subreg.
4826 * combine.c (simplify_set): Use paradoxical_subreg_p.
4827 (make_extraction): Likewise.
4828 (force_to_mode): Likewise.
4829 (rtx_equal_for_field_assignment_p): Likewise.
4830 (gen_lowpart_for_combine): Likewise.
4831 (simplify_comparison): Likewise.
4832 * cse.c (equiv_constant): Likewise.
4833 * expmed.c (store_bit_field_1): Likewise.
4834 * final.c (alter_subreg): Likewise.
4835 * fwprop.c (propagate_rtx): Likewise.
4836 (forward_propagate_subreg): Likewise.
4837 * ira-conflicts.c (ira_build_conflicts): Likewise.
4838 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4839 * lra-constraints.c (curr_insn_transform): Likewise.
4840 (split_reg): Likewise.
4841 * lra-eliminations.c (move_plus_up): Likewise.
4842 (lra_eliminate_regs_1): Likewise.
4843 * recog.c (general_operand): Likewise.
4844 * ree.c (combine_reaching_defs): Likewise.
4845 * reload.c (push_reload): Likewise.
4846 (find_reloads): Likewise.
4847 * reload1.c (elimination_effects): Likewise.
4848 (compute_reload_subreg_offset): Likewise.
4849 (choose_reload_regs): Likewise.
4850 * rtlanal.c (subreg_lsb_1): Likewise.
4851 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4852 (simplify_subreg): Likewise.
4853 * var-tracking.c (track_loc_p): Likewise.
4854 * emit-rtl.c (byte_lowpart_offset): Likewise.
4855 (paradoxical_subreg_p): Delete out-of-line definition.
4857 2017-08-22 Jeff Law <law@redhat.com>
4859 PR tree-optimization/81741
4860 PR tree-optimization/71947
4861 * tree-ssa-dom.c: Include tree-inline.h.
4862 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4863 equivalences if one is more expensive to compute than the other.
4864 * tree-ssa-scopedtables.h (class const_or_copies): Make
4865 record_const_or_copy_raw method private.
4866 (class avail_exprs_stack): New method simplify_binary_operation.
4867 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4868 avail_exprs_stack::simplify_binary_operation as needed.
4869 (avail_exprs_stack::simplify_binary_operation): New function.
4871 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4873 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4874 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4875 (DOT_SYMBOLS): Likewise.
4876 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4877 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4878 (RS6000_ABI_NAME): Likewise.
4879 (TARGET_CMODEL): Likewise.
4880 (TOC_SECTION_ASM_OP): Likewise.
4881 (SET_CMODEL): New macro.
4882 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4884 2017-08-22 Richard Biener <rguenther@suse.de>
4886 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4887 to allow for free-lang-data replacements similar to verify_type_variant.
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 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4894 VOIDmode for the LABEL_REF.
4896 2017-08-22 Richard Biener <rguenther@suse.de>
4898 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4900 2017-08-22 Richard Biener <rguenther@suse.de>
4902 * common.opt (feliminate-dwarf2-dups): Ignore.
4903 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4904 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4905 same_die_p_wrap, compute_section_prefix,
4906 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4907 (comdat_symbol_id, comdat_symbol_number): Likewise.
4908 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4910 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4911 (output_die): Mark unreachable path unreachable.
4912 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4913 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4914 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4915 (dwarf2out_early_finish): Likewise.
4917 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4919 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4921 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4924 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4925 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4926 (avr_attribute_table) <io, io_low, address>: Initialize
4927 .decl_required with true.
4929 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4931 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4932 undocumented debugging options.
4933 (-mvsx-scalar-double): Likewise.
4934 (-mallow-df-permute): Likewise.
4935 (-mvectorize-builtins): Likewise.
4936 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4937 (rs6000_builtin_vectorized_function): Likewise.
4938 (rs6000_builtin_md_vectorized_function): Likewise.
4939 (rs6000_opt_vars): Likewise.
4941 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4944 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4945 (*btrq_imm): Rename from *btrq.
4946 (*btcq_imm): Rename from *btcq.
4947 (btsc): New code attribute.
4948 (*<btsc><mode>): New insn pattern.
4949 (*btr<mode>): Ditto.
4950 (*<btsc><mode>_mask): New insn_and_split pattern.
4951 (*btr<mode>_mask): Ditto.
4953 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4954 Alan Hayward <alan.hayward@arm.com>
4955 David Sherwood <david.sherwood@arm.com>
4957 * function.c (pad_below): Simplify padding calculation.
4959 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4960 Alan Hayward <alan.hayward@arm.com>
4961 David Sherwood <david.sherwood@arm.com>
4963 * target.def (function_prologue): Remove frame size argument.
4964 (function_epilogue): Likewise.
4965 * doc/tm.texi: Regenerate.
4966 * final.c (final_start_function): Update call to function_prologue.
4967 (final_end_function): Update call to function_epilogue.
4968 (default_function_pro_epilogue): Remove frame size argument.
4969 * output.h (default_function_pro_epilogue): Likewise.
4970 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4971 (arm_output_function_prologue): Likewise.
4972 * config/frv/frv.c (frv_function_prologue): Likewise.
4973 (frv_function_epilogue): Likewise.
4974 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4975 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4976 (ia64_output_function_epilogue): Likewise.
4977 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4978 (m32r_output_function_epilogue): Likewise.
4979 * config/microblaze/microblaze.c (microblaze_function_prologue)
4980 (microblaze_function_epilogue): Likewise.
4981 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4982 (mips_output_function_epilogue): Likewise.
4983 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4984 (mmix_target_asm_function_epilogue): Likewise.
4985 * config/msp430/msp430.c (msp430_start_function): Likewise.
4986 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4987 (nds32_asm_function_epilogue): Likewise.
4988 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4989 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4990 (pa_output_function_epilogue): Likewise.
4991 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4992 (rs6000_output_function_epilogue): Likewise.
4993 * config/rl78/rl78.c (rl78_start_function): Likewise.
4994 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4995 (rs6000_output_function_epilogue): Likewise.
4996 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4997 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4998 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4999 (sparc_asm_function_epilogue): Likewise.
5001 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5003 * tree.h (type_has_mode_precision_p): New function.
5004 * convert.c (convert_to_integer_1): Use it.
5005 * expr.c (expand_expr_real_2): Likewise.
5006 (expand_expr_real_1): Likewise.
5007 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5008 * match.pd: Likewise.
5009 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5010 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5011 * tree-tailcall.c (process_assignment): Likewise.
5012 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5013 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5014 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5015 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5016 (vectorizable_assignment): Likewise.
5017 (vectorizable_shift): Likewise.
5018 (vectorizable_operation): Likewise.
5019 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5021 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5023 * match.pd: Add pow (C, x) simplification.
5025 2017-08-21 Richard Biener <rguenther@suse.de>
5027 PR tree-optimization/81900
5028 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5029 for blocks with abnormal predecessors.
5030 (compute_antic): Do not set visited flag prematurely.
5032 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5035 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5037 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5039 * stor-layout.h (vector_type_mode): Move to...
5040 * tree.h (vector_type_mode): ...here.
5041 * stor-layout.c (vector_type_mode): Move to...
5042 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5044 2017-08-21 Richard Biener <rguenther@suse.de>
5046 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5047 register_external_die hooks.
5048 (debug_false_tree_charstarstar_uhwistar): Declare.
5049 (debug_nothing_tree_charstar_uhwi): Likewise.
5050 * debug.c (do_nothing_debug_hooks): Adjust.
5051 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5052 (debug_nothing_tree_charstar_uhwi): Likewise.
5053 * dbxout.c (dbx_debug_hooks): Adjust.
5054 (xcoff_debug_hooks): Likewise.
5055 * sdbout.c (sdb_debug_hooks): Likewise.
5056 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5057 * dwarf2out.c (macinfo_label_base): New global.
5058 (dwarf2out_register_external_die): New function for the
5059 register_external_die hook.
5060 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5061 (dwarf2_debug_hooks): Use them.
5062 (dwarf2_lineno_debug_hooks): Adjust.
5063 (struct die_struct): Add with_offset flag.
5064 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5065 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5066 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5067 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5068 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5069 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5070 defining section names for the early LTO debug variants.
5071 (reset_indirect_string): New helper.
5072 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5073 (print_dw_val): Add support for offsetted symbol references.
5074 (get_ultimate_context): Split out from is_cxx.
5075 (is_cxx): Use get_ultimate_context.
5076 (is_fortran): Add decl overload.
5077 (compute_comp_unit_symbol): Split out worker from
5078 compute_section_prefix.
5079 (compute_section_prefix): Call compute_comp_unit_symbol and
5080 set comdat_type_p here.
5081 (output_die): Skip DIE symbol output for the LTO added one.
5082 Handle DIE symbol references with offset.
5083 (output_comp_unit): Guard section name mangling properly.
5084 For LTO debug sections emit a symbol at the section beginning
5085 which we use to refer to its DIEs.
5086 (add_abstract_origin_attribute): For DIEs registered via
5087 dwarf2out_register_external_die directly refer to the early
5088 DIE rather than indirectly through the shadow one we created.
5089 Remove obsolete call to dwarf2out_abstract_function for
5090 non-function/block origins.
5091 (gen_array_type_die): When generating early LTO debug do
5092 not emit DW_AT_string_length.
5093 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5094 late when in LTO. As suggested place a gcc_unreachable for
5095 the DECL_ABSTRACT_P case.
5096 (gen_subprogram_die): Avoid another specification DIE
5097 for early built declarations/definitions for the late LTO case.
5098 (gen_variable_die): Add type references for late duplicated VLA dies
5100 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5101 we have the abstract instance already.
5102 (process_scope_var): Adjust decl DIE contexts in LTO which
5103 first puts them in limbo.
5104 (gen_decl_die): Do not generate type DIEs late apart from
5105 types for VLAs or for decls we do not yet have a DIE. Do not
5106 call dwarf2out_abstract_function late.
5107 (dwarf2out_early_global_decl): Make sure to create DIEs
5108 for abstract instances of a decl first.
5109 (dwarf2out_late_global_decl): Adjust comment.
5110 (output_macinfo_op): With multiple macro sections use
5111 macinfo_label_base to distinguish labels.
5112 (output_macinfo): Likewise. Update macinfo_label_base.
5113 Pass in the line info label.
5114 (note_variable_value_in_expr): When generating LTO resolve
5115 all variable values here by generating DIEs as needed.
5116 (init_sections_and_labels): Add early LTO debug flag parameter
5117 and generate different sections and names if set. Add generation
5118 counter for the labels so we can have multiple of them.
5119 (reset_dies): Helper to allow DIEs to be output multiple times.
5120 (dwarf2out_finish): When outputting DIEs to the fat part of an
5121 LTO object first reset DIEs.
5122 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5123 (modified_type_die): Check for decl_ultimate_origin being self
5125 (gen_type_die_with_usage): Likewise.
5126 (gen_typedef_die): Allow decl_ultimate_origin being self.
5127 (set_decl_abstract_flags): Remove.
5128 (set_block_abstract_flags): Likewise.
5129 (dwarf2out_abstract_function): Treat the early generated DIEs
5130 as the abstract copy and only add DW_AT_inline and
5131 DW_AT_artificial here and call set_decl_origin_self.
5132 If the DIE has an abstract origin don't do anything.
5133 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5134 if we have none yet (Go fails to build one, PR78628).
5135 (variably_modified_type_p): Prevent endless recursion for Ada
5136 cyclic pointer types.
5137 * lto-streamer-in.c: Include debug.h.
5138 (dref_queue): New global.
5139 (lto_read_tree_1): Stream in DIE references.
5140 (lto_input_tree): Register DIE references.
5141 (input_function): Stream DECL_DEBUG_ARGS.
5142 * lto-streamer-out.c: Include debug.h.
5143 (lto_write_tree_1): Output DIE references.
5144 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5145 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5146 (output_function): Stream DECL_DEBUG_ARGS.
5147 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5148 Stream DECL_ABSTRACT_ORIGIN.
5149 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5150 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5151 DECL_CONTEXT for file-scope decls.
5152 * lto-streamer.h (struct dref_entry): Declare.
5153 (dref_queue): Likewise.
5154 * cfgexpand.c (pass_expand::execute): Do not call the
5155 outlining_inline_function hook here.
5156 * lto-wrapper.c (debug_obj): New global.
5157 (tool_cleanup): Unlink it if required.
5158 (debug_objcopy): New function.
5159 (run_gcc): Handle early debug sections in the IL files by
5160 extracting them to separate files, partially linkin them and
5161 feeding the result back as result to the linker.
5162 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5163 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5164 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5165 sections into a separate segment.
5166 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5168 (darwin_asm_dwarf_section): Likewise.
5169 (darwin_asm_output_dwarf_offset): Likewise.
5170 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5172 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5173 Alan Hayward <alan.hayward@arm.com>
5174 David Sherwood <david.sherwood@arm.com>
5176 * read-md.h (md_reader::record_potential_iterator_use): Replace
5177 pointer argument with an rtx and an index.
5178 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5179 (apply_mode_iterator): Likewise.
5180 (apply_code_iterator): Likewise.
5181 (apply_int_iterator): Likewise.
5182 (apply_subst_iterator): Likewise.
5183 (record_iterator_use): Likewise.
5184 (record_attribute_use): Likewise.
5185 (md_reader::record_potential_iterator_use): Likewise. Update calls
5186 to record_iterator_use and apply_iterator.
5187 (iterator_use): Replace ptr with x and index.
5188 (attribute_use): Likewise.
5189 (apply_attribute_uses): Update calls to apply_iterator.
5190 (apply_iterators): Likewise. Update initialization of iterator_use.
5191 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5192 and record_potential_iterator_use.
5193 (rtx_reader::read_rtx_operand): Likewise.
5195 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5196 Alan Hayward <alan.hayward@arm.com>
5197 David Sherwood <david.sherwood@arm.com>
5199 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5202 2017-08-21 Richard Biener <rguenther@suse.de>
5205 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5206 at struct end conservatively when comparing common bases.
5208 2017-08-21 Richard Biener <rguenther@suse.de>
5210 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5211 (mem_ref_in_stmt): Remove.
5212 (determine_max_movement): Use ref index to get at the reference.
5213 (invariantness_dom_walker::before_dom_children): Deal with
5214 lim data already initialized.
5215 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5217 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5219 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5220 (TARGET_ISA_ROUND): Ditto.
5221 (TARGET_ROUND): Ditto.
5222 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5223 * config/i386/i386.md: Ditto.
5224 * config/i386/sse.md: Ditto.
5225 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5226 with OPTION_MASK_ISA_SSE4_1.
5228 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5231 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5232 __builtin_ia32_lzcnt_u16.
5234 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5237 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5238 (rs6000_set_current_function): Rewrite function to use it.
5240 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5243 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5244 and TYPE_WARN_IF_NOT_ALIGN.
5245 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5246 (handle_warn_if_not_align): New.
5247 (place_union_field): Call handle_warn_if_not_align.
5248 (place_field): Call handle_warn_if_not_align.
5249 Copy TYPE_WARN_IF_NOT_ALIGN.
5250 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5251 (layout_type): Likewise.
5252 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5254 (tree_decl_common): Add warn_if_not_align.
5255 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5256 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5257 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5258 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5259 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5260 * doc/extend.texi: Document warn_if_not_aligned attribute.
5261 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5263 2017-08-17 Martin Liska <mliska@suse.cz>
5266 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5267 (get_data_dependence): Use it as pointer type.
5268 (distribute_loop): Likewise.
5270 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5272 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5273 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5274 (p8_vmrgow_<mode>_direct): New define_insn.
5275 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5276 handle endianness for vmrgew and vmrgow permute patterns.
5278 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5280 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5281 * config/rs6000/rs6000-cpus.def: Remove comment.
5282 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5283 (POWERPC_MASKS): Likewise.
5284 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5285 use of TARGET_VSX_TIMODE.
5286 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5287 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5288 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5289 (rs6000_option_override_internal): Remove dead code.
5290 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5291 (rs6000_legitimize_reload_address): Likewise.
5292 (rs6000_legitimate_address_p): Likewise.
5293 (rs6000_opt_masks): Delete "vsx-timode".
5294 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5295 from function comment.
5296 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5297 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5298 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5300 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5301 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5302 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5305 (define_peephole2): Likewise.
5307 2017-08-17 Martin Sebor <msebor@redhat.com>
5310 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5311 past the end of an array.
5312 (test_pp_format): Add test cases.
5314 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5316 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5317 missing ECF_NOTHROW flags.
5319 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5322 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5323 operands residing in integer registers.
5324 (*vsx_le_perm_load_<mode>): Likewise.
5325 (*vsx_le_perm_store_<mode>): Likewise.
5326 (define_peephole2): Add peepholes to optimize the above.
5328 2017-08-17 Marek Polacek <polacek@redhat.com>
5331 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5332 to mimic what shorten_compare did. Change the return type to bool.
5333 (fold_cond_expr_with_comparison): Update call to
5334 operand_equal_for_comparison_p.
5335 (fold_ternary_loc): Likewise.
5337 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5339 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5341 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5343 2017-08-17 Richard Biener <rguenther@suse.de>
5345 * tree-ssa-structalias.c (solve_graph): When propagating
5346 to successors update the graphs succ edges and avoid duplicate work.
5348 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5351 * config/i386/i386.c (ix86_option_override_internal): Save target
5352 specific options after ix86_stack_protector_guard_reg was changed.
5354 2017-08-17 Richard Biener <rguenther@suse.de>
5356 PR tree-optimization/81827
5357 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5358 (new_var_info): Initialize it conservatively.
5359 (get_call_vi): Mark register vars.
5360 (new_scalar_tmp_constraint_exp): Likewise.
5361 (handle_rhs_call): Likewise.
5362 (handle_const_call): Likewise.
5363 (create_function_info_for): Likewise.
5364 (solve_constraints): Sort varinfos to separate register from
5365 non-register vars to pack points-to solution bitmaps during
5368 2017-08-17 Marek Polacek <polacek@redhat.com>
5370 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5372 2017-08-17 Richard Biener <rguenther@suse.de>
5374 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5375 to true when overflow is undefined and we saturated the result.
5377 2017-08-17 Alan Modra <amodra@gmail.com>
5380 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5381 Don't use store multiple if only one reg needs saving.
5382 (interesting_frame_related_regno): New function.
5383 (rs6000_frame_related): Don't emit frame info for regs that
5385 (rs6000_emit_epilogue): Likewise.
5387 2017-08-16 Nathan Sidwell <nathan@acm.org>
5389 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5390 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5391 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5392 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5393 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5394 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5395 (verify_type): Adjust for TYPE_BINFO move.
5396 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5397 process TYPE_BINFO directly.
5398 (hash_tree): Likewise.
5399 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5401 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5404 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5406 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5408 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5411 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5412 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5413 Add ix86_binary_operator_ok to insn constraint.
5414 (*iordi_1_bts): Ditto.
5415 (*xordi_1_btc): Ditto.
5416 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5417 Update corresponding peephole2 pattern.
5421 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5423 PR tree-optimization/81832
5424 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5425 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5427 2017-08-16 Marek Polacek <polacek@redhat.com>
5430 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5431 regarding size_zero_node.
5433 2017-08-16 Martin Liska <mliska@suse.cz>
5436 * config.gcc: Respect previously set extra_objs in case
5439 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5441 PR tree-optimization/81835
5442 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5443 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5444 not depend on the phi.
5446 2017-08-16 Alan Modra <amodra@gmail.com>
5448 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5451 2017-08-16 Alan Modra <amodra@gmail.com>
5453 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5454 (save_reg_p): ..into this. Update all callers.
5455 (first_reg_to_save): Simplify.
5457 2017-08-16 Alan Modra <amodra@gmail.com>
5459 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5462 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5466 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5467 constant count if that count is less than 32.
5469 2017-08-15 Nathan Sidwell <nathan@acm.org>
5471 * gcc.c (execute): Emit friendlier message if inferior is killed
5472 by an external cause.
5474 2017-08-15 Richard Biener <rguenther@suse.de>
5476 PR tree-optimization/81790
5477 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5478 CONSTRUCTORs from simplifying and VN.
5480 2017-08-14 Martin Sebor <msebor@redhat.com>
5482 * builtin-attrs.def: Add comments.
5484 2017-08-14 Martin Sebor <msebor@redhat.com>
5487 * doc/extend.texi (attribute nonstring): Document new attribute.
5489 2017-08-14 Martin Sebor <msebor@redhat.com>
5492 * tree-diagnostic.c (default_tree_printer): Handle %G.
5493 * gimple-pretty-print.h (percent_G_format): Declare new function.
5494 * gimple-pretty-print.c (percent_G_format): Define.
5495 * tree-pretty-print.c (percent_K_format): Add argument.
5497 2017-08-14 Martin Sebor <msebor@redhat.com>
5499 PR translation/79998
5500 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5501 Remove a stray space.
5503 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5506 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5507 (*iordi_1_bts): Ditto.
5508 (*xordi_1_btc): Ditto.
5510 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5513 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5514 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5516 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5517 quoted strings, and make more translator-friendly.
5518 (darwin_rs6000_override_options): Likewise.
5519 (rs6000_option_override_internal): Likewise.
5520 (rs6000_return_in_memory): Fix overlong line.
5521 (init_cmulative_args): Use quoted strings, and make more
5522 translator-friendly.
5523 (rs6000_pass_by_reference): Fix overlong line.
5524 (def_builtin): Use quoted strings.
5525 (altivec_expand_predicate_builtin): Use quoted strings, and make
5526 more translator-friendly.
5527 (htm_expand_builtin): Use quoted strings.
5528 (cpu_expand_builtin): Use quoted strings, and make more
5529 translator-friendly.
5530 (altivec_expand_builtin): Likewise.
5531 (paired_expand_predicate_builtin): Likewise.
5532 (rs6000_invalid_builtin): Likewise.
5533 (builtin_function_type): Use quoted strings.
5534 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5535 more translator-friendly.
5536 (rs6000_trampoline_init): Likewise.
5537 (rs6000_handle_altivec_attribute): Likewise.
5538 (rs6000_inner_target_options): Use quoted strings.
5539 (rs6000_disable_incompatible_switches): Likewise.
5540 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5541 strings, and make more translator-friendly.
5542 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5544 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5546 PR tree-optimization/81799
5547 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5548 cond_expr to simple gimple operand.
5550 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5553 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5555 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5559 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5560 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5561 TARGET_GASISR_PROLOGUES.
5562 * config/avr/avr.c (avr_option_override): Same.
5563 (avr_pass_pre_proep::execute): Same.
5565 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5568 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5569 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5571 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5573 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5574 define_insn_and_split. Split to a memory load from 0 in
5575 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5576 using PTR mode iterator.
5577 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5578 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5579 (*add_tp_<mode>): Redefine as define_insn_and_split.
5580 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5581 address space. Merge with *add_tp_x32 using PTR mode iterator.
5582 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5583 Split to an add with a memory load from 0 in
5584 DEFAULT_TLS_SEG_REG address space.
5586 2017-08-12 Andrew Pinski <apinski@cavium.com>
5588 * config/aarch64/aarch64-option-extensions.def (rdma):
5589 Fix feature string to what Linux prints out in /proc/cpuinfo.
5591 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5594 * dwarf2out.c (modified_type_die): For C typedef types that have
5595 an ultimate origin, process the ultimate origin instead of the
5597 (gen_typedef_die): Assert that input DECLs have no ultimate
5599 (gen_type_die_with_usage): For typedef variants that have an
5600 ultimate origin, just call gen_decl_die on the original DECL.
5601 (process_scope_var): Avoid creating DIEs for local typedefs and
5602 concrete static variables.
5604 2017-08-12 Alan Modra <amodra@gmail.com>
5608 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5609 match gnu-user.h startfile.
5610 (ENDFILE_LINUX_SPEC): Similarly.
5612 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5615 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5617 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5619 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5620 * config/aarch64/aarch64.md (mov<mode>): Change.
5621 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5622 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5623 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5625 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5627 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5628 for storage order barriers.
5630 2017-08-11 Martin Liska <mliska@suse.cz>
5633 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5634 variables of void type.
5636 2017-08-11 Martin Liska <mliska@suse.cz>
5638 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5639 TARGET_SUPPORTS_ALIASES.
5640 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5641 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5642 (optimize_weakref): Likewise.
5643 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5644 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5645 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5647 2017-08-11 Martin Liska <mliska@suse.cz>
5650 * config/i386/i386.c (make_resolver_func): Do complete
5651 refactoring of the function.
5653 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5656 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5657 * config/i386/i386.c (ix86_stack_protect_guard): Use
5658 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5659 * doc/invoke.texi (x86 Options): Document
5660 -mstack-protector-guard-symbol= option.
5662 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5664 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5665 * config/i386/i386.c (ix86_split_stack_guard): New function.
5666 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5667 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5668 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5669 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5670 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5671 (split_stack_space_check): Call ix86_split_stack_guard.
5673 2017-08-10 Martin Sebor <msebor@redhat.com>
5675 * print-tree.c (print_node): Print location using the established
5677 Replace spaces with colons.
5678 (debug_raw, debug): Ditto.
5680 2017-08-10 Martin Sebor <msebor@redhat.com>
5683 * pretty-print.c (pp_format): Correct the handling of %s precision.
5685 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5688 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5690 (ix86_finalize_stack_frame_flags): This. Also clear
5691 frame_pointer_needed if -fno-omit-frame-pointer is used without
5693 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5694 with ix86_finalize_stack_frame_flags.
5695 (ix86_expand_epilogue): Likewise.
5696 (ix86_expand_split_stack_prologue): Likewise.
5697 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5699 2017-08-10 Martin Liska <mliska@suse.cz>
5702 * c-attribs.c (handle_target_attribute):
5703 Report warning for an empty string argument of target attribute.
5705 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5708 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5710 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5711 or DECL_NONLOCAL labels.
5712 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5713 or DECL_NONLOCAL labels here.
5715 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5717 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5718 to indicate when early gimple folding has been disabled.
5719 (rs6000_gimple_fold_builtin): Add debug content.
5720 (rs6000_invalid_builtin): Fix whitespace.
5721 (rs6000_expand_builtin): Fix whitespace.
5722 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5724 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5727 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5728 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5730 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5732 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5733 (qdf24xx): Likewise.
5734 * config/aarch64/aarch64.md: Include falkor.md.
5735 * config/aarch64/falkor.md: New.
5737 2017-08-09 Marek Polacek <polacek@redhat.com>
5740 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5741 * diagnostic.c (emit_diagnostic): Add a comment.
5742 (emit_diagnostic_valist): New function.
5744 2017-08-09 Marek Polacek <polacek@redhat.com>
5747 * input.c (make_location): New overload.
5748 * input.h (make_location): Declare.
5750 2017-08-08 Alan Modra <amodra@gmail.com>
5751 H.J. Lu <hongjiu.lu@intel.com>
5754 * gcc.c (NO_PIE_SPEC): Delete.
5755 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5757 (LINK_PIE_SPEC): ..to here.
5758 (LINK_COMMAND_SPEC): Support -no-pie.
5759 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5760 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5762 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5763 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5764 (ENDFILE_CRTEND_SPEC): Similarly.
5766 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5769 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5770 (mstack-protector-guard-offset=): Ditto.
5771 * config/i386/i386.c (ix86_option_override): Handle
5772 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5774 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5775 ix86_stack_protect_guard_offset variables.
5776 (TARGET_STACK_PROTECT_GUARD): Always define.
5777 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5778 and -mstack-protector-guard-offset= options.
5780 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5782 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5783 boundary case for the last candidate.
5785 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5787 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5788 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5790 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5793 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5794 * config/aarch64/aarch64-builtins.c
5795 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5796 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5797 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5799 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5800 Andrew Pinski <pinskia@gmail.com>
5803 * internal-fn.def (XORSIGN): New.
5804 * optabs.def (xorsign_optab): New.
5805 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5806 (convert_expand_mult_copysign): New.
5807 (pass_optimize_widening_mul::execute): Call
5808 convert_expand_mult_copysign.
5810 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5812 PR tree-optimization/81354
5813 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5814 Insert on edges rather than explicitly creating landing pads.
5815 (analyze_candidates_and_replace): Commit edge inserts.
5817 2017-08-08 Richard Biener <rguenther@suse.de>
5820 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5821 (expand_simple_operations): Also look through ADDR_EXPRs with
5822 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5824 2017-08-08 Richard Biener <rguenther@suse.de>
5826 PR tree-optimization/81723
5827 * tree-vect-slp.c (struct bst_traits): New hash traits.
5828 (bst_fail): New global.
5829 (vect_build_slp_tree_2): New worker, split out from ...
5830 (vect_build_slp_tree): ... this now wrapping it with using
5831 bst_fail set to cache SLP tree build fails. Properly handle
5833 (vect_analyze_slp_instance): Allocate and free bst_fail.
5835 2017-08-08 Martin Liska <mliska@suse.cz>
5838 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5839 LABEL_DECLs that can be from a different function.
5841 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5843 PR tree-optimization/81744
5844 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5845 loop's number of iterations.
5847 2017-08-08 Martin Liska <mliska@suse.cz>
5849 * asan.c: Include header files.
5850 * attribs.c (build_decl_attribute_variant): New function moved
5852 (build_type_attribute_qual_variant): Likewise.
5853 (cmp_attrib_identifiers): Likewise.
5854 (simple_cst_list_equal): Likewise.
5855 (omp_declare_simd_clauses_equal): Likewise.
5856 (attribute_value_equal): Likewise.
5857 (comp_type_attributes): Likewise.
5858 (build_type_attribute_variant): Likewise.
5859 (lookup_ident_attribute): Likewise.
5860 (remove_attribute): Likewise.
5861 (merge_attributes): Likewise.
5862 (merge_type_attributes): Likewise.
5863 (merge_decl_attributes): Likewise.
5864 (merge_dllimport_decl_attributes): Likewise.
5865 (handle_dll_attribute): Likewise.
5866 (attribute_list_equal): Likewise.
5867 (attribute_list_contained): Likewise.
5868 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5869 (lookup_attribute_by_prefix): Likewise.
5870 * bb-reorder.c: Include header files.
5871 * builtins.c: Likewise.
5872 * calls.c: Likewise.
5873 * cfgexpand.c: Likewise.
5874 * cgraph.c: Likewise.
5875 * cgraphunit.c: Likewise.
5876 * convert.c: Likewise.
5877 * dwarf2out.c: Likewise.
5878 * final.c: Likewise.
5879 * fold-const.c: Likewise.
5880 * function.c: Likewise.
5881 * gimple-expr.c: Likewise.
5882 * gimple-fold.c: Likewise.
5883 * gimple-pretty-print.c: Likewise.
5884 * gimple.c: Likewise.
5885 * gimplify.c: Likewise.
5886 * hsa-common.c: Likewise.
5887 * hsa-gen.c: Likewise.
5888 * internal-fn.c: Likewise.
5889 * ipa-chkp.c: Likewise.
5890 * ipa-cp.c: Likewise.
5891 * ipa-devirt.c: Likewise.
5892 * ipa-fnsummary.c: Likewise.
5893 * ipa-inline.c: Likewise.
5894 * ipa-visibility.c: Likewise.
5896 * lto-cgraph.c: Likewise.
5897 * omp-expand.c: Likewise.
5898 * omp-general.c: Likewise.
5899 * omp-low.c: Likewise.
5900 * omp-offload.c: Likewise.
5901 * omp-simd-clone.c: Likewise.
5902 * opts-global.c: Likewise.
5903 * passes.c: Likewise.
5904 * predict.c: Likewise.
5905 * sancov.c: Likewise.
5906 * sanopt.c: Likewise.
5907 * symtab.c: Likewise.
5908 * toplev.c: Likewise.
5909 * trans-mem.c: Likewise.
5910 * tree-chkp.c: Likewise.
5911 * tree-eh.c: Likewise.
5912 * tree-into-ssa.c: Likewise.
5913 * tree-object-size.c: Likewise.
5914 * tree-parloops.c: Likewise.
5915 * tree-profile.c: Likewise.
5916 * tree-ssa-ccp.c: Likewise.
5917 * tree-ssa-live.c: Likewise.
5918 * tree-ssa-loop.c: Likewise.
5919 * tree-ssa-sccvn.c: Likewise.
5920 * tree-ssa-structalias.c: Likewise.
5921 * tree-ssa.c: Likewise.
5922 * tree-streamer-in.c: Likewise.
5923 * tree-vectorizer.c: Likewise.
5924 * tree-vrp.c: Likewise.
5926 * ubsan.c: Likewise.
5927 * varasm.c: Likewise.
5928 * varpool.c: Likewise.
5929 * tree.c: Remove functions moved to attribs.[ch].
5931 * config/aarch64/aarch64.c: Add attrs.h header file.
5932 * config/alpha/alpha.c: Likewise.
5933 * config/arc/arc.c: Likewise.
5934 * config/arm/arm.c: Likewise.
5935 * config/avr/avr.c: Likewise.
5936 * config/bfin/bfin.c: Likewise.
5937 * config/c6x/c6x.c: Likewise.
5938 * config/cr16/cr16.c: Likewise.
5939 * config/cris/cris.c: Likewise.
5940 * config/darwin.c: Likewise.
5941 * config/epiphany/epiphany.c: Likewise.
5942 * config/fr30/fr30.c: Likewise.
5943 * config/frv/frv.c: Likewise.
5944 * config/ft32/ft32.c: Likewise.
5945 * config/h8300/h8300.c: Likewise.
5946 * config/i386/winnt.c: Likewise.
5947 * config/ia64/ia64.c: Likewise.
5948 * config/iq2000/iq2000.c: Likewise.
5949 * config/lm32/lm32.c: Likewise.
5950 * config/m32c/m32c.c: Likewise.
5951 * config/m32r/m32r.c: Likewise.
5952 * config/m68k/m68k.c: Likewise.
5953 * config/mcore/mcore.c: Likewise.
5954 * config/microblaze/microblaze.c: Likewise.
5955 * config/mips/mips.c: Likewise.
5956 * config/mmix/mmix.c: Likewise.
5957 * config/mn10300/mn10300.c: Likewise.
5958 * config/moxie/moxie.c: Likewise.
5959 * config/msp430/msp430.c: Likewise.
5960 * config/nds32/nds32-isr.c: Likewise.
5961 * config/nds32/nds32.c: Likewise.
5962 * config/nios2/nios2.c: Likewise.
5963 * config/nvptx/nvptx.c: Likewise.
5964 * config/pa/pa.c: Likewise.
5965 * config/pdp11/pdp11.c: Likewise.
5966 * config/powerpcspe/powerpcspe.c: Likewise.
5967 * config/riscv/riscv.c: Likewise.
5968 * config/rl78/rl78.c: Likewise.
5969 * config/rx/rx.c: Likewise.
5970 * config/s390/s390.c: Likewise.
5971 * config/sh/sh.c: Likewise.
5972 * config/sol2.c: Likewise.
5973 * config/sparc/sparc.c: Likewise.
5974 * config/spu/spu.c: Likewise.
5975 * config/stormy16/stormy16.c: Likewise.
5976 * config/tilegx/tilegx.c: Likewise.
5977 * config/tilepro/tilepro.c: Likewise.
5978 * config/v850/v850.c: Likewise.
5979 * config/vax/vax.c: Likewise.
5980 * config/visium/visium.c: Likewise.
5981 * config/xtensa/xtensa.c: Likewise.
5983 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5986 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5987 constraints since the -mupper-regs-* switches have been
5989 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5990 into a vector from a double word element that was extracted from
5991 another vector, and eliminate extra XXPERMDI instructions.
5992 (vsx_concat_<mode>_2): Likewise.
5993 (vsx_concat_<mode>_3): Likewise.
5994 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5995 concat to allow optimizing inserts from previous extracts.
5997 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5999 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6000 memory reference to a SSP offset in TLS address space.
6001 (ix86_print_operand) <case '@'>: Remove.
6002 (ix86_print_operand_punct_valid_p): Remove '@' code.
6003 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6005 (stack_tls_protect_set_<mode>): Remove.
6006 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6007 (stack_tls_protect_test_<mode>): Remove.
6008 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6010 2017-08-07 Olivier Hainque <hainque@adacore.com>
6013 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6015 2017-08-07 Douglas Rupp <rupp@adacore.com>
6017 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6018 variable was referenced as multidir in command.
6020 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6023 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6026 2017-08-07 Martin Liska <mliska@suse.cz>
6028 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6029 * config/rl78/rl78.c: Add include of attribs.h.
6030 * config/sh/sh.c: Likewise.
6031 * config/v850/v850.c: Likewise.
6033 2017-08-07 Tom de Vries <tom@codesourcery.com>
6036 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6038 2017-08-07 Martin Liska <mliska@suse.cz>
6040 * config/mips/mips.c: Include attribs.h.
6042 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6045 * doc/invoke.texi: Document change in behvaior for -Ofast for
6048 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6050 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6052 (aarch64_pop_regs): Likewise.
6053 (aarch64_gen_load_pair): Likewise.
6054 (aarch64_save_callee_saves): Likewise.
6055 (aarch64_restore_callee_saves): Likewise.
6057 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6059 * config/i386/i386.c: Revert the last change.
6061 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6064 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6066 (ix86_finalize_stack_frame_flags): This. Also clear
6067 frame_pointer_needed if -fno-omit-frame-pointer is used without
6069 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6070 with ix86_finalize_stack_frame_flags.
6071 (ix86_expand_epilogue): Likewise.
6072 (ix86_expand_split_stack_prologue): Likewise.
6074 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6077 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6078 to P_AES for Westmere.
6080 2017-08-07 Jonathan Yong <10walls@gmail.com>
6082 * config/i386/mingw.opt (fset-stack-executable): Removed.
6083 * config/i386/cygming.opt (fset-stack-executable): Moved
6085 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6087 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6089 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6091 2017-08-07 Marek Polacek <polacek@redhat.com>
6094 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6096 2017-08-07 Martin Liska <mliska@suse.cz>
6098 * attribs.h (canonicalize_attr_name): New function.
6099 (cmp_attribs): Move from c-format.c and adjusted.
6100 (is_attribute_p): Moved from tree.h.
6101 * tree-inline.c: Add new includes.
6102 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6103 (private_is_attribute_p): Remove.
6104 (private_lookup_attribute): Likewise.
6105 (private_lookup_attribute_by_prefix): Simplify.
6106 (remove_attribute): Use is_attribute_p.
6107 * tree.h: Remove removed declarations.
6109 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6112 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6113 instead of computing it in the function. Formatting fix.
6114 (expand_case): Don't rely on default_edge being the first edge,
6115 clear it if removing it, pass default_edge to
6116 emit_case_dispatch_table.
6117 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6120 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6122 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6123 insn in the function, emit NOP after the insn.
6125 2017-08-06 Tom de Vries <tom@codesourcery.com>
6127 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6130 2017-08-06 Tom de Vries <tom@codesourcery.com>
6132 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6135 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6137 PR tree-optimization/57371
6138 * match.pd: New pattern.
6140 2017-08-04 Marek Polacek <polacek@redhat.com>
6143 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6144 perform the computation in offset_int.
6146 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6148 PR tree-optimization/81136
6149 * tree-vectorizer.h: Include tree-hash-traits.h.
6150 (vec_base_alignments): New typedef.
6151 (vec_info): Add a base_alignments field.
6152 (vect_record_base_alignments): Declare.
6153 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6155 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6156 (create_data_ref): Add an is_conditional_in_stmt argument.
6157 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6158 the is_conditional_in_stmt field.
6159 (data_ref_loc): Add an is_conditional_in_stmt field.
6160 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6161 (find_data_references_in_stmt): Update call to create_data_ref.
6162 (graphite_find_data_references_in_stmt): Likewise.
6163 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6164 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6165 (vect_record_base_alignment): New function.
6166 (vect_record_base_alignments): Likewise.
6167 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6168 for nested statements even if we fail to compute a misalignment.
6169 Use pooled base alignments for unconditional references.
6170 (vect_find_same_alignment_drs): Compare base addresses instead
6172 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6173 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6175 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6177 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6178 Add an explicit name for the enum. Use auto_vec for slp_instances
6180 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6182 (_bb_vec_info): Add a constructor and destructor.
6183 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6184 (destroy_loop_vec_info): Delete.
6185 (vect_destroy_datarefs): Likewise.
6186 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6187 (vec_info::vec_info): New function.
6188 (vec_info::~vec_info): Likewise.
6189 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6190 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6191 destroy_loop_vec_info.
6192 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6193 (_loop_vec_info::_loop_vec_info): ...this.
6194 (destroy_loop_vec_info): Replace with...
6195 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6196 the stmt_vec_infos. Leave handling of vec_info information to its
6197 destructor. Remove explicit vector releases.
6198 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6199 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6200 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6201 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6202 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6203 (destroy_bb_vec_info): Replace with...
6204 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6205 information to its destructor.
6206 (vect_slp_analyze_bb_1): Use new and delete instead of
6207 new_bb_vec_info and destroy_bb_vec_info.
6208 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6211 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6213 * tree-data-ref.h (subscript): Add access_fn field.
6214 (data_dependence_relation): Add could_be_independent_p.
6215 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6216 (same_access_functions): Move to tree-data-ref.c.
6217 * tree-data-ref.c (ref_contains_union_access_p): New function.
6218 (access_fn_component_p): Likewise.
6219 (access_fn_components_comparable_p): Likewise.
6220 (dr_analyze_indices): Add a reference to access_fn_component_p.
6221 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6223 (constant_access_functions): Likewise.
6224 (add_other_self_distances): Likewise.
6225 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6226 (initialize_data_dependence_relation): Use XCNEW and remove
6227 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6228 of access functions that have the same type. Allow the
6229 subsequence to end with different bases in some circumstances.
6230 Record the chosen access functions in SUB_ACCESS_FN.
6231 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6232 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6233 (subscript_dependence_tester_1): Likewise dra and drb.
6234 (build_classic_dist_vector): Update calls accordingly.
6235 (subscript_dependence_tester): Likewise.
6236 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6237 DDR_COULD_BE_INDEPENDENT_P.
6238 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6239 comp_alias_ddrs instead of may_alias_ddrs.
6240 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6242 (vect_analyze_data_ref_dependence): Use it if
6243 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6244 distance vectors if that fails.
6245 (dependence_distance_ge_vf): New function.
6246 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6247 LOOP_VINFO_MAY_ALIAS_DDRS.
6249 2017-08-04 Richard Biener <rguenther@suse.de>
6252 * fold-const.c (fold_binary_loc): Properly restrict
6253 minus_var0 && minus_var1 case when associating undefined overflow
6256 2017-08-04 Tom de Vries <tom@codesourcery.com>
6258 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6260 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6262 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6263 Don't start diagnostic messages with a capital letter.
6264 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6266 (rs6000_invalid_builtin): Likewise.
6267 (rs6000_trampoline_init): Likewise.
6269 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6272 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6273 after setting changeable df flags.
6275 2017-08-03 Richard Biener <rguenther@suse.de>
6277 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6278 up if the use is in USE - X.
6280 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6282 * toplev.c (dumpfile.h): New include.
6283 (internal_error_reentered): New static function. Use it...
6284 (internal_error_function): ...here to handle reentered internal_error.
6286 2017-08-03 Richard Biener <rguenther@suse.de>
6289 * fold-const.c (split_tree): Add minus_var and minus_con
6290 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6291 here but always use minus_*.
6292 (associate_trees): Assert we never associate with MINUS_EXPR
6293 and NULL first operand. Do not recurse for PLUS_EXPR operands
6294 when associating as MINUS_EXPR either.
6295 (fold_binary_loc): Track minus_var and minus_con.
6297 2017-08-03 Tom de Vries <tom@codesourcery.com>
6300 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6301 ACCEL_COMPILER, apply finish_options on
6302 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6304 2017-08-03 Tom de Vries <tom@codesourcery.com>
6307 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6308 function_entry_patch_area_size > 0.
6310 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6313 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6314 instead of 10??LU, perform unit multiplication in wide_int,
6315 don't change alloc_object_size_limit if the limit is larger
6318 PR tree-optimization/81655
6319 PR tree-optimization/81588
6320 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6321 the case when ranges[i].low and high are 1 for unsigned type with
6325 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6326 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6328 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6330 * tree-vrp.h: Add include guard.
6332 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6335 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6336 (ud2): New insn pattern.
6337 * config/i386/i386.c (ix86_expand_epilogue):
6338 For naked functions, generate ud2 instead of trap insn.
6340 2017-08-02 Marek Polacek <polacek@redhat.com>
6343 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6345 2017-08-02 Tom de Vries <tom@codesourcery.com>
6346 Cesar Philippidis <cesar@codesourcery.com>
6348 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6349 Add missing edge probabilities.
6351 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6353 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6356 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6359 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6360 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6361 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6362 prologue_seq sequences - if any.
6364 2017-08-02 Richard Biener <rguenther@suse.de>
6366 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6367 via vectors if supported, integer extracts via punning if supported
6368 or otherwise vector extracts.
6370 2017-08-02 Richard Biener <rguenther@suse.de>
6372 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6374 (bitmap_insert_into_set): ... this.
6376 2017-08-02 Richard Biener <rguenther@suse.de>
6378 PR tree-optimization/81633
6380 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6382 PR tree-optimization/71752
6383 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6385 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6387 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6388 (machine_function::call_ms2sysv_pad_out): Remove field.
6389 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6390 (ix86_compute_frame_layout): Likewise.
6392 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6395 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6396 attribute with interrupt attribute.
6398 2017-08-01 Andrew Pinski <apinski@cavium.com>
6400 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6401 BIT_INSERT_EXPR's operand 1
6402 to see if the types precision matches.
6404 2017-08-01 Martin Liska <mliska@suse.cz>
6407 * builtins.c (expand_builtin_memcpy_args): Remove.
6408 (expand_builtin_memcpy): Call newly added function
6409 expand_builtin_memory_copy_args.
6410 (expand_builtin_memcpy_with_bounds): Likewise.
6411 (expand_builtin_mempcpy): Remove last argument.
6412 (expand_builtin_mempcpy_with_bounds): Likewise.
6413 (expand_builtin_memory_copy_args): New function created from
6414 expand_builtin_mempcpy_args with small modifications.
6415 (expand_builtin_mempcpy_args): Remove.
6416 (expand_builtin_stpcpy): Remove unused argument.
6417 (expand_builtin): Likewise.
6418 (expand_builtin_with_bounds): Likewise.
6420 2017-08-01 Martin Liska <mliska@suse.cz>
6423 Make mempcpy more optimal (PR middle-end/70140).
6425 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6428 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6429 __builtin_vec_cmpne verify both arguments are compatible vectors
6430 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6431 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6432 move computation of aligned to after checking the argument types.
6436 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6437 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6440 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6441 Doug Rupp <rupp@adacore.com>
6442 Olivier Hainque <hainque@adacore.com>
6444 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6445 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6447 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6448 for TARGET_OS_CPP_BUILTIN.
6449 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6450 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6452 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6453 passing required abi options to the assembler for EABI configurations.
6454 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6455 of .text.hot and .text.unlikely sections for kernel modules when
6456 using ARM style exceptions.
6457 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6458 options. Add EXTRA_CC1_SPEC.
6459 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6461 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6463 (ARM_TARGET2_DWARF_FORMAT): Define.
6464 * config/arm/t-vxworks: Adjust multilib control to removal of the
6465 Diab command line options.
6467 2017-08-01 Martin Liska <mliska@suse.cz>
6469 PR gcov-profile/81561
6470 * gcov.c (unblock): Make unblocking safe as we need to preserve
6471 index correspondence of blocks and block_lists.
6473 2017-08-01 Richard Biener <rguenther@suse.de>
6475 PR tree-optimization/81181
6476 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6477 (compute_antic): ... end of iteration here.
6479 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6481 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6482 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6483 (ftree-slp-vectorize): Likewise.
6484 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6485 can no longer be set independent of flag_tree_loop_vectorize.
6486 * omp-general.c (emp_max_vf): Likewise.
6487 * opts.c (enable_fdo_optimizations): Remove references to
6488 flag_tree_vectorize, these are now implicit.
6489 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6490 and leave it for the options machinery.
6492 2017-08-01 Martin Liska <mliska@suse.cz>
6495 * builtins.c (expand_builtin_memcpy_args): Remove.
6496 (expand_builtin_memcpy): Call newly added function
6497 expand_builtin_memory_copy_args.
6498 (expand_builtin_memcpy_with_bounds): Likewise.
6499 (expand_builtin_mempcpy): Remove last argument.
6500 (expand_builtin_mempcpy_with_bounds): Likewise.
6501 (expand_builtin_memory_copy_args): New function created from
6502 expand_builtin_mempcpy_args with small modifications.
6503 (expand_builtin_mempcpy_args): Remove.
6504 (expand_builtin_stpcpy): Remove unused argument.
6505 (expand_builtin): Likewise.
6506 (expand_builtin_with_bounds): Likewise.
6508 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6511 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6512 print "ds:" only for immediates in generic address space.
6514 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6517 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6518 (ix86_function_ok_for_sibcall): Return false for naked functions.
6520 2017-08-01 Richard Biener <rguenther@suse.de>
6522 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6523 (compute_antic): Seed worklist with exit block predecessors.
6524 * cfganal.c (dfs_find_deadend): For a cycle return the source
6525 of the edge closing it.
6527 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6529 * config/aarch64/aarch64.c
6530 (aarch64_can_const_movi_rtx_p): Move 0 check.
6532 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6534 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6535 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6537 * match.pd: Ditto in address comparison pattern.
6539 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6541 PR tree-optimization/81627
6542 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6543 closed ssa form for store-store chain.
6545 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6547 PR tree-optimization/81620
6548 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6549 for store-store chain.
6551 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6553 PR tree-optimization/81588
6554 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6555 ranges[i].in_p, invert comparison code ccode. For >/>=,
6556 swap rhs1 and rhs2 and comparison code unconditionally,
6557 for </<= don't do that. Don't swap rhs1/rhs2 again if
6558 ranges[i].in_p, instead invert comparison code ccode if
6559 opcode or oe->rank is BIT_IOR_EXPR.
6562 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6563 a direct optab to conversion optab.
6564 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6565 with GET_MODE_INNER as last argument instead of optab_handler.
6566 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6567 vector extraction if possible and optab is available.
6568 * expr.c (store_constructor): Use convert_optab_handler instead
6569 of optab_handler. Use vector initialization from smaller
6570 vectors if possible and optab is available.
6571 * tree-vect-stmts.c (vectorizable_load): Likewise.
6572 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6574 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6575 of vec_init from half-sized vectors with the same element mode.
6576 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6577 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6578 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6579 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6580 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6581 after mode in gen_vec_extract* calls.
6582 (vec_extract<mode>): Renamed to ...
6583 (vec_extract<mode><ssescalarmodelower>): ... this.
6584 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6585 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6586 element mode after mode in gen_vec_init* calls.
6587 (VEC_INIT_HALF_MODE): New mode iterator.
6588 (vec_init<mode>): Renamed to ...
6589 (vec_init<mode><ssescalarmodelower>): ... this.
6590 (vec_init<mode><ssehalfvecmodelower>): New expander.
6591 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6592 (vec_extractv2sfsf): ... this.
6593 (vec_initv2sf): Renamed to ...
6594 (vec_initv2sfsf): ... this.
6595 (vec_extractv2si): Renamed to ...
6596 (vec_extractv2sisi): ... this.
6597 (vec_initv2si): Renamed to ...
6598 (vec_initv2sisi): ... this.
6599 (vec_extractv4hi): Renamed to ...
6600 (vec_extractv4hihi): ... this.
6601 (vec_initv4hi): Renamed to ...
6602 (vec_initv4hihi): ... this.
6603 (vec_extractv8qi): Renamed to ...
6604 (vec_extractv8qiqi): ... this.
6605 (vec_initv8qi): Renamed to ...
6606 (vec_initv8qiqi): ... this.
6607 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6608 (vec_init<mode>): Renamed to ...
6609 (vec_init<mode><VEC_base_l>): ... this.
6610 (vec_extract<mode>): Renamed to ...
6611 (vec_extract<mode><VEC_base_l>): ... this.
6612 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6613 (vec_initv2sfsf): ... this.
6614 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6615 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6616 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6617 element mode after mode in gen_vec_init* calls.
6618 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6619 (vec_init<mode><Vel>): ... this.
6620 (vec_extract<mode>): Renamed to ...
6621 (vec_extract<mode><Vel>): ... this.
6622 * config/aarch64/iterators.md (Vel): New mode attribute.
6623 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6624 Add element mode after mode in gen_vec_extract* calls.
6625 * config/s390/vector.md (non_vec_l): New mode attribute.
6626 (vec_extract<mode>): Renamed to ...
6627 (vec_extract<mode><non_vec_l>): ... this.
6628 (vec_init<mode>): Renamed to ...
6629 (vec_init<mode><non_vec_l>): ... this.
6630 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6631 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6633 * config/arm/iterators.md (V_elem_l): New mode attribute.
6634 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6635 (vec_extract<mode><V_elem_l>): ... this.
6636 (vec_extractv2di): Renamed to ...
6637 (vec_extractv2didi): ... this.
6638 (vec_init<mode>): Renamed to ...
6639 (vec_init<mode><V_elem_l>): ... this.
6640 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6641 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6642 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6643 Add element mode after gen_vec_extract* calls.
6644 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6645 (vec_init<mode><unitmode>): ... this.
6646 (vec_extract<mode>): Renamed to ...
6647 (vec_extract<mode><unitmode>): ... this.
6648 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6649 (vec_init<mode><unitmode>): ... this.
6650 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6651 (vec_initv2sfsf): ... this.
6652 (vec_extractv2sf): Renamed to ...
6653 (vec_extractv2sfsf): ... this.
6654 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6655 Add element mode after gen_vec_extract* calls.
6656 * config/mips/mips.md (unitmode): New mode iterator.
6657 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6658 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6659 * config/spu/spu.md (inner_l): New mode attribute.
6660 (vec_init<mode>): Renamed to ...
6661 (vec_init<mode><inner_l>): ... this.
6662 (vec_extract<mode>): Renamed to ...
6663 (vec_extract<mode><inner_l>): ... this.
6664 * config/sparc/sparc.md (veltmode): New mode iterator.
6665 (vec_init<VMALL:mode>): Renamed to ...
6666 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6667 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6668 (vec_initv2sisi): ... this.
6669 (vec_initv2sf): Renamed to ...
6670 (vec_initv2sfsf): ... this.
6671 (vec_extractv2sf): Renamed to ...
6672 (vec_extractv2sfsf): ... this.
6673 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6674 (vec_init<mode>): Renamed to ...
6675 (vec_init<mode><VEC_base_l>): ... this.
6676 (vec_extract<mode>): Renamed to ...
6677 (vec_extract<mode><VEC_base_l>): ... this.
6678 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6679 (vec_initv2sfsf): ... this.
6680 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6681 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6682 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6683 gen_vec_init* calls.
6685 2017-08-01 Richard Biener <rguenther@suse.de>
6687 PR tree-optimization/81297
6688 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6689 TREE_OVERFLOW from INTEGER_CSTs.
6691 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6693 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6695 2017-07-31 Carl Love <cel@us.ibm.com>
6697 * config/rs6000/rs6000-c: Add support for built-in functions
6698 vector signed char vec_xl_be (signed long long, signed char *);
6699 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6700 vector signed int vec_xl_be (signed long long, signed int *);
6701 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6702 vector signed long long vec_xl_be (signed long long, signed long long *);
6703 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6704 vector signed short vec_xl_be (signed long long, signed short *);
6705 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6706 vector double vec_xl_be (signed long long, double *);
6707 vector float vec_xl_be (signed long long, float *);
6708 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6709 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6710 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6712 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6713 (altivec_expand_builtin): Add switch statement to call
6714 altivec_expand_xl_be for each builtin.
6715 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6716 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6717 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6718 __builtin_vsx_le_be_v16qi.
6719 * doc/extend.texi: Update the built-in documentation file for the
6720 new built-in functions.
6722 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6725 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6727 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6729 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6731 * config.gcc: Add z14.
6732 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6733 CPU model numbers for z13s and z14.
6734 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6736 * config/s390/s390-opts.h (enum processor_type): Rename
6737 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6738 * config/s390/s390.c (processor_table): Add field for CPU name to
6739 be passed to Binutils.
6740 (s390_asm_output_machine_for_arch): Use the new field in
6741 processor_table for Binutils.
6742 (s390_expand_builtin): Replace arch12 with z14.
6743 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6744 (s390_get_sched_attrmask): Likewise.
6745 (s390_get_unit_mask): Likewise.
6746 * config/s390/s390.opt: Add z14 to processor_type enum.
6748 2017-07-31 Martin Jambor <mjambor@suse.cz>
6751 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6752 regardless of optimization level.
6754 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6755 Martin Liska <mliska@suse.cz>
6757 * predict.def: Remove old comment and adjust probability.
6758 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6761 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6764 * config/i386/i386.c (ix86_function_naked): New function.
6765 (ix86_can_use_return_insn_p): Return false for naked functions.
6766 (ix86_expand_prologue): Skip prologue for naked functions.
6767 (ix86_expand_epilogue): Skip epilogue for naked functions
6768 and emit trap instruction.
6769 (ix86_warn_func_return): New function.
6770 (ix86_attribute_table): Add "naked" attribute specification.
6771 (TARGET_WARN_FUNC_RETURN): Define.
6772 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6774 2017-07-31 Martin Liska <mliska@suse.cz>
6776 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6777 (dump_gimple_bb_header): Always dump BB info.
6778 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6780 2017-07-31 Martin Liska <mliska@suse.cz>
6783 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6784 also with current_function_decl non-null equality.
6786 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6789 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6790 change type to the element type, instead add eltype variable and
6791 use it where we are interested in the element type.
6793 PR tree-optimization/81603
6794 * ipa-polymorphic-call.c
6795 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6796 offset arithmetic in offset_int, bail out if the resulting bit offset
6797 doesn't fit into shwi.
6799 2017-07-31 Martin Liska <mliska@suse.cz>
6801 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6802 (gimplify_save_expr): Fix comment.
6804 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6807 * config/i386/i386.c (ix86_function_arg): Update arguments for
6809 (ix86_compute_frame_layout): Set the initial stack offset to
6810 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6811 INCOMING_FRAME_SP_OFFSET.
6812 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6813 stack before exception handler returns.
6814 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6815 the 'ERROR_CODE' for exception handler.
6817 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6819 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6820 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6821 (ASM_OUTPUT_REG_POP): Ditto.
6822 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6823 instead of asm_fprintf to output pure string.
6825 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6827 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6828 to imported_module_or_decl hook.
6829 (debug_nothing_tree_tree_tree_bool): Remove.
6830 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6831 * debug.c (do_nothing_debug_hooks): Use
6832 debug_nothing_tree_tree_tree_bool_bool instead of
6833 debug_nothing_tree_tree_tree_bool.
6834 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6835 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6836 * sdbout.c (sdb_debug_hooks): Likewise.
6837 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6838 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6840 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6841 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6842 attribute, don't add anything.
6844 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6846 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6847 (fold_build2_stat_loc): Likewise.
6848 (fold_build3_stat_loc): Likewise.
6849 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6850 (fold_build1_loc): Remove macro.
6851 (fold_build2_loc): Likewise.
6852 (fold_build3_loc): Likewise.
6854 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6856 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6857 (gimple_build_debug_bind_source_stat): Likewise.
6858 * gimple.h (gimple_build_debug_bind): Remove macro.
6859 (gimple_build_debug_bind_source): Likewise.
6861 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6863 * bitmap.c (bitmap_alloc): Adjust.
6864 (bitmap_gc_alloc): Likewise.
6865 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6869 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6870 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6871 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6872 (bitmap_gc_alloc_stat): Likewise.
6873 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6875 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6877 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6878 * rtl.h (shallow_copy_rtx): Remove macro.
6880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6882 * emit-rtl.c (gen_raw_REG): Adjust.
6883 * gengenrtl.c (gendef): Likewise.
6884 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6885 * rtl.h (rtx_alloc): Remove macro.
6887 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6889 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6890 (build_tree_list_stat): Likewise.
6891 * tree.h (build_tree_list): Remove macro.
6892 (build_tree_list_vec): Likewise.
6894 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6896 * tree.c (make_vector_stat): Remove _stat from name.
6897 (build_vector_stat): Likewise.
6898 * tree.h (make_vector_stat): Remove macro.
6899 (build_vector_stat): Likewise.
6901 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6903 * tree.h (build_var_debug_value): Remove prototype.
6905 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6907 * tree.c (tree_cons_stat): Remove _stat from name.
6908 * tree.h (tree_cons): Remove macro.
6910 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6912 * tree.c (build_vl_exp_stat): Remove _stat from name.
6913 * tree.h (build_vl_exp): Remove macro.
6915 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6917 * tree.c (build_decl_stat): Remove _stat from name.
6918 * tree.h (build_decl): Remove macro.
6920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6922 * gimple.c (gimple_build_with_ops_stat): Adjust.
6923 (gimple_alloc_stat): Remove _stat from name.
6924 * gimple.h (gimple_alloc): Remove macro.
6926 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6928 * tree.c (make_tree_vec_stat): Remove _stat from name.
6929 (grow_tree_vec_stat): Likewise.
6930 * tree.h (make_tree_vec_stat): Adjust prototype.
6931 (grow_tree_vec_stat): Likewise.
6932 (make_tree_vec): Remove macro.
6933 (grow_tree_vec): Likewise.
6935 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6937 * fold-const.c (fold_build1_stat_loc): Adjust.
6938 (fold_build2_stat_loc): Likewise.
6939 (fold_build3_stat_loc): Likewise.
6940 * tree.c (build0_stat): Remove _stat from name.
6941 (build1_stat): Likewise.
6942 (build2_stat): Likewise.
6943 (build3_stat): Likewise.
6944 (build4_stat): Likewise.
6945 (build5_stat): Likewise.
6946 * tree.h (build1_loc): Remove macro, and rename _stat function
6948 (build2_loc): Likewise.
6949 (build3_loc): Likewise.
6950 (build4_loc): Likewise.
6951 (build5_loc): Likewise.
6953 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6955 * tree.c (make_int_cst_stat): Remove _stat from name.
6956 * tree.h (make_int_cst_stat): Adjust prototype.
6957 (make_int_cst): Remove macro.
6959 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6961 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6962 * tree.h (make_tree_binfo_stat): Adjust prototype.
6963 (make_tree_binfo): Remove.
6965 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6967 * tree.c (copy_node_stat): Rename to copy_node.
6968 (build_distinct_type_copy): Adjust.
6969 * tree.h (copy_node_stat): Adjust prototype.
6970 (copy_node): Remove macro.
6972 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6974 * tree.c (make_node_stat): rename to make_node.
6975 (build_tree_list_stat): Adjust.
6976 (build0_stat): Likewise.
6977 (build2_stat): Likewise.
6978 (build3_stat): Likewise.
6979 (build4_stat): Likewise.
6980 (build5_stat): Likewise.
6981 (build_decl_stat): Likewise.
6982 * tree.h (make_node_stat): Adjust prototype.
6983 (make_node): remove macro.
6985 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6987 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6988 (PPC_FEATURE2_SCV): Likewise.
6989 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6991 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6993 * config/aarch64/aarch64.c
6994 (aarch64_internal_mov_immediate): Add new special pattern.
6995 * config/aarch64/aarch64.md (*movdi_aarch64):
6996 Add reg/32bit const mov case.
6998 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6999 Richard Sandiford <richard.sandiford@linaro.org>
7001 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7002 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7003 Add integer and movi cases.
7004 (movi-split-hf-df-sf split, fp16): New.
7005 (enabled): Added TARGET_FP_F16INST.
7006 * config/aarch64/iterators.md (GPF_HF): New.
7007 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7009 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7011 * config/aarch64/aarch64.c
7012 (aarch64_simd_container_mode): Add prototype.
7013 (aarch64_expand_mov_immediate): Add HI support.
7014 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7015 (aarch64_can_const_movi_rtx_p): New.
7016 (aarch64_preferred_reload_class):
7017 Remove restrictions of using FP registers for certain SIMD operations.
7018 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7019 (aarch64_valid_floating_const): Add integer move validation.
7020 (aarch64_simd_imm_scalar_p): Remove.
7021 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7022 (aarch64_legitimate_constant_p): Expand list of supported cases.
7023 * config/aarch64/aarch64-protos.h
7024 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7025 (aarch64_reinterpret_float_as_int): New.
7026 (aarch64_simd_imm_scalar_p): Remove.
7027 * config/aarch64/constraints.md (Uvi): New.
7028 (Dd): Split into Ds and new Dd.
7029 * config/aarch64/aarch64.md (*movsi_aarch64):
7031 (*movdi_aarch64): Add SIMD mov case.
7033 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7035 * tree-predcom.c: (struct chain): Handle store-store chain in which
7036 stores for elimination only store loop invariant values.
7037 (execute_pred_commoning_chain): Ditto.
7038 (prepare_initializers_chain_store_elim): Ditto.
7039 (prepare_finalizers): Ditto.
7040 (is_inv_store_elimination_chain): New function.
7041 (initialize_root_vars_store_elim_1): New function.
7043 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7045 * tree-predcom.c: Revise general description of the pass.
7046 (enum chain_type): New enum type for store elimination.
7047 (struct chain): New field supporting store elimination.
7048 (struct component): Ditto.
7049 (dump_chain): Dump store-stores chain.
7050 (release_chain): Release resources.
7051 (split_data_refs_to_components): Compute and create component
7052 contains only stores for elimination.
7053 (get_chain_last_ref_at): New function.
7054 (make_invariant_chain): Initialization.
7055 (make_rooted_chain): Specify chain type in parameter and record it.
7056 (add_looparound_copies): Skip for store-stores chain.
7057 (determine_roots_comp): Compute type of chain and pass it to
7059 (initialize_root_vars_store_elim_2): New function.
7060 (finalize_eliminated_stores): New function.
7061 (remove_stmt): Handle store for elimination.
7062 (execute_pred_commoning_chain): Execute predictive commoning on
7064 (determine_unroll_factor): Skip unroll for store-stores chain.
7065 (prepare_initializers_chain_store_elim): New function.
7066 (prepare_initializers_chain): Hanlde store-store chain.
7067 (prepare_finalizers_chain, prepare_finalizers): New function.
7068 (tree_predictive_commoning_loop): Return integer value indicating
7069 if loop is unrolled or lcssa form is corrupted.
7070 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7072 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7074 * tree-predcom.c (initialize_root): Delete.
7075 (execute_pred_commoning_chain): Initialize root vars and replace
7076 reference of non-combined chain directly, rather than call above
7079 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7081 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7082 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7084 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7086 * tree-predcom.c (struct chain): New field init_seq.
7087 (release_chain): Release init_seq.
7088 (prepare_initializers_chain): Record intialization stmts in above
7090 (insert_init_seqs): New function.
7091 (tree_predictive_commoning_loop): Call insert_init_seqs.
7093 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7095 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7097 2017-07-28 Richard Biener <rguenther@suse.de>
7099 * match.pd: Remove superfluous :c.
7100 * genmatch.c (simplify::id): Add member.
7101 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7103 (current_id): New global.
7104 (dt_node::parent): Move from ...
7105 (dt_operand::parent): ... here. Add for_id member.
7106 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7107 (decision_tree::find_node): Relax order requirement when
7108 merging DT_TRUE nodes to ones inbetween the current simplify
7109 and the one we try to merge with. Add diagnostic whenever
7110 we need to enforce pattern order by not merging.
7111 (decision_tree::insert): Set current_id.
7112 (decision_tree::print_node): Dump parent node and for_id.
7113 (parser::last_id): Add member.
7114 (parser::push_simplify): Assign unique id.
7115 (parser::parser): Initialize last_id.
7117 2017-07-28 Martin Liska <mliska@suse.cz>
7120 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7121 gimple_build_debug_bind.
7123 2017-07-28 Richard Biener <rguenther@suse.de>
7125 PR tree-optimization/81502
7126 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7128 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7130 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7131 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7132 for BIT_FIELD_REF args.
7133 * fold-const.c (make_bit_field_ref): Likewise.
7134 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7136 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7139 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7140 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7141 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7142 Or it into SANITIZER_UNDEFINED.
7143 * ubsan.c: Include gimple-fold.h and varasm.h.
7144 (ubsan_expand_ptr_ifn): New function.
7145 (instrument_pointer_overflow): New function.
7146 (maybe_instrument_pointer_overflow): New function.
7147 (instrument_object_size): Formatting fix.
7148 (pass_ubsan::execute): Call instrument_pointer_overflow
7149 and maybe_instrument_pointer_overflow.
7150 * internal-fn.c (expand_UBSAN_PTR): New function.
7151 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7152 * sanitizer.def (__ubsan_handle_pointer_overflow,
7153 __ubsan_handle_pointer_overflow_abort): New builtins.
7154 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7155 * internal-fn.def (UBSAN_PTR): New internal function.
7156 * opts.c (sanitizer_opts): Add pointer-overflow.
7157 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7158 * fold-const.c (build_range_check): Compute pointer range check in
7159 integral type if pointer arithmetics would be needed. Formatting
7162 2017-07-28 Martin Liska <mliska@suse.cz>
7165 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7166 parameters that are of a variable-length.
7168 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7170 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7172 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7173 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7174 (CRT_CALL_STATIC_FUNCTION): Likewise.
7175 (ASM_DEFAULT_SPEC): New define.
7176 (ASM_SPEC32): Likewise.
7177 (ASM_SPEC64): Likewise.
7178 (ASM_SPEC_COMMON): Likewise.
7179 (ASM_SPEC): Likewise.
7180 (INVALID_64BIT): Likewise.
7181 (LINK_OS_DEFAULT_SPEC): Likewise.
7182 (LINK_OS_SPEC32): Likewise.
7183 (LINK_OS_SPEC64): Likewise.
7184 (POWERPC_LINUX): Likewise.
7185 (PTRDIFF_TYPE): Likewise.
7186 (RESTORE_FP_PREFIX): Likewise.
7187 (RESTORE_FP_SUFFIX): Likewise.
7188 (SAVE_FP_PREFIX): Likewise.
7189 (SAVE_FP_SUFFIX): Likewise.
7190 (SIZE_TYPE): Likewise.
7191 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7192 (TARGET_64BIT): Likewise.
7193 (TARGET_64BIT): Likewise.
7194 (TARGET_AIX): Likewise.
7195 (WCHAR_TYPE_SIZE): Likewise.
7196 (WCHAR_TYPE): Undefine.
7197 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7198 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7199 (CPP_OS_RTEMS_SPEC): Delete.
7200 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7201 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7203 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7205 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7207 PR tree-optimization/81578
7208 * tree-parloops.c (build_new_reduction): Bail out if
7209 reduction_code isn't one of the standard OpenMP reductions.
7210 Move the details printing after that decision.
7212 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7214 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7215 related to reload_in_progress.
7216 (splat_input_operand): Likewise.
7217 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7219 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7221 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7222 (TARGET_INSTANTIATE_DECLS): Likewise.
7223 (legitimate_indexed_address_p): Delete reload_in_progress code.
7224 (rs6000_debug_legitimate_address_p): Likewise.
7225 (rs6000_eliminate_indexed_memrefs): Likewise.
7226 (rs6000_emit_le_vsx_store): Likewise.
7227 (rs6000_emit_move_si_sf_subreg): Likewise.
7228 (rs6000_emit_move): Likewise.
7229 (register_to_reg_type): Likewise.
7230 (rs6000_pre_atomic_barrier): Likewise.
7231 (rs6000_machopic_legitimize_pic_address): Likewise.
7232 (rs6000_allocate_stack_temp): Likewise.
7233 (rs6000_address_for_fpconvert): Likewise.
7234 (rs6000_address_for_altivec): Likewise.
7235 (rs6000_secondary_memory_needed_rtx): Delete function.
7236 (rs6000_check_sdmode): Likewise.
7237 (rs6000_alloc_sdmode_stack_slot): Likewise.
7238 (rs6000_instantiate_decls): Likewise.
7239 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7240 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7241 Delete reload_in_progress.
7242 (*vec_reload_and_plus_<mptrsize>): Likewise.
7243 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7244 (vsx_div_v2di): Likewise.
7245 (vsx_udiv_v2di): Likewise.
7247 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7249 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7250 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7251 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7252 (rs6000_debug_reg_global): Delete print of LRA status.
7253 (rs6000_option_override_internal): Delete dead LRA related code.
7254 (rs6000_lra_p): Delete function.
7255 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7257 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7259 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7260 * config/riscv/rtems.h: New file.
7262 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7263 Sudakshina Das <sudi.das@arm.com>
7265 * config/aarch64/aarch64.md
7266 (define_split for and<mode>3nr_compare): Move
7267 non aarch64_logical_operand to a register.
7268 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7269 register immediate operand to a register.
7270 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7272 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7275 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7277 2017-07-27 Richard Biener <rguenther@suse.de>
7279 PR tree-optimization/81573
7280 PR tree-optimization/81494
7281 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7282 multi defuse cycle case.
7284 2017-07-27 Richard Biener <rguenther@suse.de>
7286 PR tree-optimization/81571
7287 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7290 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7292 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7293 earlier and only if MASK_FPU is set. Adjust formatting.
7295 2017-07-27 Martin Liska <mliska@suse.cz>
7297 * opt-functions.awk: Add validation of value of Init.
7298 * optc-gen.awk: Pass new argument.
7300 2017-07-27 Martin Liska <mliska@suse.cz>
7302 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7303 Fix wrong condition.
7305 2017-07-27 Martin Liska <mliska@suse.cz>
7307 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7308 BBs and edges seen by autoFDO.
7310 2017-07-27 Richard Biener <rguenther@suse.de>
7312 PR tree-optimization/81502
7313 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7314 with incompatible but same sized type.
7315 (execute_update_addresses_taken): Likewise.
7317 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7319 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7320 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7322 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7325 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7326 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7327 Change s_operand to memory_operand.
7329 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7331 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7332 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7333 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7334 Emit instructions rather than returning an expression. Handle TFmode
7335 and KFmode by casting to TImode.
7336 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7337 (rs6000_emit_le_vsx_store): Likewise.
7338 * config/rs6000/vsx.md (VSX_TI): New iterator.
7339 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7340 (*vsx_le_undo_permute_<mode>): Likewise.
7341 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7342 emit the split sequence.
7343 (*vsx_le_perm_store_<mode>): Likewise.
7345 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7347 PR tree-optimization/81555
7348 PR tree-optimization/81556
7349 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7350 if true, force CHANGED for the recursive invocation.
7351 (reassociate_bb): Remember original length of ops array, pass
7352 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7354 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7355 attributes for noipa attribute. For naked attribute use
7356 lookup_attribute first before lookup_attribute_spec.
7357 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7359 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7360 for functions with noipa attribute.
7361 (cgraph_externally_visible_p): Return true for functions with noipa
7363 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7364 for functions with noipa attribute.
7365 * doc/extend.texi: Document noipa function attribute.
7366 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7367 also for functions with noipa attribute.
7368 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7370 2017-07-26 Andrew Pinski <apinski@cavium.com>
7372 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7373 vec_unalign_load_cost and vec_unalign_store_cost.
7375 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7377 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7378 -mvsx-small-integer option.
7379 (ISA_3_0_MASKS_IEEE): Likewise.
7380 (OTHER_VSX_VECTOR_MASKS): Likewise.
7381 (POWERPC_MASKS): Likewise.
7382 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7383 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7384 code, only testing for DImode being allowed in non-VSX floating
7386 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7387 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7389 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7390 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7391 TARGET_P8_VECTOR test.
7392 (rs6000_secondary_reload_simple_move): Likewise.
7393 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7394 since TARGET_P9_VECTOR was already tested.
7395 (rs6000_opt_masks): Remove -mvsx-small-integer.
7396 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7397 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7399 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7400 test for TARGET_VEXTRACTUB was used, and that uses
7402 (p9 extract splitter): Likewise.
7403 (vsx_extract_<mode>_di_p9): Likewise.
7404 (vsx_extract_<mode>_store_p9): Likewise.
7405 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7406 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7407 the elimination of TARGET_VSX_SMALL_INTEGER.
7408 (vsx_extract_<mode>_p8): Likewise.
7409 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7410 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7411 (vsx_set_<mode>_p9): Likewise.
7412 (vsx_set_v4sf_p9): Likewise.
7413 (vsx_set_v4sf_p9_zero): Likewise.
7414 (vsx_insert_extract_v4sf_p9): Likewise.
7415 (vsx_insert_extract_v4sf_p9_2): Likewise.
7416 * config/rs6000/rs6000.md (sign extend splitter): Change
7417 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7418 (floatsi<mode>2_lfiwax_mem): Likewise.
7419 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7420 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7421 since a test for TARGET_P9_VECTOR was used.
7422 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7423 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7424 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7425 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7426 TARGET_P8_VECTOR test.
7427 (fix_trunc<mode>si2_stfiwx): Likewise.
7428 (fix_trunc<mode>si2_internal): Likewise.
7429 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7430 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7432 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7433 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7434 TARGET_P8_VECTOR test.
7435 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7436 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7437 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7439 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7440 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7441 since a test for TARGET_P9_VECTOR was used.
7442 (splitter for loading small constants): Likewise.
7444 2017-07-26 Andrew Pinski <apinski@cavium.com>
7446 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7449 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7452 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7453 (fp_valid_at): Likewise.
7455 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7457 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7458 (qdf24xx_addrcost_table): Likewise.
7459 (cortexa57_tunings): Update to use generic_branch_cost.
7460 (cortexa72_tunings): Likewise.
7461 (cortexa73_tunings): Likewise.
7462 (qdf24xx_tunings): Likewise.
7464 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7466 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7467 (thunderx2t99_branch_cost): Likewise.
7468 (cortexa35_tunings): Update to use generic_branch_cost.
7469 (cortexa53_tunings): Likewise.
7470 (cortexa57_tunings): Likewise.
7471 (cortexa72_tunings): Likewise.
7472 (cortexa73_tunings): Likewise.
7473 (thunderx2t99_tunings): Likewise.
7475 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7477 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7478 (sparc_option_override): Honour MASK_FSMULD.
7479 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7480 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7481 * config/sparc/sparc.opt (mfsmuld): New option.
7482 * doc/invoke.texi (mfsmuld): Document option.
7484 2017-07-26 Marek Polacek <polacek@redhat.com>
7487 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7489 2017-07-26 Richard Biener <rguenther@suse.de>
7491 * gimple-match-head.c (do_valueize): Return OP if valueize
7493 (get_def): New helper to get at the def stmt of a SSA name
7495 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7496 do_valueize to get at the def stmt.
7497 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7499 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7502 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7504 2017-07-26 Martin Liska <mliska@suse.cz>
7507 * function.c (expand_function_start): Make expansion of
7508 nonlocal_goto_save_area after parm_birth_insn.
7510 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7512 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7513 from all CPU target flags enable members.
7515 2017-07-26 Richard Biener <rguenther@suse.de>
7517 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7518 (decision_tree::gen): Make 'type' const.
7519 (write_predicate): Likewise.
7521 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7523 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7524 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7525 (rs6000_option_override_internal): Likewise.
7526 (rs6000_expand_vector_set): Likewise.
7527 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7528 (TARGET_UPPER_REGS_SF): Likewise.
7529 (TARGET_UPPER_REGS_DI): Likewise.
7530 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7531 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7532 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7533 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7534 (Splitters for DI constants in Altivec registers): Likewise.
7535 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7536 (vsx_set_v4sf_p9): Likewise.
7537 (vsx_set_v4sf_p9_zero): Likewise.
7538 (vsx_insert_extract_v4sf_p9): Likewise.
7539 (vsx_insert_extract_v4sf_p9_2): Likewise.
7541 2017-07-25 Carl Love <cel@us.ibm.com>
7543 * doc/extend.texi: Update the built-in documentation file for the
7544 existing built-in functions
7545 vector signed char vec_cnttz (vector signed char);
7546 vector unsigned char vec_cnttz (vector unsigned char);
7547 vector signed short vec_cnttz (vector signed short);
7548 vector unsigned short vec_cnttz (vector unsigned short);
7549 vector signed int vec_cnttz (vector signed int);
7550 vector unsigned int vec_cnttz (vector unsigned int);
7551 vector signed long long vec_cnttz (vector signed long long);
7552 vector unsigned long long vec_cnttz (vector unsigned long long);
7554 2017-07-25 Andrew Pinski <apinski@cavium.com>
7556 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7557 accesses where the use is for the first operand of a BIT_INSERT.
7559 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7562 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7563 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7565 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7567 * config/i386/gstabs.h: Delete.
7568 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7570 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7572 * config/i386/i386.c (ix86_decompose_address): Do not check for
7573 register RTX when looking at index_reg or base_reg.
7574 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7576 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7578 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7579 to update EH info here.
7581 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7583 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7585 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7587 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7589 2017-07-25 Torsten Duwe <duwe@suse.de>
7591 * common.opt: Introduce -fpatchable-function-entry
7592 command line option, and its variables function_entry_patch_area_size
7593 and function_entry_patch_area_start.
7594 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7595 including a two-value parser.
7596 * target.def (print_patchable_function_entry): New target hook.
7597 * targhooks.h (default_print_patchable_function_entry): New function.
7598 * targhooks.c (default_print_patchable_function_entry): Likewise.
7599 * toplev.c (process_options): Switch off IPA-RA if
7600 patchable function entries are being generated.
7601 * varasm.c (assemble_start_function): Look at the
7602 patchable-function-entry command line switch and current
7603 function attributes and maybe generate NOP instructions by
7604 calling the print_patchable_function_entry hook.
7605 * doc/extend.texi: Document patchable_function_entry attribute.
7606 * doc/invoke.texi: Document -fpatchable_function_entry
7607 command line option.
7608 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7610 * doc/tm.texi: Re-generate.
7612 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7615 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7616 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7618 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7620 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7622 2017-07-25 Richard Biener <rguenther@suse.de>
7624 PR tree-optimization/81455
7625 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7626 not walk in cycles when looking for guards.
7628 2017-07-25 Richard Biener <rguenther@suse.de>
7630 PR tree-optimization/81529
7631 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7632 when optimizing backedge uses.
7634 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7636 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7638 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7639 to dl_section_ref. On AIX, append an expression to subtract
7640 the size of the section length to dl_section_ref.
7642 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7644 * configure.ac: If any of the config.* scripts fail, exit 1.
7645 * configure: Regenerate.
7647 2017-07-25 Richard Biener <rguenther@suse.de>
7650 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7651 of immediate uses, be more verbose on errors.
7653 2017-07-25 Richard Biener <rguenther@suse.de>
7655 PR tree-optimization/81510
7656 * tree-vect-loop.c (vect_is_simple_reduction): When the
7657 reduction stmt is not inside the loop bail out.
7659 2017-07-25 Richard Biener <rguenther@suse.de>
7661 PR tree-optimization/81303
7662 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7663 profitability check against LOOP_VINFO_NITERSM1.
7665 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7667 * domwalk.c (cmp_bb_postorder): Simplify.
7668 (sort_bbs_postorder): New function. Use it...
7669 (dom_walker::walk): ...here to optimize common cases.
7671 2017-07-25 Martin Liska <mliska@suse.cz>
7674 * ipa-visibility.c (function_and_variable_visibility): Make the
7675 redirection just on target that supports aliasing.
7676 Fix GNU coding style.
7678 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7681 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7683 * config/arm/rtems.h: Likewise.
7684 * config/bfin/rtems.h: Likewise.
7685 * config/i386/rtemself.h: Likewise.
7686 * config/lm32/rtems.h: Likewise.
7687 * config/m32c/rtems.h: Likewise.
7688 * config/m68k/rtemself.h: Likewise.
7689 * config/microblaze/rtems.h: Likewise.
7690 * config/mips/rtems.h: Likewise.
7691 * config/moxie/rtems.h: Likewise.
7692 * config/nios2/rtems.h: Likewise.
7693 * config/powerpcspe/rtems.h: Likewise.
7694 * config/rs6000/rtems.h: Likewise.
7695 * config/rtems.h: Likewise.
7696 * config/sh/rtems.h: Likewise.
7697 * config/sh/rtemself.h: Likewise.
7698 * config/sparc/rtemself.h: Likewise.
7700 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7703 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7704 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7705 * tree-ssa-structalias.c (alias_get_name): Same.
7707 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7710 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7711 instructions if no du chain is found.
7713 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7715 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7717 2017-07-25 Richard Biener <rguenther@suse.de>
7720 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7723 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7725 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7727 (ISA_2_7_MASKS_SERVER): Likewise.
7728 (ISA_3_0_MASKS_IEEE): Likewise.
7729 (OTHER_P8_VECTOR_MASKS): Likewise.
7730 (OTHER_VSX_VECTOR_MASKS): Likewise.
7731 (POWERPC_MASKS): Likewise.
7732 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7733 duplicate list of options.
7734 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7735 explicit -mupper-regs options.
7736 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7737 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7738 alias for -mupper-regs-df.
7739 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7740 (rs6000_init_hard_regno_mode_ok): Likewise.
7741 (rs6000_option_override_internal): Likewise.
7742 (rs6000_opt_masks): Likewise.
7743 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7744 options in terms of whether -mvsx or -mpower8-vector was used.
7745 (TARGET_UPPER_REGS_DI): Likewise.
7746 (TARGET_UPPER_REGS_SF): Likewise.
7747 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7748 -mupper-regs-* options.
7750 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7752 * passes.c (emergency_dump_function): Print some empty lines and a
7753 header before the RTL dump.
7755 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7757 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7759 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7762 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7763 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7765 2017-07-24 Carl Love <cel@us.ibm.com>
7767 * config/rs6000/rs6000-c.c: Add support for built-in functions
7768 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7769 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7770 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7771 vec_extract_fp_from_shortl): Add defines for the two builtins.
7772 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7773 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7775 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7776 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7777 * doc/extend.texi: Update the built-in documentation file for the
7778 new built-in function.
7780 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7783 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7785 * doc/generic.texi: Likewise.
7786 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7787 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7789 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7791 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7792 (aarch64_mls_elt_merge<mode>): Likewise.
7794 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7796 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7797 having __cxa_atexit.
7799 2017-07-23 Michael Collison <michael.collison@arm.com>
7801 * config/arm/arm.c (arm_option_override): Deprecate
7802 use of -mstructure-size-boundary.
7803 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7804 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7806 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7809 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7810 Reduce cost estimate for direct moves.
7812 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7815 * config/i386/i386.c (ix86_option_override_internal): Disable
7816 BMI, BMI2 and TBM instructions for -m16.
7818 2017-07-21 Carl Love <cel@us.ibm.com>
7820 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7821 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7822 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7823 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7824 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7825 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7826 VMULOSW): New enum "unspec" values.
7827 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7828 altivec_vmulosw): New patterns.
7829 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7830 VMULOSW): Add definitions.
7832 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7834 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7835 (qdf24xx): Likewise.
7836 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7837 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7838 (AARCH64_FL_V8_1): Renumber.
7839 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7840 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7841 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7842 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7843 rdma to feature modifiers list.
7845 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7848 * ipa-visibility (function_and_variable_visibility): Convert
7849 recursive PLT call to direct call if appropriate.
7851 2017-07-21 Andrew Pinski <apinski@cavium.com>
7853 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7854 operand 1 to see if the types precision matches.
7855 * fold-const.c (operand_equal_p): Likewise.
7857 2017-07-21 Richard Biener <rguenther@suse.de>
7859 PR tree-optimization/81303
7860 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7861 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7862 (vect_peeling_hash_get_lowest_cost): Adjust.
7863 (vect_enhance_data_refs_alignment): Likewise. Use
7864 vect_get_peeling_costs_all_drs to compute the penalty for no
7865 peeling to match up costs.
7867 2017-07-21 Richard Biener <rguenther@suse.de>
7869 PR tree-optimization/81500
7870 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7871 we didn't identify a reduction path.
7873 2017-07-21 Tom de Vries <tom@codesourcery.com>
7874 Cesar Philippidis <cesar@codesourcery.com>
7876 PR gcov-profile/81442
7877 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7880 2017-07-21 Tom de Vries <tom@codesourcery.com>
7883 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7885 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7886 nvptx_override_options_after_change.
7888 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7890 * dwarf2out.c (output_file_names): Avoid double testing for
7893 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7895 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7897 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7899 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7901 (try_crossjump_to_edge): Do not punt on partitioned functions.
7903 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7905 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7906 Put all BBs reachable only via paths crossing cold region to cold
7908 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7910 2016-07-21 Richard Biener <rguenther@suse.de>
7912 PR tree-optimization/81303
7913 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7914 into account prologue and epilogue iterations when raising
7915 min_profitable_iters to sth at least covering one vector iteration.
7917 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7919 * config/arm/arm.c (arm_test_cpu_arch_dat):
7922 2017-07-20 Nathan Sidwell <nathan@acm.org>
7924 Remove TYPE_METHODS.
7925 * tree.h (TYPE_METHODS): Delete.
7926 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7927 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7928 (dbxout_type_methods): Scan TYPE_FIELDS.
7929 (dbxout_type): Don't check TYPE_METHODS here.
7930 * function.c (use_register_for_decl): Always ignore register for
7931 class types when not optimizing.
7932 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7933 * tree.c (free_lang_data_in_type): Stitch out member functions and
7934 templates from TYPE_FIELDS.
7935 (build_distinct_type_copy, verify_type_variant,
7936 verify_type): Member fns are on TYPE_FIELDS.
7937 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7938 * tree-pretty-print.c (dump_generic_node): Likewise.
7940 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7943 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7944 V2TImode and V4TImode.
7945 (ix86_expand_vector_extract): Likewise.
7946 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7947 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7948 (ssescalarmode): Handle V4TImode and V2TImode.
7949 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7950 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7951 (VEXTRACTI128_MODE): New mode iterator.
7952 (splitter for *vec_extractv?ti first element): New.
7953 (VEC_INIT_MODE): New mode iterator.
7954 (vec_init<mode>): Consolidate 3 expanders into one using
7955 VEC_INIT_MODE mode iterator.
7957 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7959 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7960 non_spilled_static_chain_regno_p.
7962 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7964 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7966 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7968 * bb-reorder.c (connect_traces): Allow copying of blocks within
7971 2017-07-20 Richard Biener <rguenther@suse.de>
7973 * gimple.h (gimple_phi_result): Add gphi * overload.
7974 (gimple_phi_result_ptr): Likewise.
7975 (gimple_phi_arg): Likewise. Adjust index assert to only
7976 allow actual argument accesses rather than all slots available
7978 (gimple_phi_arg_def): Add gphi * overload.
7979 * tree-phinodes.c (make_phi_node): Initialize only actual
7981 (resize_phi_node): Clear memory not covered by old node,
7982 do not initialize excess argument slots.
7983 (reserve_phi_args_for_new_edge): Initialize new argument slot
7986 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7988 PR tree-optimization/81388
7990 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7992 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7993 by removing computation of may_be_zero.
7995 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7996 Tom de Vries <tom@codesourcery.com>
7999 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8000 when gimple level profile disagrees with what RTL expander did.
8002 2017-07-20 Richard Biener <rguenther@suse.de>
8004 PR tree-optimization/61171
8005 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8006 (vect_analyze_stmt): Add slp instance parameter.
8007 (vectorizable_reduction): Likewise.
8008 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8009 (vect_is_simple_reduction): Deal with chains not detected
8010 as SLP reduction chain, specifically not properly associated
8011 chains containing a mix of plus/minus.
8012 (get_reduction_op): Remove.
8013 (get_initial_defs_for_reduction): Simplify, pass in whether
8014 this is a reduction chain, pass in the SLP node for the PHIs.
8015 (vect_create_epilog_for_reduction): Get the SLP instance as
8017 (vectorizable_reduction): Get the SLP instance as arg.
8018 During analysis remember the SLP node with the PHIs in the
8019 instance. Simplify getting at the vectorized reduction PHIs.
8020 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8021 through SLP instance.
8022 (vect_slp_analyze_operations): Likewise.
8023 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8024 (vect_transform_stmt): Likewise.
8026 2017-07-20 Tom de Vries <tom@codesourcery.com>
8028 PR tree-optimization/81489
8029 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8030 read of phi arg location to before loop that modifies phi.
8032 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8034 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8037 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8040 * except.c (execute): Fix ordering issue.
8042 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8044 PR rtl-optimization/81423
8045 * combine.c (make_compound_operation_int): Don't try to optimize
8046 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8048 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8050 PR rtl-optimization/81423
8051 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8052 with a constant that is -1 in the truncated to mode.
8054 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8056 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8057 (determine_unlikely_bbs): ... here.
8058 * predict.h (propagate_unlikely_bbs_forward): Declare.
8059 * cfgexpand.c (pass_expand::execute): Use it.
8060 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8062 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8063 propagate_unlikely_bbs_forward.
8065 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8068 * except.c (maybe_add_nop_after_section_switch): New function.
8071 2017-07-19 Tom de Vries <tom@codesourcery.com>
8073 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8075 2017-07-19 Tom de Vries <tom@codesourcery.com>
8077 * gimple.h (gimple_phi_arg): Make assert more strict.
8079 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8081 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8082 * config/rs6000/mmintrin.h: New file.
8083 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8085 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8087 PR tree-optimization/81346
8088 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8090 2017-07-19 Tom de Vries <tom@codesourcery.com>
8092 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8094 2017-07-19 Tom de Vries <tom@codesourcery.com>
8096 * config/nvptx/nvptx-modes.def: Add V2DImode.
8097 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8098 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8099 (nvptx_output_mov_insn): Handle lack of mov.b128.
8100 (nvptx_print_operand): Handle 'H' and 'L' codes.
8101 (nvptx_vector_mode_supported): Allow V2DImode.
8102 (nvptx_preferred_simd_mode): New function.
8103 (nvptx_data_alignment): New function.
8104 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8105 nvptx_preferred_simd_mode.
8106 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8108 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8110 2017-07-19 Tom de Vries <tom@codesourcery.com>
8112 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8113 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8114 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8115 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8116 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8117 (mov<VECIM>_insn): New define_insn.
8118 (define_expand "mov<VECIM>): New define_expand.
8120 2017-07-19 Tom de Vries <tom@codesourcery.com>
8122 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8124 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8126 PR tree-optimization/81346
8127 * fold-const.h (fold_div_compare, range_check_type): Declare.
8128 * fold-const.c (range_check_type): New function.
8129 (build_range_check): Use range_check_type.
8130 (fold_div_compare): No longer static, rewritten into
8131 a match.pd helper function.
8132 (fold_comparison): Don't call fold_div_compare here.
8133 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8136 2017-07-19 Nathan Sidwell <nathan@acm.org>
8138 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8139 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8140 * tree.c (find_decls_types_r, verify_type): Use
8141 TYPE_{MIN,MAX}_VALUE_RAW.
8142 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8143 (hash_tree): Likewise.
8144 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8146 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8149 2017-07-18 Tom de Vries <tom@codesourcery.com>
8152 * omp-expand.c (expand_omp_for_static_chunk): Handle
8153 equal-argument loop exit phi.
8155 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8158 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8159 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8160 operand 2 predicate.
8161 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8162 operand 2 predicate.
8163 (ror,rol -> rorx splitters): Use const_int_operand as
8164 operand 2 predicate.
8166 2017-06-18 Richard Biener <rguenther@suse.de>
8168 PR tree-optimization/81410
8169 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8170 the gap in the ! slp_perm SLP case after each group.
8172 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8175 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8178 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8181 * predict.c (set_even_probabilities): Cleanup; do not affect
8182 probabilities that are already known.
8183 (combine_predictions_for_bb): Call even when count is set.
8185 2017-07-18 Nathan Sidwell <nathan@acm.org>
8187 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8190 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8193 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8194 optimization for loop niter analysis.
8196 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8199 * config/avr/avr.c (avr_optimize_casesi): Don't use
8200 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8202 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8204 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8205 body_cost_vec from _vect_peel_extended_info.
8206 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8207 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8210 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8212 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8214 2017-07-18 Richard Biener <rguenther@suse.de>
8216 PR tree-optimization/80620
8217 PR tree-optimization/81403
8218 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8219 info when re-using a VN table entry.
8221 2017-07-18 Richard Biener <rguenther@suse.de>
8223 PR tree-optimization/81418
8224 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8225 vectype_in. Verify that with lane-reducing reduction operations
8226 we have a single def-use cycle.
8228 2017-07-17 Carl Love <cel@us.ibm.com>
8230 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8232 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8233 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8234 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8235 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8236 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8237 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8238 VMULOSW): New enum "unspec" values.
8239 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8240 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8241 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8242 altivec_vmulosw): New patterns.
8243 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8244 VMULOSW): Add definitions.
8245 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8247 * config/alpha/alpha.c: Include predict.h.
8249 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8251 * tree-vrp.c (compare_assert_loc): Fix comparison function
8252 to return predictable results.
8254 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8256 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8259 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8260 * doc/invoke.texi (mexpand-adddi): Update text.
8262 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8264 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8265 that also clobbers the CC register. The old expand code is moved
8267 (*arc_clzsi2): ... here.
8268 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8269 the CC register. The old expand code is moved to ...
8270 (arc_ctzsi2): ... here.
8272 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8274 * config/arc/arc.opt (mindexed-loads): Use initial value
8275 TARGET_INDEXED_LOADS_DEFAULT.
8276 (mauto-modify-reg): Use initial value
8277 TARGET_AUTO_MODIFY_REG_DEFAULT.
8278 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8279 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8280 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8281 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8283 2017-07-17 Martin Liska <mliska@suse.cz>
8286 * opts.c (finish_options): Do not allow -fgnu-tm
8287 w/ -fsanitize={kernel-,}address. Say sorry.
8289 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8292 * tree-loop-distribution.c (classify_partition): Only assert on
8293 numer of iterations.
8294 (merge_dep_scc_partitions): Delete prameter. Update function call.
8295 (distribute_loop): Remove code handling loop with unknown niters.
8296 (pass_loop_distribution::execute): Skip loop with unknown niters.
8298 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8301 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8302 function sort_partitions_by_post_order.
8304 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8306 PR tree-optimization/81374
8307 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8308 the max index of basic blocks, rather than number of basic blocks.
8310 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8312 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8314 (arc_legitimate_pic_operand_p): Likewise.
8315 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8317 (arc_needs_pcl_p): Likewise.
8318 (arc_legitimate_pc_offset_p): Likewise.
8319 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8320 function is also used in constrains.md.
8321 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8322 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8323 PLUS. Only return true/false in known cases, otherwise assert.
8324 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8325 is already called in arc_legitimate_constant_p.
8326 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8328 (LEGITIMATE_PIC_OPERAND_P): Use
8329 arc_raw_symbolic_reference_mentioned_p function.
8330 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8335 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8336 Andrew Burgess <andrew.burgess@embecosm.com>
8338 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8339 (arc_return_address_register): New function.
8340 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8341 (arc_handle_fndecl_attribute): Add naked attribute.
8342 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8343 (TARGET_WARN_FUNC_RETURN): Likewise.
8344 (arc_allocate_stack_slots_for_args): New function.
8345 (arc_warn_func_return): Likewise.
8346 (machine_function): Change type fn_type.
8347 (arc_compute_function_type): Consider new naked function type,
8348 change function return type.
8349 (arc_must_save_register): Adapt to handle new
8350 arc_compute_function_type's return type.
8351 (arc_expand_prologue): Likewise.
8352 (arc_expand_epilogue): Likewise.
8353 (arc_return_address_regs): Delete.
8354 (arc_return_address_register): New function.
8355 (arc_epilogue_uses): Use above function.
8356 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8357 (arc_function_type): Change encoding, add naked type.
8358 (ARC_INTERRUPT_P): Change to handle the new encoding.
8359 (ARC_FAST_INTERRUPT_P): Likewise.
8360 (ARC_NORMAL_P): Define.
8361 (ARC_NAKED_P): Likewise.
8362 (arc_compute_function_type): Delete prototype.
8363 * config/arc/arc.md (in_ret_delay_slot): Use
8364 arc_return_address_register function.
8365 (simple_return): Likewise.
8366 (p_return_i): Likewise.
8368 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8370 PR tree-optimization/81428
8371 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8372 can't be built for those types.
8374 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8376 Remove stuff dead since r239246.
8378 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8379 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8380 (avr_inform_devices): Remove dead stuff.
8382 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8384 * config/arm/arm_neon.h: Fix softp typo.
8386 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8388 PR tree-optimization/81365
8389 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8390 aggregate moves onto bb predecessor edges, make sure there are no
8391 loads that could alias the lhs in between the start of bb and the
8394 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8397 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8398 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8399 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8401 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8403 PR tree-optimization/81396
8404 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8405 (init_symbolic_number): Initialize it to 1.
8406 (perform_symbolic_merge): Add n_ops from both operands into the new
8408 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8409 without base_addr as useless if they need more than one operation.
8410 (bswap_replace): Handle !bswap case for NULL base_addr.
8412 2017-07-17 Tom de Vries <tom@codesourcery.com>
8415 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8418 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8420 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8421 conditional builtin define __FIX_LEON3FT_B2BST.
8423 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8425 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8426 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8429 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8431 PR rtl-optimization/81424
8432 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8433 to remove potential trapping from operands if -fnon-call-exceptions.
8435 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8437 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8438 profile_proability for scalling.
8439 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8441 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8443 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8445 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8447 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8448 fixpoint arithmetics.
8450 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8452 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8453 fixpoint arithmetics.
8455 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8457 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8458 fixpoint arithmetics.
8460 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8462 * profile-count.h (profile_probability::from_reg_br_prob_note,
8463 profile_probability::to_reg_br_prob_note): New functions.
8464 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8465 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8466 * predict.c (probability_reliable_p): Update.
8467 (edge_probability_reliable_p): Update.
8468 (br_prob_note_reliable_p): Update.
8469 (invert_br_probabilities): Update.
8470 (add_reg_br_prob_note): New function.
8471 (combine_predictions_for_insn): Update.
8472 * asan.c (asan_clear_shadow): Update.
8473 * cfgbuild.c (compute_outgoing_frequencies): Update.
8474 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8475 (update_br_prob_note): Update.
8476 (rtl_verify_edges): Update.
8477 (purge_dead_edges): Update.
8478 (fixup_reorder_chain): Update.
8479 * emit-rtl.c (try_split): Update.
8480 * ifcvt.c (cond_exec_process_insns): Update.
8481 (cond_exec_process_if_block): Update.
8482 (dead_or_predicable): Update.
8483 * internal-fn.c (expand_addsub_overflow): Update.
8484 (expand_neg_overflow): Update.
8485 (expand_mul_overflow): Update.
8486 * loop-doloop.c (doloop_modify): Update.
8487 * loop-unroll.c (compare_and_jump_seq): Update.
8488 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8489 * predict.h: Update.
8490 * reorg.c (mostly_true_jump): Update.
8492 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8493 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8494 * config/arc/arc.c: (emit_unlikely_jump): Update.
8495 * config/arm/arm.c: (emit_unlikely_jump): Update.
8496 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8497 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8498 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8499 (ix86_print_operand): Update.
8500 (ix86_split_fp_branch): Update.
8501 (predict_jump): Update.
8502 * config/ia64/ia64.c (ia64_print_operand): Update.
8503 * config/mmix/mmix.c (mmix_print_operand): Update.
8504 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8505 (rs6000_expand_split_stack_prologue): Update.
8506 * config/rs6000/rs6000.c: Update.
8507 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8508 (s390_expand_vec_movstr): Update.
8509 (s390_expand_cs_tdsi): Update.
8510 (s390_expand_split_stack_prologue): Update.
8511 * config/sh/sh.c (sh_print_operand): Update.
8512 (expand_cbranchsi4): Update.
8513 (expand_cbranchdi4): Update.
8514 * config/sparc/sparc.c (output_v9branch): Update.
8515 * config/spu/spu.c (get_branch_target): Update.
8516 (ea_load_store_inline): Update.
8517 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8518 * config/tilepro/tilepro.c: Update.
8520 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8522 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8523 (gimplify_save_expr): Likewise.
8525 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8527 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8529 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8531 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8533 * timevar.def (TV_IPA_FNSUMMARY): Define.
8535 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8537 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8538 to back store errata sensitive sequence from being generated.
8539 (sqrtdf2_fix): Likewise.
8541 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8543 * tree-ssa-threadupdate.c (compute_path_counts,
8544 update_joiner_offpath_counts): Use profile_probability.
8546 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8549 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8551 * config/arm/arm-c.c (arm_cpu_builtins): Define
8552 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8554 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8556 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8557 array entries to represent __ieee128 versions of the
8558 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8559 scalar_extract_sig, and scalar_insert_exp built-in functions.
8560 (altivec_resolve_overloaded_builtin): Add special case handling
8561 for the __builtin_scalar_insert_exp function, as represented by
8562 the P9V_BUILTIN_VEC_VSIEDP constant.
8563 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8564 exponent support for __ieee128 argument.
8565 (VSESQP): Add scalar extract signature support for __ieee128
8567 (VSTDCNQP): Add scalar test negative support for __ieee128
8569 (VSIEQP): Add scalar insert exponent support for __int128 argument
8570 with __ieee128 result.
8571 (VSIEQPF): Add scalar insert exponent support for __ieee128
8572 argument with __ieee128 result.
8573 (VSTDCQP): Add scalar test data class support for __ieee128
8575 (VSTDCNQP): Add overload support for scalar test negative with
8577 (VSTDCQP): Add overload support for scalar test data class
8579 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8581 (UNSPEC_VSX_SIEXPQP): New constant.
8582 (xsxexpqp): New insn for VSX scalar extract exponent quad
8584 (xsxsigqp): New insn for VSX scalar extract significand quad
8586 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8587 precision with floating point argument.
8588 (xststdcqp): New expand for VSX scalar test data class quad
8590 (xststdcnegqp): New expand for VSX scalar test negative quad
8592 (xststdcqp): New insn to match expansions for VSX scalar test data
8593 class quad precision and VSX scalar test negative quad precision.
8594 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8595 special case operand checking to enforce that second operand of
8596 VSX scalar test data class with quad precision argument is a 7-bit
8598 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8599 prototypes and descriptions of __ieee128 versions of
8600 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8601 scalar_test_data_class, and scalar_test_neg built-in functions.
8603 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8605 PR tree-optimization/81162
8606 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8607 replace a negate with an add.
8609 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8611 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8613 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8615 * config/arm/arm-c.c (arm_cpu_builtins): Define
8616 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8618 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8620 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8621 (armv8-r): Set ARM Cortex-R52 as default CPU.
8622 * config/arm/arm-tables.opt: Regenerate.
8623 * config/arm/arm-tune.md: Regenerate.
8624 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8626 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8627 extension for -mcpu=cortex-r52.
8629 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8631 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8632 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8633 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8634 (fp-armv8): Define it as FP_ARMv8 only.
8635 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8636 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8638 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8639 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8640 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8641 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8642 than TARGET_FPU_ARMV8.
8643 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8644 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8645 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8647 * config/arm/neon.md (neon_vrint): Likewise.
8648 (neon_vcvt): Likewise.
8649 (neon_<fmaxmin_op><mode>): Likewise.
8650 (<fmaxmin><mode>3): Likewise.
8651 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8652 * config/arm/predicates.md (arm_cond_move_operator): Check against
8653 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8655 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8657 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8660 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8662 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8663 loop in comment with memset.
8665 2017-07-14 Martin Liska <mliska@suse.cz>
8667 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8668 * dwarf2out.c (is_java): Remove the function.
8669 (output_pubname): Remove usage of the function.
8670 (lower_bound_default): Remove usage of DW_LANG_Java.
8671 (gen_compile_unit_die): Likewise.
8672 * gcc.c: Remove compiler defaults for .java and .zip files.
8673 * gimple-expr.c (remove_suffix): Change as there's no longer
8674 extension than 4-letter one.
8675 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8676 (gimplify_save_expr): Likewise.
8677 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8678 as it's possible even for other languages than Java.
8679 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8680 * lto-opts.c (lto_write_options): Remove reference to Java.
8681 * opts.c (strip_off_ending): Update file extension handling.
8682 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8683 * tree-eh.c (lower_resx): Likewise.
8684 * tree.c (free_lang_data_in_type): Remove dead code.
8685 (find_decls_types_r): Likewise.
8686 (build_common_builtin_nodes): Remove Java from a comment.
8687 (verify_type): Remove dead code.
8688 * varasm.c (assemble_external): Remove Java from a comment.
8690 2017-07-14 Martin Liska <mliska@suse.cz>
8692 * opts.c (finish_options): Add quotes.
8693 (common_handle_option): Likewise.
8695 2017-07-14 Martin Liska <mliska@suse.cz>
8697 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8698 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8700 * dwarf2out.c (lower_bound_default): Do not handle
8702 (gen_compile_unit_die): Likewise.
8703 * gcc.c: Remove default extension binding for GNU Pascal.
8704 * stmt.c: Remove Pascal language from a comment.
8705 * xcoffout.c: Likewise.
8707 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8710 * diagnostic-show-locus.c (fixit_cmp): New function.
8711 (layout::layout): Sort m_fixit_hints.
8712 (column_range::column_range): Assert that the values are valid.
8713 (struct char_span): New struct.
8714 (correction::overwrite): New method.
8715 (struct source_line): New struct.
8716 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8717 calls in terms of classes source_line and char_span, and
8718 correction::overwrite.
8719 (selftest::test_overlapped_fixit_printing_2): New function.
8720 (selftest::diagnostic_show_locus_c_tests): Call it.
8722 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8724 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8725 early if there is no lhs.
8727 2017-07-13 Martin Liska <mliska@suse.cz>
8729 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8730 (gen_reference_type_die): Likewise.
8731 * stor-layout.c: Remove Pascal-related comment.
8733 2017-07-13 Martin Liska <mliska@suse.cz>
8735 * opts.c (finish_options): Add quotes to error messages.
8736 (parse_sanitizer_options): Likewise.
8738 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8740 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8742 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8744 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8746 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8748 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8750 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8752 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8755 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8756 provides the hardware capability bits, define the macro
8757 __BUILTIN_CPU_SUPPORTS__.
8758 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8759 if GLIBC does not provide the hardware capability bits. Add a
8760 gcc_unreachable call if the built-in cpu function is neither
8761 __builtin_cpu_is nor __builtin_cpu_supports.
8762 (rs6000_get_function_versions_dispatcher): Change the warning
8763 that an old GLIBC is used which does not export the capability
8764 bits to be an error.
8765 * doc/extend.texi (target_clones attribute): Document the
8766 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8767 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8768 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8769 the macros defined by GCC if the newer GLIBC is available.
8771 2017-07-12 Jeff Law <law@redhat.com>
8773 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8774 remaining includes slightly.
8775 * config/riscv/riscv-builtins.c: Include profile-count.h.
8777 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8780 * config/avr/avr.c (avr_set_current_function): In diagnostic
8781 messages: Quote keywords and (parts of) identifiers.
8782 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8785 2017-07-12 Carl Love <cel@us.ibm.com>
8787 * config/rs6000/rs6000-c.c: Add support for built-in functions
8788 vector bool char vec_revb (vector bool char);
8789 vector bool short vec_revb (vector short char);
8790 vector bool int vec_revb (vector bool int);
8791 vector bool long long vec_revb (vector bool long long);
8792 * doc/extend.texi: Update the built-in documentation file for the
8793 new built-in functions.
8795 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8797 * config/s390/s390.md: Remove movcc splitter.
8799 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8801 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8802 load/store on condition.
8804 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8807 * config/avr/avr.c (avr_encode_section_info)
8808 [progmem && !TREE_READONLY]: Error if progmem object needs
8811 2017-07-11 Michael Collison <michael.collison@arm.com>
8813 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8816 2017-07-11 Carl Love <cel@us.ibm.com>
8818 * config/rs6000/rs6000-c.c: Add support for builtins
8819 vector unsigned int vec_parity_lsbb (vector signed int);
8820 vector unsigned int vec_parity_lsbb (vector unsigned int);
8821 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8822 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8823 vector unsigned long long vec_parity_lsbb (vector signed long long);
8824 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8825 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8826 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8827 * doc/extend.texi: Update the built-in documentation file for the
8828 new built-in functions.
8830 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8832 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8833 (layout::m_primary_loc): New field.
8834 (layout::layout): Initialize new field. Move location filtering
8835 logic from here to...
8836 (layout::maybe_add_location_range): ...this new method. Add
8837 support for filtering to just the lines already specified by other
8839 (layout::will_show_line_p): New method.
8840 (gcc_rich_location::add_location_if_nearby): New method.
8841 (selftest::test_add_location_if_nearby): New test function.
8842 (selftest::diagnostic_show_locus_c_tests): Call it.
8843 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8846 2017-07-11 Tom de Vries <tom@codesourcery.com>
8848 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8849 (bb_first_real_insn): New function.
8850 (nvptx_single): Add extra initialization of broadcasted condition
8853 2017-07-11 Nathan Sidwell <nathan@acm.org>
8855 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8857 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8859 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8860 Binutils doc as TEXI will mess them up.
8861 * doc/invoke.texi (AVR Options): Same here.
8863 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8865 * config/sparc/sparc.opt (mfix-ut700): New option.
8866 (mfix-gr712rc): Likewise.
8867 (sparc_fix_b2bst): New variable.
8868 * doc/invoke.texi (SPARC options): Document them.
8869 (ARM options): Fix warnings.
8870 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8871 instructions to prevent sequences that can trigger the store-store
8872 errata for certain LEON3FT processors.
8873 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8874 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8875 * config/sparc/sparc.md (fix_b2bst): New attribute.
8876 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8878 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8881 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8883 (*rsqrtsf2_sse): Ditto.
8885 (div<mode>3): Macroize insn from divdf3 and divsf3
8886 using MODEF mode iterator.
8888 2017-07-10 Martin Sebor <msebor@redhat.com>
8890 PR tree-optimization/80397
8891 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8892 instead of testing for equality to INTEGER_TYPE.
8894 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8896 * config.gcc: Remove uclibc from arc target spec.
8898 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8900 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8902 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8905 * lto-wrapper.c (remove_option): New function.
8906 (merge_and_complain): Merge PIC/PIE options more realistically.
8908 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8910 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8914 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8915 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8916 * config.in: Regenerate.
8917 * configure: Regenerate.
8918 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8919 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8920 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8921 (TARGET_GASISR_PROLOGUES): ...target mask.
8922 * common/config/avr/avr-common.c
8923 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8924 Set -mgas-isr-prologues.
8925 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8926 INSERT_PASS_BEFORE for it.
8927 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8928 * config/avr/avr.c (avr_option_override)
8929 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8930 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8931 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8932 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8933 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8935 (make_avr_pass_pre_proep): New function.
8936 (emit_push_sfr) <treg>: Add argument to function and use it
8938 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8939 and set machine->gasisr.yes.
8940 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8941 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8942 __gcc_isr.n_pushed to .L__stack_usage.
8943 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8944 (avr_asm_final_postscan_insn): ...this new static function.
8945 * config/avr/avr.h (machine_function)
8946 <is_no_gccisr, use_L__stack_usage>: New fields.
8947 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8948 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8949 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8950 (gasisr, *gasisr): New expander and insn.
8951 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8952 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8953 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8955 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8957 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8960 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8962 Move jump-tables out of .text again.
8965 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8966 (ASM_OUTPUT_ADDR_VEC): New function.
8967 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8968 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8969 INSN_ADDRESSes as asm comment.
8970 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8971 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8972 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8973 * config/avr/avr.md (*tablejump): Adjust comment.
8974 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8975 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8977 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8978 (avr_output_addr_vec): New proto.
8979 (avr_log_t) <insn_addresses>: New field.
8981 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8984 * config/i386/i386.c (ix86_function_arg_advance): Set
8985 outgoing_args_on_stack to true if there are outgoing arguments
8987 (ix86_function_arg): Likewise.
8988 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8989 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8990 * config/i386/i386.h (machine_function): Add
8991 outgoing_args_on_stack.
8993 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8995 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8997 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8999 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9001 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9002 (REAL_H): Remove $(MACHMODE_H).
9003 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9005 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9006 $(MACHMODE_H) and double-int.h.
9007 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9009 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9012 2017-07-07 Andrew Pinski <apinski@cavium.com>
9014 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9015 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9017 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9019 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9020 Add warning if GCC was not configured to link against a GLIBC that
9021 exports the hardware capability bits.
9022 (make_resolver_func): Make resolver function private and not a
9023 COMDAT function. Create the name with clone_function_name instead
9024 of make_unique_name.
9027 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9028 correct operand in doing the split.
9030 2017-07-07 Carl Love <cel@us.ibm.com>
9032 * config/rs6000/rs6000-c: Add support for built-in function
9033 vector unsigned short vec_pack_to_short_fp32 (vector float,
9035 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9036 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9037 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9038 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9039 (convert_4f32_8i16): Add define_expand.
9040 * doc/extend.texi: Update the built-in documentation file for the
9041 new built-in function.
9043 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9045 * config/sparc/m8.md: New file.
9046 * config/sparc/sparc.md: Include m8.md.
9048 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9050 * config/sparc/sparc.opt: New option -mvis4b.
9051 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9052 (sparc_option_override): Handle VIS4B.
9053 (enum sparc_builtins): Define
9054 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9055 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9056 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9057 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9058 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9059 (check_constant_argument): New function.
9060 (sparc_vis_init_builtins): Define builtins
9061 __builtin_vis_dictunpack{8,16,32},
9062 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9063 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9064 __builtin_vis_fpcmpde{8,16,32}shl and
9065 __builtin_vis_fpcmpur{8,16,32}shl.
9066 (sparc_expand_builtin): Check that the constant operands to
9067 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9068 constant and in range.
9069 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9071 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9072 (SPARC_IMM5_P): Likewise.
9073 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9074 (enabled): Handle vis4b.
9075 (UNSPEC_DICTUNPACK): New unspec.
9076 (UNSPEC_FPCMPSHL): Likewise.
9077 (UNSPEC_FPUCMPSHL): Likewise.
9078 (UNSPEC_FPCMPDESHL): Likewise.
9079 (UNSPEC_FPCMPURSHL): Likewise.
9080 (cpu_feature): New CPU feature `vis4b'.
9081 (dictunpack{8,16,32}): New insns.
9082 (FPCSMODE): New mode iterator.
9083 (fpcscond): New code iterator.
9084 (fpcsucond): Likewise.
9085 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9086 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9087 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9088 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9089 * config/sparc/constraints.md: Define constraints `q' for unsigned
9090 2-bit integer constants and `t' for unsigned 5-bit integer
9092 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9094 (imm5_operand_dictunpack16): Likewise.
9095 (imm5_operand_dictunpack32): Likewise.
9096 (imm2_operand): Likewise.
9097 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9098 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9099 ditunpack* and fpcmp*shl builtins.
9101 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9103 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9104 * config.in: Add HAVE_AS_SPARC6 define.
9105 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9107 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9109 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9110 (CPP_CPU_SPEC): Handle m8.
9111 (ASM_CPU_SPEC): Likewise.
9112 * config/sparc/sparc-opts.h (enum processor_type): Add
9114 * config/sparc/sparc.c (m8_costs): New struct.
9115 (sparc_option_override): Handle TARGET_CPU_m8.
9116 (sparc32_initialize_trampoline): Likewise.
9117 (sparc64_initialize_trampoline): Likewise.
9118 (sparc_issue_rate): Likewise.
9119 (sparc_register_move_cost): Likewise.
9120 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9121 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9122 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9123 (CPP_CPU_SPEC): Handle M8.
9124 (ASM_CPU_SPEC): Likewise.
9125 (AS_M8_FLAG): Define.
9126 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9127 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9128 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9130 * configure: Regenerate.
9131 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9134 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9136 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9138 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9139 ("*movdi_insn_sp32"): Do not set v3pipe.
9140 ("*movsi_insn"): Likewise.
9141 ("*movdi_insn_sp64"): Likewise.
9142 ("*movsf_insn"): Likewise.
9143 ("*movdf_insn_sp32"): Likewise.
9144 ("*movdf_insn_sp64"): Likewise.
9145 ("*zero_extendsidi2_insn_sp64"): Likewise.
9146 ("*sign_extendsidi2_insn"): Likewise.
9147 ("*mov<VM32:mode>_insn"): Likewise.
9148 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9149 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9150 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9151 ("<vlop:code><VL:mode>3"): Likewise.
9152 ("*not_<vlop:code><VL:mode>3"): Likewise.
9153 ("*nand<VL:mode>_vis"): Likewise.
9154 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9155 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9156 ("one_cmpl<VL:mode>2"): Likewise.
9157 ("faligndata<VM64:mode>_vis"): Likewise.
9158 ("alignaddrsi_vis"): Likewise.
9159 ("alignaddrdi_vis"): Likweise.
9160 ("alignaddrlsi_vis"): Likewise.
9161 ("alignaddrldi_vis"): Likewise.
9162 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9163 ("bmaskdi_vis"): Likewise.
9164 ("bmasksi_vis"): Likewise.
9165 ("bshuffle<VM64:mode>_vis"): Likewise.
9166 ("cmask8<P:mode>_vis"): Likewise.
9167 ("cmask16<P:mode>_vis"): Likewise.
9168 ("cmask32<P:mode>_vis"): Likewise.
9169 ("pdistn<P:mode>_vis"): Likewise.
9170 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9172 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9174 * config/sparc/sparc.md ("subtype"): New insn attribute.
9175 ("*wrgsr_sp64"): Set insn subtype.
9176 ("*rdgsr_sp64"): Likewise.
9177 ("alignaddrsi_vis"): Likewise.
9178 ("alignaddrdi_vis"): Likewise.
9179 ("alignaddrlsi_vis"): Likewise.
9180 ("alignaddrldi_vis"): Likewise.
9181 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9182 ("fexpand_vis"): Likewise.
9183 ("fpmerge_vis"): Likewise.
9184 ("faligndata<VM64:mode>_vis"): Likewise.
9185 ("bshuffle<VM64:mode>_vis"): Likewise.
9186 ("cmask8<P:mode>_vis"): Likewise.
9187 ("cmask16<P:mode>_vis"): Likewise.
9188 ("cmask32<P:mode>_vis"): Likewise.
9189 ("fchksm16_vis"): Likewise.
9190 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9191 ("fmean16_vis"): Likewise.
9192 ("fp<plusminus_insn>64_vis"): Likewise.
9193 ("<plusminus_insn>v8qi3"): Likewise.
9194 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9195 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9196 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9197 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9198 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9199 ("*movqi_insn"): Likewise.
9200 ("*movhi_insn"): Likewise.
9201 ("*movsi_insn"): Likewise.
9202 ("movsi_pic_gotdata_op"): Likewise.
9203 ("*movdi_insn_sp32"): Likewise.
9204 ("*movdi_insn_sp64"): Likewise.
9205 ("movdi_pic_gotdata_op"): Likewise.
9206 ("*movsf_insn"): Likewise.
9207 ("*movdf_insn_sp32"): Likewise.
9208 ("*movdf_insn_sp64"): Likewise.
9209 ("*zero_extendhisi2_insn"): Likewise.
9210 ("*zero_extendqihi2_insn"): Likewise.
9211 ("*zero_extendqisi2_insn"): Likewise.
9212 ("*zero_extendqidi2_insn"): Likewise.
9213 ("*zero_extendhidi2_insn"): Likewise.
9214 ("*zero_extendsidi2_insn_sp64"): Likewise.
9215 ("ldfsr"): Likewise.
9216 ("prefetch_64"): Likewise.
9217 ("prefetch_32"): Likewise.
9218 ("tie_ld32"): Likewise.
9219 ("tie_ld64"): Likewise.
9220 ("*tldo_ldub_sp32"): Likewise.
9221 ("*tldo_ldub1_sp32"): Likewise.
9222 ("*tldo_ldub2_sp32"): Likewise.
9223 ("*tldo_ldub_sp64"): Likewise.
9224 ("*tldo_ldub1_sp64"): Likewise.
9225 ("*tldo_ldub2_sp64"): Likewise.
9226 ("*tldo_ldub3_sp64"): Likewise.
9227 ("*tldo_lduh_sp32"): Likewise.
9228 ("*tldo_lduh1_sp32"): Likewise.
9229 ("*tldo_lduh_sp64"): Likewise.
9230 ("*tldo_lduh1_sp64"): Likewise.
9231 ("*tldo_lduh2_sp64"): Likewise.
9232 ("*tldo_lduw_sp32"): Likewise.
9233 ("*tldo_lduw_sp64"): Likewise.
9234 ("*tldo_lduw1_sp64"): Likewise.
9235 ("*tldo_ldx_sp64"): Likewise.
9236 ("*mov<VM32:mode>_insn"): Likewise.
9237 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9238 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9240 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9242 * config/sparc/sparc.md ("type"): New insn type viscmp.
9243 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9245 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9246 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9247 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9248 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9250 ("n7_vis_logical_11cycle"): Likewise.
9251 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9252 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9253 * config/sparc/niagara.md ("niag_vis"): Likewise.
9254 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9255 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9257 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9259 * config/sparc/sparc.md: New instruction type `bmask'.
9260 (bmaskdi_vis): Use the `bmask' type.
9261 (bmasksi_vis): Likewise.
9262 * config/sparc/ultra3.md (us3_array): Likewise.
9263 * config/sparc/niagara7.md (n7_array): Likewise.
9264 * config/sparc/niagara4.md (n4_array): Likewise.
9265 * config/sparc/niagara2.md (niag2_vis): Likewise.
9266 (niag3_vis): Likewise.
9267 * config/sparc/niagara.md (niag_vis): Likewise.
9269 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9271 * ipa-comdats.c: Remove optimize check from gate.
9272 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9273 for functions not optimized.
9274 (ipa_fn_summary_read): Skip optimize check.
9275 (ipa_fn_summary_write): Likewise.
9276 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9278 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9280 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9281 (check_callers): Give up on caller which is not optimized.
9282 (inline_small_functions): Likewise.
9283 (ipa_inline): Do not give up when not optimizing.
9284 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9285 away unoptimizes cdtors.
9286 (whole_program_function_and_variable_visibility): Do
9287 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9288 * ipa.c (process_references): Do not check optimize.
9289 (symbol_table::remove_unreachable_nodes): Update optimize check.
9290 (set_writeonly_bit): Update optimize check.
9291 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9292 (pass_ipa_single_use::gate): Remove.
9294 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9296 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9297 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9298 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9299 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9300 permute_load, permute_store, adjust_extract, adjust_splat,
9301 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9302 replace_swap_with_copy, dump_swap_insn_table,
9303 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9304 recombine_lvx_pattern, recombine_stvx_pattern,
9305 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9306 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9307 to file rs6000-p8swap.c.
9308 * config/rs6000/rs6000-p8swap.c: New file.
9309 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9310 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9311 and rs6000*-*-* targets.
9313 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9315 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9317 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9319 * lto-wrapper.c (merge_and_complain): Do not merge
9320 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9321 fsigned_zeros, ftrapping_math, fwrapv.
9322 (append_compiler_options): Do not track these options.
9323 (append_linker_options): Likewie
9325 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9327 * cgraphunit.c (cgraph_node::finalize_function): When
9328 !flag_toplevel_reorde set no_reorder flag.
9329 (varpool_node::finalize_decl): Likewise.
9330 (symbol_table::compile): Drop no toplevel reorder path.
9332 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9334 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9335 edges; zero probability is not better than uninitialized.
9337 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9339 * asan.h (asan_sanitize_allocas_p): Declare.
9340 * asan.c (asan_sanitize_allocas_p): New function.
9341 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9342 (handle_builtin_alloca): Likewise.
9343 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9344 if !asan_sanitize_allocas_p.
9345 * params.def (asan-instrument-allocas): Add new option.
9346 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9347 * opts.c (common_handle_option): Disable allocas sanitization for
9350 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9352 * asan.c: Include gimple-fold.h.
9353 (get_last_alloca_addr): New function.
9354 (handle_builtin_stackrestore): Likewise.
9355 (handle_builtin_alloca): Likewise.
9356 (asan_emit_allocas_unpoison): Likewise.
9357 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9358 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9359 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9360 (instrument_builtin_call): Pass gimple iterator to
9361 get_mem_refs_of_builtin_call.
9362 (last_alloca_addr): New global.
9363 * asan.h (asan_emit_allocas_unpoison): Declare.
9364 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9365 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9366 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9367 if function calls alloca.
9368 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9369 * gimple-fold.h (replace_call_with_value): Declare.
9370 * internal-fn.c: Include asan.h.
9371 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9372 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9374 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9376 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9377 (C_SELFTEST_FLAGS): New.
9378 (CPP_SELFTEST_FLAGS): New.
9379 (SELFTEST_DEPS): New, from deps of s-selftest.
9380 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9381 (CPP_SELFTEST_DEPS): New.
9382 (selftest): Add dependency on s-selftest-c++.
9383 (s-selftest): Rename to...
9384 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9385 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9386 than SELFTEST_FLAGS.
9387 (selftest-gdb): Rename to...
9388 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9390 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9391 (selftest-valgrind): Rename to...
9392 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9394 (selftest-valgrind): Reintroduce as an alias for
9395 selftest-c-valgrind.
9396 (s-selftest-c++): New.
9397 (selftest-c++-gdb): New.
9398 (selftest-c++-valgrind): New.
9400 2017-07-06 Olivier Hainque <hainque@adacore.com>
9402 * gcc.c (process_command): When deciding if undefined variables
9403 should be ignored when processing specs, accept "gcc -v" as well.
9405 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9407 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9408 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9410 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9412 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9413 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9414 * config/arm/arm-tables.opt: Regenerate.
9415 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9417 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9419 2017-07-06 Carl Love <cel@us.ibm.com>
9421 * ChangeLog: Clean up from mid air collision
9423 2017-07-06 Carl Love <cel@us.ibm.com>
9425 * config/rs6000/rs6000-c.c: Add support for built-in functions
9426 vector signed int vec_subc (vector signed int, vector signed int);
9427 vector signed __int128 vec_subc (vector signed __int128,
9428 vector signed __int128);
9429 vector unsigned __int128 vec_subc (vector unsigned __int128,
9430 vector unsigned __int128);
9431 vector signed int vec_sube (vector signed int, vector signed int,
9433 vector unsigned int vec_sube (vector unsigned int,
9434 vector unsigned int,
9435 vector unsigned int);
9436 vector signed __int128 vec_sube (vector signed __int128,
9437 vector signed __int128,
9438 vector signed__int128);
9439 vector unsigned __int128 vec_sube (vector unsigned __int128,
9440 vector unsigned __int128,
9441 vector unsigned __int128);
9442 vector signed int vec_subec (vector signed int, vector signed int,
9444 vector unsigned int vec_subec (vector unsigned int,
9445 vector unsigned int,
9446 vector unsigned int);
9447 vector signed __int128 vec_subec (vector signed __int128,
9448 vector signed __int128,
9449 vector signed__int128);
9450 vector unsigned __int128 vec_subec (vector unsigned __int128,
9451 vector unsigned __int128,
9452 vector unsigned __int128);
9453 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9454 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9455 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9456 BU_ALTIVEC_OVERLOAD_X definitions.
9457 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9458 * doc/extend.texi: Update the built-in documentation file for the new
9461 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9464 * diagnostic-show-locus.c (layout::layout): Use start and finish
9465 spelling location for the start and finish of each range.
9466 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9467 Add unused aspect param.
9468 * input.c (expand_location_1): Add "aspect" param, and use it
9469 to access the correct part of the location.
9470 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9472 (expand_location_to_spelling_point): Likewise.
9473 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9474 param, and pass it to expand_location_1.
9476 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9478 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9479 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9480 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9481 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9482 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9483 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9484 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9485 _mm_maskz_getmant_ss): New intrinsics.
9486 (__builtin_ia32_getexpss128_mask): Changed to ...
9487 __builtin_ia32_getexpss128_round ... this.
9488 (__builtin_ia32_getexpsd128_mask): Changed to ...
9489 __builtin_ia32_getexpsd128_round ... this.
9490 * config/i386/i386-builtin-types.def
9491 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9492 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9493 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9494 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9495 __builtin_ia32_getmantss_mask_round): New builtins.
9496 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9497 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9498 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9499 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9500 * config/i386/sse.md
9501 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9502 avx512f_sgetexp<mode><mask_scalar_name>
9503 <round_saeonly_scalar_name> ... this.
9504 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9505 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9506 vgetexp<ssescalarmodesuffix>
9507 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9508 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9509 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9510 avx512f_vgetmant<mode><mask_scalar_name>
9511 <round_saeonly_scalar_name> ... this.
9512 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9513 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9514 vgetmant<ssescalarmodesuffix>
9515 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9516 %0<mask_scalar_operand4>, %1, %2
9517 <round_saeonly_scalar_mask_op4>, %3} ... this.
9518 * config/i386/subst.md (mask_scalar_operand4,
9519 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9520 round_saeonly_scalar_nimm_predicate): New subst attributes.
9522 2017-07-06 Julia Koval <julia.koval@intel.com>
9524 * config/i386/i386.c (ix86_erase_embedded_rounding):
9525 Remove code for old rounding pattern.
9527 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9529 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9531 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9533 * doc/sourcebuild.texi (Test Directives, Variants of
9534 dg-require-support): Add documentation for dg-require-stack-check.
9536 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9538 * config/i386/subst.md (mask_scalar, round_scalar,
9539 round_saeonly_scalar): New meta-templates.
9540 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9541 round_scalar_mask_operand3, round_scalar_mask_op3,
9542 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9543 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9544 round_saeonly_scalar_constraint,
9545 round_saeonly_scalar_prefix): New subst attribute.
9546 * config/i386/sse.md
9547 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9548 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9549 <round_scalar_name> ... this.
9550 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9551 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9552 <round_scalar_name> ... this.
9553 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9554 <sse>_vm<code><mode>3<mask_scalar_name>
9555 <round_saeonly_scalar_name> ... this.
9556 (v<plusminus_mnemonic><ssescalarmodesuffix>
9557 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9558 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9559 v<plusminus_mnemonic><ssescalarmodesuffix>
9560 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9561 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9562 (v<multdiv_mnemonic><ssescalarmodesuffix>
9563 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9564 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9565 v<multdiv_mnemonic><ssescalarmodesuffix>
9566 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9567 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9568 (v<maxmin_float><ssescalarmodesuffix>
9569 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9570 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9571 v<maxmin_float><ssescalarmodesuffix>
9572 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9573 %0<mask_scalar_operand3>, %1, %<iptr>2
9574 <round_saeonly_scalar_mask_op3>} ... this.
9576 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9578 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9579 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9581 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9582 Alan Hayward <alan.hayward@arm.com>
9583 David Sherwood <david.sherwood@arm.com>
9585 * combine.c (simplify_if_then_else): Remove "enum" before
9587 * compare-elim.c (can_eliminate_compare): Likewise.
9588 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9590 (aarch64_lookup_simd_builtin_type): Likewise.
9591 (aarch64_simd_builtin_type): Likewise.
9592 (aarch64_init_simd_builtin_types): Likewise.
9593 (aarch64_simd_expand_args): Likewise.
9594 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9596 (aarch64_reverse_mask): Likewise.
9597 (aarch64_simd_emit_reg_reg_move): Likewise.
9598 (aarch64_gen_adjusted_ldpstp): Likewise.
9599 (aarch64_ccmp_mode_to_code): Likewise.
9600 (aarch64_operands_ok_for_ldpstp): Likewise.
9601 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9602 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9604 (aarch64_min_divisions_for_recip_mul): Likewise.
9605 (aarch64_reassociation_width): Likewise.
9606 (aarch64_get_condition_code_1): Likewise.
9607 (aarch64_simd_emit_reg_reg_move): Likewise.
9608 (aarch64_simd_attr_length_rglist): Likewise.
9609 (aarch64_reverse_mask): Likewise.
9610 (aarch64_operands_ok_for_ldpstp): Likewise.
9611 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9612 (aarch64_gen_adjusted_ldpstp): Likewise.
9613 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9615 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9616 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9617 (arm_lookup_simd_builtin_type): Likewise.
9618 (arm_simd_builtin_type): Likewise.
9619 (arm_init_simd_builtin_types): Likewise.
9620 (arm_expand_builtin_args): Likewise.
9621 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9622 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9623 (ft32_setup_incoming_varargs): Likewise.
9624 (ft32_function_arg): Likewise.
9625 (ft32_function_arg_advance): Likewise.
9626 (ft32_pass_by_reference): Likewise.
9627 (ft32_arg_partial_bytes): Likewise.
9628 (ft32_valid_pointer_mode): Likewise.
9629 (ft32_addr_space_pointer_mode): Likewise.
9630 (ft32_addr_space_legitimate_address_p): Likewise.
9631 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9633 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9634 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9635 (iamcu_alignment): Likewise.
9636 (canonicalize_vector_int_perm): Likewise.
9637 (ix86_noce_conversion_profitable_p): Likewise.
9638 (ix86_mpx_bound_mode): Likewise.
9639 (ix86_operands_ok_for_move_multiple): Likewise.
9640 * config/microblaze/microblaze-protos.h
9641 (microblaze_expand_conditional_branch_reg): Likewise.
9642 * config/microblaze/microblaze.c
9643 (microblaze_expand_conditional_branch_reg): Likewise.
9644 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9646 (rs6000_reassociation_width): Likewise.
9647 (rs6000_invalid_binary_op): Likewise.
9648 (fusion_p9_p): Likewise.
9649 (emit_fusion_p9_load): Likewise.
9650 (emit_fusion_p9_store): Likewise.
9651 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9653 (riscv_hard_regno_mode_ok_p): Likewise.
9654 (riscv_address_insns): Likewise.
9655 (riscv_split_symbol): Likewise.
9656 (riscv_legitimize_move): Likewise.
9657 (riscv_function_value): Likewise.
9658 (riscv_hard_regno_nregs): Likewise.
9659 (riscv_expand_builtin): Likewise.
9660 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9661 (riscv_build_integer): Likewise.
9662 (riscv_split_integer): Likewise.
9663 (riscv_legitimate_constant_p): Likewise.
9664 (riscv_cannot_force_const_mem): Likewise.
9665 (riscv_regno_mode_ok_for_base_p): Likewise.
9666 (riscv_valid_base_register_p): Likewise.
9667 (riscv_valid_offset_p): Likewise.
9668 (riscv_valid_lo_sum_p): Likewise.
9669 (riscv_classify_address): Likewise.
9670 (riscv_legitimate_address_p): Likewise.
9671 (riscv_address_insns): Likewise.
9672 (riscv_load_store_insns): Likewise.
9673 (riscv_force_binary): Likewise.
9674 (riscv_split_symbol): Likewise.
9675 (riscv_force_address): Likewise.
9676 (riscv_legitimize_address): Likewise.
9677 (riscv_move_integer): Likewise.
9678 (riscv_legitimize_const_move): Likewise.
9679 (riscv_legitimize_move): Likewise.
9680 (riscv_address_cost): Likewise.
9681 (riscv_subword): Likewise.
9682 (riscv_output_move): Likewise.
9683 (riscv_canonicalize_int_order_test): Likewise.
9684 (riscv_emit_int_order_test): Likewise.
9685 (riscv_function_arg_boundary): Likewise.
9686 (riscv_pass_mode_in_fpr_p): Likewise.
9687 (riscv_pass_fpr_single): Likewise.
9688 (riscv_pass_fpr_pair): Likewise.
9689 (riscv_get_arg_info): Likewise.
9690 (riscv_function_arg): Likewise.
9691 (riscv_function_arg_advance): Likewise.
9692 (riscv_arg_partial_bytes): Likewise.
9693 (riscv_function_value): Likewise.
9694 (riscv_pass_by_reference): Likewise.
9695 (riscv_setup_incoming_varargs): Likewise.
9696 (riscv_print_operand): Likewise.
9697 (riscv_elf_select_rtx_section): Likewise.
9698 (riscv_save_restore_reg): Likewise.
9699 (riscv_for_each_saved_reg): Likewise.
9700 (riscv_register_move_cost): Likewise.
9701 (riscv_hard_regno_mode_ok_p): Likewise.
9702 (riscv_hard_regno_nregs): Likewise.
9703 (riscv_class_max_nregs): Likewise.
9704 (riscv_memory_move_cost): Likewise.
9705 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9706 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9707 (rl78_addr_space_address_mode): Likewise.
9708 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9710 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9711 (rs6000_reassociation_width): Likewise.
9712 (rs6000_invalid_binary_op): Likewise.
9713 (fusion_p9_p): Likewise.
9714 (emit_fusion_p9_load): Likewise.
9715 (emit_fusion_p9_store): Likewise.
9716 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9717 (ok_for_simple_move_operands): Likewise.
9718 (ok_for_simple_move_strict_operands): Likewise.
9719 (ok_for_simple_arith_logic_operands): Likewise.
9720 (visium_legitimize_reload_address): Likewise.
9721 (visium_select_cc_mode): Likewise.
9722 (output_cbranch): Likewise.
9723 (visium_split_double_move): Likewise.
9724 (visium_expand_copysign): Likewise.
9725 (visium_expand_int_cstore): Likewise.
9726 (visium_expand_fp_cstore): Likewise.
9727 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9728 (visium_function_arg): Likewise.
9729 (visium_function_arg_advance): Likewise.
9730 (visium_libcall_value): Likewise.
9731 (visium_setup_incoming_varargs): Likewise.
9732 (visium_legitimate_constant_p): Likewise.
9733 (visium_legitimate_address_p): Likewise.
9734 (visium_legitimize_address): Likewise.
9735 (visium_secondary_reload): Likewise.
9736 (visium_register_move_cost): Likewise.
9737 (visium_memory_move_cost): Likewise.
9738 (prepare_move_operands): Likewise.
9739 (ok_for_simple_move_operands): Likewise.
9740 (ok_for_simple_move_strict_operands): Likewise.
9741 (ok_for_simple_arith_logic_operands): Likewise.
9742 (visium_function_value_1): Likewise.
9743 (rtx_ok_for_offset_p): Likewise.
9744 (visium_legitimize_reload_address): Likewise.
9745 (visium_split_double_move): Likewise.
9746 (visium_expand_copysign): Likewise.
9747 (visium_expand_int_cstore): Likewise.
9748 (visium_expand_fp_cstore): Likewise.
9749 (visium_split_cstore): Likewise.
9750 (visium_select_cc_mode): Likewise.
9751 (visium_split_cbranch): Likewise.
9752 (output_cbranch): Likewise.
9753 (visium_print_operand_address): Likewise.
9754 * expmed.c (flip_storage_order): Likewise.
9755 * expmed.h (emit_cstore): Likewise.
9756 (flip_storage_order): Likewise.
9757 * genrecog.c (validate_pattern): Likewise.
9758 * hsa-gen.c (gen_hsa_addr): Likewise.
9759 * internal-fn.c (expand_arith_overflow): Likewise.
9760 * ira-color.c (allocno_copy_cost_saving): Likewise.
9761 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9762 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9763 (process_invariant_for_inheritance): Likewise.
9764 * lra-eliminations.c (move_plus_up): Likewise.
9765 * omp-low.c (lower_oacc_reductions): Likewise.
9766 * simplify-rtx.c (simplify_subreg): Likewise.
9767 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9768 (TARGET_CHKP_BOUND_MODE): Likewise..
9769 * targhooks.c (default_chkp_bound_mode): Likewise.
9770 (default_setup_incoming_vararg_bounds): Likewise.
9771 * targhooks.h (default_chkp_bound_mode): Likewise.
9772 (default_setup_incoming_vararg_bounds): Likewise.
9773 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9774 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9775 (have_whole_vector_shift): Likewise.
9776 * tree-vect-stmts.c (vectorizable_load): Likewise.
9777 * doc/tm.texi: Regenerate.
9779 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9781 Graceful degrade if Binutils PR21472 is not available.
9784 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9785 .rodata in flash test fails.
9786 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9787 * confgure: Regenerate.
9788 * config.in: Regenerate.
9789 * config/avr/avr.c (avr_asm_named_section)
9790 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9791 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9792 (avr_asm_init_sections): Same.
9794 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9796 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9797 (fma<VH:mode>4_intrinsic): Likewise.
9798 (*fmsub<VCVTF:mode>4): Likewise.
9799 (*fmsub<VH:mode>4_intrinsic): Likewise.
9801 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9804 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9805 Don't depend on "optimize > 0".
9806 (out_movhi_r_mr, out_movqi_mr_r): Same.
9807 (out_movhi_mr_r, out_movqi_r_mr): Same.
9808 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9809 io_address_operand on "optimize > 0".
9811 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9813 * tree-loop-distribution.c: Add general explanantion on the pass.
9814 (generate_loops_for_partition): Mark distributed loop.
9815 (pg_add_dependence_edges): New parameter. Handle alias data
9816 dependence specially and record it in the parameter if asked.
9817 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9818 (init_partition_graph_vertices, add_partition_graph_edge): New.
9819 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9820 (free_partition_graph_vdata, build_partition_graph): New.
9821 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9822 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9823 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9824 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9825 (version_for_distribution_p, finalize_partitions): New.
9826 (distribute_loop): Handle alias data dependence specially. Factor
9827 out loop fusion code as functions and call these functions.
9829 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9831 * tree-loop-distribution.c (classify_partition): New parameter and
9832 better handle reduction statement.
9833 (rdg_build_partitions): Revise comment.
9834 (distribute_loop): Compute statements in all partitions and pass it
9835 to classify_partition.
9837 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9839 * tree-loop-distribution.c (enum partition_type): New.
9840 (struct partition): New field type.
9841 (partition_merge_into): Add parameter. Update partition type.
9842 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9843 (build_rdg_partition_for_vertex): Compute partition type.
9844 (rdg_build_partitions): Dump partition type.
9845 (distribute_loop): Update calls to partition_merge_into.
9847 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9849 * tree-loop-distribution.c (struct ddr_hasher): New.
9850 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9852 (classify_partition): Call get_data_dependence.
9853 (pg_add_dependence_edges): Ditto.
9854 (distribute_loop): Release data dependence hash table.
9856 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9858 * tree-loop-distribution.c (ref_base_address): Delete.
9859 (similar_memory_accesses): Rename ...
9860 (share_memory_accesses): ... to this. Check if partitions access
9861 the same memory reference.
9862 (distribute_loop): Call share_memory_accesses.
9864 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9866 * tree-loop-distribution.c (struct partition): New field recording
9868 (partition_alloc, partition_free): Init and release data refs.
9869 (partition_merge_into): Merge data refs.
9870 (build_rdg_partition_for_vertex): Collect data refs for partition.
9871 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9873 (distribute_loop): Remve data refs from vertice data of partition
9876 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9878 * tree-loop-distribution.c (params.h): Include header file.
9879 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9880 (datarefs_vec): New global var.
9881 (create_rdg_vertices): Use datarefs_vec directly.
9882 (free_rdg): Don't free data references.
9883 (build_rdg): Update use. Don't free data references.
9884 (distribute_loop): Compute global variable for data references.
9885 Bail out if there are too many data references.
9887 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9889 * tree-loop-distribution.c (loop_nest): New global var.
9890 (build_rdg): Use loop directly, rather than loop nest.
9891 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9893 (distribute_loop): Compute global variable loop nest. Update use.
9895 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9897 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9898 (partition_merge_into): New parameter. Dump reason for fusion.
9899 (distribute_loop): Update use of partition_merge_into.
9901 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9903 * tree-loop-distribution.c (bb_top_order_index): New.
9904 (bb_top_order_index_size, bb_top_order_cmp): New.
9905 (stmts_from_loop): Use topological order.
9906 (pass_loop_distribution::execute): Compute and release topological
9907 order for basic blocks.
9909 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9911 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9914 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9916 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9917 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9918 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9919 * internal-fn.def (LOOP_DIST_ALIAS): New.
9920 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9921 (fold_loop_internal_call): ... this.
9922 (vect_loop_dist_alias_call): New function.
9923 (set_uid_loop_bbs): Call fold_loop_internal_call.
9924 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9927 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9930 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9931 Require dead FLAGS_REG at the beginning of a peephole.
9933 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9936 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9937 arguments in the call to __builtin_ia32_sbb_u32.
9938 (_subborrow_u64): Swap _X and _Y arguments in the call to
9939 __builtin_ia32_sbb_u64.
9941 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9944 * tree-vrp.c (compare_assert_loc): Turn into a function template
9945 with stable template parameter. Only test if a->e is NULL,
9946 !a->e == !b->e has been verified already. Use e == NULL or
9947 e != NULL instead of e or ! e tests. If stable is true, don't use
9948 iterative_hash_expr, on the other side allow a or b or both NULL
9949 and sort the NULLs last.
9950 (process_assert_insertions): Sort using compare_assert_loc<false>
9951 instead of compare_assert_loc, later sort using
9952 compare_assert_loc<true> before calling process_assert_insertions_for
9953 in a loop. Use break instead of continue once seen NULL pointer.
9955 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9957 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9958 Cortex-R7 and Cortex-R8 processors.
9960 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9962 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9963 uninitialized while src is not.
9965 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9967 * common/config/arm/arm-common.c: Adjust include path for
9969 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9970 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9971 (arm-cpu-data.h): Likewise.
9972 (arm-cpu-cdata.h): Likewise.
9973 * config/arm/arm-cpu.h: Delete.
9974 * config/arm/arm-cpu-cdata.h: Delete.
9975 * config/arm/arm-cpu-data.h: Delete.
9977 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9979 * config/arm/arm-cpus.in (cortex-a55): New.
9980 (cortex-a75): Likewise.
9981 (cortex-a75.cortex-a55): Likewise.
9982 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9984 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9985 * config/arm/arm-cpu-cdata.h: Regenerate.
9986 * config/arm/arm-cpu-data.h: Regenerate.
9987 * config/arm/arm-cpu.h: Regenerate.
9988 * config/arm/arm-tables.opt: Regenerate.
9989 * config/arm/arm-tune.md: Regenerate.
9991 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9993 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9995 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9997 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10000 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10002 PR tree-optimization/81292
10003 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10004 full_string_p, also call adjust_related_strinfos if the adjustment
10005 is simple, otherwise invalidate related strinfos.
10007 2017-07-04 Martin Liska <mliska@suse.cz>
10010 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10011 newly created variable as DECL_IGNORED_P.
10013 2017-07-04 Martin Liska <mliska@suse.cz>
10016 * ipa-inline.c (inline_small_functions):
10017 Use xstrdup_for_dump.
10019 2017-07-04 Tom de Vries <tom@codesourcery.com>
10021 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10023 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10026 * config/darwin.c (darwin_function_switched_text_sections):
10029 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10031 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10033 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10035 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10037 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10039 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10040 min_profitable_iters, and th as inclusive lower bounds.
10041 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10042 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10043 for min_profitable_iters and min_profitable_estimate.
10044 (vect_transform_loop): Treat th as an inclusive lower bound.
10045 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10047 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10050 * config/darwin.c (darwin_function_switched_text_sections):
10051 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10052 in two pieces, and suppress the use of buf.
10054 2017-07-03 Nathan Sidwell <nathan@acm.org>
10056 * hash-table.h (hash_table_mod1): Fix indentation.
10058 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10060 PR middle-end/81290
10061 * predict.c (force_edge_cold): Be more careful about propagation
10063 * profile-count.h (profile_probability::guessed,
10064 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10066 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10068 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10070 * doc/invoke.texi (rcpc architecture extension): Document it.
10072 2017-07-03 Richard Biener <rguenther@suse.de>
10074 PR tree-optimization/60510
10075 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10076 the scalar reduction PHI and use it.
10077 (vectorizable_reduction): Properly guard the single_defuse_cycle
10078 path for non-SLP reduction chains where we cannot use it.
10079 Rework reduc_def/index and vector type deduction. Rework
10080 vector operand gathering during reduction op code-gen.
10081 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10082 chains dissolve the chain and leave it to non-SLP reduction
10085 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10087 * tree-data-ref.h (dr_alignment): Declare.
10088 * tree-data-ref.c (dr_alignment): New function.
10089 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10090 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10092 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10094 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10096 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10097 and base_misalignment fields.
10098 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10099 * tree-data-ref.c: Include builtins.h.
10100 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10101 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10102 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10103 * tree-vect-data-refs.c: Include tree-cfg.h.
10104 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10105 fields instead of calculating an alignment here.
10106 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10107 innermost_loop_behavior fields.
10109 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10111 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10113 (DR_STEP_ALIGNMENT): New macro.
10114 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10115 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10116 (create_data_ref): Print it.
10117 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10118 to tell whether the step preserves vector (mis)alignment.
10119 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10120 Move the check for an integer step and generalise to all INTEGER_CST.
10121 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10122 Print the outer step alignment.
10124 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10126 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10127 with offset_alignment.
10128 (DR_ALIGNED_TO): Delete.
10129 (DR_OFFSET_ALIGNMENT): New macro.
10130 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10131 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10132 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10133 (create_data_ref): Likewise.
10134 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10135 (vect_analyze_data_refs): Likewise.
10136 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10137 creating dummy innermost behavior.
10139 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10141 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10142 with a "innermost_loop_behavior *" and refeence tree.
10143 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10144 (create_data_ref): Update call accordingly.
10145 * tree-predcom.c (find_looparound_phi): Likewise.
10147 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10149 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10150 fields with dr_wrt_vec_loop.
10151 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10152 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10153 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10154 (vect_dr_behavior): New function.
10155 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10156 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10157 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10158 track whether the step preserves the misalignment.
10159 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10160 Use vect_dr_behavior.
10161 (vect_setup_realignment): Update call accordingly.
10162 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10163 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10164 call to vect_create_addr_base_for_vector_ref.
10165 (vect_create_cond_for_align_checks): Likewise.
10166 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10167 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10168 (vect_recog_mask_conversion_pattern): Likewise.
10169 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10170 (new_stmt_vec_info): Remove redundant zeroing.
10172 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10174 * common/config/arm/arm-common.c (arm_be8_option): New function.
10175 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10176 (ISA_ARMv6): Add isa_bit_be8.
10177 * config/arm/arm.h (arm_be8_option): Add prototype.
10178 (BE8_SPEC_FUNCTION): New define.
10179 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10180 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10181 (mlittle-endian): Similarly.
10182 (mbe8, mbe32): New options.
10183 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10184 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10186 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10188 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10190 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10192 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10193 (cleanup_tree_cfg_bb): Use it.
10194 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10196 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10198 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10201 * loop-doloop.c (add_test): Update profile.
10203 2017-07-03 Martin Liska <mliska@suse.cz>
10206 * sanopt.c (rewrite_usage_of_param): New function.
10207 (sanitize_rewrite_addressable_params): Likewise.
10208 (pass_sanopt::execute): Call rewrite_usage_of_param.
10210 2017-07-03 Richard Biener <rguenther@suse.de>
10212 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10213 back to using VIEW_CONVERT_EXPR.
10215 2017-07-03 Martin Liska <mliska@suse.cz>
10218 * doc/extend.texi: Document when a resolver function is
10219 generated for target_clones.
10221 2017-07-03 Martin Liska <mliska@suse.cz>
10223 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10224 and shadow memory of auto variables which are subject of
10225 use-after-scope sanitization.
10226 (asan_expand_mark_ifn): Add do set only when is_poison.
10228 2016-07-03 Richard Biener <rguenther@suse.de>
10230 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10232 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10233 (vectorizable_reduction): Perform reduction PHI creation when
10234 visiting a reduction PHI and adjust and simplify code generation
10235 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10236 (vect_transform_loop): Visit reduction PHIs.
10237 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10238 defs into the SLP tree.
10239 (vect_build_slp_tree): Reduction defs terminate the recursion.
10240 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10242 (vect_get_vec_defs_for_stmt_copy): Export.
10243 (vect_get_vec_defs): Likewise.
10244 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10246 (vect_get_vec_defs_for_stmt_copy): Declare.
10247 (vect_get_vec_defs): Likewise.
10249 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10251 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10252 parameter with a "loop" parameter and use it instead of the
10253 loop containing DR_STMT. Don't check simple_iv when doing
10254 BB analysis. Describe the two analysis modes in the comment.
10256 2017-07-03 Tom de Vries <tom@codesourcery.com>
10258 PR tree-optimization/69468
10259 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10260 (find_same_succ_bb): Handle ignore_edge_flags.
10262 2017-07-03 Tom de Vries <tom@codesourcery.com>
10264 PR tree-optimization/81192
10265 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10267 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10269 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10271 2017-07-03 Tom de Vries <tom@codesourcery.com>
10273 PR tree-optimization/81192
10274 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10275 BB_SAME_SUCC (bb) == NULL.
10277 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10279 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10282 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10284 * dumpfile.c: Include profile-count.h
10285 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10287 (insert_cond_bb): Update profile.
10288 * tree-cfg.h (insert_cond_bb): Update prototype.
10289 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10290 * tree-dump.c: Do not include tree-cfg.
10292 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10294 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10296 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10298 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10301 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10303 * tree-complex.c (expand_complex_div_wide): update profile.
10305 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10306 Alan Hayward <alan.hayward@arm.com>
10307 David Sherwood <david.sherwood@arm.com>
10309 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10310 (CORETYPES_H): New define.
10311 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10312 (insn-modes-inline.h, s-modes-inline-h): New rules.
10313 (generated_files): Add insn-modes-inline.h.
10314 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10315 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10316 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10317 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10318 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10319 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10320 (build/gencodes.o, build/genconditions.o): Likewise.
10321 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10322 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10323 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10324 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10325 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10326 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10327 * coretypes.h: Include everything up to real.h for generators.
10328 Include insn-modes.h first. Include wide-int-print.h after
10329 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10330 * machmode.h: Don't include insn-modes.h here.
10331 * function-tests.c: Remove includes of signop.h, machmode.h,
10332 double-int.h and wide-int.h.
10334 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10336 * optc-save-gen.awk: Likewise.
10337 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10338 * godump.c: Remove include of wide-int-print.h.
10339 * pretty-print.h: Likewise.
10340 * wide-int-print.cc: Likewise.
10341 * wide-int.cc: Likewise.
10342 * hash-map-tests.c: Remove include of signop.h.
10343 * hash-set-tests.c: Likewise.
10344 * rtl-tests.c: Likewise.
10345 * mkconfig.sh: Remove include of machmode.h.
10346 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10348 (emit_insn_modes_inline_h): ...this new function. Emit the code
10349 into an insn-modes-inline.h header file, adding appropriate
10350 include guards and end comments.
10351 (emit_insn_modes_c_header): Remove include of machmode.h.
10352 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10354 (main): Handle -i flag and call emit_insn_modes_inline_h when
10357 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10359 * tree-ssa-strlen.c (strinfo): Rename the length field to
10360 nonzero_chars. Add a full_string_p field.
10361 (compare_nonzero_chars, zero_length_string_p): New functions.
10362 (get_addr_stridx): Add an offset_out parameter.
10363 Use compare_nonzero_chars.
10364 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10365 (new_strinfo): Update after above changes to strinfo.
10366 (set_endptr_and_length): Set full_string_p.
10367 (get_string_length): Update after above changes to strinfo.
10368 (unshare_strinfo): Update call to new_strinfo.
10369 (maybe_invalidate): Likewise.
10370 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10371 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10372 as a uhwi instead of an shwi. Update after above changes to
10373 strinfo and new_strinfo.
10374 (zero_length_string): Assert that chainsi contains full strings.
10375 Use zero_length_string_p. Update call to new_strinfo.
10376 (adjust_related_strinfos): Update after above changes to strinfo.
10377 Copy full_string_p from origsi.
10378 (adjust_last_stmt): Use zero_length_string_p.
10379 (handle_builtin_strlen): Update after above changes to strinfo and
10380 new_strinfo. Install the lhs as the string length if the previous
10381 entry didn't describe a full string.
10382 (handle_builtin_strchr): Update after above changes to strinfo
10384 (handle_builtin_strcpy): Likewise.
10385 (handle_builtin_strcat): Likewise.
10386 (handle_builtin_malloc): Likewise.
10387 (handle_pointer_plus): Likewise.
10388 (handle_builtin_memcpy): Likewise. Track nonzero characters
10389 that aren't necessarily followed by a nul terminator.
10390 (handle_char_store): Likewise.
10392 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10394 PR tree-optimization/80769
10395 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10396 for malloc and calloc. Document the new invariant that all related
10397 strinfos have delayed lengths or none do.
10398 (verify_related_strinfos): Move earlier in file.
10399 (set_endptr_and_length): New function, split out from...
10400 (get_string_length): ...here. Also set the lengths of related
10402 (zero_length_string): Assert that chainsi has known (rather than
10404 (adjust_related_strinfos): Likewise.
10406 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10408 PR tree-optimization/81136
10409 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10410 assert that two references with the same misalignment have the same
10411 compile-time misalignment if those compile-time misalignments
10414 2017-07-01 Andi Kleen <ak@linux.intel.com>
10416 * print-tree.c (print_node): Print all attributes.
10418 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10420 * cfg.c (scale_bbs_frequencies): New function.
10421 * cfg.h (scale_bbs_frequencies): Declare it.
10422 * cfgloopanal.c (single_likely_exit): Cleanup.
10423 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10425 (scale_loop_profile): Likewise.
10426 (loop_version): Likewise.
10427 (create_empty_loop_on_edge): Update.
10428 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10429 scale_loop_frequencies, scale_loop_profile, loopify,
10430 loop_version): Update prototypes.
10431 * modulo-sched.c (sms_schedule): Update.
10432 * predict.c (unlikely_executed_edge_p): Also check probability.
10433 (probably_never_executed_edge_p): Fix typo.
10434 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10435 * tree-parloops.c (gen_parallel_loop): Update.
10436 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10437 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10438 * tree-ssa-loop-split.c (split_loop): Update.
10439 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10440 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10441 (vect_loop_versioning): Update.
10442 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10444 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10446 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10448 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10450 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10451 to keep profile consistent.
10453 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10455 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10456 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10457 * profile-count.h (max_safe_multiplier): Make unsigned.
10458 (profile_count::guessed_zero): New.
10460 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10462 * bb-reorder.c (fix_up_crossing_landing_pad,
10463 fix_crossing_conditional_branches): Use make_single_succ_edge
10464 to keep profile consistent.
10466 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10468 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10471 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10474 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10475 the right scopes, make sure cond_jump isn't preserved between multiple
10476 iterations. Search for fallthru edge whenever there are 3+ edges and
10477 use find_fallthru_edge for it.
10479 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10481 Patch by Alexander Monakov <amonakov@ispras.ru>
10482 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10483 probabilities consistently.
10485 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10487 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10488 emit_cmp_and_jump_insns.
10490 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10493 * tree-inline.c (expand_call_inline): Combine profile statuses.
10495 2017-06-30 Andrew Pinski <apinski@cavium.com>
10497 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10498 fold_stmt returned true.
10500 2017-06-30 Nathan Sidwell <nathan@acm.org>
10502 * ggc.h (empty_string): Delete.
10503 * cfgexpand.c (expand_asm_stmt): Use plain "".
10504 * optabs.c (expand_asm_memory_barrier): Likewise.
10505 * stringpool.c (empty_string): Delete.
10506 (digit_vector, digit_string): Delete.
10507 (ggc_alloc_string): Use plain "", don't optimize single digit
10508 strings. Use ggc_alloc_atomic.
10510 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10512 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10513 comparison set and one other set, use the cost of the non-comparison
10516 2017-06-30 Nathan Sidwell <nathan@acm.org>
10518 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10521 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10523 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10524 loops. Remove now unneeded calls to gimple_switch_set_label() that
10525 just set removed labels to NULL_TREE.
10527 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10529 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10530 (set_range_info): ...here. Only call set_range_info_raw if domain
10532 (set_nonzero_bits): Call set_range_info_raw.
10533 * tree-ssanames.h (set_range_info_raw): New.
10535 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10538 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10539 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10540 of nonimmediate_operand and <store_mask_constraint> instead of m
10541 for the input operand. For V8FI iterator, always split if input
10542 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10543 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10544 <store_mask_predicate> instead of register_operand and
10545 <store_mask_constraint> instead of v for the input operand. Make
10546 sure both operands aren't MEMs for if not <mask_applied>.
10548 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10550 * lto-wrapper.c (copy_file) Close both file descriptors before
10553 2017-06-30 Martin Liska <mliska@suse.cz>
10556 * multiple_target.c (create_dispatcher_calls): Make ifunc
10557 also for function that don't have calls or are not referenced.
10559 2017-06-30 Richard Biener <rguenther@suse.de>
10561 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10562 analyze the first scalar stmt. Move vector type computation
10563 for the BB case here from ...
10564 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10565 live operation processing in the SLP case properly.
10567 2017-06-30 Richard Biener <rguenther@suse.de>
10569 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10571 2017-06-30 Martin Liska <mliska@suse.cz>
10574 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10575 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10577 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10579 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10582 2017-06-30 Richard Biener <rguenther@suse.de>
10584 PR tree-optimization/81249
10585 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10586 condition reduction result to original scalar type.
10588 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10590 * profile-count.h (enum profile_quality): Fix typos and whitespace
10593 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10595 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10596 type for branch probabilities.
10598 2017-06-29 Julian Brown <julian@codesourcery.com>
10599 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10601 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10602 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10603 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10604 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10606 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10608 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10609 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10610 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10611 CC usage from generic code to here.
10612 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10613 CC usage into the target macros.
10615 2017-06-29 Maya Rashish <coypu@sdf.org>
10617 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10620 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10622 * arm/arm-builtins.c: Include profile-count.h
10623 * except.c (sjlj_emit_function_enter): Use
10624 profile_probability::unlikely.
10626 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10628 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10629 and tocrel_offset be pointer args rather than implicitly using
10631 (legitimate_constant_pool_address_p, rs6000_emit_move,
10632 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10633 tocrel_offset and use in toc_relative_expr_p call.
10634 (print_operand, print_operand_address): Use static tocrel_base_oac
10635 and tocrel_offset_oac.
10636 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10639 2017-06-29 Maya Rashish <coypu@sdf.org>
10641 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10643 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10645 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10646 objects, take into account only the alignment of 'op0' and 'mode1' if
10649 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10651 * ccmp.c (ccmp_tree_comparison_p): New function.
10652 (ccmp_candidate_p): Update to use above function.
10653 (get_compare_parts): New function.
10654 (expand_ccmp_next): Update to use new functions.
10655 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10657 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10658 take mode as argument.
10659 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10660 * expr.c (expand_expr_real_1): Pass mode as argument.
10662 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10664 * combine.c (combine_instructions): Print insns to dump_file, together
10667 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10669 * asan.c (asan_emit_stack_protection): Update.
10670 (create_cond_insert_point): Update.
10671 * auto-profile.c (afdo_propagate_circuit): Update.
10672 * basic-block.h (struct edge_def): Turn probability to
10673 profile_probability.
10674 (EDGE_FREQUENCY): Update.
10675 * bb-reorder.c (find_traces_1_round): Update.
10676 (better_edge_p): Update.
10677 (sanitize_hot_paths): Update.
10678 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10679 (make_single_succ_edge): Update.
10680 (check_bb_profile): Update.
10681 (dump_edge_info): Update.
10682 (update_bb_profile_for_threading): Update.
10683 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10684 probabilitycount to 0.
10685 * cfgbuild.c (compute_outgoing_frequencies): Update.
10686 * cfgcleanup.c (try_forward_edges): Update.
10687 (outgoing_edges_match): Update.
10688 (try_crossjump_to_edge): Update.
10689 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10690 (expand_gimple_tailcall): Update.
10691 (construct_init_block): Use make_single_succ_edge.
10692 (construct_exit_block): Use make_single_succ_edge.
10693 * cfghooks.c (verify_flow_info): Update.
10694 (redirect_edge_succ_nodup): Update.
10695 (split_edge): Update.
10696 (account_profile_record): Update.
10697 * cfgloopanal.c (single_likely_exit): Update.
10698 * cfgloopmanip.c (scale_loop_profile): Update.
10699 (set_zero_probability): Remove.
10700 (duplicate_loop_to_header_edge): Update.
10701 * cfgloopmanip.h (loop_version): Update prototype.
10702 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10703 (force_nonfallthru_and_redirect): Update.
10704 (update_br_prob_note): Update.
10705 (rtl_verify_edges): Update.
10706 (purge_dead_edges): Update.
10707 (rtl_lv_add_condition_to_bb): Update.
10708 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10709 * cgraphunit.c (init_lowered_empty_function): Update.
10710 (cgraph_node::expand_thunk): Update.
10711 * cilk-common.c: Include profile-count.h
10712 * dojump.c (inv): Remove.
10713 (jumpifnot): Update.
10714 (jumpifnot_1): Update.
10715 (do_jump_1): Update.
10717 (do_jump_by_parts_greater_rtx): Update.
10718 (do_compare_rtx_and_jump): Update.
10719 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10720 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10721 * dwarf2cfi.c: Include profile-count.h
10722 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10723 (sjlj_emit_dispatch_table): Likewise.
10724 * explow.c: Include profile-count.h
10725 * expmed.c (emit_store_flag_force): Update.
10726 (do_cmp_and_jump): Update.
10727 * expr.c (compare_by_pieces_d::generate): Update.
10728 (compare_by_pieces_d::finish_mode): Update.
10729 (emit_block_move_via_loop): Update.
10730 (store_expr_with_bounds): Update.
10731 (store_constructor): Update.
10732 (expand_expr_real_2): Update.
10733 (expand_expr_real_1): Update.
10734 * expr.h (try_casesi, try_tablejump): Update prototypes.
10735 * gimple-pretty-print.c (dump_probability): Update.
10736 (dump_profile): New.
10737 (dump_gimple_label): Update.
10738 (dump_gimple_bb_header): Update.
10739 * graph.c (draw_cfg_node_succ_edges): Update.
10740 * hsa-gen.c (convert_switch_statements): Update.
10741 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10742 (find_if_case_1): Update.
10743 (find_if_case_2): Update.
10744 * internal-fn.c (expand_arith_overflow_result_store): Update.
10745 (expand_addsub_overflow): Update.
10746 (expand_neg_overflow): Update.
10747 (expand_mul_overflow): Update.
10748 (expand_vector_ubsan_overflow): Update.
10749 * ipa-cp.c (good_cloning_opportunity_p): Update.
10750 * ipa-split.c (split_function): Use make_single_succ_edge.
10751 * ipa-utils.c (ipa_merge_profiles): Update.
10752 * loop-doloop.c (add_test): Update.
10753 (doloop_modify): Update.
10754 * loop-unroll.c (compare_and_jump_seq): Update.
10755 (unroll_loop_runtime_iterations): Update.
10756 * lra-constraints.c (lra_inheritance): Update.
10757 * lto-streamer-in.c (input_cfg): Update.
10758 * lto-streamer-out.c (output_cfg): Update.
10759 * mcf.c (adjust_cfg_counts): Update.
10760 * modulo-sched.c (sms_schedule): Update.
10761 * omp-expand.c (expand_omp_for_init_counts): Update.
10762 (extract_omp_for_update_vars): Update.
10763 (expand_omp_ordered_sink): Update.
10764 (expand_omp_for_ordered_loops): Update.
10765 (expand_omp_for_generic): Update.
10766 (expand_omp_for_static_nochunk): Update.
10767 (expand_omp_for_static_chunk): Update.
10768 (expand_cilk_for): Update.
10769 (expand_omp_simd): Update.
10770 (expand_omp_taskloop_for_outer): Update.
10771 (expand_omp_taskloop_for_inner): Update.
10772 * omp-simd-clone.c (simd_clone_adjust): Update.
10773 * optabs.c (expand_doubleword_shift): Update.
10774 (expand_abs): Update.
10775 (emit_cmp_and_jump_insn_1): Update.
10776 (expand_compare_and_swap_loop): Update.
10777 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10778 * predict.c (predictable_edge_p): Update.
10779 (edge_probability_reliable_p): Update.
10780 (set_even_probabilities): Update.
10781 (combine_predictions_for_insn): Update.
10782 (combine_predictions_for_bb): Update.
10783 (propagate_freq): Update.
10784 (estimate_bb_frequencies): Update.
10785 (force_edge_cold): Update.
10786 * profile-count.c (profile_count::dump): Add missing space into dump.
10787 (profile_count::debug): Add newline.
10788 (profile_count::differs_from_p): Explicitly convert to unsigned.
10789 (profile_count::stream_in): Update.
10790 (profile_probability::dump): New member function.
10791 (profile_probability::debug): New member function.
10792 (profile_probability::differs_from_p): New member function.
10793 (profile_probability::differs_lot_from_p): New member function.
10794 (profile_probability::stream_in): New member function.
10795 (profile_probability::stream_out): New member function.
10796 * profile-count.h (profile_count_quality): Rename to ...
10797 (profile_quality): ... this one.
10798 (profile_probability): New.
10799 (profile_count): Update.
10800 * profile.c (compute_branch_probabilities): Update.
10801 * recog.c (peep2_attempt): Update.
10802 * sched-ebb.c (schedule_ebbs): Update.
10803 * sched-rgn.c (find_single_block_region): Update.
10804 (compute_dom_prob_ps): Update.
10805 (schedule_region): Update.
10806 * sel-sched-ir.c (compute_succs_info): Update.
10807 * stmt.c (struct case_node): Update.
10808 (do_jump_if_equal): Update.
10809 (get_outgoing_edge_probs): Update.
10810 (conditional_probability): Update.
10811 (emit_case_dispatch_table): Update.
10812 (expand_case): Update.
10813 (expand_sjlj_dispatch_table): Update.
10814 (emit_case_nodes): Update.
10815 * targhooks.c: Update.
10816 * tracer.c (better_p): Update.
10817 (find_best_successor): Update.
10818 * trans-mem.c (expand_transaction): Update.
10819 * tree-call-cdce.c: Update.
10820 * tree-cfg.c (gimple_split_edge): Upate.
10821 (move_sese_region_to_fn): Upate.
10822 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10823 * tree-eh.c (lower_resx): Upate.
10824 (cleanup_empty_eh_move_lp): Upate.
10825 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10826 * tree-inline.c (copy_edges_for_bb): Update.
10827 (copy_cfg_body): Update.
10828 * tree-parloops.c (gen_parallel_loop): Update.
10829 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10830 (gimple_gen_time_profiler): Update.
10831 * tree-ssa-dce.c (remove_dead_stmt): Update.
10832 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10833 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10834 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10835 (unloop_loops): Update.
10836 (try_peel_loop): Update.
10837 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10838 * tree-ssa-loop-split.c (connect_loops): Update.
10839 (split_loop): Update.
10840 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10841 (hoist_guard): Update.
10842 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10843 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10844 (value_replacement): Update.
10845 * tree-ssa-reassoc.c (branch_fixup): Update.
10846 * tree-ssa-tail-merge.c (replace_block_by): Update.
10847 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10848 (create_edge_and_update_destination_phis): Update.
10849 (compute_path_counts): Update.
10850 (recompute_probabilities): Update.
10851 (update_joiner_offpath_counts): Update.
10852 (freqs_to_counts_path): Update.
10853 (duplicate_thread_path): Update.
10854 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10855 (struct switch_conv_info): Update.
10856 (gen_inbound_check): Update.
10857 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10858 (vect_do_peeling): Update.
10859 (vect_loop_versioning): Update.
10860 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10861 (optimize_mask_stores): Update.
10862 * ubsan.c (ubsan_expand_null_ifn): Update.
10863 * value-prof.c (gimple_divmod_fixed_value): Update.
10864 (gimple_divmod_fixed_value_transform): Update.
10865 (gimple_mod_pow2): Update.
10866 (gimple_mod_pow2_value_transform): Update.
10867 (gimple_mod_subtract): Update.
10868 (gimple_mod_subtract_transform): Update.
10869 (gimple_ic): Update.
10870 (gimple_stringop_fixed_value): Update.
10871 (gimple_stringops_transform): Update.
10872 * value-prof.h: Update.
10874 2017-06-29 Carl Love <cel@us.ibm.com>
10876 * config/rs6000/rs6000-c.c: Add support for built-in functions
10877 vector signed int vec_signed (vector float);
10878 vector signed long long vec_signed (vector double);
10879 vector signed int vec_signed2 (vector double, vector double);
10880 vector signed int vec_signede (vector double);
10881 vector signed int vec_signedo (vector double);
10882 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10883 instruction generator.
10884 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10885 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10886 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10888 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10889 vunsignede_v2df): Add define_expands.
10890 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10891 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10892 VEC_UNSIGNEDO): Add definitions.
10893 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10894 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10895 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10896 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10897 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10898 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10899 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10900 vec_unsignede, vec_unsignedo): Add builtin defines.
10901 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10903 * doc/extend.texi: Update the built-in documentation file for the
10904 new built-in functions.
10906 2017-06-29 Richard Biener <rguenther@suse.de>
10908 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10909 reduction chains to LOOP_VINFO_REDUCTIONS.
10910 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10911 SLP reductions after processing reduction chains.
10913 2017-06-29 Nathan Sidwell <nathan@acm.org>
10915 * builtins.c (fold_builtin_FUNCTION): Use
10916 lang_hooks.decl_printable_name.
10918 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10920 PR middle-end/81194
10921 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10922 with only one label.
10923 * stmt.c (expand_case): Assert NCASES is greater than one.
10925 2017-06-29 Richard Biener <rguenther@suse.de>
10927 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10929 (group_case_labels): Likewise.
10930 (find_taken_edge): Push sanity checking on val to workers...
10931 (find_taken_edge_cond_expr): ... here
10932 (find_taken_edge_switch_expr): ... and here, handle cases
10933 with just a default label.
10934 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10935 (group_case_labels): Likewise.
10936 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10937 group_case_labels does anything cleanup the CFG again.
10939 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10941 PR tree-optimization/81196
10942 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10943 exit condition comparing two IVs.
10945 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10947 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10948 profile to the dummy entry at the end of the list of architectures.
10949 * config/arm/arm-cpu-cdata.h: Regenerated.
10951 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10952 Michael Collison <michael.collison@arm.com>
10955 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10957 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10958 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10959 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10960 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10961 of shift when the shift amount is masked with constant equal to
10962 the size of the mode.
10963 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10966 2017-06-29 Martin Liska <mliska@suse.cz>
10968 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10970 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10973 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10976 (edge_growth_cache_entry::edge_growth_cache_entry): New
10978 (reset_edge_growth_cache): Update to use constructor.
10980 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10982 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10983 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10984 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10986 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10988 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10989 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10991 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10993 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10994 (*-linux-uclibc*): Add t-uclibc tmake_file.
10995 * config/t-musl: New.
10996 * config/t-uclibc: New.
10998 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11000 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11002 (gen_comm_data): Emit architectural setting of arch_prof.
11003 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11005 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11006 (armv8-m.base, armv8-m.main): Likewise.
11007 * arm-protos.h (arm_build_target): Add profile field.
11008 (arch_option): Likewise.
11009 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11011 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11012 arm_active_target.profile.
11014 2017-06-28 Richard Biener <rguenther@suse.de>
11016 PR middle-end/81227
11017 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11018 TYPE_OVERFLOW_WRAPS.
11019 * match.pd (negate_expr_p): Likewise.
11020 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11021 fold_build2, not fold_binary.
11023 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11025 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11026 Convert memory address to Pmode.
11027 (aarch64_print_operand): Assert MEM operands are always Pmode.
11029 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11032 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11033 Remove redundant if.
11034 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11035 * config/arm/aarch-common-protos.h
11036 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11037 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11039 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11042 * multiple_target.c (create_dispatcher_calls): Set the default
11043 clone to be static, not public.
11045 2017-06-28 Richard Biener <rguenther@suse.de>
11047 * tree-vect-loop.c (vectorizable_reduction): Move special
11048 cond reduction IV var creation ...
11049 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11050 parameter. Use STMT_VINFO_VECTYPE.
11051 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11054 2017-06-28 Martin Liska <mliska@suse.cz>
11057 * ipa-visibility.c (non_local_p): Handle visibility.
11059 2017-06-28 Martin Liska <mliska@suse.cz>
11062 * common.opt: Add IntegerRange to various options.
11063 * opt-functions.awk (integer_range_info): New function.
11064 * optc-gen.awk: Add integer_range_info to cl_options struct.
11065 * opts-common.c (decode_cmdline_option): Handle
11066 CL_ERR_INT_RANGE_ARG.
11067 (cmdline_handle_error): Likewise.
11068 * opts.c (print_filtered_help): Show valid interval in
11069 when --help is provided.
11070 * opts.h (struct cl_option): Add range_min and range_max fields.
11071 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11073 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11075 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11076 (x * C EQ/NE y * C): New transformation.
11078 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11080 * genmultilib (combination_space): Accept '+' in option names.
11082 2017-06-28 Martin Liska <mliska@suse.cz>
11085 * asan.c (instrument_derefs): Bail out inner references
11086 that are hard register variables.
11088 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11091 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11092 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11094 2017-06-28 Richard Biener <rguenther@suse.de>
11096 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11097 (vect_get_slp_defs): Adjust.
11098 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11100 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11102 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11103 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11104 (vectorizable_reduction): Adjust.
11105 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11107 (vect_get_slp_defs): Likewise.
11108 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11109 (vectorizable_bswap): Adjust.
11110 (vectorizable_call): Likewise.
11111 (vectorizable_conversion): Likewise.
11112 (vectorizable_assignment): Likewise.
11113 (vectorizable_shift): Likewise.
11114 (vectorizable_operation): Likewise.
11115 (vectorizable_store): Likewise.
11116 (vectorizable_condition): Likewise.
11117 (vectorizable_comparison): Likewise.
11119 2017-06-28 Michael Collison <michael.collison@arm.com>
11122 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11124 (arm_gen_movmemqi): Removed unused variable 'i'.
11125 Convert 'for' loop into 'while' loop.
11126 (arm_expand_prologue): Remove last unnecessary set of insn.
11127 (thumb_pop): Remove unused variable 'pushed_words'.
11128 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11130 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11132 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11133 * config/s390/s390-protos.h: Add prototype of
11134 s390_rel_address_ok_p.
11135 * config/s390/s390.c (s390_got_symbol): New function.
11136 (s390_rel_address_ok_p): New function.
11137 (legitimize_pic_address): Use s390_rel_address_ok_p.
11138 (s390_load_got): Use s390_got_symbol.
11139 (s390_option_override): Issue error if
11140 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11141 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11143 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11145 2017-06-27 Andrew Pinski <apinski@cavium.com>
11147 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11148 (X * copysign (1.0, X)): New pattern.
11149 (X * copysign (1.0, -X)): New pattern.
11150 (copysign (-1.0, CST)): New pattern.
11152 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11154 * genmultilib (combination_space): Remove variable.
11155 Validate reuse rules against regular expression for any sequence
11156 of multilib options in any order.
11158 2017-06-27 Michael Collison <michael.collison@arm.com>
11160 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11161 call aarch64_split_simd_combine.
11162 * (aarch64_combine_internal<mode>): Delete pattern.
11163 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11164 Allow register and subreg operands.
11166 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11168 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11169 specific need, just fallback on defaults.
11170 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11172 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11173 Olivier Hainque <hainque@adacore.com>
11175 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11177 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11178 targets. Pick a default if no particular attempt applied.
11179 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11182 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11184 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11185 (x86_64-wrs-vxworks7): Likewise.
11187 2017-06-27 Marek Polacek <polacek@redhat.com>
11190 * ubsan.c (instrument_null): Check get_base_address's result for null.
11192 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11194 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11196 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11198 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11199 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11200 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11201 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11202 New function types.
11203 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11204 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11205 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11206 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11207 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11208 BUILT_IN_FEUPDATEENV): New builtins.
11209 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11210 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11211 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11212 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11214 (builtin_structptr_types): Adjust size.
11215 * tree.c (builtin_structptr_types): Add four entries.
11217 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11218 Olivier Hainque <hainque@adacore.com>
11220 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11221 (TLS_SYM): New local macro, forcing reference to __tls__ on
11222 link command lines for VxWorks 7 RTPs, triggering initialization
11224 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11225 OS features TLS support, true for RTPs on VxWorks 7.
11226 * config/vxworks.c (vxworks_override_options): Setup emutls
11229 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11231 * predict.c (test_prediction_value_range): Use -1U instead of -1
11232 to avoid narrowing conversion warning.
11233 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11234 to avoid narrowing conversion warning.
11235 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11237 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11239 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11241 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11242 64bit configurations.
11243 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11244 (SIZE_TYPE): Likewise.
11245 * config/vxworks.c (vxworks_emutls_var_fields): Use
11246 long_unsigned_type_node instead of unsigned_type_node as the offset
11247 field type, which is "pointer" mode in emutls.c.
11249 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11252 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11254 PR middle-end/81207
11255 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11256 gimple_vuse copying separately from gimple_vdef copying.
11258 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11260 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11263 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11264 Olivier Hainque <hainque@adacore.com>
11266 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11267 for all vxworks7 targets.
11268 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11269 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11270 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11271 variations for VX6/VX7 and 32/64bits later on in ...
11272 (VXWORKS_LIB_SPEC): Leverage new macros.
11273 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11274 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11276 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11278 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11279 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11281 2017-06-26 Carl Love <cel@us.ibm.com>
11283 * config/rs6000/rs6000-c.c: Add support for built-in functions
11284 vector bool char vec_reve (vector bool char);
11285 vector signed char vec_reve (vector signed char);
11286 vector unsigned char vec_reve (vector unsigned char);
11287 vector bool int vec_reve (vector bool int);
11288 vector signed int vec_reve (vector signed int);
11289 vector unsigned int vec_reve (vector unsigned int);
11290 vector bool long long vec_reve (vector bool long long);
11291 vector signed long long vec_reve (vector signed long long);
11292 vector unsigned long long vec_reve (vector unsigned long long);
11293 vector bool short vec_reve (vector bool short);
11294 vector signed short vec_reve (vector signed short);
11295 vector double vec_reve (vector double);
11296 vector float vec_reve (vector float);
11297 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11298 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11299 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11300 (altivec_vreve): New pattern.
11301 * config/rs6000/altivec.h (vec_reve): New define.
11302 * doc/extend.texi (vec_rev): Update the built-in documentation file
11303 for the new built-in functions.
11305 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11307 PR tree-optimization/71815
11308 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11310 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11312 (slsr_process_phi): Likewise.
11313 (replace_uncond_cands_and_profitable_phis): Don't replace a
11314 multiply candidate with a stride of 1 (copy or cast).
11315 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11317 (lowest_cost_path): Likewise.
11318 (total_savings): Likewise.
11320 2017-06-26 Richard Biener <rguenther@suse.de>
11323 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11324 Use def_builtin_pure for all gather builtins.
11326 2017-06-26 Richard Biener <rguenther@suse.de>
11328 PR tree-optimization/81203
11329 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11330 non-dominating BBs.
11332 2017-06-26 Marek Polacek <polacek@redhat.com>
11335 * doc/invoke.texi: Document -Wmultistatement-macros.
11337 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11339 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11340 arm_neon_ok_no_float_abi effective target.
11342 2017-06-26 Richard Biener <rguenther@suse.de>
11344 PR tree-optimization/80928
11345 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11346 (copy_bbs): Set BB_DUPLICATED flag early.
11347 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11349 (execute_on_shrinking_pred): Likewise.
11350 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11351 BB_DUPLICATED blocks.
11352 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11353 iterate over all PHIs considering removal of *gsi.
11355 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11357 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11360 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11362 * config/rs6000/rs6000-string.c: (expand_block_clear,
11363 do_load_for_compare, select_block_compare_mode,
11364 compute_current_alignment, expand_block_compare,
11365 expand_strncmp_align_check, expand_strn_compare,
11366 expand_block_move, rs6000_output_load_multiple)
11367 Move functions related to string/block move/compare
11368 to a separate file.
11369 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11370 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11371 for this function which is now used in two files.
11372 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11373 * config.gcc: Add rs6000-string.o to extra_objs for
11374 targets powerpc*-*-* and rs6000*-*-*.
11376 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11379 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11380 32-bit, since indexed is not valid for DImode.
11381 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11382 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11383 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11384 (define_peephole2 for Altivec d-form store): Likewise.
11387 * multiple_target.c (create_dispatcher_calls): Only create the
11388 dispatcher call if the function is the default clone of a
11389 versioned function.
11391 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11393 PR middle-end/80902
11394 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11395 a call, force the call to not be a tail call.
11397 2017-06-23 Jeff Law <law@redhat.com>
11399 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11402 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11404 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11405 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11406 handling for early expansion of vector shifts (sl,sr,sra,rl).
11407 (builtin_function_type): Add vector shift right instructions
11408 to the unsigned argument list.
11410 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11412 rtl-optimizatoin/79286
11413 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11414 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11415 trap. PIC register plus a const unspec without offset can never trap.
11417 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11419 * tree.h (builtin_structptr_type): New type.
11420 (builtin_structptr_types): Declare new array.
11421 * tree.c (builtin_structptr_types): New array.
11422 (free_lang_data, build_common_tree_nodes): Use it.
11424 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11427 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11430 2017-06-22 Matt Turner <mattst88@gmail.com>
11432 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11433 Lake models to skylake case. Assume skylake for unknown
11434 models with clflushopt.
11436 2017-06-22 Jeff Law <law@redhat.com>
11438 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11439 frame sizes that do not satisfy aarch64_uimm12_shift.
11441 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11443 * profile-count.h (apply_probability,
11444 apply_scale, probability_in): Fix checks for zero.
11446 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11448 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11449 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11451 2016-06-22 Richard Biener <rguenther@suse.de>
11453 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11454 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11455 REDUC_MAX_EXPR support.
11456 (vectorizable_reduction): Likewise.
11457 (vect_create_epilog_for_reduction): Likewise.
11459 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11461 * match.pd (A / (1 << B) -> A >> B): New.
11462 * generic-match-head.c: Include optabs-tree.h.
11463 * gimple-match-head.c: Likewise.
11464 * optabs-tree.h (target_supports_op_p): New.
11465 * optabs-tree.c (target_supports_op_p): New.
11467 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11469 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11470 $gcc_cv_ld --help output.
11471 (gcc_cv_ld_demangle): Likewise.
11472 (gcc_cv_ld_eh_frame_hdr): Likewise.
11473 (gcc_cv_ld_pie): Likewise.
11474 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11475 (gcc_cv_ld_buildid): Likewise.
11476 (gcc_cv_ld_sysroot): Likewise.
11477 (ld_bndplt_support): Likewise.
11478 (ld_pushpopstate_support): Likewise.
11479 * configure: Regenerate.
11480 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11482 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11485 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11486 operands indexes to avoid gap between operands and match_dups.
11488 2017-06-21 Andrew Pinski <apinski@cavium.com>
11490 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11491 Increment Arith_shift and Arith_shift_reg by 1.
11492 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11494 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11495 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11496 (aarch64_strip_extend): Add new argument and test for it.
11497 (aarch64_cheap_mult_shift_p): New function.
11498 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11499 add a cost if it is true.
11500 Update calls to aarch64_strip_extend.
11501 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11503 2017-06-21 Andrew Pinski <apinski@cavium.com>
11505 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11507 (thunderxt88): Likewise.
11508 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11509 (thunderx_prefetch_tune): New variable.
11510 (thunderx2t99_prefetch_tune): Update for the correct values.
11511 (thunderxt88_tunings): New variable.
11512 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11513 generic_prefetch_tune.
11514 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11516 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11518 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11519 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11520 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11521 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11522 (aarch64_atomic_cas<mode>, GPI): Likewise.
11524 2017-06-21 Martin Liska <mliska@suse.cz>
11526 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11527 statements on cold and hot labels.
11528 * predict.c (tree_estimate_probability_bb): Remove the
11529 prediction from this place.
11531 2017-06-21 Martin Liska <mliska@suse.cz>
11533 PR tree-optimization/79489
11534 * gimplify.c (maybe_add_early_return_predict_stmt): New
11536 (gimplify_return_expr): Call the function.
11537 * predict.c (tree_estimate_probability_bb): Remove handling
11539 * predict.def: Update comment about early return predictor.
11540 * gimple-predict.h (is_gimple_predict): New function.
11541 * predict.def: Change default value of early return to 66.
11542 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11544 * passes.def: Put pass_strip_predict_hints to the beginning of
11547 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11549 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11550 FUNCTION_DECL declarations.
11551 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11553 (dwaf2out_decl): Likewise.
11554 * godump.c (go_early_global_decl): Skip call to the real debug hook
11555 for FUNCTION_DECL declarations.
11556 * passes.c (rest_of_decl_compilation): Skip call to the
11557 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11558 -fdump-go-spec is passed.
11560 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11562 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11563 Reorder for compactness.
11564 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11565 (def_builtin_pure, def_builtin_pure2): New functions.
11566 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11568 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11570 * match.pd (nop_convert): New predicate.
11571 ((A +- CST1) +- CST2): Allow some NOP conversions.
11573 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11576 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11577 with ctors/dtors if GOVD_SHARED is set.
11579 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11581 * config/aarch64/aarch64.md (movti_aarch64):
11582 Emit mov rather than orr.
11583 (movtf_aarch64): Likewise.
11584 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11585 Emit mov rather than orr.
11587 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11589 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11590 Swap alternatives, make integer dup more expensive.
11592 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11594 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11595 Return true for non-tls symbols.
11597 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11599 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11600 (cortex-a75): Likewise.
11601 (cortex-a75.cortex-a55): Likewise.
11602 * config/aarch64/aarch64-tune.md: Regenerate.
11603 * doc/invoke.texi (-mtune): Document new values for -mtune.
11605 2017-06-21 Tom de Vries <tom@codesourcery.com>
11607 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11608 stack_size feature.
11609 (Effective-Target Keywords, Other attributes): Suggest using
11610 dg-add-options stack_size feature to get stack limit in stack_size
11611 effective target documentation.
11613 2017-06-21 Julian Brown <julian@codesourcery.com>
11614 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11616 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11617 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11618 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11620 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11621 attribute type list for neon_multiply.
11622 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11623 attribute type list for neon_multiply.
11624 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11625 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11626 attribute type list for neon_multiply.
11627 * config/arm/types.md (crypto_pmull): Add.
11628 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11629 attribute type list.
11631 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11633 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11636 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11638 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11639 to make sure not to dereference a NULL cost_classes_ptr pointer.
11641 2017-06-20 Carl Love <cel@us.ibm.com>
11643 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11644 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11645 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11646 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11647 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11648 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11649 VMULOSW): New enum "unspec" values.
11650 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11651 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11652 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11653 altivec_vmulosw): New patterns.
11654 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11655 VMULOSW): Add definitions.
11657 2017-06-20 Julia Koval <julia.koval@intel.com>
11659 * config/i386/i386.c: Fix rounding expand for new pattern.
11660 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11662 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11664 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11665 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11667 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11669 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11672 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11674 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11675 architecture, then by implementer ID.
11676 * config/aarch64/aarch64-tune.md: Regenerate.
11678 2017-06-20 Richard Biener <rguenther@suse.de>
11680 PR middle-end/81097
11681 * fold-const.c (split_tree): Fold to type before negating.
11683 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11685 * diagnostic-show-locus.c
11686 (selftest::test_fixit_deletion_affecting_newline): New function.
11687 (selftest::diagnostic_show_locus_c_tests): Call it.
11689 2017-06-20 Andreas Schwab <schwab@suse.de>
11692 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11695 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11697 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11698 __ARM_FEATURE_COPROC according to support.
11700 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11702 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11703 Rewritten to avoid overflow for > 32-bit pointers.
11706 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11707 by removing enum keyword.
11708 (ubsan_type_descriptor): Likewise. Formatting fix.
11711 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11712 splitter): Require TARGET_SSE2 in the condition.
11714 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11717 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11718 for doing vector set of SFmode on ISA 3.0.
11719 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11720 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11722 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11723 SFmode value into a V4SF variable that was extracted from another
11724 V4SF variable without converting the element to double precision
11725 and back to single precision vector format.
11726 (vsx_insert_extract_v4sf_p9_2): Likewise.
11728 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11730 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11731 in UWHI to avoid undefined overflow.
11734 * ubsan.h (enum ubsan_encode_value_phase): New.
11735 (ubsan_encode_value): Change second argument to
11736 enum ubsan_encode_value_phase with default value of
11737 UBSAN_ENCODE_VALUE_GENERIC.
11738 * ubsan.c (ubsan_encode_value): Change second argument to
11739 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11740 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11741 create_tmp_var_raw instead of create_tmp_var and use a
11743 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11744 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11745 ubsan_encode_value callers.
11748 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11749 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11750 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11752 2017-06-19 Richard Biener <rguenther@suse.de>
11754 PR middle-end/81118
11755 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11756 estimates if we changed anything.
11758 2017-06-19 Richard Biener <rguenther@suse.de>
11760 PR tree-optimization/80887
11761 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11762 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11763 simplified lookups, then reset mprts_hook.
11764 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11766 (try_to_simplify): Likewise.
11768 2017-06-19 Martin Liska <mliska@suse.cz>
11771 * gimplify.c (gimplify_switch_expr):
11772 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11774 2017-06-19 Martin Liska <mliska@suse.cz>
11776 * doc/install.texi: Document that PGO runs in 4 stages.
11778 2017-06-19 Martin Liska <mliska@suse.cz>
11781 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11782 to dispatcher function name.
11783 * multiple_target.c (replace_function_decl): New function.
11784 (create_dispatcher_calls): Redirect both edges and references.
11786 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11788 * profile-count.c (profile_count::dump): Dump quality.
11789 (profile_count::differs_from_p): Update for unsigned val.
11790 * profile-count.h (profile_count_quality): New enum.
11791 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11793 2017-06-19 Richard Biener <rguenther@suse.de>
11795 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11796 struct function as arg.
11797 (estimate_numbers_of_iterations): Export overload with loop arg.
11798 (free_numbers_of_iterations_estimates_loop): Use an overload of
11799 free_numbers_of_iterations_estimates instead.
11800 * tree-cfg.c (remove_bb): Adjust.
11801 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11802 * tree-parloops.c (gen_parallel_loop): Likewise.
11803 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11805 (tree_unroll_loops_completely): Likewise.
11806 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11807 Use an overload instead and export.
11808 (estimated_loop_iterations): Adjust.
11809 (max_loop_iterations): Likewise.
11810 (likely_max_loop_iterations): Likewise.
11811 (estimate_numbers_of_iterations): Take struct function as arg
11813 (loop_exits_before_overflow): Adjust.
11814 (free_numbers_of_iterations_estimates_loop): Use an overload.
11815 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11816 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11818 2017-06-19 Richard Biener <rguenther@suse.de>
11821 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11822 RANGE_EXPR conservatively.
11824 2017-06-16 Carl Love <cel@us.ibm.com>
11826 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11827 definitions for vec_float, vec_float2, vec_floato,
11828 vec_floate built-ins.
11829 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11830 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11832 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11833 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11834 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11835 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11836 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11837 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11838 vec_floato): Add builtin defines.
11839 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11840 Update the built-in documentation file for the new built-in
11843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11845 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11846 (mthumb): Mark as the negative of -marm.
11848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11850 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11852 (ARM Options, -mtune): Document that this accepts the same
11853 extension options as -mcpu.
11854 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11858 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11859 permitted extensions.
11861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11863 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11864 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11865 (armv8-m.main): Add option +nodsp.
11866 * config/arm/arm-cpu-cdata.h: Regenerated.
11868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11870 * config/arm/t-fuchsia: New file.
11871 * config.gcc (arm*-*-fuchsia*): Use it.
11873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11875 * config/arm/t-symbian: Rewrite for new option infrastructure.
11877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11879 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11880 (MULTILIB_REQUIRED): Likewise.
11882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11884 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11885 (MULTILIB_RESUE): Likewise.
11886 (MULTILIB_MATCHES): Likewise.
11887 (MULTLIB_REQUIRED): Likewise.
11889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11891 * config/arm/t-rtems: Rewrite for new option framework.
11893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11895 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11896 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11897 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11898 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11899 * config/arm/t-multilib: ... here.
11900 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11901 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11902 armv7-a and armv8*-a when A-profile libraries have not been built.
11903 * config/arm/t-rmprofile: Rewrite.
11905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11907 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11908 with a backslash. Remove the backslash after substituting unescaped
11910 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11912 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11914 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11915 the driver to pass through the default setting of -mfloat-abi.
11916 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11918 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11919 (all_feat_combs): New rule.
11920 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11922 * config/arm/t-aprofile: Rewrite.
11924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11926 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11927 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11928 fpu is not specified by the user/command-line.
11929 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11930 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11931 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11932 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11933 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11934 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11938 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11939 * config/arm/t-arm-elf: Rewritten.
11941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11943 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11944 have some floating-point instructions.
11945 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11946 (TARGET_MAYBE_HARD_FLOAT): New macro.
11947 * config/arm/arm-builtins.c (arm_init_builtins): Use
11948 TARGET_MAYBE_HARD_FLOAT.
11949 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11951 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11953 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11954 (configargs.h): Include it.
11955 (arm_print_hint_for_fpu_option): New function.
11956 (arm_parse_fpu_option): New function.
11957 (candidate_extension): New class.
11958 (arm_canon_for_multilib): New function.
11959 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11960 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11961 (ARCH_CANONICAL_SPECS): New macro.
11962 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11966 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11967 are set after handling multilib fragments. Set target_cpu_default2
11970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11972 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11974 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11975 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11977 (check_arch): Likewise.
11978 * config/arm/arm.c (arm_configure_build_target): Handle
11979 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11980 options in the default.
11982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11984 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11985 when an option is an alias of another.
11986 * config/arm/parsecpu.awk (optalias): New parser token.
11987 (gen_comm_data): Mark non-alias options as such. Emit entries
11988 for extension aliases.
11989 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11990 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11991 (armv6kz, armv6zk, armv6t2): Likewise.
11992 (armv7): Make vfpv3-d16 an alias.
11993 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11995 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11996 Sort in canonical order.
11997 (armv8-a): Sort in canonical order.
11998 (armv8.1-a, armv8.2-a): Likewise.
11999 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12001 (cortex-a9): Sort in canonical order.
12002 * config/arm/arm.c (selftests.h): Include it.
12003 (arm_test_cpu_arch_data): New function.
12004 (arm_run_self_tests): New function.
12005 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12006 (targetm): Move declaration to the end of the file.
12007 * arm-cpu-cdata.h: Regenerated.
12009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12011 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12012 call to target_mode_check describing the type of option passed.
12013 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12014 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12015 arm_parse_cpu_option_name to match parameters against list of
12017 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12018 arm_arch_core_flags data structure.
12019 * config/arm/arm-cpu_cdata.h: Regenerated.
12021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12023 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12025 (arm_print_hint_for_cpu_option): Likewise.
12026 (arm_print_hint_for_arch_option): Likewise.
12027 (arm_parse_cpu_option_name): Likewise.
12028 (arm_parse_arch_option_name): Likewise.
12029 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12030 of entries in the all_fpus list.
12031 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12032 (arm_parse_cpu_option_name): Declare.
12033 (arm_parse_arch_option_name): Declare.
12034 (arm_parse_option_features): Declare.
12035 (arm_intialize_isa): Declare.
12036 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12038 (gen_comm_data): ... here. Make definitions non-static.
12039 * config/arm/arm-cpu-data.h: Regenerated.
12040 * config/arm/arm-cpu-cdata.h: Regenerated.
12042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12044 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12045 (cpu_arch_extension): New structure.
12046 (cpu_arch_option, arch_option, cpu_option): New structures.
12047 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12048 architecture types.
12049 (gen_data): Generate new format data tables.
12050 * config/arm/arm.c (cpu_tune): New structure.
12051 (cpu_option, processors): Delete.
12052 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12053 (arm_print_hint_for_cpu_option): ... this and ...
12054 (arm_print_hint_for_arch_option): ... this.
12055 (arm_parse_arch_cpu_name): Delete. Replace with ...
12056 (arm_parse_cpu_option_name): ... this and ...
12057 (arm_parse_arch_option_name): ... this.
12058 (arm_unrecognized_feature): Change type of target parameter to
12060 (arm_parse_arch_cpu_features): Delete. Replace with ...
12061 (arm_parse_option_features): ... this.
12062 (arm_configure_build_target): Rework to use new configuration data
12064 (arm_print_tune_info): Rework for new configuration data tables.
12065 * config/arm/arm-cpu-data.h: Regenerated.
12066 * config/arm/arm-cpu.h: Regenerated.
12068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12070 * Makefile.in (OBJS): Move sbitmap.o from here ...
12071 (OBJS-libcommon): ... to here.
12073 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12075 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12076 (ISA_ALL_CRYPTO): New macro.
12077 (ISA_ALL_SIMD): New macro
12078 (ISA_ALL_FP): New macro.
12079 * config/arm/arm.c (fpu_bitlist): Update initializer.
12080 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12082 (arm9e): Add fpu. Add option for nofp
12083 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12084 (arm926ej-s, arm1026ej-s): Likewise.
12085 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12086 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12087 neon-fp16, neon-vfpv4, nofp and nosimd.
12088 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12089 (cortex-a8): Add fpu. Add option for nofp.
12090 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12091 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12092 (cortex-r4f): Add fpu.
12093 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12094 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12096 (cortex-r8): Likewise.
12097 (cortex-m4): Add fpu. Add option for nofp.
12098 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12099 (cortex-a17.cortex-a7): Likewise.
12100 (cortex-a32): Add fpu. Add options for crypto and nofp.
12101 (cortex-a35, cortex-a53): Likewise.
12102 (cortex-a57): Add fpu. Add option for crypto.
12103 (cortex-a72, cortex-a73): Likewise.
12104 (exynos-m1): Likewise.
12105 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12106 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12107 (cortex-m33): Add fpu. Add option for nofp.
12108 * config/arm/arm-cpu-cdata.h: Regenerated
12109 * config/arm/arm-cpu-data.h: Regenerated.
12111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12113 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12114 (armv5te, armv5tej): Likewise.
12115 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12116 (armv7): Add options fp and vfpv3-d16.
12117 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12118 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12120 (armv7ve): Likewise.
12121 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12122 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12123 (armv8-a): Add nocrypto option.
12124 (armv8.1-a, armv8.2-a): Likewise.
12125 (armv8-m.main): add options fp, fp.dp and nofp.
12127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12129 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12131 (armv8-a+crc): Delete.
12132 (armv8.1-a): Add options simd, crypto and nofp.
12133 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12134 (armv8.2-a+fp16): Delete.
12135 (armv8-m.main): Add option dsp.
12136 (armv8-m.main+dsp): Delete.
12137 (cortex-a8): Add fpu. Add nofp option.
12138 (cortex-a9): Add fpu. Add nofp and nosimd options.
12139 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12140 link to main cpu and architecture data structures.
12141 (gen_comm_data): Only put isa attributes from the main architecture
12143 (option): New statement for architecture and CPU entries.
12144 * arm.c (struct cpu_option): New structure.
12145 (struct processors): Add entry for options.
12146 (arm_unrecognized_feature): New function.
12147 (arm_parse_arch_cpu_name): Ignore any characters after the first
12149 (arm_parse_arch_cpu_feature): New function.
12150 (arm_configure_build_target): Separate out any CPU and architecture
12151 features and parse separately. Don't error out if -mfpu=auto is
12152 used with only an architecture string.
12153 (arm_print_asm_arch_directives): New function.
12154 (arm_file_start): Call it.
12155 * config/arm/arm-cpu-cdata.h: Regenerated.
12156 * config/arm/arm-cpu-data.h: Likewise.
12157 * config/arm/arm-tables.opt: Likewise.
12159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12161 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12162 assembler when it is not -mfpu=auto.
12164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12166 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12167 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12168 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12169 (ASM_CPU_SPEC): Rewrite.
12170 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12171 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12172 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12173 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12174 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12175 copied string is NUL-terminated. Also strip any characters prefixed
12177 (arm_rewrite_selected_arch): New function.
12178 (arm_rewrite_march): New function.
12180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12182 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12183 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12184 (march, mcpu, mtune): Convert to string-based options.
12185 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12186 (arm_parse_arch_cpu_name): New function.
12187 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12188 identify selected architecture or CPU.
12189 (arm_option_save): New function.
12190 (TARGET_OPTION_SAVE): Redefine.
12191 (arm_option_restore): Restore string options.
12192 (arm_option_print): Print string options.
12194 2017-06-16 Martin Sebor <msebor@redhat.com>
12196 PR tree-optimization/80933
12197 PR tree-optimization/80934
12198 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12199 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12200 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12201 (gimple_fold_builtin): Call them.
12203 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12205 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12206 as unlikely; update profile.
12208 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12210 * predict.c (force_edge_cold): Handle declaring edges impossible
12213 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12215 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12217 (try_unroll_loop_completely): Fix reporting.
12219 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12221 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12223 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12226 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12227 if given a non-constant argument for an intrinsic which requires a
12230 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12232 * profile.c (compare_freqs): New function.
12233 (branch_prob): Sort edge list.
12234 (find_spanning_tree): Assume that the list is priority sorted.
12236 2017-06-16 Richard Biener <rguenther@suse.de>
12238 PR tree-optimization/81090
12239 * passes.def (pass_record_bounds): Remove.
12240 * tree-pass.h (make_pass_record_bounds): Likewise.
12241 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12242 make_pass_record_bounds): Likewise.
12243 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12244 not free niter estimates at the beginning but at the end.
12245 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12247 2017-06-16 Richard Biener <rguenther@suse.de>
12249 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12250 initializer to workaround ICE in host GCC 4.8.
12252 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12254 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12256 (clone_inlined_nodes): Update.
12258 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12260 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12261 prefetch settings, and enable prefetching by default at -O3.
12263 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12265 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12266 Set flag_prefetch_loop_arrays according to tuning data.
12268 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12270 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12271 New tune structure.
12272 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12273 [Unrelated to main purpose of the patch] Place the pointer field last
12274 to enable type checking errors when tune structure are wrongly merged.
12275 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12276 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12277 (thunderx2t99_prefetch_tune): New tune constants.
12278 (tune_params *_tunings): Update all tunings (no functional change).
12279 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12280 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12281 from tunings structures.
12283 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12286 * ubsan.c (instrument_null): Add T argument, use it instead
12287 of computing it based on IS_LHS.
12288 (instrument_object_size): Likewise.
12289 (pass_ubsan::execute): Adjust instrument_null and
12290 instrument_object_size callers to pass gimple_get_lhs or
12291 gimple_assign_rhs1 result to it. Use instrument_null instead of
12292 calling get_base_address and instrument_mem_ref. Handle
12293 aggregate call arguments for object-size sanitization.
12295 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12297 PR tree-optimization/81089
12298 * tree-vrp.c (is_masked_range_test): Validate operands of
12301 2017-06-15 Martin Sebor <msebor@redhat.com>
12304 * dumpfile.c (dump_register): Avoid calling memset to initialize
12305 a class with a default ctor.
12306 * gcc.c (struct compiler): Remove const qualification.
12307 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12308 * hash-table.h: Ditto.
12309 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12311 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12312 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12314 * params.h (struct param_info): Make struct members non-const.
12315 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12316 with default initialization.
12317 * vec.h (vec_copy_construct, vec_default_construct): New helper
12319 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12320 with vec_copy_construct.
12321 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12322 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12323 * doc/invoke.texi (-Wclass-memaccess): Document.
12325 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12327 * emit-rtl.h (is_leaf): Update comment about local
12328 register allocator.
12330 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12333 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12334 for a variable to have a section before checking if the section has a
12336 Set section to.persistent if persistent attribute is set.
12337 Warn if .persistent attribute is used on an automatic variable.
12339 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12341 PR rtl-optimization/80474
12342 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12344 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12346 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12349 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12351 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12352 (casesi): Emit bounds checking as RTL.
12353 (casesi_internal_mips16_<mode>): Remove bounds checking.
12355 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12357 * config/xtensa/xtensa.c (xtensa_option_override): Append
12358 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12359 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12360 xtensa_doloop_hooks): Define unconditionally.
12361 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12363 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12364 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12365 for it in xtensa_option_override.
12366 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12367 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12369 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12371 * doc/cppopts.texi: Document '-' special value to -MF.
12373 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12375 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12376 (cortex_a53_fconst): Likewise.
12377 (cortex_a53_fpmul): Likewise.
12378 (cortex_a53_f_load_64): Likewise.
12379 (cortex_a53_f_load_many): Likewise.
12380 (cortex_a53_advsimd_alu): Likewise.
12381 (cortex_a53_advsimd_alu_q): Likewise.
12382 (cortex_a53_advsimd_mul): Likewise.
12383 (cortex_a53_advsimd_mul_q): Likewise.
12384 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12385 Add missing fmul, r2f_cvt and fconst cases.
12387 2017-06-14 Richard Biener <rguenther@suse.de>
12389 PR middle-end/81088
12390 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12392 (fold_binary_loc): When associating do not treat pre-existing
12393 TREE_OVERFLOW on literal constants as a reason to allow
12394 TREE_OVERFLOW on associated literal constants.
12396 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12398 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12399 (MASK_FEATURES): New macro.
12400 * config/sparc/sparc.c (sparc_option_override): Remove the special
12401 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12403 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12405 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12406 a division of 0 if non-call exceptions are enabled.
12408 2017-06-14 Andrew Pinski <apinski@cavium.com>
12409 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12412 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12413 Improve vector initialization code gen for only variable case.
12415 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12417 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12419 2017-06-14 Richard Biener <rguenther@suse.de>
12421 PR tree-optimization/81083
12422 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12425 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12427 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12428 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12429 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12430 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12431 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12432 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12434 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12436 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12437 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12439 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12441 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12443 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12445 * config/rs6000/t-rtems: Don't handle SPE.
12447 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12449 * config/rs6000/t-linux: Don't handle SPE.
12451 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12453 * config/rs6000/eabispe.h: Delete file.
12455 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12457 * config/rs6000/t-spe: Delete file.
12459 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12461 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12462 (rs6000_legitimate_offset_address_p): Return false for anything in
12463 V2SImode or V2SFmode.
12465 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12467 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12468 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12469 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12471 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12472 (rs6000_legitimate_offset_address_p): Ditto.
12473 (rs6000_emit_move): Ditto.
12474 (rs6000_init_builtins): Remove V4HI_type_node.
12476 2017-06-13 Martin Liska <mliska@suse.cz>
12479 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12480 (gate_asan): Likewise.
12481 * asan.h (asan_no_sanitize_address_p): Remove the function.
12482 (sanitize_flags_p): New function.
12483 * builtins.def: Fix coding style.
12484 * common.opt: Use renamed enum value.
12485 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12486 * doc/extend.texi: Document no_sanitize attribute.
12487 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12488 to SANITIZE_UNDEFINED_NONDEFAULT.
12489 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12490 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12491 Use sanitize_flags_p.
12492 * gimplify.c (gimplify_function_tree): Likewise.
12493 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12494 * opts.c (parse_no_sanitize_attribute): New function.
12495 (common_handle_option): Use renamed enum value.
12496 * opts.h (parse_no_sanitize_attribute): Declare.
12497 * tree.c (sanitize_flags_p): New function.
12498 * tree.h: Declared here.
12499 * tsan.c: Use sanitize_flags_p.
12500 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12501 (instrument_mem_ref): Likewise.
12502 (instrument_bool_enum_load): Likewise.
12503 (do_ubsan_in_current_function): Remove the function.
12504 (pass_ubsan::execute): Use sanitize_flags_p.
12505 * ubsan.h: Remove do_ubsan_in_current_function
12506 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12507 (dump_function_to_file): Use it here.
12509 2017-06-13 Martin Jambor <mjambor@suse.cz>
12511 PR tree-optimization/80803
12512 PR tree-optimization/81063
12513 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12514 (propagate_subaccesses_across_link): Enqueue subtree whenever
12515 necessary instead of relying on the caller.
12517 2017-06-13 Martin Jambor <mjambor@suse.cz>
12519 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12520 that have a first_link.
12521 (sort_and_splice_var_accesses): Do not check first_link before
12523 (subtree_mark_written_and_enqueue): Likewise.
12524 (propagate_all_subaccesses): Likewise and do not stop at first
12525 parent with a first_link.
12527 2017-06-13 Martin Jambor <mjambor@suse.cz>
12529 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12532 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12534 * match.pd: New pattern.
12536 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12538 * tree-vrp.c (is_masked_range_test): New function.
12539 (register_edge_assert_for): Determine ranges for
12542 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12544 PR tree-optimization/67328
12545 * fold-const.c (maskable_range_p): New function.
12546 (build_range_check): Generate bittests if possible.
12548 2017-06-13 Martin Liska <mliska@suse.cz>
12550 * gimple-pretty-print.c (dump_probability): Add new argument.
12551 (dump_edge_probability): Dump both probability and count.
12552 (dump_gimple_label): Likewise.
12553 (dump_gimple_bb_header): Likewise.
12555 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12558 * config/avr/avr-devices.c: Fix indentation.
12559 * config/avr/gen-avr-mmcu-specs.c: Dito.
12561 2017-06-13 Richard Biener <rguenther@suse.de>
12563 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12564 instead get vector type from stmt_info.
12565 (vectorizable_reduction): Adjust. Remove dead code.
12567 2017-06-13 Richard Biener <rguenther@suse.de>
12569 PR middle-end/81065
12570 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12571 case of C * (x * C2 + C3).
12572 (fold_addr_of_array_ref_difference): Properly fold index difference.
12574 2017-06-12 David S. Miller <davem@davemloft.net>
12577 * config/sparc/sparc.md (return expander): Emit frame blockage if
12578 function uses alloca.
12580 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12582 * combine.c (make_field_assignment): Check len rather than the mode
12583 precision when calling force_to_mode.
12585 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12587 Support multilibs and devices that see flash in RAM address range.
12590 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12591 (avr_mcu_t) <flash_pm_offset>: New field.
12592 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12593 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12594 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12595 (AVR_TINY_PM_OFFSET): Remove macro.
12596 * config/avr/avr.opt (-mshort-calls): New option.
12597 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12598 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12599 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12600 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12601 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12602 instead of avr_arch->have_jmp_call.
12603 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12604 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12605 avr_arch->flash_pm_offset to define.
12606 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12607 new field flash_pm_offset. Add entry for avrxmega3.
12608 (avr_texinfo): Add entry for avrxmega3.
12609 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12610 attiny212, attiny214,
12611 attiny412, attiny414, attiny416, attiny417,
12612 attiny814, attiny816, attiny817,
12613 attiny1614, attiny1616, attiny1617,
12614 attiny3214, attiny3216, attiny3217.
12615 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12616 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12617 (avr_print_operand_address) [AVR_TINY]: Same.
12618 (avr_asm_init_sections) <readonly_data_section>: Only patch
12619 callback if avr_arch->flash_pm_offset = 0.
12620 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12621 for rodata if avr_arch->flash_pm_offset != 0.
12622 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12623 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12624 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12625 (m_options): Append opt_rcall.
12626 (m_dirnames): Append dir_rcall.
12627 * config/avr/t-multilib: Regenerate.
12629 * configure.ac [target=avr]: Check whether avrxmega3 default
12630 linker description file works as needed.
12631 * configure: Regenerate.
12632 * doc/avr-mmcu.texi: Regenerate.
12633 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12634 <__AVR_ARCH__>: Document avrxmega3 and 103.
12635 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12636 <__AVR_SHORT_CALLS__>: Document it.
12637 <__AVR_PM_BASE_ADDRESS__>: Document it.
12638 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12639 (AVR Variable Attributes) <progmem>: Document this is
12640 not needed for avrxmega3.
12641 (AVR Named Address Spaces) <__flash>: Dito.
12643 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12645 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12647 2017-06-12 Doug Rupp <rupp@adacore.com>
12649 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12650 Append vxworks-stdint.h to the tm_file list.
12651 * config/vxworks-stdint.h: New file.
12653 2017-06-12 Martin Liska <mliska@suse.cz>
12655 PR tree-optimization/81041
12656 * tree-profile.c (gimple_gen_ic_func_profiler):
12657 Create an extra BB in profile-generate
12658 (gimple_gen_time_profiler): Likewise.
12660 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12662 PR tree-optimization/81003
12663 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12664 (update_range_test): Use it instead of force_gimple_operand_gsi.
12666 2017-06-12 Richard Biener <rguenther@suse.de>
12668 PR tree-optimization/81053
12669 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12670 with backedge value not defined in loop. Simplify def stmt
12673 2017-06-11 Tom de Vries <tom@codesourcery.com>
12676 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12678 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12679 nvptx_cannot_force_const_mem.
12681 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12683 * opts.c (finish_options): Move test for flag_split_stack after
12684 it has been initialized.
12686 2017-06-11 Jason Merrill <jason@redhat.com>
12688 * tree.h (id_equal): New.
12689 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12690 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12691 instead of strcmp of IDENTIFIER_POINTER.
12693 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12695 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12696 (mark_all_inlined_calls_cdtor): Fix formating.
12697 (inline_transform): Rescale profile before inlining.
12699 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12701 * cgraph.h (cgraph_edge::clone): Update prototype.
12702 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12703 (cgraph_node::create_clone): Update.
12704 (cgraph_node::create_version_clone): Update.
12705 * tree-inline.c (copy_bb): Update.
12706 (expand_call_inline): Update.
12708 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12710 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12711 factored out from ...
12712 (rs6000_emit_prologue): ... here.
12714 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12716 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12717 factored out from ...
12718 (rs6000_emit_prologue): ... here.
12720 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12722 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12724 (handle_missing_profiles): Fix computation of tp_first_run.
12725 (counts_to_freqs): Do not touch freqs when count is 0.
12727 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12729 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12732 2017-06-10 Tom de Vries <tom@codesourcery.com>
12734 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12735 attributes): Document signal effective target.
12737 2017-06-10 Tom de Vries <tom@codesourcery.com>
12739 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12740 Document effective target stack_size.
12742 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12744 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12745 to the edit_context if they can be auto-applied.
12747 2017-06-9 Ian Lance Taylor <iant@golang.org>
12749 * opts.c (finish_options): If -fsplit-stack, disable implicit
12750 -forder-blocks-and-partition.
12751 * doc/invoke.texi (Optimize Options): Document that when using
12752 -fsplit-stack -forder-blocks-and-partition is not implicitly
12755 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12757 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12758 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12759 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12760 * builtins.def (abort, trap, unreachable): Declare cold.
12761 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12762 * tree-core.h (ECF_COLD): New.
12763 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12764 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12766 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12768 * predict.c (unlikely_executed_stmt_p): Cleanup.
12770 2017-06-09 Richard Biener <rguenther@suse.de>
12772 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12773 model if the ref is always written to.
12775 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12777 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12779 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12781 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12784 2017-06-09 Tom de Vries <tom@codesourcery.com>
12787 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12788 "target cannot support label values" when encountering LABEL_REF.
12790 2017-06-09 Martin Liska <mliska@suse.cz>
12792 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12793 (gimple_gen_ic_func_profiler): Emit direct comparison
12794 of __gcov_indirect_call_callee with NULL.
12795 (gimple_gen_time_profiler): Change probability from
12796 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12798 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12800 * profile.c (edge_gcov_counts): Turn to pointer.
12801 (compute_branch_probabilities, compute_branch_probabilities): Update.
12802 (branch_prob): Do not clear edge_gcov_count.
12803 * profile.h (edge_gcov_counts): Turn to pointer.
12804 (edge_gcov_count): Update.
12806 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12808 * gimple.h (gimple_check_failed): Mark cold.
12810 2017-06-09 Richard Biener <rguenther@suse.de>
12812 PR tree-optimization/66623
12813 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12814 refactor check_reduction into two parts, properly computing
12815 whether we have to check reduction validity for outer loop
12818 2017-06-09 Richard Biener <rguenther@suse.de>
12820 PR tree-optimization/79483
12821 * graphite-scop-detection.c (order): New global.
12822 (get_order): Compute bb to order mapping that satisfies code
12823 generation constraints.
12824 (cmp_pbbs): New helper.
12825 (build_scops): Start domwalk at entry block, sort generated
12828 2017-06-09 Richard Biener <rguenther@suse.de>
12830 PR middle-end/81007
12831 * ipa-polymorphic-call.c
12832 (ipa_polymorphic_call_context::restrict_to_inner_class):
12833 Skip FIELD_DECLs with error_mark_node type.
12834 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12837 2017-06-09 Martin Liska <mliska@suse.cz>
12839 * predict.c (struct branch_predictor): New struct.
12840 (test_prediction_value_range): New test.
12841 (predict_c_tests): New function.
12842 * selftest-run-tests.c (selftest::run_tests): Run the function.
12843 * selftest.h: Declare new tests.
12845 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12848 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12849 gen_add3_insn did not fail.
12850 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12851 r0, construct that number in a temporary reg and add that reg to r0.
12852 If asked to put the result in r0 as well, fail.
12854 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12856 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12857 for early expansion of vec_eqv.
12859 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12861 PR middle-end/81005
12862 * ubsan.c (instrument_null): Avoid pointless code temporary.
12863 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12865 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12869 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12872 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12875 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12877 * predict.c (unlikely_executed_edge_p): Move ahead.
12878 (probably_never_executed_edge_p): Use it.
12880 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12882 PR middle-end/79988
12883 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12884 gimple_call_builtin_p call.
12886 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12888 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12889 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12890 rtl_check_failed_type2, rtl_check_failed_code1,
12891 rtl_check_failed_code2, rtl_check_failed_code_mode,
12892 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12893 rtvec_check_failed_bounds, rtl_check_failed_flag,
12894 _fatal_insn_not_found, _fatal_insn): Likewise.
12895 * tree.h (tree_contains_struct_check_failed,
12896 tree_check_failed, tree_not_check_failed,
12897 tree_class_check_failed, tree_range_check_failed,
12898 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12899 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12900 tree_operand_check_failed, omp_clause_check_failed,
12901 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12904 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12906 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12907 flag_branch_probabilities.
12908 * ipa-inline.c (edge_badness): Likewise.
12909 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12910 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12911 * predict.c (maybe_hot_frequency_p): Likewise.
12912 (probably_never_executed): Likewise.
12913 * sched-ebb.c (schedule_ebbs): Likewise.
12914 * sched-rgn.c (find_single_block_region): Likewise.
12915 * tracer.c (tail_duplicate): Likewise.
12917 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12919 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12920 longer requires x_flag_profile_use.
12922 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12924 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12925 instead of flag_reorder_blocks_and_partition.
12926 * dbxout.c (dbxout_function_end): Likewise.
12927 * dwarf2out.c (gen_subprogram_die): Likewise.
12928 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12929 * hw-doloop.c (reorg_loops): Likewise.
12930 * varasm.c (assemble_start_function,
12931 assemble_end_function): Likewise.
12932 (decide_function_section): Do not check for
12933 flag_reorder_blocks_and_partition.
12935 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12937 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12939 (chkp_get_hard_register_fake_addr_expr): Ditto.
12940 (chkp_build_addr_expr): Add check for hard reg case.
12941 (chkp_parse_array_and_component_ref): Ditto.
12942 (chkp_find_bounds_1): Ditto.
12943 (chkp_process_stmt): Don't generate bounds store for
12946 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12948 * predict.c (maybe_hot_bb_p): Do not check profile status.
12949 (maybe_hot_edge_p): Likewise.
12950 (probably_never_executed): Check for zero counts even if profile
12952 (unlikely_executed_edge_p): New function.
12953 (unlikely_executed_stmt_p): New function.
12954 (unlikely_executed_bb_p): New function.
12955 (set_even_probabilities): Use unlikely predicates.
12956 (combine_predictions_for_bb): Likewise.
12957 (predict_paths_for_bb): Likewise.
12958 (predict_paths_leading_to_edge): Likewise.
12959 (determine_unlikely_bbs): New function.
12960 (estimate_bb_frequencies): Use it.
12961 (compute_function_frequency): Use zero counts even if profile is
12963 * profile-count.h: Fix typo.
12965 2017-08-08 Julia Koval <julia.koval@intel.com>
12967 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12968 _mm512_mask_cvtsepi16_storeu_epi8,
12969 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12970 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12971 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12972 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12973 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12974 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12975 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12976 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12977 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12978 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12979 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12980 __builtin_ia32_pmovuswb256mem_mask,
12981 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12982 __builtin_ia32_pmovwb512mem_mask): New builtins.
12984 2017-08-08 Julia Koval <julia.koval@intel.com>
12986 PR target/73350,80862
12987 * config/i386/subst.md (round): Fix round pattern.
12988 * config/i386/i386.c (ix86_erase_embedded_rounding):
12989 Fix erasing rounding for the fixed pattern.
12991 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12993 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12995 2017-06-08 Martin Liska <mliska@suse.cz>
12997 PR gcov-profile/80911
12998 * gcov.c (block_info::block_info): New constructor.
13000 2017-06-07 Carl Love <cel@us.ibm.com>
13002 * config/rs6000/rs6000-c: The return type of the following
13003 built-in functions was implemented as int not long long. Fix sign
13004 of return value for the unsigned version of vec_mulo and vec_mule.
13005 vector unsigned long long vec_bperm (vector unsigned long long,
13006 vector unsigned char)
13007 vector signed long long vec_mule (vector signed int,
13009 vector unsigned long long vec_mule (vector unsigned int,
13010 vector unsigned int)
13011 vector signed long long vec_mulo (vector signed int,
13013 vector unsigned long long vec_mulo (vector unsigned int,
13014 vector unsigned int)
13015 * doc/extend.texi: Fix the documentation for the built-in
13018 2017-06-07 Carl Love <cel@us.ibm.com>
13021 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13024 2017-06-07 Carl Love <cel@us.ibm.com>
13026 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13027 support, Generate doublehv for signed int/float for BE case only.
13029 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13031 * doc/invoke.texi (mcx16): Rewrite.
13033 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13035 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13036 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13037 *mov<mode>_softfloat, and an anonymous splitter): Use
13038 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13042 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13044 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13045 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13046 (rs6000_debug_reg_global): Adjust.
13047 (rs6000_init_hard_regno_mode_ok): Adjust.
13048 (rs6000_dbx_register_number): Adjust.
13049 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13050 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13051 Remove SPE_ACC and SPEFSCR.
13052 (REG_ALLOC_ORDER): Ditto.
13053 (FRAME_POINTER_REGNUM): Change to 111.
13054 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13055 (REG_CLASS_NAMES): Ditto.
13056 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13057 (REGISTER_NAMES): Ditto.
13058 (ADDITIONAL_REG_NAMES): Ditto.
13059 (rs6000_reg_names): Ditto.
13060 * config/rs6000/rs6000.md: Renumber some register number
13063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13065 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13067 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13068 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13070 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13071 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13072 Delete the SPE high registers.
13073 (REG_ALLOC_ORDER): Ditto.
13074 (enum reg_class): Remove SPE_HIGH_REGS.
13075 (REG_CLASS_NAMES): Ditto.
13076 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13077 (REGISTER_NAMES): Ditto.
13078 (rs6000_reg_names): Ditto.
13079 * doc/tm.texi.in: Remove SPE as example.
13080 * doc/tm.texi: Regenerate.
13082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13084 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13085 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13086 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13087 * config/rs6000/rs6000.md (type): Remove "brinc".
13089 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13091 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13092 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13093 * config/rs6000/linuxspe.h: Delete file.
13094 * config/rs6000/rs6000.md: Don't include spe.md.
13095 * config/rs6000/spe.h: Delete file.
13096 * config/rs6000/spe.md: Delete file.
13097 * config/rs6000/t-rs6000: Remove spe.md.
13099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13101 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13102 (reg_or_none500mem_operand): Delete.
13103 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13104 instead of reg_or_none500mem_operand.
13106 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13108 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13109 handling of SPE flags.
13110 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13112 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13114 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13116 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13117 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13118 paired_divv2sf3): Similar.
13119 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13120 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13121 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13123 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13124 Rename the paired_* instruction patterns.
13125 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13127 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13128 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13129 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13130 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13131 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13132 PAIRED_VECTOR_MODE.
13133 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13134 (spe_func_has_64bit_regs_p): Delete.
13135 (spe_expand_predicate_builtin): Delete.
13136 (spe_expand_evsel_builtin): Delete.
13137 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13138 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13139 (invalid_e500_subreg): Delete.
13140 (rs6000_legitimize_address): Always force_reg op2 as well, for
13141 paired single memory accesses.
13142 (rs6000_member_type_forces_blk): Delete.
13143 (rs6000_spe_function_arg): Delete.
13144 (rs6000_expand_unop_builtin): Delete SPE handling.
13145 (rs6000_expand_binop_builtin): Ditto.
13146 (spe_expand_stv_builtin): Delete.
13147 (bdesc_2arg_spe): Delete.
13148 (spe_expand_builtin): Delete.
13149 (spe_expand_predicate_builtin): Delete.
13150 (spe_expand_evsel_builtin): Delete.
13151 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13152 (spe_init_builtins): Delete.
13153 (spe_func_has_64bit_regs_p): Delete.
13154 (savres_routine_name): Delete "info" parameter. Adjust callers.
13155 (rs6000_emit_stack_reset): Ditto.
13156 (rs6000_dwarf_register_span): Delete.
13157 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13158 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13159 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13161 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13163 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13164 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13165 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13166 mulv2sf3, divv2sf3): Delete expanders.
13168 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13170 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13172 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13174 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13175 * config/rs6000/rs6000.c: Ditto.
13177 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13179 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13180 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13181 comparison_operator.
13183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13185 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13186 * config/rs6000/rs6000.opt: Ditto.
13187 * config/rs6000/t-rtems: Ditto.
13189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13191 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13192 TARGET_E500_SINGLE by 0, simplify.
13193 * config/rs6000/rs6000.c: Ditto.
13194 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13195 (spe_build_register_parallel): Delete.
13196 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13197 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13198 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13199 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13200 (E500_CONVERT): Delete.
13201 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13205 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13206 * config/rs6000/dfp.md: Ditto.
13207 (negdd2, *negdd2_fpr): Merge.
13208 (absdd2, *absdd2_fpr): Merge.
13209 (negtd2, *negtd2_fpr): Merge.
13210 (abstd2, *abstd2_fpr): Merge.
13211 * config/rs6000/e500.h: Delete file.
13212 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13213 TARGET_FPRS by 1 and simplify.
13214 * config/rs6000/rs6000-c.c: Ditto.
13215 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13216 TARGET_DF_SPE by 0.
13217 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13219 * config/rs6000/rs6000.md: Ditto.
13220 (floatdidf2, *floatdidf2_fpr): Merge.
13221 (move_from_CR_gt_bit): Delete.
13222 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13223 (E500_CR_IOR_COMPARE): Delete.
13224 (All patterns that require !TARGET_FPRS): Delete.
13225 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13227 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13229 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13231 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13233 * graphds.c (add_edge): Intitialize edge's attached data.
13234 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13235 pointer parameter. Call pointed function on each edge during
13236 graph traversing. Skip traversing the edge when the function
13238 (graphds_dfs, graphds_scc): Ditto.
13239 (for_each_edge): New parameter. Pass the new parameter to callback
13241 * graphds.h (skip_edge_callback): New function pointer type.
13242 (graphds_dfs, graphds_scc): New function pointer parameter.
13243 (graphds_edge_callback, for_each_edge): New parameter.
13245 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13247 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13248 out code checking if runtime alias check is possible to below ...
13249 Call the new function.
13250 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13251 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13253 2017-06-07 Marek Polacek <polacek@redhat.com>
13256 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13257 TYPE_OVERFLOW_WRAPS check.
13259 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13261 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13262 if versioning is required.
13263 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13264 peeling with the check for versioning.
13266 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13268 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13269 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13270 Set true to new parameter if new ssa variable is defined.
13271 (vect_gen_vector_loop_niters): Refactor. Set range information
13272 for the new vector loop bound variable.
13273 (vect_do_peeling): Ditto.
13275 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13277 * tree-affine.c (ssa.h): Include header file.
13278 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13279 has wrapping overflow behavior.
13281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13283 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13285 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13287 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13288 (tree_to_aff_combination): ... here.
13290 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13292 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13293 reg_pressure model function.
13294 (ivopts_global_cost_for_size): Delete.
13295 (determine_set_costs, iv_ca_recount_cost): Call new model function
13296 ivopts_estimate_reg_pressure.
13298 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13300 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13301 expensive than udiv. Remove floating point cases from mod.
13303 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13305 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13306 Increase idiv cost.
13308 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13310 * config/aarch64/aarch64.md
13311 (copysignsf3): Fix mask generation.
13313 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13315 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13317 (class dump_manager): Add register_dumps method.
13318 * dumpfile.c: Include langhooks.h.
13319 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13320 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13321 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13322 (dump_manager::dump_register): Start with 512 entries instead of 32.
13323 (dump_manager::register_dumps): New method.
13324 * toplev.c (general_init): Instead of invoking register_dumps
13325 langhook, invoke register_dumps method on the dump manager.
13326 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13329 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13331 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13332 Say that # requires an associated define_split to exist, and that
13333 the define_split must be suitable for use after register allocation.
13335 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13337 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13338 (compute_outgoing_frequencies): Also initialize zero counts.
13339 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13340 around loops; preserve more of profile when nothing changes.
13342 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13344 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13346 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13347 * config/arm/arm-cpu-cdata.h: Regenerate.
13348 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13349 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13350 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13351 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13352 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13354 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13356 * config/arm/t-rmprofile: Likewise.
13357 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13359 2017-06-06 David S. Miller <davem@davemloft.net>
13362 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13363 blockage if function uses alloca.
13365 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13367 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13368 New "uid" fields to hold pretty-print IDs of group and ref.
13369 Memory references are now identified as <group_id>:<ref_id>
13370 instead of using [random] addresses.
13371 (dump_mem_details): Simplify, no functional change.
13372 (dump_mem_ref): Simplify and make output more concise.
13373 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13374 (find_or_create_group): Initialize group uid.
13375 (record_ref): Initialize ref uid. Improve debug output.
13376 (prune_group_by_reuse, should_issue_prefetch_p,)
13377 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13378 (mark_nontemporal_store, determine_loop_nest_reuse):
13379 Improve debug output.
13381 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13383 * dbgcnt.def (prefetch): New debug counter.
13384 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13385 (schedule_prefetches): Stop issueing prefetches if debug counter
13388 2017-06-06 Tom de Vries <tom@codesourcery.com>
13390 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13391 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13393 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13395 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13396 Use aarch64_reg_or_zero predicate for operand 4.
13397 (aarch64_compare_and_swap<mode> define_insn_and_split):
13398 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13399 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13401 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13403 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13404 (arm_compute_save_core_reg_mask): This.
13405 (thumb1_compute_save_reg_mask): Rename into ...
13406 (thumb1_compute_save_core_reg_mask): This.
13407 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13408 (arm_compute_frame_layout): Likewise.
13410 2017-06-06 Richard Biener <rguenther@suse.de>
13412 PR tree-optimization/80974
13413 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13414 keep or clear leaders SSA info.
13416 2017-06-06 Tom de Vries <tom@codesourcery.com>
13418 * config/nvptx/nvptx.c (split_mode_p): New function.
13419 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13421 2017-06-06 Tom de Vries <tom@codesourcery.com>
13423 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13425 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13428 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13431 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13433 * shrink-wrap.c (handle_simple_exit): Update profile.
13434 (try_shrink_wrapping): Upate profile.
13436 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13438 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13439 (tree_guess_outgoing_edge_probabilities): New.
13440 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13441 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13443 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13445 * ipa-split.c (split_function): Initialize return bb profile.
13447 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13449 * profile.c (compute_branch_probabilities): Also initialize
13450 EXIT_BLOCK profile.
13452 2017-06-06 Richard Biener <rguenther@suse.de>
13454 PR tree-optimization/80928
13455 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13456 (vect_analyze_loop_operations): Properly guard analysis for
13458 (vect_transform_loop): Likewise.
13459 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13460 (vect_model_induction_cost): Do not cost for pure SLP.
13461 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13462 of induction in inner loop vectorization.
13463 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13464 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13465 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13467 (vect_analyze_slp_cost_1): Cost induction.
13468 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13469 (vect_get_slp_vect_defs): Likewise.
13470 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13471 (vect_transform_stmt): Handle SLP reductions.
13472 * tree-vectorizer.h (vectorizable_induction): Adjust.
13474 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13476 * config/rs6000/rs6000.c (make_resolver_func): Update
13477 init_lowered_empty_function call.
13479 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13481 * doc/invoke.texi: Document the -fprofile-abs-path option.
13482 * common.opt (fprofile-abs-path): New option.
13483 * gcov-io.h (gcov_write_filename): Declare.
13484 * gcov-io.c (gcov_write_filename): New function.
13485 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13486 * profile.c (output_location): Likewise.
13488 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13490 * shring-wrap.c: Revert accidental commit.
13492 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13494 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13496 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13498 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13500 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13501 profile in callgraph edge.
13502 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13503 (apply_scale): Likewise.
13504 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13505 Also scale profile when inlining function with zero profile.
13506 (initialize_cfun): Update exit block profile even when it is zero.
13507 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13508 when profile is read.
13510 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13512 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13513 (CLONE_*): New constants to define the processors we can generate
13514 code for with the target_clone attribute.
13515 (rs6000_clone_map): New array to identify which clone processors
13516 the current program is running on.
13517 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13518 target_clone attribute.
13519 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13520 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13521 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13522 (cpu_expand_builtin): Add support for target_clone attribute.
13523 (rs6000_valid_attribute_p): Allow "default" attribute.
13524 (get_decl_name): New debug function to simplify printing the
13525 current function name in debugging statements.
13526 (rs6000_clone_priority): New functions to support the target_clone
13527 attribute, and be able to generate code to switch between ISA 2.05
13528 through ISA 3.0 (power6 through power9).
13529 (rs6000_compare_version_priority): Likewise.
13530 (rs6000_get_function_versions_dispatcher): Likewise.
13531 (make_resolver_func): Likewise.
13532 (add_condition_to_bb): Likewise.
13533 (dispatch_function_versions): Likewise.
13534 (rs6000_generate_version_dispatcher_body): Likewise.
13535 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13536 (fusion_gpr_load_p): Fix a spacing issue.
13537 * doc/extend.texi (Common Function Attributes): Document that the
13538 PowerPC supports the target_clone attribute.
13540 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13542 * config/arm/arm.h: explain F symbol found in description of ARM
13543 register allocation in its legend.
13545 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13547 * config/mips/frame-header-opt.c: Include profile-count.h.
13548 * config/riscv/riscv.c: Include profile-count.h
13550 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13552 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13554 (sm_set_flag_if_changed): Add bbs field.
13555 (execute_sm_if_changed_flag_set): Pass BBS.
13556 (execute_sm): Update.
13558 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13560 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13563 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13565 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13566 (peephole2): New peephole2 to emit the above.
13567 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13569 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13571 * config/aarch64/aarch64.c (define_peephole2 above
13572 *sub_<shift>_<mode>): New peephole.
13574 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13576 * config/i386/i386.c (make_resolver_func): Update.
13577 * Makefile.in: Add profile-count.h and profile-count.o
13578 * auto-profile.c (afdo_indirect_call): Update to new API.
13579 (afdo_set_bb_count): Update.
13580 (afdo_propagate_edge): Update.
13581 (afdo_propagate_circuit): Update.
13582 (afdo_calculate_branch_prob): Update.
13583 (afdo_annotate_cfg): Update.
13584 * basic-block.h: Include profile-count.h
13585 (struct edge_def): Turn count to profile_count.
13586 (struct basic_block_def): Likewie.
13587 (REG_BR_PROB_BASE): Move to profile-count.h
13588 (RDIV): Move to profile-count.h
13589 * bb-reorder.c (max_entry_count): Turn to profile_count.
13590 (find_traces): Update.
13591 (rotate_loop):Update.
13592 (connect_traces):Update.
13593 (sanitize_hot_paths):Update.
13594 * bt-load.c (migrate_btr_defs): Update.
13595 * cfg.c (RDIV): Remove.
13596 (init_flow): Use alloc_block.
13597 (alloc_block): Uninitialize count.
13598 (unchecked_make_edge): Uninitialize count.
13599 (check_bb_profile): Update.
13600 (dump_edge_info): Update.
13601 (dump_bb_info): Update.
13602 (update_bb_profile_for_threading): Update.
13603 (scale_bbs_frequencies_int): Update.
13604 (scale_bbs_frequencies_gcov_type): Update.
13605 (scale_bbs_frequencies_profile_count): New.
13606 * cfg.h (update_bb_profile_for_threading): Update.
13607 (scale_bbs_frequencies_profile_count): Declare.
13608 * cfgbuild.c (compute_outgoing_frequencies): Update.
13609 (find_many_sub_basic_blocks): Update.
13610 * cfgcleanup.c (try_forward_edges): Update.
13611 (try_crossjump_to_edge): Update.
13612 * cfgexpand.c (expand_gimple_tailcall): Update.
13613 (construct_exit_block): Update.
13614 * cfghooks.c (verify_flow_info): Update.
13615 (dump_bb_for_graph): Update.
13616 (split_edge): Update.
13617 (make_forwarder_block): Update.
13618 (duplicate_block): Update.
13619 (account_profile_record): Update.
13620 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13621 (get_estimated_loop_iterations): Update.
13622 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13623 (single_likely_exit): Update.
13624 * cfgloopmanip.c (scale_loop_profile): Update.
13626 (set_zero_probability): Update.
13627 (lv_adjust_loop_entry_edge): Update.
13628 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13629 (purge_dead_edges): Update.
13630 (rtl_account_profile_record): Update.
13631 * cgraph.c (cgraph_node::create): Uninitialize count.
13632 (symbol_table::create_edge): Uninitialize count.
13633 (cgraph_update_edges_for_call_stmt_node): Update.
13634 (cgraph_edge::dump_edge_flags): Update.
13635 (cgraph_node::dump): Update.
13636 (cgraph_edge::maybe_hot_p): Update.
13637 * cgraph.h: Include profile-count.h
13638 (create_clone), create_edge, create_indirect_edge): Update.
13639 (cgraph_node): Turn count to profile_count.
13640 (cgraph_edge0: Likewise.
13641 (make_speculative, clone): Update.
13642 (create_edge): Update.
13643 (init_lowered_empty_function): Update.
13644 * cgraphclones.c (cgraph_edge::clone): Update.
13645 (duplicate_thunk_for_node): Update.
13646 (cgraph_node::create_clone): Update.
13647 * cgraphunit.c (cgraph_node::analyze): Update.
13648 (cgraph_node::expand_thunk): Update.
13649 * final.c (dump_basic_block_info): Update.
13650 * gimple-streamer-in.c (input_bb): Update.
13651 * gimple-streamer-out.c (output_bb): Update.
13652 * graphite.c (print_global_statistics): Update.
13653 (print_graphite_scop_statistics): Update.
13654 * hsa-brig.c: Include basic-block.h.
13655 * hsa-dump.c: Include basic-block.h.
13656 * hsa-gen.c (T sum_slice): Update.
13657 (convert_switch_statements):Update.
13658 * hsa-regalloc.c: Include basic-block.h.
13659 * ipa-chkp.c (chkp_produce_thunks): Update.
13660 * ipa-cp.c (struct caller_statistics): Update.
13661 (init_caller_stats): Update.
13662 (gather_caller_stats): Update.
13663 (ipcp_cloning_candidate_p): Update.
13664 (good_cloning_opportunity_p): Update.
13665 (get_info_about_necessary_edges): Update.
13666 (dump_profile_updates): Update.
13667 (update_profiling_info): Update.
13668 (update_specialized_profile): Update.
13669 (perhaps_add_new_callers): Update.
13670 (decide_about_value): Update.
13671 (ipa_cp_c_finalize): Update.
13672 * ipa-devirt.c (struct odr_type_warn_count): Update.
13673 (struct decl_warn_count): Update.
13674 (struct final_warning_record): Update.
13675 (possible_polymorphic_call_targets): Update.
13676 (ipa_devirt): Update.
13677 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13678 * ipa-icf.c (sem_function::merge): Update.
13679 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13680 * ipa-inline.c (compute_uninlined_call_time): Update.
13681 (compute_inlined_call_time): Update.
13682 (want_inline_small_function_p): Update.
13683 (want_inline_self_recursive_call_p): Update.
13684 (edge_badness): Update.
13685 (lookup_recursive_calls): Update.
13686 (recursive_inlining): Update.
13687 (inline_small_functions): Update.
13688 (dump_overall_stats): Update.
13689 (dump_inline_stats): Update.
13690 * ipa-profile.c (ipa_profile_generate_summary): Update.
13691 (ipa_propagate_frequency): Update.
13692 (ipa_profile): Update.
13693 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13694 * ipa-utils.c (ipa_merge_profiles): Update.
13695 * loop-doloop.c (doloop_modify): Update.
13696 * loop-unroll.c (report_unroll): Update.
13697 (unroll_loop_runtime_iterations): Update.
13698 * lto-cgraph.c (lto_output_edge): Update.
13699 (lto_output_node): Update.
13700 (input_node): Update.
13701 (input_edge): Update.
13702 (merge_profile_summaries): Update.
13703 * lto-streamer-in.c (input_cfg): Update.
13704 * lto-streamer-out.c (output_cfg): Update.
13705 * mcf.c (create_fixup_graph): Update.
13706 (adjust_cfg_counts): Update.
13707 (sum_edge_counts): Update.
13708 * modulo-sched.c (sms_schedule): Update.
13709 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13710 * predict.c (maybe_hot_count_p): Update.
13711 (probably_never_executed): Update.
13712 (dump_prediction): Update.
13713 (combine_predictions_for_bb): Update.
13714 (propagate_freq): Update.
13715 (handle_missing_profiles): Update.
13716 (counts_to_freqs): Update.
13717 (rebuild_frequencies): Update.
13718 (force_edge_cold): Update.
13719 * predict.h: Include profile-count.h
13720 (maybe_hot_count_p, counts_to_freqs): UPdate.
13721 * print-rtl-function.c: Do not include cfg.h
13722 * print-rtl.c: Include basic-block.h
13723 * profile-count.c: New file.
13724 * profile-count.h: New file.
13725 * profile.c (is_edge_inconsistent): Update.
13726 (correct_negative_edge_counts): Update.
13727 (is_inconsistent): Update.
13728 (set_bb_counts): Update.
13729 (read_profile_edge_counts): Update.
13730 (compute_frequency_overlap): Update.
13731 (compute_branch_probabilities): Update; Initialize and deinitialize
13733 (branch_prob): Update.
13734 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13735 (edge_gcov_count): New.
13736 (bb_gcov_count): New.
13737 * shrink-wrap.c (try_shrink_wrapping): Update.
13738 * tracer.c (better_p): Update.
13739 * trans-mem.c (expand_transaction): Update.
13740 (ipa_tm_insert_irr_call): Update.
13741 (ipa_tm_insert_gettmclone_call): Update.
13742 * tree-call-cdce.c: Update.
13743 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13744 (gimple_duplicate_sese_tail): Update.
13745 (gimple_account_profile_record): Update.
13746 (execute_fixup_cfg): Update.
13747 * tree-inline.c (copy_bb): Update.
13748 (copy_edges_for_bb): Update.
13749 (initialize_cfun): Update.
13750 (freqs_to_counts): Update.
13751 (copy_cfg_body): Update.
13752 (expand_call_inline): Update.
13753 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13754 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13755 (try_unroll_loop_completely): Update.
13756 (try_peel_loop): Update.
13757 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13758 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13759 * tree-ssa-loop-split.c (connect_loops): Update.
13760 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13761 * tree-ssa-reassoc.c (branch_fixup): Update.
13762 * tree-ssa-tail-merge.c (replace_block_by): Update.
13763 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13764 (compute_path_counts): Update.
13765 (update_profile): Update.
13766 (recompute_probabilities): Update.
13767 (update_joiner_offpath_counts): Update.
13768 (estimated_freqs_path): Update.
13769 (freqs_to_counts_path): Update.
13770 (clear_counts_path): Update.
13771 (ssa_fix_duplicate_block_edges): Update.
13772 (duplicate_thread_path): Update.
13773 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13774 (struct switch_conv_info): Update.
13775 * tree-tailcall.c (decrease_profile): Update.
13776 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13777 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13778 * value-prof.c (check_counter): Update.
13779 (gimple_divmod_fixed_value): Update.
13780 (gimple_mod_pow2): Update.
13781 (gimple_mod_subtract): Update.
13782 (gimple_ic_transform): Update.
13783 (gimple_stringop_fixed_value): Update.
13784 * value-prof.h (gimple_ic): Update.
13786 2017-06-02 Carl Love <cel@us.ibm.com>
13788 * config/rs6000/rs6000-c: Add support for built-in functions
13789 vector double vec_doublee (vector signed int);
13790 vector double vec_doublee (vector unsigned int);
13791 vector double vec_doublee (vector float);
13792 vector double vec_doubleh (vector signed int);
13793 vector double vec_doubleh (vector unsigned int);
13794 vector double vec_doubleh (vector float);
13795 vector double vec_doublel (vector signed int);
13796 vector double vec_doublel (vector unsigned int);
13797 vector double vec_doublel (vector float);
13798 vector double vec_doubleo (vector signed int);
13799 vector double vec_doubleo (vector unsigned int);
13800 vector double vec_doubleo (vector float);.
13801 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13802 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13804 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13805 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13806 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13808 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13809 vec_doublel, vec_doubleh.
13810 * doc/extend.texi: Update the built-in documentation file for the
13811 new built-in functions.
13813 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13816 * ipa-inline-analysis.c (free_growth_caches): Set
13817 edge_removal_hook_holder to NULL after removing it.
13819 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13821 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13822 comparision with zero.
13824 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13826 for early expansion of vec_min and vec_max builtins.
13827 (builtin_function_type): Add min/max unsigned variants to those
13828 identified as having unsigned arguments.
13830 2017-06-02 Olivier Hainque <hainque@adacore.com>
13832 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13834 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13836 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13837 Use VALL_F16 iterator rather than VALL.
13839 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13841 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13842 Emit CBNZ inside loop when doing a strong exchange and comparing
13843 against zero. Generate the CC flags after the loop.
13845 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13847 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13848 (dl_section_ref): New.
13849 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13850 On AIX, append an expression to subtract the size of the
13851 section length to dl_section_ref.
13853 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13855 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13856 for early expansion of vector absolute builtins.
13858 2017-06-02 Richard Biener <rguenther@suse.de>
13860 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13861 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13863 2017-06-02 Richard Biener <rguenther@suse.de>
13865 PR tree-optimization/80948
13866 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13867 stmt order as well.
13869 2017-06-02 Richard Biener <rguenther@suse.de>
13871 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13873 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13874 for inductions as relevant.
13876 2017-06-02 Richard Biener <rguenther@suse.de>
13878 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13879 (vectorizable_induction): ... this. Remove dead code.
13881 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13883 * builtins. (expand_builtin_alloca): Remove second parameter and
13884 infer its value from the first parameter instead.
13885 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13887 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13889 PR rtl-optimization/80903
13890 * loop-doloop.c (add_test): Unshare sequence.
13892 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13894 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13896 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13898 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13900 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13901 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13902 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13903 (xlogue_layout::get_stub_rtx): Make static.
13904 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13905 (xlogue_layout::compute_stub_managed_regs): Rename to...
13906 (xlogue_layout::count_stub_managed_regs): ...this.
13907 (xlogue_layout::is_stub_managed_reg): New function.
13908 (xlogue_layout::m_stub_names): Rename to...
13909 (xlogue_layout::s_stub_names): ...this, make static.
13910 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13911 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13912 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13913 xlogue_layout::s_stub_names): Instantiate statics.
13914 (stub_managed_regs): Remove.
13915 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13916 (disable_call_ms2sysv_xlogues): Rename to...
13917 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13918 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13920 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13921 change after reload_completed.
13922 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13924 (ix86_expand_prologue): Likewise.
13925 (ix86_expand_epilogue): Likewise.
13926 (ix86_expand_split_stack_prologue): Likewise.
13927 (ix86_compute_frame_layout): Remove frame parameter ...
13928 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13929 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13931 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13932 (ix86_frame): Move from here ...
13933 * config/i386/i386.h (ix86_frame): ... to here.
13934 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13935 complete ix86_frame data structure instead. Remove some_ld_name.
13937 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13939 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13940 symbols that hold a DECL_VALUE_EXPR.
13942 2017-06-01 Martin Jambor <mjambor@suse.cz>
13944 PR tree-optimization/80898
13945 * tree-sra.c (process_subtree_disqualification): Removed.
13946 (disqualify_candidate): Do not acll
13947 process_subtree_disqualification.
13948 (subtree_mark_written_and_enqueue): New function.
13949 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13950 RHS has been disqualified and re-queue LHS if necessary. Apart
13951 from that, ignore disqualified RHS.
13953 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13955 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13956 address fetch for z10 or later.
13958 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13960 * config/arc/arc.md (tst_movb): Add guard when splitting.
13962 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13964 * config/arc/arc.c (arc_can_eliminate): Test against
13965 arc_frame_pointer_needed.
13967 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13969 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13970 to prevent store reordering.
13971 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13972 (type): Add block type.
13973 (stack_tie): Define special instruction to be used in
13976 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13978 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13979 constraint. It is not valid for the pattern.
13980 (noncommutative_binary_comparison): Likewise.
13982 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13984 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13987 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13989 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13990 be used by the reg-alloc.
13992 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13994 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13995 reg-alloc when having mul64 or mul32x16 instructions.
13996 (mulsidi3): Likewise.
13997 (umulsidi3): Likewise.
13998 (mulsi32x16): New pattern.
13999 (mulsi64): Likewise.
14000 (mulsidi64): Likewise.
14001 (umulsidi64): Likewise.
14002 (MUL32x16_REG): Define.
14003 (mul64_600): Use MUL32x16_REG.
14004 (mac64_600): Likewise.
14005 (umul64_600): Likewise.
14006 (umac64_600): Likewise.
14008 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14010 * config/arc/arc.md (mulsi3_700): Make it commutative.
14012 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14014 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14016 (*sign_extendsidi2_insn): Likewise for movstosw.
14018 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14020 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14021 the type of the input discriminant value. Convert the
14022 discriminant value of signedness vary.
14024 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14026 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14027 Add to -Wall section.
14029 2017-06-01 Richard Biener <rguenther@suse.de>
14031 PR middle-end/66313
14032 * fold-const.c (fold_plusminus_mult_expr): If the factored
14033 factor may be zero use a wrapping type for the inner operation.
14034 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14035 and handle moved defs.
14036 (process_assignment): Properly guard the unary op case. Return a
14037 tri-state indicating that moving the stmt before the call may allow
14038 to continue. Pass through to_move.
14039 (find_tail_calls): Handle moving unrelated defs before
14042 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14045 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14046 splitter result in the canonical way.
14048 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14050 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14051 also for 32bit target. Update insn attributes.
14052 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14054 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14056 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14057 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14058 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14059 (_mm_maskz_min_ss): New intrinsics.
14061 2017-05-31 Martin Liska <mliska@suse.cz>
14063 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14064 Change comment style to one we normally use.
14065 (vectorizable_reduction): Likewise.
14066 (vectorizable_induction): Likewise.
14067 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14068 (vectorizable_call): Likewise.
14069 (vectorizable_simd_clone_call): Likewise.
14070 (vectorizable_conversion): Likewise.
14071 (vectorizable_assignment): Likewise.
14072 (vectorizable_shift): Likewise.
14073 (vectorizable_operation): Likewise.
14074 (vectorizable_store): Likewise.
14075 (vectorizable_load): Likewise.
14076 * tree-vectorizer.h: Likewise.
14078 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14080 * passes.c (emergency_dump_function): New.
14081 * tree-pass.h (emergency_dump_function): Declare.
14082 * plugin.c (plugins_internal_error_function): Remove.
14083 * plugin.h (plugins_internal_error_function): Remove declaration.
14084 * toplev.c (internal_error_function): New static function. Use it...
14085 (general_init): ...here.
14087 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14089 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14090 (arc_rtx_costs): Add costs for new patterns.
14091 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14092 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14094 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14096 * tree-ssa-strlen.c (get_next_strinfo): New function.
14097 (get_stridx_plus_constant): Use it.
14098 (zero_length_string): Likewise.
14099 (adjust_related_strinfos): Likewise.
14100 (adjust_last_stmt): Likewise.
14102 2017-05-31 Richard Biener <rguenther@suse.de>
14105 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14106 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14108 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14110 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14111 loop_vinfo argument and use of dependence distance vectors.
14112 Check instead whether the two references differ only in their
14113 initial value and assume that they have the same alignment if the
14114 difference is a multiple of the vector alignment.
14115 (vect_analyze_data_refs_alignment): Update call accordingly.
14117 2017-05-31 Martin Liska <mliska@suse.cz>
14120 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14122 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14124 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14125 (create_intersect_range_checks): Move from ...
14126 * tree-data-ref.c (create_intersect_range_checks_index)
14127 (create_intersect_range_checks): ... to here.
14128 (create_runtime_alias_checks): New function factored from ...
14129 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14130 here. Call above function.
14131 * tree-data-ref.h (create_runtime_alias_checks): New function.
14133 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14135 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14136 segment length for dr_b and compute it in wide_int.
14138 2017-05-31 Richard Biener <rguenther@suse.de>
14140 PR tree-optimization/80906
14141 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14142 and pass through iv_map.
14143 (copy_bb_and_scalar_dependences): Adjust.
14144 (translate_pending_phi_nodes): Likewise.
14145 (copy_loop_close_phi_args): Handle code-generating IVs instead
14148 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14150 * diagnostic-color.c (color_dict): Add "type-diff".
14151 (parse_gcc_colors): Update comment.
14152 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14153 -fdiagnostics-show-template-tree and -fno-elide-type.
14154 (GCC_COLORS): Add type-diff to example.
14156 (-fdiagnostics-show-template-tree): New.
14157 (-fno-elide-type): New.
14158 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14159 the pp_format_decoder callback. Call any m_format_postprocessor's
14161 (pretty_printer::pretty_printer): Initialize
14162 m_format_postprocessor.
14163 (pretty_printer::~pretty_printer): Delete any
14164 m_format_postprocessor.
14165 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14166 (class format_postprocessor): New class.
14167 (struct pretty_printer::format_decoder): Document the new parameters.
14168 (struct pretty_printer::m_format_postprocessor): New field.
14169 * tree-diagnostic.c (default_tree_printer): Update for new
14170 bool and const char ** params.
14171 * tree-diagnostic.h (default_tree_printer): Likewise.
14173 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14175 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14176 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14177 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14178 rs6000_gen_cell_microcode code.
14179 (rs6000_final_prescan_insn): Delete.
14180 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14181 "warn-cell-microcode" entries.
14182 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14183 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14184 throughout. Change cc_reg_not_micro_cr0_operand to
14185 cc_reg_not_cr0_operand throughout.
14186 (*extendhi<mode>2_noload): Delete.
14187 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14188 (mwarn-cell-microcode): Delete.
14189 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14190 -mgen-cell-microcode and -mwarn-cell-microcode.
14192 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14195 * config/i386/constraints.md (Yd): New constraint.
14197 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14198 and (?Yd, r) alternatives. Update insn attributes.
14199 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14200 and (?*Yd, r) alternatives. Update insn attributes.
14201 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14203 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14205 * gimplify.c (gimplify_modify_expr): Don't create a
14206 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14209 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14211 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14213 2017-05-30 Richard Biener <rguenther@suse.de>
14215 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14216 and reduc_def fields.
14217 (STMT_VINFO_REDUC_TYPE): New define.
14218 (STMT_VINFO_REDUC_DEF): Likewise.
14219 (vect_force_simple_reduction): Adjust prototype.
14220 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14221 (vect_is_simple_reduction): Remove check_reduction argument.
14222 (vect_force_simple_reduction): Adjust and set
14223 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14224 (vectorizable_reduction): Do not re-do reduction analysis
14225 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14226 * tree-parloops.c (gather_scalar_reductions): Adjust.
14228 2017-05-30 Richard Biener <rguenther@suse.de>
14230 PR middle-end/80901
14231 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14234 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14236 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14237 Introduce unknown_misalignment parameter and remove vf.
14238 (vect_peeling_hash_get_lowest_cost):
14239 Pass unknown_misalignment parameter.
14240 (vect_enhance_data_refs_alignment):
14241 Fix unsupportable data ref treatment.
14243 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14245 * tree-vect-data-refs.c (vect_get_data_access_cost):
14246 Workaround for SLP handling.
14247 (vect_enhance_data_refs_alignment):
14248 Compute costs for doing no peeling at all, compare to the best
14249 peeling costs so far and avoid peeling if cheaper.
14251 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14253 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14254 Return peeling info and set costs to zero for unlimited cost
14256 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14257 with unknown misalignment. Compute and costs for unknown
14258 misalignment, compare them to the costs for known misalignment
14259 and choose the cheapest for peeling.
14261 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14263 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14264 (vect_get_peeling_costs_all_drs): Create function.
14265 (vect_peeling_hash_get_lowest_cost):
14266 Use vect_get_peeling_costs_all_drs.
14267 (vect_peeling_supportable): Create function.
14268 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14270 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14272 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14273 DR_HAS_NEGATIVE_STEP.
14274 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14275 (vect_enhance_data_refs_alignment): Use.
14276 (vect_duplicate_ssa_name_ptr_info): Use.
14277 * tree-vectorizer.h (dr_misalignment): Use.
14278 (known_alignment_for_access_p): Use.
14280 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14283 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14285 (has_section_name): New function.
14287 2017-05-30 Martin Liska <mliska@suse.cz>
14290 * auto-profile.c (get_function_decl_from_block): Fix
14293 2017-05-30 Richard Biener <rguenther@suse.de>
14295 PR middle-end/80876
14296 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14298 2017-05-30 Martin Liska <mliska@suse.cz>
14300 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14301 * dumpfile.h (struct dump_file_info): Remove ctors.
14303 2017-05-30 Martin Liska <mliska@suse.cz>
14305 * predict.def: Fix GNU coding style.
14307 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14309 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14310 Mark 'to' argument with ATTRIBUTE_UNUSED.
14312 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14314 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14315 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14316 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14319 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14321 * doc/install.texi (Options specification): Restore entry of
14322 --enable-sjlj-exceptions.
14324 2017-05-27 Michael Eager <eager@eagercon.com>
14327 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14329 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14331 * config/microblaze/microblaze.h
14332 (FIXED_REGISTERS): Update in macro.
14333 (CALL_USED_REGISTERS): Update in macro.
14335 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14337 * doc/install.texi: Add links to macOS binary distributions.
14339 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14343 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14345 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14347 2017-05-26 Martin Liska <mliska@suse.cz>
14349 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14351 2017-05-26 Martin Liska <mliska@suse.cz>
14353 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14354 always leading ';; '.
14355 (dump_bb_info): Likewise.
14356 (brief_dump_cfg): Likewise.
14357 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14358 * dumpfile.c: Remove usage of TDF_VERBOSE.
14359 * dumpfile.h (enum dump_kind): Likewise.
14360 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14361 * print-tree.c (debug_verbose): Remove.
14362 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14363 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14364 * tree-diagnostic.c (default_tree_printer): Replace
14365 TDF_DIAGNOSTIC with TDF_SLIM.
14367 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14369 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14370 in parameter loop, rather than loop_vinfo.
14371 (create_intersect_range_checks): Ditto.
14372 (vect_create_cond_for_alias_checks): Update call to above functions.
14374 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14376 PR tree-optimization/80815
14377 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14378 for merging runtime alias checks. Handle negative DR_STEPs.
14380 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14382 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14384 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14385 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14386 out code pruning runtime alias checks.
14387 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14388 factored out from above.
14389 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14391 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14393 (prune_runtime_alias_test_list): New decalaration.
14395 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14397 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14398 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14399 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14400 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14401 (operator==, comp_dr_with_seg_len_pair): Ditto.
14402 (vect_prune_runtime_alias_test_list): Ditto.
14404 2017-05-26 Martin Liska <mliska@suse.cz>
14407 * params.def: Bound partial-inlining-entry-probability param.
14409 2017-05-26 Marek Polacek <polacek@redhat.com>
14412 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14415 2017-05-26 Richard Biener <rguenther@suse.de>
14417 PR tree-optimization/80842
14418 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14421 2017-05-26 Richard Biener <rguenther@suse.de>
14423 PR tree-optimization/80844
14424 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14426 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14428 * doc/md.texi (Machine Constraints): Update x86 family
14429 machine constraints section to match 'config/i386/constraints.md'.
14431 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14433 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14435 2017-05-25 Nathan Sidwell <nathan@acm.org>
14437 * doc/invoke.texi (--enable-languages): Update documentation.
14439 2017-05-25 Martin Liska <mliska@suse.cz>
14441 * dumpfile.c: Add TDF_FOLDING.
14442 * dumpfile.h (enum dump_kind): Likewise.
14443 * genmatch.c (dt_simplify::gen_1): Use it.
14445 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14447 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14449 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14451 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14452 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14454 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14456 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14457 * match.pd (X == C): Rewrite it here.
14458 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14459 with_certain_nonzero_bits2): New predicates.
14460 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14462 2017-05-24 Nathan Sidwell <nathan@acm.org>
14464 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14467 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14470 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14472 * config/powerpcspe: New port. Files are copied from the rs6000
14473 port, with "rs6000" in filenames replaced by "powerpcspe".
14475 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14477 PR rtl-optimization/80754
14478 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14480 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14482 * config/sparc/sparc.md (length): Return the correct value for -mflat
14483 sibcalls to match output_sibcall.
14485 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14489 * config/rs6000/rs6000.c (struct machine_function): Add new field
14491 (rs6000_get_separate_components): Init that field, use it.
14492 (rs6000_components_for_bb): Use the field.
14494 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14496 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14498 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14500 PR middle-end/80823
14501 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14503 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14506 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14507 address against address_operand predicate.
14508 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14510 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14512 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14513 parameters passed indirectly.
14515 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14517 * config/i386/i386.md (*movdi_internal): Remove SSE4
14518 alternative 18 (?r, *v). Update insn attributes.
14519 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14520 Update insn attributes.
14521 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14522 Update insn attributes.
14523 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14524 alternative 1 (r, v). Remove isa attribute.
14525 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14526 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14527 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14529 2017-05-23 Tom de Vries <tom@codesourcery.com>
14531 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14534 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14536 * cgraphunit.c (symbol_table::process_new_functions): Update.
14537 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14538 (inline_generate_summary): Rename to ...
14539 (ipa_fn_summary_generate): ... this one.
14540 (inline_read_summary): Rename to ...
14541 (ipa_fn_summary_read): ... this one.
14542 (inline_write_summary): Rename to ...
14543 (ipa_fn_summary_write): ... this one.
14544 (inline_free_summary): Rename to ...
14545 (ipa_free_fn_summary): ... this one.
14546 (pass_data_local_fn_summary, pass_local_fn_summary,
14547 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14548 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14549 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14550 make_pass_ipa_fn_summary): New.
14551 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14552 inline_write_summary, inline_free_summary): Remove.
14553 (ipa_free_fn_summary) : New.
14554 * ipa-inline.c (ipa_inline): Update.
14555 (pass_ipa_inline): Do not generate summaries.
14556 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14558 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14559 and add pass_ipa_fn_summary.
14560 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14562 (make_pass_inline_parameters): Remove.
14564 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14566 * omp-low.c (struct omp_context): Remove "default_kind" member.
14569 * omp-offload.c (execute_oacc_device_lower): Remove the
14570 parallelism dimensions function attributes for unparallelized
14571 OpenACC kernels constructs.
14573 2017-05-23 Martin Liska <mliska@suse.cz>
14575 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14577 (cgraph_edge::make_speculative): Likewise.
14578 (cgraph_edge::resolve_speculation): Likewise.
14579 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14580 (cgraph_node::dump): Likewise.
14581 * cgraph.h: Likewise.
14582 * cgraphunit.c (analyze_functions): Likewise.
14583 (symbol_table::compile): Likewise.
14584 * ipa-cp.c (print_all_lattices): Likewise.
14585 (determine_versionability): Likewise.
14586 (initialize_node_lattices): Likewise.
14587 (ipcp_verify_propagated_values): Likewise.
14588 (estimate_local_effects): Likewise.
14589 (update_profiling_info): Likewise.
14590 (create_specialized_node): Likewise.
14591 (perhaps_add_new_callers): Likewise.
14592 (decide_about_value): Likewise.
14593 (decide_whether_version_node): Likewise.
14594 (identify_dead_nodes): Likewise.
14595 (ipcp_store_bits_results): Likewise.
14596 * ipa-devirt.c (dump_targets): Likewise.
14597 (ipa_devirt): Likewise.
14598 * ipa-icf.c (sem_item::dump): Likewise.
14599 (sem_function::equals): Likewise.
14600 (sem_variable::equals): Likewise.
14601 (sem_item_optimizer::read_section): Likewise.
14602 (sem_item_optimizer::execute): Likewise.
14603 (congruence_class::dump): Likewise.
14604 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14605 (dump_inline_summary): Likewise.
14606 (estimate_node_size_and_time): Likewise.
14607 (inline_analyze_function): Likewise.
14608 * ipa-inline-transform.c (inline_call): Likewise.
14609 * ipa-inline.c (report_inline_failed_reason): Likewise.
14610 (want_early_inline_function_p): Likewise.
14611 (edge_badness): Likewise.
14612 (update_edge_key): Likewise.
14613 (inline_small_functions): Likewise.
14614 * ipa-profile.c (ipa_profile): Likewise.
14615 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14616 (ipa_make_edge_direct_to_target): Likewise.
14617 (remove_described_reference): Likewise.
14618 (ipa_impossible_devirt_target): Likewise.
14619 (propagate_controlled_uses): Likewise.
14620 (ipa_print_node_params): Likewise.
14621 (ipcp_transform_function): Likewise.
14622 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14623 (propagate_pure_const): Likewise.
14624 * ipa-reference.c (generate_summary): Likewise.
14625 (read_write_all_from_decl): Likewise.
14626 (propagate): Likewise.
14627 (ipa_reference_read_optimization_summary): Likewise.
14628 * ipa-utils.c (ipa_merge_profiles): Likewise.
14629 * ipa.c (walk_polymorphic_call_targets): Likewise.
14630 (symbol_table::remove_unreachable_nodes): Likewise.
14631 (ipa_single_use): Likewise.
14632 * passes.c (execute_todo): Likewise.
14633 * predict.c (drop_profile): Likewise.
14634 * symtab.c (symtab_node::get_dump_name): New function.
14635 (symtab_node::dump_name): Likewise.
14636 (symtab_node::dump_asm_name): Likewise.
14637 (symtab_node::dump_references): Likewise.
14638 (symtab_node::dump_referring): Likewise.
14639 (symtab_node::dump_base): Likewise.
14640 (symtab_node::debug_symtab): Likewise.
14641 * tree-sra.c (convert_callers_for_node): Likewise.
14642 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14643 * value-prof.c (init_node_map): Likewise.
14645 2017-05-23 Martin Liska <mliska@suse.cz>
14647 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14648 and symtab_node::debug_symtab to symbol_table::debug.
14649 * cgraphunit.c (analyze_functions): Use the renamed function.
14650 (symbol_table::compile): Likewise.
14651 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14652 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14653 * passes.c (execute_todo): Likewise.
14654 * symtab.c (symbol_table::dump): New function.
14655 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14657 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14659 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14660 that nonconst implies exec.
14662 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14664 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14665 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14666 (inline_edge_summary_vec): Turn into ...
14667 (ipa_call_summaries): ... this one.
14668 (redirect_to_unreachable, edge_set_predicate,
14669 evaluate_properties_for_edge, inline_summary_alloc,
14670 reset_ipa_call_summary, reset_inline_summary,
14671 inline_summary_t::duplicate): Update.
14672 (inline_edge_duplication_hook): Turn to ...
14673 (ipa_call_summary_t::duplicate): ... this one.
14674 (inline_edge_removal_hook): Turn to ...
14675 (ipa_call_summary_t::remove): ... this one.
14676 (dump_inline_edge_summary): Turn to ...
14677 (dump_ipa_call_summary): ... this one.
14678 (estimate_function_body_sizes): Update.
14679 (inline_update_callee_summaries): Update.
14680 (remap_edge_change_prob): Update.
14681 (remap_edge_summaries): Update.
14682 (inline_merge_summary): Update.
14683 (do_estimate_edge_time): Update.
14684 (inline_generate_summary): Update.
14685 (inline_read_section): Update.
14686 (inline_read_summary): Update.
14687 (inline_free_summary): Update.
14688 * ipa-inline.c (can_inline_edge_p): Update.
14689 (compute_inlined_call_time): Update.
14690 (want_inline_small_function_p): Update.
14691 (edge_badness): Update.
14692 (early_inliner): Update.
14693 * ipa-inline.h (inline_edge_summary): Turn to ...
14694 (ipa_call_summary): ... this one.
14695 (ipa_call_summary_t): New class.
14696 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14697 (ipa_call_summaries): New.
14698 (inline_edge_summary): Remove.
14699 (estimate_edge_growth): Update.
14700 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14701 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14702 * ipa-split.c (execute_split_functions): Update.
14703 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14705 2017-05-23 Tom de Vries <tom@codesourcery.com>
14707 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14708 attributes): Document rdrand effective target.
14710 2017-05-23 Tom de Vries <tom@codesourcery.com>
14712 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14713 attributes): Sort alphabetically.
14715 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14717 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14719 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14722 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14723 V2DF/V2DI splat into two separate patterns, one that handles
14724 registers, and the other that only handles memory. Drop support
14725 for splatting from a GPR on ISA 2.07 and then splitting the
14726 splat into direct move and splat.
14727 (vsx_splat_<mode>_reg): Likewise.
14728 (vsx_splat_<mode>_mem): Likewise.
14730 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14732 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14734 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14736 PR middle-end/80809
14737 * omp-low.c (finish_taskreg_remap): New function.
14738 (finish_taskreg_scan): If unit size of ctx->record_type
14739 is non-constant, unshare the size expression and replace
14740 decls in it with possible outer var refs.
14742 PR middle-end/80809
14743 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14744 GOVD_SHARED rather than GOVD_PRIVATE with it.
14745 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14746 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14748 PR middle-end/80853
14749 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14750 as last argument to build_outer_var_ref for pointer bases of array
14751 section reductions.
14753 2017-05-19 Martin Sebor <msebor@redhat.com>
14755 * print-tree.c (print_node): Print DECL_READ_P flag.
14757 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14759 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14760 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14761 * cgraph.c: Likewise.
14762 * cgraphunit.c: Likewise.
14763 * gengtype.c: Likewise.
14764 * ipa-cp.c: Likewise.
14765 * ipa-devirt.c: Likewise.
14766 * ipa-icf.c: Likewise.
14767 * ipa-predicate.c: Likewise.
14768 * ipa-profile.c: Likewise.
14769 * ipa-prop.c: Likewise.
14770 * ipa-split.c: Likewise.
14772 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14773 edge_predicate_pool, dump_inline_hints,
14774 inline_summary::account_size_time, redirect_to_unreachable,
14775 edge_set_predicate, set_hint_predicate,
14776 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14777 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14778 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14779 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14780 ipa_call_summary_t::remove, initialize_growth_caches,
14781 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14782 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14783 mark_modified, unmodified_parm_1, unmodified_parm,
14784 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14785 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14786 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14787 will_be_nonconstant_predicate, record_modified_bb_info,
14788 get_minimal_bb, record_modified, param_change_prob,
14789 phi_result_unknown_predicate, predicate_for_phi_result,
14790 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14791 estimate_function_body_sizes, compute_inline_parameters,
14792 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14793 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14794 inline_update_callee_summaries, remap_edge_change_prob,
14795 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14796 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14797 inline_analyze_function, inline_summary_t::insert,
14798 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14799 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14800 inline_free_summary): Move to ipa-fnsummary.h
14801 (predicate_t): Remove.
14802 * ipa-fnsummary.c: New file.
14803 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14804 (enum inline_hints_vals, inline_hints, agg_position_info,
14805 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14806 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14807 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14808 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14809 inline_read_summary, inline_write_summary, inline_free_summary,
14810 inline_analyze_function, initialize_inline_failed,
14811 inline_merge_summary, inline_update_overall_summary,
14812 compute_inline_parameters): Move to ipa-fnsummary.h
14813 * ipa-fnsummary.h: New file.
14814 * ipa-inline-transform.h: Include ipa-inline.h.
14815 * ipa-inline.c: LIkewise.
14817 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14819 * ipa-inline.c (edge_badness): Use inlined_time instead of
14820 inline_summaries->get.
14822 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14824 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14826 2017-05-22 Nathan Sidwell <nathan@acm.org>
14828 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14829 (fdump-lang): Document 'raw' option.
14830 * dumpfile.h (TDI_tu): Delete.
14831 * dumpfile.c (dump_files): Remove translation-unit.
14832 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14834 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14836 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14837 command option from $(AWK) call.
14838 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14839 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14840 [FORMAT]: Remove handling of variable.
14841 * config/avr/t-multilib: Regenerate.
14843 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14845 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14847 (dump_inline_summary): Do not print self_time.
14848 (estimate_function_body_sizes): Do not set self_time.
14849 (compute_inline_parameters): Likewise.
14850 (inline_read_section, inline_write_summary): Do not stream self_time.
14851 * ipa-inline.h (inline_summary): Drop self_time.
14853 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14855 * ipa-inline-analysis.c (account_size_time): Rename to ...
14856 (inline_summary::account_size_time): ... this one.
14857 (reset_ipa_call_summary): Turn to ...
14858 (ipa_call_summary::reset): ... this one.
14859 (reset_inline_summary): Turn to ...
14860 (inline_summary::reset): ... this one.
14861 (inline_summary_t::remove): Update.
14862 (inline_summary_t::duplicate): Update.
14863 (ipa_call_summary_t::remove): Update.
14864 (dump_inline_summary): Update.
14865 (estimate_function_body_sizes): Update.
14866 (compute_inline_parameters): Update.
14867 (estimate_node_size_and_time): Update.
14868 (inline_merge_summary): Update.
14869 (inline_update_overall_summary): Update.
14870 (inline_read_section): Update.
14871 (inline_write_summary): Update.
14872 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14873 add account_size_time and reset member functions.
14874 (ipa_call_summary): Add reset function.
14875 * ipa-predicate.h (predicate::operator &): Constify.
14877 2017-05-22 Richard Biener <rguenther@suse.de>
14879 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14881 2017-05-19 Jason Merrill <jason@redhat.com>
14883 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14885 2017-05-19 Marek Polacek <polacek@redhat.com>
14888 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14889 TYPE_OVERFLOW_WRAPS checks.
14891 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14893 * tree-core.h (enum omp_clause_default_kind): Add
14894 "OMP_CLAUSE_DEFAULT_PRESENT".
14895 * tree-pretty-print.c (dump_omp_clause): Handle it.
14896 * gimplify.c (enum gimplify_omp_var_data): Add
14897 "GOVD_MAP_FORCE_PRESENT".
14898 (gimplify_adjust_omp_clauses_1): Map it to
14899 "GOMP_MAP_FORCE_PRESENT".
14900 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14902 * gimplify.c (oacc_default_clause): Clarify.
14904 2017-05-19 Nathan Sidwell <nathan@acm.org>
14906 LANG_HOOK_REGISTER_DUMPS
14907 * toplev.c (general_init): Call register dump lang hook.
14908 * doc/invoke.texi: Document -fdump-lang option family.
14909 * dumpfile.c (dump_files): Remove class dump here.
14910 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14911 * dumpfile.h (tree_dump_index): Remove TDI_class.
14912 * langhooks-def.h (lhd_register_dumps): Declare.
14913 (LANG_HOOKS_REGISTER_DUMPS): Define.
14914 (LANG_HOOKS_INITIALIZER): Add it.
14915 * langhooks.c (lhd_register_dumps): Define.
14916 * langhooks.h (struct lang_hooks): Add register_dumps.
14918 2017-05-19 Nathan Sidwell <nathan@acm.org>
14920 * context.h (context::set_passes): New.
14921 * context.c (context::context): Do not create pass manager.
14922 * toplev.c (general_init): Create pass manager here.
14924 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14926 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14927 use this splitter if two add or or instructions would also work for
14928 the constant we want to generate.
14930 2017-05-19 Richard Biener <rguenther@suse.de>
14933 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14934 predicate evaluation.
14936 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14938 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14940 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14943 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14945 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14946 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14947 (inline_edge_summary_vec): Turn into ...
14948 (ipa_call_summaries): ... this one.
14949 (redirect_to_unreachable, edge_set_predicate,
14950 evaluate_properties_for_edge, inline_summary_alloc,
14951 reset_ipa_call_summary, reset_inline_summary,
14952 inline_summary_t::duplicate): Update.
14953 (inline_edge_duplication_hook): Turn to ...
14954 (ipa_call_summary_t::duplicate): ... this one.
14955 (inline_edge_removal_hook): Turn to ...
14956 (ipa_call_summary_t::remove): ... this one.
14957 (dump_inline_edge_summary): Turn to ...
14958 (dump_ipa_call_summary): ... this one.
14959 (estimate_function_body_sizes): Update.
14960 (inline_update_callee_summaries): Update.
14961 (remap_edge_change_prob): Update.
14962 (remap_edge_summaries): Update.
14963 (inline_merge_summary): Update.
14964 (do_estimate_edge_time): Update.
14965 (inline_generate_summary): Update.
14966 (inline_read_section): Update.
14967 (inline_read_summary): Update.
14968 (inline_free_summary): Update.
14969 * ipa-inline.c (can_inline_edge_p): Update.
14970 (compute_inlined_call_time): Update.
14971 (want_inline_small_function_p): Update.
14972 (edge_badness): Update.
14973 (early_inliner): Update.
14974 * ipa-inline.h (inline_edge_summary): Turn to ...
14975 (ipa_call_summary): ... this one.
14976 (ipa_call_summary_t): New class.
14977 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14978 (ipa_call_summaries): New.
14979 (inline_edge_summary): Remove.
14980 (estimate_edge_growth): Update.
14981 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14982 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14983 * ipa-split.c (execute_split_functions): Update.
14984 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14986 2017-05-19 Richard Biener <rguenther@suse.de>
14988 PR middle-end/80764
14989 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14991 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14993 * config/rs6000/rs6000.c (struct machine_function): Add field
14994 fpr_is_wrapped_separately.
14995 (rs6000_get_separate_components): Use 64 components. Handle the
14996 new FPR components.
14997 (rs6000_components_for_bb): Handle the FPR components.
14998 (rs6000_emit_prologue_components): Handle the FPR components.
14999 (rs6000_emit_epilogue_components): Handle the FPR components.
15000 (rs6000_set_handled_components): Handle the FPR components.
15001 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15002 that are already separately shrink-wrapped.
15003 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15004 that are already separately shrink-wrapped.
15006 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15009 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15012 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15013 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15014 cases where the register allocator uses a move and an offsettable
15015 memory operation to/from a FPR register on ISA 2.06/2.07.
15016 (define_peephole2 for Altivec d-form store): Likewise.
15018 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15021 * config/i386/mmx.md (*mov<mode>_internal): Enable
15022 alternatives 11, 12, 13 and 14 also for 32bit targets.
15023 Remove alternatives 15, 16, 17 and 18.
15024 * config/i386/sse.md (vec_concatv2di): Change
15025 alternative (!x, *y) to (x, ?!*Yn).
15027 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15029 * dumpfile.h (enum dump_kind): Remove stray comma.
15031 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15033 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15034 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15035 predicate::num_conditions
15036 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15037 (CHANGED): turn into predicate::changed.
15038 (agg_position_info): Move to ipa-predicate.h
15039 (add_condition, predicate::add_clause, predicate::operator &=,
15040 predicate::or_with, predicate::evaluate, predicate::probability,
15041 dump_condition, dump_clause, predicate::dump,
15042 predicate::remap_after_duplication, predicate::remap_after_inlining,
15043 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15044 (evaluate_conditions_for_known_args): Update.
15045 (set_cond_stmt_execution_predicate): Update.
15046 * ipa-inline.h: Include ipa-predicate.h
15047 (condition, inline_param_summary, conditions, agg_position_info,
15048 predicate): Move to ipa-predicate.h
15049 * ipa-predicate.c: New file.
15050 * ipa-predicate.h: New file.
15052 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15054 * final.c (leaf_function_p): Check we are not in a sequence.
15056 2017-05-18 Martin Liska <mliska@suse.cz>
15058 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15059 * dumpfile.c (dump_register): Use new enum dump_kind.
15060 (get_dump_file_name): Likewise.
15061 (dump_enable_all): Likewise.
15062 (dump_switch_p_1): Likewise.
15063 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15064 * dumpfile.h (enum dump_kind): New enum type.
15065 (struct dump_file_info): Create constructor and
15066 format fields and comments.
15067 * passes.c (pass_manager::register_one_dump_file):
15069 * statistics.c (statistics_early_init): Likewise.
15070 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15071 TDF_TREE with TDF_SLIM.
15072 (gather_memory_references_ref): Likewise.
15074 2017-05-18 Martin Liska <mliska@suse.cz>
15076 * vec.h (struct vnull): Use it.
15078 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15080 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15081 (true_predicate, false_predicate, true_predicate_p,
15082 false_predicate_p): Remove.
15083 (single_cond_predicate, not_inlined_predicate): Turn to member function
15085 (add_condition): Update.
15086 (add_clause): Turn to...
15087 (predicate::add_clause): ... this one; update; allow passing NULL
15089 (and_predicates): Turn to ...
15090 (predicate::operator &=): ... this one.
15091 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15092 (or_predicates): Turn to ...
15093 (predicate::or_with): ... this one.
15094 (evaluate_predicate): Turn to ...
15095 (predicate::evaluate): ... this one.
15096 (predicate_probability): Turn to ...
15097 (predicate::probability): ... this one.
15098 (dump_condition): Update.
15099 (dump_predicate): Turn to ...
15100 (predicate::dump): ... this one.
15101 (account_size_time): Update.
15102 (edge_set_predicate): Update.
15103 (set_hint_predicate): UPdate.
15104 (evaluate_conditions_for_known_args): Update.
15105 (evaluate_properties_for_edge): Update.
15106 (remap_predicate_after_duplication): Turn to...
15107 (predicate::remap_after_duplication): ... this one.
15108 (remap_hint_predicate_after_duplication): Update.
15109 (inline_summary_t::duplicate): UPdate.
15110 (dump_inline_edge_summary): Update.
15111 (dump_inline_summary): Update.
15112 (set_cond_stmt_execution_predicate): Update.
15113 (set_switch_stmt_execution_predicate): Update.
15114 (compute_bb_predicates): Update.
15115 (will_be_nonconstant_expr_predicate): Update.
15116 (will_be_nonconstant_predicate): Update.
15117 (phi_result_unknown_predicate): Update.
15118 (predicate_for_phi_result): Update.
15119 (array_index_predicate): Update.
15120 (estimate_function_body_sizes): Update.
15121 (estimate_node_size_and_time): Update.
15122 (estimate_ipcp_clone_size_and_time): Update.
15123 (remap_predicate): Rename to ...
15124 (predicate::remap_after_inlining): ... this one.
15125 (remap_hint_predicate): Update.
15126 (inline_merge_summary): Update.
15127 (inline_update_overall_summary): Update.
15128 (estimate_size_after_inlining): Update.
15129 (read_predicate): Rename to ...
15130 (predicate::stream_in): ... this one.
15131 (read_inline_edge_summary): Update.
15132 (write_predicate): Rename to ...
15133 (predicate::stream_out): ... this one.
15134 (write_inline_edge_summary): Update.
15135 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15136 (clause_t): Turn to uint32_t
15137 (predicate): Turn to class; implement constructor and operators
15139 (size_time_entry): Update.
15140 (inline_summary): Update.
15141 (inline_edge_summary): Update.
15143 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15145 * fold-const.c (fold_binary_loc): Move transformation...
15146 * match.pd (C - X CMP X): ... here.
15148 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15150 * config/sparc/sparc.c (sparc_option_override): Set function
15151 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15152 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15154 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15156 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15158 2017-05-18 Marek Polacek <polacek@redhat.com>
15161 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15162 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15163 gimple_assign_load_p.
15165 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15167 PR middle-end/80692
15168 * real.c (do_compare): Give decimal_do_compare preference over
15169 comparing just the signs.
15171 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15173 * doc/md.texi (Canonicalization of Instructions): Describe the
15174 canonical form of instructions that inherently set a condition
15177 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15179 PR middle-end/80775
15180 * tree-cfg.c: Move deletion of unreachable case statements to after
15181 the merging of consecutive case labels.
15183 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15185 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15186 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15187 restoring of callee-saved registers.
15189 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15191 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15192 * config/visium/visium.c (single_set_and_flags): Likewise.
15193 * config/visium/visium.md (Substitutions): Likewise.
15195 2017-05-17 Martin Liska <mliska@suse.cz>
15197 * cfg.c: Introduce dump_flags_t type and
15198 use it instead of int type.
15200 * cfghooks.c: Likewise.
15201 * cfghooks.h (struct cfg_hooks): Likewise.
15202 * cfgrtl.c: Likewise.
15203 * cfgrtl.h: Likewise.
15204 * cgraph.c (cgraph_node::get_body): Likewise.
15205 * coretypes.h: Likewise.
15206 * domwalk.c: Likewise.
15207 * domwalk.h: Likewise.
15208 * dumpfile.c (struct dump_option_value_info): Likewise.
15209 (dump_enable_all): Likewise.
15210 (dump_switch_p_1): Likewise.
15211 (opt_info_switch_p): Likewise.
15212 * dumpfile.h (enum tree_dump_index): Likewise.
15213 (struct dump_file_info): Likewise.
15214 * genemit.c: Likewise.
15215 * generic-match-head.c: Likewise.
15216 * gengtype.c (open_base_files): Likewise.
15217 * gimple-pretty-print.c: Likewise.
15218 * gimple-pretty-print.h: Likewise.
15219 * graph.c (print_graph_cfg): Likewise.
15220 * graphite-scop-detection.c (dot_all_sese): Likewise.
15221 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15222 * loop-unroll.c (report_unroll): Likewise.
15223 * passes.c (pass_manager::register_one_dump_file): Likewise.
15224 * print-tree.c: Likewise.
15225 * statistics.c: Likewise.
15226 * tree-cfg.c: Likewise.
15227 * tree-cfg.h: Likewise.
15228 * tree-dfa.c: Likewise.
15229 * tree-dfa.h: Likewise.
15230 * tree-dump.c (dump_function): Likewise.
15231 * tree-dump.h (struct dump_info): Likewise.
15232 * tree-pretty-print.c: Likewise.
15233 * tree-pretty-print.h: Likewise.
15234 * tree-ssa-live.c: Likewise.
15235 * tree-ssa-live.h: Likewise.
15236 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15237 * tree-vect-loop.c: Likewise.
15238 * tree-vect-slp.c: Likewise.
15240 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15241 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15243 PR tree-optimization/80457
15244 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15245 of all arguments to a statement as scalar_to_vec operations.
15246 (vectorizable_call): Adjust call to vect_model_simple_cost for
15248 (vectorizable_conversion): Likewise.
15249 (vectorizable_assignment): Likewise.
15250 (vectorizable_shift): Likewise.
15251 (vectorizable_operation): Likewise.
15252 (vectorizable_comparison): Likewise.
15253 (vect_is_simple_cond): Record the def types for operands.
15254 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15255 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15256 for statement argument count.
15258 2017-05-16 Carl Love <cel@us.ibm.com>
15260 * config/rs6000/rs6000-c: Add support for built-in functions
15261 vector unsigned long long vec_bperm (vector unsigned long long,
15262 vector unsigned char)
15263 vector signed long long vec_mule (vector signed int,
15265 vector unsigned long long vec_mule (vector unsigned int,
15266 vector unsigned int)
15267 vector signed long long vec_mulo (vector signed int,
15269 vector unsigned long long vec_mulo (vector unsigned int,
15270 vector unsigned int)
15271 vector signed char vec_sldw (vector signed char,
15272 vector signed char,
15274 vector unsigned char vec_sldw (vector unsigned char,
15275 vector unsigned char,
15277 vector signed short vec_sldw (vector signed short,
15278 vector signed short,
15280 vector unsigned short vec_sldw (vector unsigned short,
15281 vector unsigned short,
15283 vector signed int vec_sldw (vector signed int,
15286 vector unsigned int vec_sldw (vector unsigned int,
15287 vector unsigned int,
15289 vector signed long long vec_sldw (vector signed long long,
15290 vector signed long long,
15292 vector unsigned long long vec_sldw (vector unsigned long long,
15293 vector unsigned long long,
15295 * config/rs6000/rs6000-c: Add support for built-in functions
15296 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15297 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15298 * doc/extend.texi: Update the built-in documentation for the
15299 new built-in functions.
15301 2017-05-16 Marek Polacek <polacek@redhat.com>
15305 * tree.c (save_expr): Don't fold the expression.
15307 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15309 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15310 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15311 and (?*y,m). Update insn attributes.
15313 2017-05-16 Martin Liska <mliska@suse.cz>
15315 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15316 flags argument of print_gimple_stmt, print_gimple_expr,
15317 print_generic_stmt and print_generic_expr.
15318 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15319 * coretypes.h: Likewise.
15320 * except.c (dump_eh_tree): Likewise.
15321 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15322 * gimple-pretty-print.h: Likewise.
15323 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15324 (backprop::push_to_worklist): Likewise.
15325 (backprop::pop_from_worklist): Likewise.
15326 (backprop::process_use): Likewise.
15327 (backprop::intersect_uses): Likewise.
15328 (note_replacement): Likewise.
15329 * gimple-ssa-store-merging.c
15330 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15331 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15332 (pass_store_merging::execute): Likewise.
15333 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15334 (ssa_base_cand_dump_callback): Likewise.
15335 (dump_incr_vec): Likewise.
15336 (replace_refs): Likewise.
15337 (replace_mult_candidate): Likewise.
15338 (create_add_on_incoming_edge): Likewise.
15339 (create_phi_basis): Likewise.
15340 (insert_initializers): Likewise.
15341 (all_phi_incrs_profitable): Likewise.
15342 (introduce_cast_before_cand): Likewise.
15343 (replace_one_candidate): Likewise.
15344 * gimplify.c (gimplify_expr): Likewise.
15345 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15346 (set_rename): Likewise.
15347 (rename_uses): Likewise.
15348 (copy_loop_phi_nodes): Likewise.
15349 (add_close_phis_to_merge_points): Likewise.
15350 (copy_loop_close_phi_args): Likewise.
15351 (copy_cond_phi_args): Likewise.
15352 (graphite_copy_stmts_from_block): Likewise.
15353 (translate_pending_phi_nodes): Likewise.
15354 * graphite-poly.c (print_pdr): Likewise.
15355 (dump_gbb_cases): Likewise.
15356 (dump_gbb_conditions): Likewise.
15357 (print_scop_params): Likewise.
15358 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15359 (build_cross_bb_scalars_use): Likewise.
15360 (gather_bbs::before_dom_children): Likewise.
15361 * hsa-dump.c (dump_hsa_immed): Likewise.
15362 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15363 (get_replacement_map): Likewise.
15364 * ipa-inline-analysis.c (dump_condition): Likewise.
15365 (estimate_function_body_sizes): Likewise.
15366 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15367 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15368 * ipa-prop.c (ipa_dump_param): Likewise.
15369 (ipa_print_node_jump_functions_for_edge): Likewise.
15370 (ipa_modify_call_arguments): Likewise.
15371 (ipa_modify_expr): Likewise.
15372 (ipa_dump_param_adjustments): Likewise.
15373 (ipa_dump_agg_replacement_values): Likewise.
15374 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15375 * ipa-pure-const.c (check_stmt): Likewise.
15376 (pass_nothrow::execute): Likewise.
15377 * ipa-split.c (execute_split_functions): Likewise.
15378 * omp-offload.c (dump_oacc_loop_part): Likewise.
15379 (dump_oacc_loop): Likewise.
15380 * trans-mem.c (tm_log_emit): Likewise.
15381 (tm_memopt_accumulate_memops): Likewise.
15382 (dump_tm_memopt_set): Likewise.
15383 (dump_tm_memopt_transform): Likewise.
15384 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15385 (print_loop): Likewise.
15386 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15387 (chkp_gather_checks_info): Likewise.
15388 (chkp_get_check_result): Likewise.
15389 (chkp_remove_check_if_pass): Likewise.
15390 (chkp_use_outer_bounds_if_possible): Likewise.
15391 (chkp_reduce_bounds_lifetime): Likewise.
15392 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15393 (chkp_mark_completed_bounds): Likewise.
15394 (chkp_register_incomplete_bounds): Likewise.
15395 (chkp_mark_invalid_bounds): Likewise.
15396 (chkp_maybe_copy_and_register_bounds): Likewise.
15397 (chkp_build_returned_bound): Likewise.
15398 (chkp_get_bound_for_parm): Likewise.
15399 (chkp_build_bndldx): Likewise.
15400 (chkp_get_bounds_by_definition): Likewise.
15401 (chkp_generate_extern_var_bounds): Likewise.
15402 (chkp_get_bounds_for_decl_addr): Likewise.
15403 * tree-chrec.c (chrec_apply): Likewise.
15404 * tree-data-ref.c (dump_data_reference): Likewise.
15405 (dump_subscript): Likewise.
15406 (dump_data_dependence_relation): Likewise.
15407 (analyze_overlapping_iterations): Likewise.
15408 * tree-inline.c (expand_call_inline): Likewise.
15409 (tree_function_versioning): Likewise.
15410 * tree-into-ssa.c (dump_defs_stack): Likewise.
15411 (dump_currdefs): Likewise.
15412 (dump_names_replaced_by): Likewise.
15413 (dump_update_ssa): Likewise.
15414 (update_ssa): Likewise.
15415 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15416 * tree-parloops.c (build_new_reduction): Likewise.
15417 (try_create_reduction_list): Likewise.
15418 (ref_conflicts_with_region): Likewise.
15419 (oacc_entry_exit_ok_1): Likewise.
15420 (oacc_entry_exit_single_gang): Likewise.
15421 * tree-pretty-print.h: Likewise.
15422 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15423 (get_scalar_evolution): Likewise.
15424 (add_to_evolution): Likewise.
15425 (get_loop_exit_condition): Likewise.
15426 (analyze_evolution_in_loop): Likewise.
15427 (analyze_initial_condition): Likewise.
15428 (analyze_scalar_evolution): Likewise.
15429 (instantiate_scev): Likewise.
15430 (number_of_latch_executions): Likewise.
15431 (gather_chrec_stats): Likewise.
15432 (final_value_replacement_loop): Likewise.
15433 (scev_const_prop): Likewise.
15434 * tree-sra.c (dump_access): Likewise.
15435 (disqualify_candidate): Likewise.
15436 (create_access): Likewise.
15437 (reject): Likewise.
15438 (maybe_add_sra_candidate): Likewise.
15439 (create_access_replacement): Likewise.
15440 (analyze_access_subtree): Likewise.
15441 (analyze_all_variable_accesses): Likewise.
15442 (sra_modify_assign): Likewise.
15443 (initialize_constant_pool_replacements): Likewise.
15444 (find_param_candidates): Likewise.
15445 (decide_one_param_reduction): Likewise.
15446 (replace_removed_params_ssa_names): Likewise.
15447 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15448 * tree-ssa-copy.c (dump_copy_of): Likewise.
15449 (copy_prop_visit_cond_stmt): Likewise.
15450 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15451 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15452 (record_equivalences_from_stmt): Likewise.
15453 * tree-ssa-dse.c (compute_trims): Likewise.
15454 (delete_dead_call): Likewise.
15455 (delete_dead_assignment): Likewise.
15456 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15457 (forward_propagate_into_cond): Likewise.
15458 (pass_forwprop::execute): Likewise.
15459 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15460 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15462 (move_computations_worker): Likewise.
15463 (execute_sm): Likewise.
15464 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15465 (remove_exits_and_undefined_stmts): Likewise.
15466 (remove_redundant_iv_tests): Likewise.
15467 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15468 (adjust_iv_update_pos): Likewise.
15469 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15470 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15471 (value_replacement): Likewise.
15472 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15473 * tree-ssa-pre.c (print_pre_expr): Likewise.
15474 (get_representative_for): Likewise.
15475 (create_expression_by_pieces): Likewise.
15476 (insert_into_preds_of_block): Likewise.
15477 (eliminate_insert): Likewise.
15478 (eliminate_dom_walker::before_dom_children): Likewise.
15479 (eliminate): Likewise.
15480 (remove_dead_inserted_code): Likewise.
15481 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15482 * tree-ssa-reassoc.c (get_rank): Likewise.
15483 (eliminate_duplicate_pair): Likewise.
15484 (eliminate_plus_minus_pair): Likewise.
15485 (eliminate_not_pairs): Likewise.
15486 (undistribute_ops_list): Likewise.
15487 (eliminate_redundant_comparison): Likewise.
15488 (update_range_test): Likewise.
15489 (optimize_range_tests_var_bound): Likewise.
15490 (optimize_vec_cond_expr): Likewise.
15491 (rewrite_expr_tree): Likewise.
15492 (rewrite_expr_tree_parallel): Likewise.
15493 (linearize_expr): Likewise.
15494 (break_up_subtract): Likewise.
15495 (linearize_expr_tree): Likewise.
15496 (attempt_builtin_powi): Likewise.
15497 (attempt_builtin_copysign): Likewise.
15498 (transform_stmt_to_copy): Likewise.
15499 (transform_stmt_to_multiply): Likewise.
15500 (dump_ops_vector): Likewise.
15501 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15502 (print_scc): Likewise.
15503 (set_ssa_val_to): Likewise.
15504 (visit_reference_op_store): Likewise.
15505 (visit_use): Likewise.
15506 (sccvn_dom_walker::before_dom_children): Likewise.
15507 (run_scc_vn): Likewise.
15508 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15510 (expr_hash_elt::print): Likewise.
15511 (const_and_copies::pop_to_marker): Likewise.
15512 (const_and_copies::record_const_or_copy_raw): Likewise.
15513 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15514 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15515 (dump_predicates): Likewise.
15516 (find_uninit_use): Likewise.
15517 (warn_uninitialized_phi): Likewise.
15518 (pass_late_warn_uninitialized::execute): Likewise.
15519 * tree-ssa.c (verify_vssa): Likewise.
15520 (verify_ssa): Likewise.
15521 (maybe_optimize_var): Likewise.
15522 * tree-vrp.c (dump_value_range): Likewise.
15523 (dump_all_value_ranges): Likewise.
15524 (dump_asserts_for): Likewise.
15525 (register_edge_assert_for_2): Likewise.
15526 (vrp_visit_cond_stmt): Likewise.
15527 (vrp_visit_switch_stmt): Likewise.
15528 (vrp_visit_stmt): Likewise.
15529 (vrp_visit_phi_node): Likewise.
15530 (simplify_cond_using_ranges_1): Likewise.
15531 (fold_predicate_in): Likewise.
15532 (evrp_dom_walker::before_dom_children): Likewise.
15533 (evrp_dom_walker::push_value_range): Likewise.
15534 (evrp_dom_walker::pop_value_range): Likewise.
15535 (execute_early_vrp): Likewise.
15537 2017-05-16 Richard Biener <rguenther@suse.de>
15539 * dwarf2out.c (loc_list_from_tree_1): Do not create
15540 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15542 2017-05-16 Richard Biener <rguenther@suse.de>
15544 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15546 (note_variable_value_in_expr): If we resolved the decl ref
15547 do not push to the stack.
15549 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15551 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15552 operations in fast-math mode.
15553 (vaddq_f16): Likewise.
15554 (vmul_f16): Likewise.
15555 (vmulq_f16): Likewise.
15556 (vsub_f16): Likewise.
15557 (vsubq_f16): Likewise.
15558 * config/arm/neon.md (add<mode>3): New.
15560 (fma:<VH:mode>3): New. Also remove outdated comment.
15563 2017-05-16 Martin Liska <mliska@suse.cz>
15567 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15568 (odr_types_equivalent_p): Likewise.
15570 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15572 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15574 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15577 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15578 non-interunit SSE move alternatives with '?'.
15579 (zero-extendsidi peephole2): New peephole to skip intermediate
15580 general register in SSE zero-extend sequence.
15582 2017-05-15 Jeff Law <law@redhat.com>
15584 * reorg.c (relax_delay_slots): Create a new variable to hold
15585 the temporary target rather than clobbering TARGET_LABEL.
15587 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15588 missing argument to extract_bit_field call.
15589 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15591 2017-05-15 Martin Liska <mliska@suse.cz>
15594 * gcc.c (process_command): Do not allow empty argument of -o option.
15596 2017-05-15 Renlin Li <renlin.li@arm.com>
15598 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15599 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15600 * config/aarch64/constraints.md (Usf): Add long call check.
15601 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15602 (call_value): Likewise.
15603 (sibcall): Likewise.
15604 (sibcall_value): Likewise.
15606 (call_value_insn): New.
15607 (sibcall_insn): Update rtx pattern.
15608 (sibcall_value_insn): Likewise.
15609 (call_internal): Remove.
15610 (call_value_internal): Likewise.
15611 (sibcall_internal): Likewise.
15612 (sibcall_value_internal): Likewise.
15613 (call_reg): Likewise.
15614 (call_symbol): Likewise.
15615 (call_value_reg): Likewise.
15616 (call_value_symbol): Likewise.
15618 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15621 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15623 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15625 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15626 compatible with CCGOCmode and with CCZmode.
15628 2017-05-14 Martin Sebor <msebor@redhat.com>
15630 PR middle-end/77671
15631 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15632 (gimple_fold_builtin_snprintf): Same.
15633 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15634 (gimple_fold_builtin_snprintf): Same.
15635 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15636 of character types.
15637 (is_call_safe): New function.
15638 (try_substitute_return_value): Call it.
15639 (try_simplify_call): New function.
15640 (pass_sprintf_length::handle_gimple_call): Call it.
15642 2017-05-14 Martin Sebor <msebor@redhat.com>
15644 PR middle-end/80669
15645 * builtins.c (expand_builtin_stpncpy): Simplify.
15647 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15649 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15650 * config/i386/i386.h
15651 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15652 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15653 (struct machine_function): Add new members call_ms2sysv,
15654 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15655 (struct machine_frame_state): New fields sp_realigned and
15656 sp_realigned_offset.
15657 * config/i386/i386.c
15658 (enum xlogue_stub): New enum.
15659 (enum xlogue_stub_sets): New enum.
15660 (class xlogue_layout): New class.
15661 (struct ix86_frame): New fields stack_realign_allocate_offset,
15662 stack_realign_offset and outlined_save_offset. Modify comments to
15663 detail stack layout when using out-of-line stubs.
15664 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15665 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15666 -mcall-ms2sysv-xlogues.
15667 (stub_managed_regs): New static variable.
15668 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15669 registers managed by out-of-line stub.
15670 (disable_call_ms2sysv_xlogues): New function.
15671 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15672 m->call_ms2sysv when appropriate and compute frame layout for
15674 (sp_valid_at, fp_valid_at): New inline functions.
15675 (choose_basereg): New function.
15676 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15678 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15679 Use align parameter of choose_baseaddr to generated aligned SSE movs
15681 (pro_epilogue_adjust_stack): Modify to track
15682 machine_frame_state::sp_realigned.
15683 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15684 (ix86_nsaved_sseregs): Likewise.
15685 (ix86_emit_save_regs): Likewise.
15686 (ix86_emit_save_regs_using_mov): Likewise.
15687 (ix86_emit_save_sse_regs_using_mov): Likewise.
15688 (get_scratch_register_on_entry): Likewise.
15689 (gen_frame_set): New function.
15690 (gen_frame_load): Likewise.
15691 (gen_frame_store): Likewise.
15692 (emit_outlined_ms2sysv_save): Likewise.
15693 (emit_outlined_ms2sysv_restore): Likewise.
15694 (ix86_expand_prologue): Modify stack re-alignment code and call
15695 emit_outlined_ms2sysv_save when appropriate.
15696 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15697 parameter rtx_insn *insn, which allows the function to be used to only
15698 generate the notes.
15699 (ix86_expand_epilogue): Modify validity checks of frame and stack
15700 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15701 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15702 * config/i386/predicates.md
15703 (save_multiple): New predicate.
15704 (restore_multiple): Likewise.
15705 * config/i386/sse.md
15706 (save_multiple<mode>): New pattern.
15707 (save_multiple_realign<mode>): Likewise.
15708 (restore_multiple<mode>): Likewise.
15709 (restore_multiple_and_return<mode>): Likewise.
15710 (restore_multiple_leave_return<mode>): Likewise.
15711 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15713 2017-05-14 Julia Koval <julia.koval@intel.com>
15715 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15716 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15717 (__builtin_ia32_xsetbv): New builtins.
15718 * config/i386/i386.c (ix86_expand_special_args_builtin):
15720 (ix86_expand_builtin): Special expand for new intrinsics.
15721 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15722 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15723 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15725 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15727 * cfganal.c (inverted_post_order_compute): Change argument type
15729 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15730 * df-core.c (rest_of_handle_df_initialize): Adjust.
15731 (rest_of_handle_df_finish): Likewise.
15732 (df_analyze_1): Likewise.
15733 (df_analyze): Likewise.
15734 (loop_inverted_post_order_compute): Change argument to be a vec *.
15735 (df_analyze_loop): Adjust.
15736 (df_get_n_blocks): Likewise.
15737 (df_get_postorder): Likewise.
15738 * df.h (struct df_d): Change field to be a vec.
15739 * lcm.c (compute_laterin): Adjust.
15740 (compute_available): Likewise.
15741 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15742 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15743 * tree-ssa-pre.c (compute_antic): Likewise.
15745 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15747 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15748 (depth_first_search::depth_first_search): Change structure init
15749 function to this constructor.
15750 (depth_first_search::add_bb): Rename function to this member.
15751 (depth_first_search::execute): Likewise.
15752 (flow_dfs_compute_reverse_finish): Adjust.
15754 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15756 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15757 (longest_simple_path): Likewise.
15758 * shrink-wrap.c (spread_components): Likewise.
15759 (disqualify_problematic_components): Likewise.
15760 (emit_common_heads_for_components): Likewise.
15761 (emit_common_tails_for_components): Likewise.
15762 (insert_prologue_epilogue_for_components): Likewise.
15764 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15766 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15769 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15771 * df-core.c (df_set_blocks): Start using auto_bitmap.
15772 (df_compact_blocks): Likewise.
15773 * df-problems.c (df_rd_confluence_n): Likewise.
15774 * df-scan.c (df_insn_rescan_all): Likewise.
15775 (df_process_deferred_rescans): Likewise.
15776 (df_update_entry_block_defs): Likewise.
15777 (df_update_exit_block_uses): Likewise.
15778 (df_entry_block_bitmap_verify): Likewise.
15779 (df_exit_block_bitmap_verify): Likewise.
15780 (df_scan_verify): Likewise.
15781 * lra-constraints.c (lra_constraints): Likewise.
15782 (undo_optional_reloads): Likewise.
15783 (lra_undo_inheritance): Likewise.
15784 * lra-remat.c (calculate_gen_cands): Likewise.
15785 (do_remat): Likewise.
15786 * lra-spills.c (assign_spill_hard_regs): Likewise.
15787 (spill_pseudos): Likewise.
15788 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15789 (bitmap_set_subtract_values): Likewise.
15791 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15793 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15794 management with auto_bitmap.
15795 (fix_inter_tick): Likewise.
15796 (fix_recovery_deps): Likewise.
15797 * ira.c (add_store_equivs): Likewise.
15798 (find_moveable_pseudos): Likewise.
15799 (split_live_ranges_for_shrink_wrap): Likewise.
15800 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15801 (rtx_reuse_manager::seen_def_p): Likewise.
15802 (rtx_reuse_manager::set_seen_def): Likewise.
15803 * print-rtl.h (class rtx_reuse_manager): Likewise.
15805 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15807 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15809 (migrate_btr_def): Likewise.
15810 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15811 * df-core.c (loop_post_order_compute): Likewise.
15812 (loop_inverted_post_order_compute): Likewise.
15813 * hsa-common.h: Likewise.
15814 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15815 * init-regs.c (initialize_uninitialized_regs): Likewise.
15816 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15817 (inline_small_functions): Likewise.
15818 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15819 * ira.c (combine_and_move_insns): Likewise.
15820 (build_insn_chain): Likewise.
15821 * loop-invariant.c (find_invariants): Likewise.
15822 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15823 * predict.c (tree_predict_by_opcode): Likewise.
15824 (predict_paths_leading_to): Likewise.
15825 (predict_paths_leading_to_edge): Likewise.
15826 (estimate_loops_at_level): Likewise.
15827 (estimate_loops): Likewise.
15828 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15829 (spread_components): Likewise.
15830 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15831 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15832 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15833 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15834 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15835 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15836 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15837 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15838 (mark_threaded_blocks): Likewise.
15839 (thread_through_all_blocks): Likewise.
15840 * tree-ssa.c (verify_ssa): Likewise.
15841 (execute_update_addresses_taken): Likewise.
15842 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15844 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15846 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15848 (post_order_compute): Likewise.
15849 (inverted_post_order_compute): Likewise.
15850 (pre_and_rev_post_order_compute_fn): Likewise.
15852 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15854 * genrecog.c (int_set::int_set): Explicitly construct our
15855 auto_vec base class.
15856 * vec.h (auto_vec::auto_vec): New constructor.
15858 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15860 * bitmap.h (class auto_bitmap): New constructor taking
15861 bitmap_obstack * argument.
15863 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15865 * bitmap.h (class auto_bitmap): Change type of m_bits to
15866 bitmap_head, and adjust ctor / dtor and member operators.
15868 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15870 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15871 when returned register mode doesn't match original mode.
15873 2017-05-12 Jeff Law <law@redhat.com>
15874 Jakub Jelinek <jakub@redhat.com>
15876 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15877 we look for cc setter after the compare-elim changes.
15878 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15879 within the vector to match what compare-elim now expects.
15880 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15881 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15883 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15884 after the compare-elim changes.
15885 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15886 the vector to match what compare-elim now expects.
15887 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15888 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15889 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15890 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15891 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15893 * config/visium/visium.c (single_set_and_flags): Fix where
15894 we look for cc setter after the compare-elim changes.
15895 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15896 with the vector to match what compare-elim now expects.
15897 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15898 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15899 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15900 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15901 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15902 (neg<mode>2_insn_set_overflow): Likewise.
15904 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15906 PR middle-end/79794
15907 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15908 maybe_expand_insn call, set ops[0].target. If still set after call,
15909 set alt_rtl. Add extra arg to recursive calls.
15910 (extract_bit_field): Add alt_rtl argument. Pass to
15912 * expmed.h (extract_bit_field): Fix prototype.
15913 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15914 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15915 to extract_bit_field_calls.
15916 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15917 Pass alt_rtl to extract_bit_field calls.
15918 * calls.c (store_unaligned_arguments_into_psuedos)
15919 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15920 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15922 * optabs.h (struct expand_operand): Add target bitfield.
15924 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15926 * compare-elim.c (try_eliminate_compare): Canonicalize
15927 operation with embedded compare to
15928 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15929 (set (reg) (operation)].
15931 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15933 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15936 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15937 cost of adding a carry flag for ADC instruction.
15938 [case MINUS]: Ignore the cost of subtracting a carry flag
15939 for SBB instruction.
15941 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15943 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15945 * config/rs6000/bmiintrin.h: New file.
15946 * config/rs6000/bmi2intrin.h: New file.
15947 * config/rs6000/x86intrin.h: New file.
15949 2017-05-12 Jeff Law <law@redhat.com>
15951 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15954 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15956 PR middle-end/80707
15957 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15959 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15961 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15962 early expansion of vector divide builtins.
15963 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15964 builtins identified as having unsigned arguments.
15966 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15968 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15969 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15970 expansion of vector logical operations (and, andc, or, xor,
15973 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15975 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15976 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15978 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15981 early GIMPLE expansion of vector multiplies.
15983 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15985 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15986 TARGET_HAVE_MOVT conditional.
15987 (movt splitter): Likewise.
15989 2017-05-12 Richard Biener <rguenther@suse.de>
15991 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15992 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15993 Fold all stmts not inplace.
15995 2017-05-12 Richard Biener <rguenther@suse.de>
15997 PR tree-optimization/80713
15998 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15999 inserted_exprs bit for not removed stmts.
16001 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16003 PR middle-end/69921
16004 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16005 parallelized" attribute for parallelized OpenACC kernels.
16006 * omp-offload.c (execute_oacc_device_lower): Use it.
16008 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16009 Set "oacc kernels" attribute.
16010 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16011 parameter. Adjust all users.
16012 (oacc_fn_attrib_kernels_p): Remove function.
16013 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16014 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16015 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16016 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16017 assert "oacc kernels" attribute is set.
16019 2017-05-11 Carl Love <cel@us.ibm.com>
16021 * config/rs6000/rs6000-c: Add support for built-in functions
16022 vector unsigned char vec_popcnt (vector signed char)
16023 vector unsigned char vec_popcnt (vector unsigned char)
16024 vector unsigned short vec_popcnt (vector signed short)
16025 vector unsigned short vec_popcnt (vector unsigned short)
16026 vector unsigned int vec_popcnt (vector signed int)
16027 vector unsigned int vec_popcnt (vector unsigned int)
16028 vector unsigned long long vec_popcnt (vector signed long long)
16029 vector unsigned long long vec_popcnt (vector unsigned long long)
16030 vector signed long long vec_slo (vector signed long long,
16031 vector signed char)
16032 vector signed long long vec_slo (vector signed long long,
16033 vector unsigned char)
16034 vector unsigned long long vec_slo (vector unsigned long long,
16035 vector signed char)
16036 vector unsigned long long vec_slo (vector unsigned long long,
16037 vector unsigned char)
16038 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16039 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16040 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16041 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16042 * doc/extend.texi: Update the built-in documentation file for the
16043 new built-in functions.
16045 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16047 * attribs.h (sorted_attr_string): Move machine independent
16048 functions for target clone support from the i386 port to common
16049 code. Rename ix86_function_versions to common_function_versions.
16050 Rename make_name to make_unique_name.
16051 (common_function_versions): Likewise.
16052 (make_unique_name): Likewise.
16053 (make_dispatcher_decl): Likewise.
16054 (is_function_default_version): Likewise.
16055 * attribs.c (attr_strcmp): Likewise.
16056 (sorted_attr_string): Likewise.
16057 (common_function_versions): Likewise.
16058 (make_unique_name): Likewise.
16059 (make_dispatcher_decl): Likewise.
16060 (is_function_default_version): Likewise.
16061 * config/i386/i386.c (attr_strcmp): Likewise.
16062 (sorted_attr_string): Likewise.
16063 (ix86_function_versions): Likewise.
16064 (make_name): Likewise.
16065 (make_dispatcher_decl): Likewise.
16066 (is_function_default_version): Likewise.
16067 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16069 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16072 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16073 Account for direct move costs for vec_construct of integer
16076 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16079 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16080 (UNSPEC_STX_ATOMIC): Ditto.
16081 (loaddi_via_sse): New insn.
16082 (storedi_via_sse): Ditto.
16083 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16084 Update corresponding peephole2 patterns.
16085 (atomic_storedi_fpu): Ditto.
16087 2017-05-11 Julia Koval <julia.koval@intel.com>
16089 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16090 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16092 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16093 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16094 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16096 2017-05-11 Nathan Sidwell <nathan@acm.org>
16098 * graphite-poly.c: Include dumpfile.h.
16100 * dumpfle.h (dump_function): Declare here ...
16101 * tree-dump.h (dump_function): ... not here.
16102 * dumpfile.c: #include tree-cfg.h.
16103 (dump_function): Move here from ...
16104 * tree-dump.c (dump_function): ... here.
16105 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16106 * graphite-poly.c: Don't include tree-dump.h.
16107 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16108 * print-tree.c: Likewise.
16109 * stor-layout.c: Likewise.
16110 * tree-nested.c: Likewise.
16112 * dumpfile.c (dump_start): Use TDF_FLAGS.
16113 (dump_enable_all): Fix TDF_KIND check thinko.
16115 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16117 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16118 array entries to represent two legal parameterizations of the
16119 overloaded __builtin_cmpb function, as represented by the
16120 P6_OV_BUILTIN_CMPB constant.
16121 (altivec_resolve_overloaded_builtin): Add special case handling
16122 for the __builtin_cmpb function, as represented by the
16123 P6_OV_BUILTIN_CMPB constant.
16124 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16125 (BU_P6_64BIT_2): New macro.
16126 (BU_P6_OVERLOAD_2): New macro
16127 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16128 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16129 (CMPB): Add overload support to represent both 32-bit and 64-bit
16130 compare-bytes function.
16131 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16132 support for TARGET_CMPB.
16133 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16134 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16135 documentation of the __builtin_cmpb overloaded built-in function.
16137 2017-05-11 Richard Biener <rguenther@suse.de>
16139 PR tree-optimization/80705
16140 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16141 bases are not vectorizable.
16143 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16145 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16146 when counting register pressure.
16148 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16150 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16151 (iv_ca_more_deps): Renamed to ...
16152 (iv_ca_compare_deps): ... this.
16153 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16157 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16159 (determine_group_iv_costs): ... here.
16160 (find_inv_vars_cb): Record inv var if it's not recorded before.
16162 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16164 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16165 (get_shiftadd_cost): Ditto.
16167 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16169 * tree-ssa-address.c: Include header file.
16170 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16172 (add_to_parts): Refactor.
16173 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16174 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16177 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16179 PR tree-optimization/53090
16180 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16182 (extract_cond_operands): Detect condition with IV on both sides
16183 and return COMP_IV_EXPR_2.
16184 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16185 (rewrite_use_compare): Simplify by removing call to function
16186 extract_cond_operands.
16188 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16190 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16191 (extract_cond_operands): Detect condition comparing against non-
16192 invariant bound and return appropriate enum value.
16193 (find_interesting_uses_cond): Update use of extract_cond_operands.
16194 Handle its return value accordingly.
16195 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16197 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16199 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16200 nonlinear iv_use computation in loop invariant sensitive way.
16202 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16204 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16205 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16207 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16209 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16210 (dump_cand): Support iv_cand.inv_exprs.
16211 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16213 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16216 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16218 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16220 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16221 as local function. Include necessary header files.
16222 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16224 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16226 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16228 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16230 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16231 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16232 RSHIFT_EXPR and BIT_NOT_EXPR.
16234 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16236 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16237 (adjust_setup_cost): New parameter supporting round up adjustment.
16238 (struct address_cost_data): Delete.
16239 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16240 (split_address_cost, ptr_difference_cost): Delete.
16241 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16242 (struct ainc_cost_data): New struct.
16243 (get_address_cost_ainc): New function.
16244 (get_address_cost, get_computation_cost): Reimplement.
16245 (determine_group_iv_cost_address): Record inv_expr for all uses of
16247 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16248 (iv_ca_has_deps): Reimplemented to ...
16249 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16251 (iv_ca_extend): Call iv_ca_more_deps.
16253 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16255 * tree-ssa-address.c (struct mem_address): Move to header file.
16256 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16257 * tree-ssa-address.h (struct mem_address): Move from C file.
16258 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16262 * tree-affine.h (aff_combination_type): New interface.
16263 (aff_combination_zero_p): Remove static.
16264 (aff_combination_const_p): New interface.
16265 (aff_combination_singleton_var_p): New interfaces.
16267 2017-05-11 Richard Biener <rguenther@suse.de>
16269 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16270 Skip unreachable blocks and destinations.
16271 (eliminate): Move stmt removal and fixup ...
16272 (fini_eliminate): ... here. Skip inserted exprs.
16273 (pass_pre::execute): Move fini_pre after fini_eliminate.
16274 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16275 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16276 PRE to get rid of dead code that has invalid SSA form and
16277 split critical edges again.
16279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16281 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16283 2017-05-11 Richard Biener <rguenther@suse.de>
16285 * passes.c (execute_function_todo): Verify loops if they are
16286 said to be up-to-date.
16287 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16288 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16290 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16293 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16294 handle calling assemble_external ourself.
16297 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16298 modes with zero size. Enhance comment.
16300 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16302 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16303 built-ins for vec_xl and vec_xst with short and char pointer
16306 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16308 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16309 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16310 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16311 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16312 (_mm_maskz_min_round_ss): New intrinsics.
16313 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16314 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16315 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16316 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16317 (__builtin_ia32_minss_mask_round): New builtins.
16318 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16319 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16320 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16322 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16323 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16325 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16328 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16330 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16331 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16332 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16333 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16334 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16335 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16336 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16337 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16338 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16339 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16340 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16341 (__builtin_ia32_mulss_mask_round): New builtins.
16342 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16343 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16344 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16346 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16347 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16349 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16352 2017-05-10 Julia Koval <julia.koval@intel.com>
16354 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16355 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16356 (_mm256_setr_m128i): New intrinsics.
16358 2017-05-10 Julia Koval <julia.koval@intel.com>
16360 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16361 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16362 (_mm_maskz_rcp14_ss): New intrinsics.
16363 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16364 (__builtin_ia32_rcp14ss_mask): New builtins.
16365 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16367 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16369 PR tree-optimization/51513
16370 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16371 (assert_unreachable_fallthru_edge_p): Use it.
16372 (group_case_labels_stmt): Likewise.
16373 * tree-cfg.h: Prototype it.
16374 * stmt.c: Include cfghooks.h and tree-cfg.h.
16375 (emit_case_dispatch_table) <gap_label>: New local variable.
16376 Use it to fill dispatch table gaps.
16377 Test for default_label before updating probabilities.
16378 (expand_case) <default_label>: Remove unneeded initialization.
16379 Test for unreachable default case statement and remove its edge.
16380 Set default_label accordingly.
16381 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16383 2017-05-10 Carl Love <cel@us.ibm.com>
16385 * config/rs6000/rs6000-c: Add support for built-in functions
16386 vector signed char vec_neg (vector signed char)
16387 vector signed short int vec_neg (vector short int)
16388 vector signed int vec_neg (vector signed int)
16389 vector signed long long vec_neg (vector signed long long)
16390 vector float vec_neg (vector float)
16391 vector double vec_neg (vector double)
16392 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16394 * config/rs6000/altivec.h: Add define for vec_neg
16395 * doc/extend.texi: Update the built-in documentation for the
16396 new built-in functions.
16398 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16400 PR tree-optimization/77644
16401 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16403 2017-05-10 Nathan Sidwell <nathan@acm.org>
16405 * dumpfile.h (TDI_lang_all): New.
16406 (TDF_KIND): New. Renumber others
16407 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16409 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16411 (get_dump_file_name): Adjust suffix generation.
16412 (dump_enable_all): Use TDF_KIND.
16413 * doc/invoke.texi (-fdump-lang-all): Document.
16415 * dumpfile.h: Tabify.
16417 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16420 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16421 Move member access before delete.
16423 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16425 * tree-inline.c (expand_call_inline): Split block at stmt
16428 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16431 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16432 are now unused after splitting mov{sf,sd}_hardfloat.
16433 (f32_lr2): Likewise.
16434 (f32_lm): Likewise.
16435 (f32_lm2): Likewise.
16436 (f32_li): Likewise.
16437 (f32_li2): Likewise.
16438 (f32_lv): Likewise.
16439 (f32_sr): Likewise.
16440 (f32_sr2): Likewise.
16441 (f32_sm): Likewise.
16442 (f32_sm2): Likewise.
16443 (f32_si): Likewise.
16444 (f32_si2): Likewise.
16445 (f32_sv): Likewise.
16446 (f32_dm): Likewise.
16447 (f32_vsx): Likewise.
16448 (f32_av): Likewise.
16449 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16450 For movsf, order stores so the VSX stores occur before the GPR
16451 store which encourages the register allocator to use a traditional
16452 FPR instead of a GPR. For movsd, order the stores so that the GPR
16453 store comes before the VSX stores to allow the power6 to work.
16454 This is due to the power6 not having a 32-bit integer store
16455 instruction from a FPR.
16456 (movsf_hardfloat): Likewise.
16457 (movsd_hardfloat): Likewise.
16459 2017-05-09 Martin Sebor <msebor@redhat.com>
16461 PR translation/80280
16462 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16465 PR translation/80280
16466 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16467 data member added in r247778.
16468 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16470 2017-05-09 Nathan Sidwell <nathan@acm.org>
16472 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16474 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16477 2017-05-09 Marek Polacek <polacek@redhat.com>
16479 * doc/invoke.texi: Fix typo.
16481 2017-05-09 Richard Biener <rguenther@suse.de>
16483 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16484 (vrp_val_is_min): Likewise.
16485 (set_value_range_to_value): Likewise.
16486 (set_value_range_to_nonnegative): Likewise.
16487 (gimple_assign_nonzero_p): Likewise.
16488 (gimple_stmt_nonzero_p): Likewise.
16489 (vrp_int_const_binop): Likewise. Remove unreachable case.
16490 (adjust_range_with_scev): Adjust comments.
16491 (compare_range_with_value): Likewise.
16492 (extract_range_from_phi_node): Likewise.
16493 (test_for_singularity): Likewise.
16495 2017-05-09 Richard Biener <rguenther@suse.de>
16497 * tree-vrp.c (get_single_symbol): Add assert that we don't
16498 get overflowed constants as invariant part.
16499 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16500 checks. Use wi::cmp instead of recursing for integer constants.
16501 (compare_values): Just ignore whether we assumed undefined
16502 overflow instead of failing the compare.
16503 (extract_range_for_var_from_comparison_expr): Add comment before the
16504 TREE_NO_WARNING sets.
16505 (test_for_singularity): Likewise.
16506 (extract_range_from_comparison): Do not disable optimization
16507 when we assumed undefined overflow.
16508 (extract_range_basic): Remove init of unused var.
16510 2017-05-09 Richard Biener <rguenther@suse.de>
16512 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16513 (extract_range_from_multiplicative_op_1): Adjust.
16514 (extract_range_from_binary_expr_1): Use int_const_binop.
16516 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16519 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16520 rs6000_store_data_bypass_p in seven define_bypass directives and
16521 in several comments.
16522 * config/rs6000/rs6000-protos.h: Add prototype for
16523 rs6000_store_data_bypass_p function.
16524 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16525 function implements slightly different (rs6000-specific) semantics
16526 than store_data_bypass_p, returning false rather than aborting
16527 with assertion error when arguments do not satisfy the
16528 requirements of store data bypass.
16529 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16530 rs6000_store_data_bypass_p.
16532 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16534 * config/xtensa/xtensa-protos.h
16535 (xtensa_initial_elimination_offset): New declaration.
16536 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16537 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16538 macro definition, add case for FRAME_POINTER_REGNUM when
16539 FRAME_GROWS_DOWNWARD.
16540 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16541 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16542 xtensa_initial_elimination_offset.
16544 2017-05-08 Nathan Sidwell <nathan@acm.org>
16546 * doc/invoke.texi: Alphabetize -fdump options.
16548 2017-05-08 Martin Sebor <msebor@redhat.com>
16550 PR translation/80280
16551 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16553 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16555 * target.def (compute_frame_layout): New optional target hook.
16556 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16557 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16558 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16560 * reload1.c (verify_initial_elim_offsets): Likewise.
16561 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16562 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16563 (arm_get_frame_offsets): Split up into this ...
16564 (arm_compute_frame_layout): ... and this function.
16566 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16568 * config/aarch64/constraints.md (Usa): New constraint.
16569 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16571 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16573 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16574 with_multilib_list after it has been checked.
16576 2017-05-08 Richard Biener <rguenther@suse.de>
16578 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16579 (bitmap_set_subtract_values): Likewise.
16581 2017-05-08 Richard Biener <rguenther@suse.de>
16583 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16584 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16586 (gimple_stmt_nonzero_warnv_p): Rename to ...
16587 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16589 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16590 (extract_range_basic): Adjust, do not disable propagation on
16591 strict overflow sensitive simplification.
16592 (vrp_visit_cond_stmt): Likewise.
16594 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16596 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16597 body size unconditionally.
16599 2017-05-07 Jeff Law <law@redhat.com>
16602 2017-05-06 Jeff Law <law@redhat.com>
16603 PR tree-optimization/78496
16604 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16607 PR tree-optimization/78496
16608 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16609 (simplify_stmt_using_ranges): Call it.
16610 (vrp_dom_walker::before_dom_children): Extract equivalences
16611 from an ASSERT_EXPR with an equality comparison against a
16614 2017-05-06 Jeff Law <law@redhat.com>
16616 PR tree-optimization/78496
16617 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16620 PR tree-optimization/78496
16621 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16622 (simplify_stmt_using_ranges): Call it.
16623 (vrp_dom_walker::before_dom_children): Extract equivalences
16624 from an ASSERT_EXPR with an equality comparison against a
16627 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16629 * lra-constraints.c (lra_copy_reg_equiv): New function.
16630 (split_reg): Use it to copy equivalence information from the
16631 original register to the spill register.
16633 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16635 PR rtl-optimization/75964
16636 * simplify-rtx.c (simplify_const_relational_operation): Remove
16637 invalid handling of comparisons of integer ABS.
16639 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16641 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16642 initialize to zero.
16643 (init_regs): Remove declaration.
16644 (function_arg_advance_32): Initialize error_p as boolean variable.
16646 2017-05-05 Nathan Sidwell <nathan@acm.org>
16648 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16649 lines. Use for (;;).
16651 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16653 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16654 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16655 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16656 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16657 VF=2 that require versioning.
16659 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16661 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16664 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16666 * diagnostic.h (diagnostic_override_option_index): Convert from
16667 macro to inline function.
16669 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16671 * diagnostic.c (last_module_changed_p): New function.
16672 (set_last_module): New function.
16673 (diagnostic_report_current_module): Convert macro usage to
16674 the above functions.
16675 * diagnostic.h (diagnostic_context::last_module): Strengthen
16676 from const line_map * to const line_map_ordinary *.
16677 (diagnostic_last_module_changed): Delete macro.
16678 (diagnostic_set_last_module): Delete macro.
16680 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16682 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16683 with diagnostic_report_diagnostic.
16684 (diagnostic_n_impl_richloc): Likewise.
16685 * diagnostic.h (report_diagnostic): Delete macro.
16686 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16687 with diagnostic_report_diagnostic.
16688 * substring-locations.c (format_warning_va): Likewise.
16690 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16692 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16693 save/restor of format_spec. Move option-printing code to...
16694 (print_option_information): ...this new function, and
16695 reimplement by simply printing to the pretty_printer,
16696 rather than appending to the format string.
16698 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16700 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16701 handling logic into...
16702 (update_effective_level_from_pragmas): ...this new function.
16704 2017-05-04 Andrew Waterman <andrew@sifive.com>
16706 * config/riscv/riscv.opt (mstrict-align): New option.
16707 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16708 (SLOW_UNALIGNED_ACCESS): Define.
16709 (riscv_slow_unaligned_access): Declare.
16710 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16712 (riscv_slow_unaligned_access): New variable.
16713 (rocket_tune_info): Set slow_unaligned_access to true.
16714 (optimize_size_tune_info): Set slow_unaligned_access to false.
16715 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16716 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16717 (riscv_option_override): Set riscv_slow_unaligned_access.
16718 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16720 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16722 * config/riscv/riscv.md: Unify indentation.
16724 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16729 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16731 (extendsi<mode>2): Add support for doing sign extension via
16732 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16733 don't have ISA 3.0 instructions.
16734 (extendsi<mode>2 splitter): Likewise.
16735 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16736 generate the normal insns since SImode can now go in vector
16737 registers. Disallow the special UNSPECs needed for previous
16738 machines to hide SImode being used. Add new insns
16739 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16740 (fix_trunc<mode>si2_stfiwx): Likewise.
16741 (fix_trunc<mode>si2_internal): Likewise.
16742 (fixuns_trunc<mode>si2): Likewise.
16743 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16744 (fctiw<u>z_<mode>_smallint): Likewise.
16745 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16746 of floating point to 32-bit integer from doing a direct move to
16747 the GPR registers to do a store.
16748 (fctiwz_<mode>): Break long line.
16750 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16752 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16753 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16754 (addr_list, addr_offset_valid_p): New.
16755 (split_address_groups): Check offset validity with above function.
16756 (gt-tree-ssa-loop-ivopts.h): Include header file.
16758 2017-05-05 Nathan Sidwell <nathan@acm.org>
16760 * config.gcc (arm*-*-*): Add missing 'fi'.
16762 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16764 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16765 included in -fopt-info does not matter.
16766 * doc/optinfo.texi (-fopt-info): Fix description of default
16767 behavour. Explicitly say order of options included in -fopt-info
16770 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16772 * config.gcc: Allow combinations of aprofile and rmprofile values for
16773 --with-multilib-list.
16774 * config/arm/t-multilib: New file.
16775 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16776 variables. Remove setting of ISA and floating-point ABI in
16777 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16778 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16779 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16780 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16782 * config/arm/t-rmprofile: Likewise except for the matches changes.
16783 * doc/install.texi (--with-multilib-list): Document the combination of
16784 aprofile and rmprofile values and warn about pitfalls in doing that.
16786 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16788 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16789 (movdi_aarch64): Likewise.
16791 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16793 PR tree-optimization/80632
16794 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16796 (build_arrays): Initialize it for virtual phis.
16797 (fix_phi_nodes): Use it for virtual phis.
16799 PR tree-optimization/80558
16800 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16801 [x, y] op z into [x op, y op z] for op & or | if conditions
16804 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16805 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16808 * config/arm/arm.md (use_literal_pool): Remove.
16809 (64-bit immediate split): No longer takes cost into consideration
16810 if arm_disable_literal_pool is enabled.
16811 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16812 used when arm_disable_literal_pool is enabled.
16813 (arm_max_const_double_inline_cost): Remove use of
16814 arm_disable_literal_pool.
16815 (push_minipool_fix): Add assert.
16816 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16817 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16818 (no_literal_pool_sf_immediate): New.
16820 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16822 PR tree-optimization/80613
16823 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16824 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16826 2017-05-05 Richard Biener <rguenther@suse.de>
16828 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16830 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16832 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16833 of this flag from insn conditions due to removal from r247495.
16835 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16837 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16839 (arm_early_store_addr_dep_ptr): Likewise.
16840 * config/arm/aarch-common-protos.h
16841 (arm_early_load_addr_dep_ptr): Add prototype.
16842 (arm_early_store_addr_dep_ptr): Likewise.
16843 * config/arm/cortex-a53.md: Add new bypasses.
16845 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16847 * tree.c (next_type_uid): Change type to unsigned.
16848 (type_hash_canon): Decrement back next_type_uid if
16849 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16850 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16853 2017-05-04 Martin Sebor <msebor@redhat.com>
16855 * builtins.c: Fix a trivial typo in a comment.
16857 PR middle-end/79234
16858 * builtins.c (check_sizes): Adjust to handle reading past the end.
16859 Avoid printing excessive upper bound of ranges. Use %E to print
16860 tree nodes instead of converting them to %wu.
16861 (expand_builtin_memchr): New function.
16862 (compute_dest_size): Rename...
16863 (compute_objsize): ...to this.
16864 (expand_builtin_memcpy): Adjust.
16865 (expand_builtin_mempcpy): Adjust.
16866 (expand_builtin_strcat): Adjust.
16867 (expand_builtin_strcpy): Adjust.
16868 (check_strncat_sizes): Adjust.
16869 (expand_builtin_strncat): Adjust.
16870 (expand_builtin_strncpy): Adjust and simplify.
16871 (expand_builtin_memset): Adjust.
16872 (expand_builtin_bzero): Adjust.
16873 (expand_builtin_memcmp): Adjust.
16874 (expand_builtin): Handle memcmp.
16875 (maybe_emit_chk_warning): Check strncat just once.
16877 2017-05-04 Martin Sebor <msebor@redhat.com>
16879 PR preprocessor/79214
16880 PR middle-end/79222
16881 PR middle-end/79223
16882 * builtins.c (check_sizes): Add inlining context and issue
16883 warnings even when -Wno-system-headers is set.
16884 (check_strncat_sizes): Same.
16885 (expand_builtin_strncat): Same.
16886 (expand_builtin_memmove): New function.
16887 (expand_builtin_stpncpy): Same.
16888 (expand_builtin): Handle memmove and stpncpy.
16890 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16892 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16893 which is not used any more.
16895 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16897 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16899 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16901 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16902 (cortexa53_tunings): Likewise.
16903 (cortexa57_tunings): Likewise.
16904 (cortexa72_tunings): Likewise.
16905 (cortexa73_tunings): Likewise.
16907 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16909 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16910 Set loop alignment to 8.
16912 2017-05-04 Martin Sebor <msebor@redhat.com>
16914 PR translation/80280
16915 * builtins.c (expand_builtin_object_size): Add missing quoting to
16916 %D and like directives.
16917 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16918 (hsa_type_for_tree_type): Same.
16919 (verify_function_arguments): Same.
16920 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16921 * varasm.c (get_section): Same.
16924 2017-05-04 Martin Sebor <msebor@redhat.com>
16926 PR translation/80280
16927 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16929 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16931 * config/aarch64/aarch64.c (generic_addrcost_table):
16932 Change HI/TI mode setting.
16934 2017-05-04 Martin Jambor <mjambor@suse.cz>
16936 PR tree-optimization/80622
16937 * tree-sra.c (comes_initialized_p): New function.
16938 (build_accesses_from_assign): Only set write lazily when
16939 comes_initialized_p is false.
16940 (analyze_access_subtree): Use comes_initialized_p.
16941 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16942 instead of testing for PARM_DECL.
16944 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16946 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16947 constraint on operand 0 to allow more general addressing modes.
16948 Adjust output template.
16949 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16951 * config/aarch64/aarch64-protos.h
16952 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16953 * config/aarch64/constraints.md (Dp): New address constraint.
16954 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16957 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16959 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16960 update use of estimate_ipcp_clone_size_and_time.
16961 (estimate_local_effects): Update use of
16962 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16963 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16964 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16965 Return nonspecialized time.
16967 2017-05-04 Richard Biener <rguenther@suse.de>
16969 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16970 for the last VUSE which def dominates the PHI. Directly call
16972 (get_continuation_for_phi_1): Remove.
16974 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16976 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16977 to explain the use of truncating division. Cap the number of
16978 iterations to the maximum given by nb_iterations_upper_bound,
16981 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16983 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16984 * configure: Regenerate.
16985 * config.in: Regenerate.
16986 * config/i386/driver-mingw32.c: new file.
16987 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16988 * config.host: Link driver-mingw32.o on MinGW host.
16989 * doc/install.texi: Document new --enable-mingw-wildcard configure
16992 2017-05-04 Marek Polacek <polacek@redhat.com>
16994 PR tree-optimization/80612
16995 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16997 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16998 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17000 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17001 (movt splitter): Likewise.
17002 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17003 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17004 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17005 block for Thumb-1 with MOVT.
17006 (thumb2_legitimate_address_p): Move code block ...
17007 (can_avoid_literal_pool_for_label_p): ... into this new function.
17008 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17010 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17011 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17012 "M-profile targets with the MOVT instruction".
17014 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17016 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17017 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17018 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17020 2017-05-04 Martin Liska <mliska@suse.cz>
17022 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17023 variable cond_code.
17025 2017-05-04 Richard Biener <rguenther@suse.de>
17027 * tree.c (array_at_struct_end_p): Handle arrays at struct
17028 end with flexarrays more conservatively. Refactor and treat
17029 arrays of arrays or aggregates more strict. Fix
17030 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17031 * tree.c (array_at_struct_end_p): Adjust prototype.
17032 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17033 * gimple-fold.c (get_range_strlen): Likewise.
17034 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17036 2017-05-04 Richard Biener <rguenther@suse.de>
17038 PR tree-optimization/31130
17039 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17041 (supports_overflow_infinity): Likewise.
17042 (is_negative_overflow_infinity): Likewise.
17043 (is_positive_overflow_infinity): Likewise.
17044 (is_overflow_infinity): Likewise.
17045 (stmt_overflow_infinity): Likewise.
17046 (overflow_infinity_range_p): Likewise.
17047 (usable_range_p): Remove as always returning true.
17048 (make_overflow_infinity): Remove.
17049 (negative_overflow_infinity): Likewise.
17050 (positive_overflow_infinity): Likewise.
17051 (avoid_overflow_infinity): Likewise.
17052 (set_value_range): Adjust accordingly.
17053 (set_value_range_to_nonnegative): Likewise, remove now unused
17054 overflow_infinity arg.
17055 (vrp_operand_equal_p): Adjust.
17056 (update_value_range): Likewise.
17057 (range_int_cst_singleton_p): Likewise.
17058 (operand_less_p): Likewise.
17059 (compare_values_warnv): Likewise.
17060 (extract_range_for_var_from_comparison_expr): Likewise.
17061 (vrp_int_const_binop): Likewise.
17062 (zero_nonzero_bits_from_vr): Likewise.
17063 (extract_range_from_multiplicative_op_1): Likewise.
17064 (extract_range_from_binary_expr_1): Likewise.
17065 (extract_range_from_unary_expr): Likewise.
17066 (extract_range_from_comparison): Likewise.
17067 (extract_range_basic): Likewise.
17068 (adjust_range_with_scev): Likewise.
17069 (compare_ranges): Likewise.
17070 (compare_range_with_value): Likewise.
17071 (dump_value_range): Likewise.
17072 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17074 (simplify_cond_using_ranges): Adjust.
17076 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17078 * brig-builtins.def: Added a builtin for class_f64.
17079 * builtin-types.def: Added a builtin type needed by class_f64.
17081 2017-05-03 Jason Merrill <jason@redhat.com>
17083 * timevar.def: Add TV_CONSTEXPR.
17085 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17087 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17089 2017-05-03 Martin Jambor <mjambor@suse.cz>
17091 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17092 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17093 * ipa-cp.c (ipcp_propagate_stage): Do not call
17094 ipa_update_after_lto_read.
17095 * ipa-inline.c (ipa_inline): Likewise.
17097 2017-05-03 Martin Jambor <mjambor@suse.cz>
17099 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17100 tag. Added a default constructor and a destructor.
17101 (ipa_edge_args_sum_t): New class;
17102 (ipa_edge_args_sum): Declare.
17103 (ipa_edge_args_vector): Remove declaration.
17104 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17105 (ipa_free_edge_args_substructures): Remove declaration.
17106 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17107 (ipa_edge_args_info_available_for_edge_p): Likewise.
17108 * ipa-prop.c (ipa_edge_args_vector): Removed.
17109 (edge_removal_hook_holder): Likewise.
17110 (edge_duplication_hook_holder): Likewise.
17111 (ipa_edge_args_sum): New variable.
17112 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17113 ipa_edge_args_vector.
17114 (ipa_free_edge_args_substructures): Likewise.
17115 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17116 ipa_edge_args_vector.
17117 (ipa_edge_removal_hook): Turned into method
17118 ipa_edge_args_sum_t::remove.
17119 (ipa_edge_duplication_hook): Turned into method
17120 ipa_edge_args_sum_t::duplicate.
17121 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17122 registering edge hooks.
17123 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17124 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17125 ipa_edge_args_sum instead of ipa_edge_args_vector.
17126 * ipa-profile.c (ipa_profile): Likewise.
17128 2017-05-03 Martin Jambor <mjambor@suse.cz>
17130 * symbol-summary.h (function_summary): New method exists.
17131 (function_summary::symtab_removal): Deallocate through release.
17132 (call_summary): New class.
17133 (gt_ggc_mx): New overload.
17134 (gt_pch_nx): Likewise.
17135 (gt_pch_nx): Likewise.
17137 2017-05-03 Jeff Law <law@redhat.com>
17139 PR tree-optimization/78496
17140 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17141 from simplify_cond_using_ranges. Split off code to walk
17142 backwards through casts into ...
17143 (simplify_cond_using_ranges_2): New function.
17144 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17145 (execute_vrp): After identifying jump threads, call
17146 simplify_cond_using_ranges_2.
17148 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17151 * ipa-inline.h (inline_summary): Add ctor.
17152 (create_ggc): Do not use ggc_cleared_alloc.
17154 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17155 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17157 * gcc.c (handle_braces): Support escaping in switch matching
17159 * doc/invoke.texi (Spec Files): Document it.
17160 Remove superfluous @code markup in items.
17162 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17164 * diagnostic-show-locus.c (struct column_range): New struct.
17165 (get_affected_columns): New function.
17166 (get_printed_columns): New function.
17167 (struct correction): New struct.
17168 (correction::ensure_capacity): New function.
17169 (correction::ensure_terminated): New function.
17170 (struct line_corrections): New struct.
17171 (line_corrections::~line_corrections): New dtor.
17172 (line_corrections::add_hint): New function.
17173 (layout::print_trailing_fixits): Reimplement in terms of the new
17175 (selftest::test_overlapped_fixit_printing): New function.
17176 (selftest::diagnostic_show_locus_c_tests): Call it.
17178 2017-05-03 Nathan Sidwell <nathan@acm.org>
17180 Canonicalize canonical type hashing
17181 * tree.h (type_hash_canon_hash): Declare.
17182 * tree.c (type_hash_list, attribute_hash_list): Move into
17183 type_hash_canon_hash.
17184 (build_type_attribute_qual_variant): Break out hash code calc into
17185 type_hash_canon_hash.
17186 (type_hash_canon_hash): New. Generic type hash computation.
17187 (build_range_type_1, build_array_type_1, build_function_type,
17188 build_method_type_directly, build_offset_type, build_complex_type,
17189 make_vector_type): Call it.
17191 2017-05-03 Richard Biener <rguenther@suse.de>
17193 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17194 When all DRs have unknown misaligned do not always peel
17195 when there is a store but apply the same costing model as if
17196 there were only loads.
17198 2017-05-03 Richard Biener <rguenther@suse.de>
17201 PR tree-optimization/80492
17202 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17203 compare_base_decls returning dont-know properly.
17205 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17207 * config/arm/iterators.md (CCSI): New mode iterator.
17208 (arch): New mode attribute.
17209 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17210 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17211 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17212 code iterator for success result mode.
17213 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17214 the corresponding new insn generators.
17216 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17219 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17220 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17222 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17224 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17225 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17226 (DDR_A): Wrap DDR argument in brackets.
17227 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17228 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17229 (DDR_REVERSED_P): Likewise.
17231 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17233 PR tree-optimization/79472
17234 * tree-switch-conversion.c (struct switch_conv_info): Add
17235 contiguous_range and default_case_nonstandard fields.
17236 (collect_switch_conv_info): Compute contiguous_range and
17237 default_case_nonstandard fields, don't clear final_bb if
17238 contiguous_range and only the default case doesn't have the required
17240 (check_all_empty_except_final): Set default_case_nonstandard instead
17241 of failing if contiguous_range and the default case doesn't have empty
17243 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17244 and only the default case doesn't have the required constants. Skip
17246 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17247 if default_case_nonstandard.
17248 (build_constructors): Build constant 1 just once. Assert that default
17249 values aren't inserted in between cases if contiguous_range. Skip
17251 (build_arrays): Skip virtual phis.
17252 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17253 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17254 Handle virtual phis.
17255 (gen_inbound_check): Handle default_case_nonstandard case.
17256 (process_switch): Adjust check_final_bb caller. Call
17257 gather_default_values with the first non-default case instead of
17258 default case if default_case_nonstandard.
17260 2017-05-02 Nathan Sidwell <nathan@acm.org>
17262 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17263 check. Fix formatting.
17265 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17267 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17268 errors when comparing specialized and unspecialized times.
17270 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17272 * diagnostic-show-locus.c
17273 (layout::should_print_annotation_line_p): Make private.
17274 (layout::print_annotation_line): Make private.
17275 (layout::annotation_line_showed_range_p): Make private.
17276 (layout::show_ruler): Make private.
17277 (layout::print_source_line): Make private. Pass in line and
17278 line_width, rather than calling location_get_source_line. Drop
17280 (layout::print_leading_fixits): New method.
17281 (layout::print_any_fixits): Rename to...
17282 (layout::print_trailing_fixits): ...this, and make private.
17283 Don't print newline fixits.
17284 (diagnostic_show_locus): Move logic for printing one row into...
17285 (layout::print_line): ...this new function. Move the
17286 location_get_source_line call and error-handling from
17287 print_source_line to here. Call print_leading_fixits, and rename
17288 print_any_fixits to print_trailing_fixits.
17289 (selftest::test_fixit_insert_containing_newline): Update now that
17290 newlines are partially supported.
17291 (selftest::test_fixit_insert_containing_newline_2): New test.
17292 (selftest::test_fixit_replace_containing_newline): Update comments.
17293 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17294 * edit-context.c (class added_line): New class.
17295 (class edited_line): Describe newline handling in comment.
17296 (edited_line::actually_edited_p): New method.
17297 (edited_line::print_content): Delete redundant decl.
17298 (edited_line::m_predecessors): New field.
17299 (edited_file::print_content): Call edited_line::print_content.
17300 (edited_file::print_diff): Update to support newlines.
17301 (edited_file::print_diff_hunk): Likewise.
17302 (edited_file::print_run_of_changed_lines): New function.
17303 (edited_file::print_diff_line): Convert to...
17304 (print_diff_line): ...this.
17305 (edited_file::get_effective_line_count): New function.
17306 (edited_line::edited_line): Initialize new field m_predecessors.
17307 (edited_line::~edited_line): Clean up m_predecessors.
17308 (edited_line::apply_fixit): Handle newlines.
17309 (edited_line::get_effective_line_count): New function.
17310 (edited_line::print_content): New function.
17311 (edited_line::print_diff_lines): New function.
17312 (selftest::test_applying_fixits_insert_containing_newline): New
17314 (selftest::test_applying_fixits_replace_containing_newline): New
17316 (selftest::insert_line): New function.
17317 (selftest::test_applying_fixits_multiple_lines): Add example of
17319 (selftest::edit_context_c_tests): Call the new tests.
17321 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17323 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17324 parameter cand. Update dump information.
17325 (get_computation_cost): Update uses.
17327 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17329 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17330 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17331 (get_computation_at, rewrite_use_address): Update use of
17332 get_computation_aff.
17334 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17336 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17337 (get_computation): Delete.
17338 (get_computation_cost): Implement like get_computation_cost_at.
17339 Use get_computation_at.
17340 (get_computation_cost_at): Delete.
17341 (rewrite_use_nonlinear_expr): Use get_computation_at.
17342 (rewrite_use_compare, remove_unused_ivs): Ditto.
17344 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17346 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17348 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17350 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17351 (ivopts_global_cost_for_size): Rename parameter and update uses.
17352 (iv_ca_recount_cost): Update uses.
17353 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17354 candidates seperately in n_invs and n_cands.
17355 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17357 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17359 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17360 (find_inv_vars_cb): New.
17361 (find_depends): Renamed to ...
17362 (find_inv_vars): ... this.
17363 (add_candidate_1, force_var_cost): Call find_inv_vars.
17364 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17366 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17368 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17369 inv_vars. Add inv_exprs.
17370 (struct iv_cand): Rename depends_on to inv_vars.
17371 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17372 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17373 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17374 (dump_cand): Dump inv_vars.
17375 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17376 (record_invariant, find_depends, add_candidate_1): Ditto.
17377 (set_group_iv_cost, force_var_cost): Ditto.
17378 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17379 (get_computation_cost_at, get_computation_cost): Ditto.
17380 (determine_group_iv_cost_generic): Ditto.
17381 (determine_group_iv_cost_address): Ditto.
17382 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17383 (determine_group_iv_costs): Ditto.
17384 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17385 (iv_ca_set_remove_invariants): Renamed to ...
17386 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17387 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17388 (iv_ca_set_add_invariants): Renamed to ...
17389 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17390 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17391 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17392 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17393 (create_new_ivs): Remove useless dump.
17395 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17397 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17399 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17400 (iv_ca_set_no_cp, create_new_iv): Ditto.
17402 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17404 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17406 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17408 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17409 function tree_check2.
17411 2017-05-02 Martin Liska <mliska@suse.cz>
17413 * doc/gcov.texi: Add missing preposition.
17414 * gcov.c (function_info::function_info): Properly fill up
17415 all member variables.
17417 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17419 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17421 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17423 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17425 2017-05-02 Martin Liska <mliska@suse.cz>
17428 * lto-streamer-in.c (lto_read_tree_1): Remove
17429 LTO_STREAMER_DEBUG.
17430 * lto-streamer.c (struct tree_hash_entry): Likewise.
17431 (struct tree_entry_hasher): Likewise.
17432 (tree_entry_hasher::hash): Likewise.
17433 (tree_entry_hasher::equal): Likewise.
17434 (lto_streamer_init): Likewise.
17435 (lto_orig_address_map): Likewise.
17436 (lto_orig_address_get): Likewise.
17437 (lto_orig_address_remove): Likewise.
17438 * lto-streamer.h: Likewise.
17439 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17440 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17442 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17444 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17445 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17446 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17447 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17448 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17449 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17450 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17451 (mm_maskz_sub_ss): New intrinsics.
17452 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17453 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17454 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17455 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17456 (__builtin_ia32_subss_mask_round): New builtins.
17457 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17458 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17459 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17461 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17462 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17464 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17467 2017-05-02 Martin Jambor <mjambor@suse.cz>
17469 PR tree-optimization/78687
17470 * tree-sra.c (access): New field parent.
17471 (process_subtree_disqualification): New function.
17472 (disqualify_candidate): Call it.
17473 (build_accesses_from_assign): Reset write flag if creating an
17475 (build_access_subtree): Fill in parent field and also prpagate
17476 down grp_write flag.
17477 (create_artificial_child_access): New parameter set_grp_write, set
17478 grp_write to its value.
17479 (propagate_subaccesses_across_link): Also propagate grp_write flag
17481 (propagate_all_subaccesses): Push the closest parent back to work
17482 queue if add_access_to_work_queue returned true.
17484 2017-05-02 Richard Biener <rguenther@suse.de>
17486 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17487 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17488 -fstrict-overflow documentation.
17489 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17490 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17491 flag_strict_overflow.
17492 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17493 * lto-opts.c (lto_write_options): Do not stream it.
17494 * lto-wrapper.c (merge_and_complain): Do not handle it.
17495 * opts.c (default_options_table): Do not set -fstrict-overflow.
17496 (finish_options): Likewise do not clear it when sanitizing.
17497 * simplify-rtx.c (simplify_const_relational_operation): Do not
17498 test flag_strict_overflow.
17500 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17502 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17503 using enabled attribute.
17504 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17505 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17506 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17507 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17508 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17509 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17510 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17511 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17512 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17513 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17515 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17517 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17519 2017-05-02 Richard Biener <rguenther@suse.de>
17521 PR tree-optimization/80591
17523 2017-04-10 Richard Biener <rguenther@suse.de>
17525 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17528 2017-05-02 Richard Biener <rguenther@suse.de>
17530 PR tree-optimization/80549
17531 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17532 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17533 headers if they do not have a preheader.
17535 2017-05-02 Martin Liska <mliska@suse.cz>
17538 * common.opt: Fix typo.
17539 * doc/invoke.texi: Likewise.
17541 2017-05-01 Jan Beulich <jbeulich@suse.com>
17543 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17544 swapping, add (x,x,m,x,n) alternative.
17546 2017-05-01 Nathan Sidwell <nathan@acm.org>
17548 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17549 unnecessary unadjusted_alignment check.
17551 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17554 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17556 * gimplify.c (gimplify_cilk_detach): New function.
17557 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17558 * tree-core.h: Document EXPR_CILK_SPAWN.
17559 * tree.h (EXPR_CILK_SPAWN): Define.
17561 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17563 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17564 to use new fixit_hint representation, using the "replace" logic.
17565 (get_line_span_for_fixit_hint): Likewise.
17566 (layout::print_any_fixits): Likewise.
17567 (selftest::test_one_liner_many_fixits): Rename to...
17568 (selftest::test_one_liner_many_fixits_1): ...this, and update
17569 comment and expected output to reflect that the multiple fix-it
17570 hints are now consolidated into one insertion.
17571 (selftest::test_one_liner_many_fixits_2): New test.
17572 (selftest::test_diagnostic_show_locus_one_liner): Update for
17574 (selftest::test_fixit_consolidation): Update for fix-it API
17576 * diagnostic.c (print_parseable_fixits): Likewise.
17577 * edit-context.c (edited_line::m_line_events): Convert from
17578 auto_vec <line_event *> to auto_vec <line_event>.
17579 (class line_event): Convert from abstract base class to a concrete
17580 class, taking over the role of replace_event.
17581 (class insert_event): Delete.
17582 (class replace_event): Rename to class line_event. Convert to
17584 (edit_context::add_fixits): Reimplement.
17585 (edit_context::apply_insert): Delete.
17586 (edit_context::apply_replace): Rename to...
17587 (edit_context::apply_fixit): ...this. Convert to half-open range.
17588 (edited_file::apply_insert): Delete.
17589 (edited_file::apply_replace): Rename to...
17590 (edited_file::apply_fixit): ...this.
17591 (edited_line::~edited_line): Drop deletion of events.
17592 (edited_line::apply_insert): Delete.
17593 (edited_line::apply_replace): Rename to...
17594 (edited_line::apply_fixit): ...this. Convert to half-open range.
17595 Update for change to type of m_line_events.
17596 * edit-context.h (edit_context::apply_insert): Delete.
17597 (edit_context::apply_replace): Rename to...
17598 (edit_context::apply_fixit): ...this.
17600 2017-05-01 Martin Sebor <msebor@redhat.com>
17602 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17605 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17608 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17609 __get_cpuid_max returns 0.
17610 (__get_cpuid_count): Ditto.
17612 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17614 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17615 replacement expression is another instance of one of its arguments.
17617 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17620 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17621 check for stack push/pop autoinc.
17622 * config/i386/i386.c (ix86_agi_dependent): Return false
17623 if the only reason why modified_in_p returned true is that
17624 addr is SP based and set_insn is a push or pop.
17626 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17628 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17631 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17634 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17635 (account_size_time): Use two predicates - exec_pred and
17637 (evaluate_conditions_for_known_args): Compute both clause and
17639 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17640 (inline_summary_t::duplicate): Update.
17641 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17643 (compute_inline_parameters): Likewise.
17644 (estimate_edge_size_and_time): Update caluclation of time.
17645 (estimate_node_size_and_time): Compute both time and nonspecialized
17647 (estimate_ipcp_clone_size_and_time): Update.
17648 (inline_merge_summary): Update.
17649 (do_estimate_edge_time): Update.
17650 (do_estimate_edge_size): Update.
17651 (do_estimate_edge_hints): Update.
17652 (inline_read_section, inline_write_summary): Stream both new predicates.
17653 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17655 (compute_inlined_call_time): Cleanup.
17656 (big_speedup_p): Update.
17657 (edge_badness): Update.
17658 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17659 (size_time_entry): Replace predicate by exec_predicate and
17660 nonconst_predicate.
17661 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17662 (estimate_edge_time): Return also nonspec_time.
17663 (reset_edge_growth_cache): Update.
17665 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17667 PR rtl-optimization/80491
17668 * ifcvt.c (noce_process_if_block): When looking for x setter
17669 with missing else_bb, don't check only the insn right before
17670 cond_earliest, but look for the last insn that x is modified in
17671 within the same bb.
17673 PR rtl-optimization/80491
17674 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17676 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17678 PR tree-optimization/80487
17679 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17681 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17683 PR tree-optimization/79697
17684 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17685 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17686 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17688 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17689 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17691 2017-04-28 Martin Sebor <msebor@redhat.com>
17693 PR tree-optimization/80523
17694 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17695 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17697 (maybe_warn, format_directive, parse_directive): Use new functions.
17698 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17700 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17702 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17704 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17706 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17707 target_header_dir): Set correctly.
17708 * configure: Regenerated.
17709 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17710 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17711 instead of SYSTEM_HEADER_DIR.
17713 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17715 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17716 (estimate_local_effects): Likewise.
17717 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17718 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17719 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17720 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17721 do_estimate_edge_time, estimate_edge_time): Likewise.
17722 * ipa-inline-analysis.c (estimate_node_size_and_time,
17723 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17724 (estimate_time_after_inlining): Remove.
17726 2017-04-28 Martin Liska <mliska@suse.cz>
17728 * doc/gcov.texi: Enhance documentation of gcov.
17730 2017-04-28 Martin Liska <mliska@suse.cz>
17732 * doc/gcov.texi: Sort options in alphabetic order.
17733 * doc/gcov-dump.texi: Likewise.
17734 * doc/gcov-tool.texi: Likewise.
17735 * gcov.c (print_usage): Likewise.
17736 * gcov-dump.c (print_usage): Likewise.
17737 * gcov-tool.c (print_merge_usage_message): Likewise.
17738 (print_rewrite_usage_message): Likewise.
17739 (print_overlap_usage_message): Likewise.
17741 2017-04-28 Martin Liska <mliska@suse.cz>
17743 PR gcov-profile/53915
17744 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17746 2017-04-28 Martin Liska <mliska@suse.cz>
17748 PR gcov-profile/79891
17749 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17750 is marked by compiler as living on a line.
17751 (get_cycles_count): Remove usage of the union.
17752 (output_intermediate_file): Likewise.
17753 (find_source): Fix GNU coding style.
17754 (accumulate_line_counts): Remove old non-all block mode.
17755 (output_lines): Remove usage of the union.
17756 * profile.c (output_location): Include all BBs, even if
17757 belonging to a same line (and file) as a previous BB.
17759 2017-04-28 Martin Liska <mliska@suse.cz>
17761 * gcov.c (process_args): Handle new argument 'w'.
17762 (read_graph_file): Assign ID to BBs.
17763 (output_branch_count): Display BB # if verbose flag is set.
17764 (output_lines): Likewise for arcs.
17765 (print_usage): Add '--verbose' option help.
17766 * doc/gcov.texi: Document --verbose (-w) option.
17768 2017-04-28 Martin Liska <mliska@suse.cz>
17770 * gcov.c (struct block_location_info): New struct.
17771 (process_file): Fill up the new structure.
17772 (read_graph_file): Replace usage of encoding by the newly added
17774 (add_line_counts): Likewise.
17775 (accumulate_line_counts): Remove usage of the union.
17776 (function_info::function_info): New function.
17777 (function_info::~function_info): Likewise.
17778 (process_file): Call delete instead of release_function.
17779 (release_function): Release the function.
17780 (release_structures): Call delete instead of release_function.
17781 (solve_flow_graph): Replace usage of num_blocks.
17782 (find_exception_blocks): Likewise.
17783 (output_lines): Fix GNU coding style.
17785 2017-04-28 Martin Liska <mliska@suse.cz>
17788 * coverage.c (coverage_remove_note_file): New function.
17789 * coverage.h: Declare the function.
17790 * toplev.c (finalize): Clean if an error has been seen.
17792 2017-04-28 Martin Liska <mliska@suse.cz>
17794 PR gcov-profile/80031
17795 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17796 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17797 * gcov.c (read_graph_file): Read just number of blocks.
17798 * profile.c (branch_prob): Do not stream 0 flags per a basic
17801 2017-04-28 Martin Liska <mliska@suse.cz>
17803 * gcov-dump.c (tag_*): Add new argument to declarations.
17804 (dump_gcov_file): Likewise.
17805 (tag_blocks): Add and use new argument depth.
17806 (tag_arcs): Likewise.
17807 (tag_lines): Likewise.
17808 (tag_counters): Likewise.
17809 (tag_summary): Likewise.
17810 (dump_working_sets): Use depth to do a proper indentation.
17812 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17815 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17816 * symtab.c (symtab_node::debug_symtab): Move definition here.
17818 2017-04-28 Richard Biener <rguenther@suse.de>
17820 * lto-streamer.h (LTO_major_version): Bump to 7.
17822 2017-04-28 Richard Biener <rguenther@suse.de>
17824 * tree-vrp.c (assert_info): New struct.
17825 (add_assert_info): New helper.
17826 (register_edge_assert_for_2): Refactor to add asserts to a vector
17828 (register_edge_assert_for_1): Likewise.
17829 (register_edge_assert_for): Likewise.
17830 (finish_register_edge_assert_for): New helper actually registering
17831 asserts where live on edge.
17832 (find_conditional_asserts): Adjust.
17833 (find_switch_asserts): Likewise.
17834 (evrp_dom_walker::try_find_new_range): Generalize.
17835 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17837 2017-04-27 Marek Polacek <polacek@redhat.com>
17840 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17841 arg10 and arg11 to itype.
17843 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17845 * doc/extend.texi (Object Size Checking): Improve grammar.
17847 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17850 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17851 that the logic for permitting reciprocal estimates matches that
17854 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17857 * tree.c (type_cache_hasher::equal): Only compare
17858 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17859 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17860 non-aggregate element types.
17861 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17862 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17864 2017-04-27 Richard Biener <rguenther@suse.de>
17866 PR middle-end/80533
17867 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17868 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17869 keeping a reference to a trailing array.
17871 2017-04-27 Richard Biener <rguenther@suse.de>
17873 PR middle-end/80539
17874 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17875 being in loop-closed SSA form conservatively.
17876 (chrec_fold_multiply_poly_poly): Likewise.
17878 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17880 PR middle-end/79665
17881 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17882 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17884 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17887 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17888 (aarch64_function_arg_alignment): Return unsigned int again, but still
17889 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17890 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17891 Don't emit -Wpsabi note.
17892 (aarch64_function_arg_boundary): Likewise.
17893 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17896 2017-04-26 Nathan Sidwell <nathan@acm.org>
17898 * tree.h (crc32_unsigned_n): Declare.
17899 (crc32_unsigned, crc32_unsigned): Make inline.
17900 * tree.c (crc32_unsigned_bits): Replace with ...
17901 (crc32_unsigned_n): ... this.
17902 (crc32_unsigned, crc32_byte): Remove.
17903 (crc32_string): Remove unnecessary braces.
17905 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17907 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17908 * ipa-inline-analysis.c (MAX_TIME): Remove.
17909 (account_size_time): Use sreal for time.
17910 (dump_inline_summary): Update.
17911 (estimate_function_body_sizes): Update.
17912 (estimate_edge_size_and_time): Update.
17913 (estimate_calls_size_and_time): Update.
17914 (estimate_node_size_and_time): Update.
17915 (inline_merge_summary): Update.
17916 (inline_update_overall_summary): Update.
17917 (estimate_time_after_inlining): Update.
17918 (inline_read_section): Update.
17919 (inline_write_summary): Update.
17920 * ipa-inline.c (compute_uninlined_call_time): Update.
17921 (compute_inlined_call_time): Update.
17922 (recursive_inlining): Update.
17923 (inline_small_functions): Update.
17924 (dump_overall_stats): Update.
17925 * ipa-inline.h: Include sreal.h.
17926 (size_time_entry): Turn time to sreal.
17927 (inline_summary): Turn self_time nad time to sreal.
17929 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17931 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17933 (sreal::stream_out, sreal::stream_in): New.
17934 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17936 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17938 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17941 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17944 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17946 (dimode_scalar_chain::compute_convert_gain): Ditto.
17947 (dimode_scalar_chain::make_vector_copies): Ditto.
17948 (dimode_scalar_chain::convert_reg): Ditto.
17949 (dimode_scalar_chain::convert_insn): Ditto.
17950 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17951 (VI248_AVX512BW_1): New mode iterator.
17952 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17953 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17956 2017-04-25 Martin Sebor <msebor@redhat.com>
17958 PR tree-optimization/80497
17959 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17960 constants are representable in HOST_WIDE_INT.
17961 (parse_directive): Ditto.
17963 2017-04-25 Martin Sebor <msebor@redhat.com>
17966 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17967 (new_zero_array): Adjust signature.
17968 (dom_info::dom_init): Used unsigned rather that size_t.
17969 (dom_info::dom_info): Same.
17971 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17972 Jakub Jelinek <jakub@redhat.com>
17975 * config/arm/arm.c: Include gimple.h.
17976 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17977 returns negative, increment ncrn only if it returned positive.
17978 (arm_needs_doubleword_align): Return int instead of bool,
17979 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17980 members, but if there is any such non-FIELD_DECL
17981 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17982 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17983 returns negative, increment nregs only if it returned positive.
17984 (arm_setup_incoming_varargs): Likewise.
17985 (arm_function_arg_boundary): Emit -Wpsabi note if
17986 arm_needs_doubleword_align returns negative, return
17987 DOUBLEWORD_ALIGNMENT only if it returned positive.
17989 2017-04-25 Marek Polacek <polacek@redhat.com>
17992 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17993 first argument to type.
17995 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17998 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17999 type checks to test for compatibility instead of equality.
18001 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18002 Jakub Jelinek <jakub@redhat.com>
18005 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18007 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18008 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18009 the alignment computation, but return their maximum in warn_alignment.
18010 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18011 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18013 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18014 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18017 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18019 * config/arc/simdext.md (dmpyh): Fix typo.
18021 2017-04-25 Richard Biener <rguenther@suse.de>
18023 PR tree-optimization/80492
18024 * alias.c (compare_base_decls): Handle registers with asm
18025 specification conservatively.
18026 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18027 compare_base_decls returning dont-know properly.
18029 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18031 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18032 (legitimate_offset_address_p): New function.
18033 (arc_legitimate_address_p): Use above function.
18035 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18037 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18039 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18041 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18042 ACCH registers whenever they are available.
18044 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18046 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18047 double regs fix when not used.
18049 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18051 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18053 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18054 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18056 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18058 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18059 when emitting short ADD instructions.
18061 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18063 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18065 (cmpsi_cc_c_insn): Likewise.
18066 (cbranchsi4_scratch): Compute proper instruction length using
18067 compact_hreg_operand.
18068 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18070 2017-04-25 Richard Biener <rguenther@suse.de>
18072 PR middle-end/80509
18073 * passes.c (pass_manager::pass_manager): Initialize
18074 m_name_to_pass_map.
18076 2017-04-25 Richard Biener <rguenther@suse.de>
18078 PR tree-optimization/79201
18079 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18081 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18084 * config/s390/vector.md: Split MEM->GPR vector moves for
18085 non-s_operand addresses.
18087 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18090 * config/s390/predicates.md (reload_const_wide_int_operand): New
18092 * config/s390/s390.md ("movti"): Remove d/P alternative.
18093 ("movti_bigconst"): New pattern definition.
18095 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18098 * s390-protos.h (s390_expand_cs_hqi): Removed.
18099 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18100 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18101 modes as well as CCZ1mode and CCZmode.
18102 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18103 signature of s390_emit_compare_and_swap.
18104 (s390_expand_cs_hqi): Likewise, make static.
18105 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18106 compare-and-swap, in some cases.
18107 (s390_expand_cs): Wrapper function.
18108 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18110 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18111 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18112 patterns for small and large integers. Forbid symref memory operands.
18113 Move expander to s390.c. Require cc register.
18114 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18115 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18116 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18117 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18118 symref memory operands. Remove CC mode and call s390_match_ccmode
18120 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18122 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18124 * config/s390/s390.md (define_peephole2): New peephole to help
18125 combining the load-and-test pattern with volatile memory.
18127 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18129 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18130 with CCZmode for TARGET_Z196.
18132 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18134 PR rtl-optimization/80501
18135 * combine.c (make_compound_operation_int): Set subreg_code to SET
18136 even for AND with mask of the sign bit of mode.
18138 PR rtl-optimization/80500
18139 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18140 sum's initial value.
18142 2017-04-25 Julian Brown <julian@codesourcery.com>
18143 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18145 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18147 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18149 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18151 2017-04-25 Julian Brown <julian@codesourcery.com>
18152 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18154 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18155 (thunderx2t99_sha): New Reservation.
18157 2017-04-25 Julian Brown <julian@codesourcery.com>
18158 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18160 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18161 type for 1-element load.
18163 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18165 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18167 2017-04-24 Martin Jambor <mjambor@suse.cz>
18169 PR tree-optimization/80293
18170 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18171 char arrays not totally scalarizable if it is false.
18172 (analyze_all_variable_accesses): Pass correct value in the new
18173 parameter. Add a statistics counter.
18175 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18177 PR middle-end/79931
18178 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18180 2017-04-24 Richard Biener <rguenther@suse.de>
18182 PR tree-optimization/80494
18183 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18184 out for complex types.
18186 2017-04-24 Richard Biener <rguenther@suse.de>
18188 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18189 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18190 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18191 (DFS): Adjust and never fail.
18192 (sccvn_dom_walker::fail): Remove.
18193 (sccvn_dom_walker::before_dom_children): Adjust.
18194 (run_scc_vn): Likewise and never fail.
18195 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18196 (pass_fre::execute): Likewise.
18198 2017-04-24 Richard Biener <rguenther@suse.de>
18200 PR tree-optimization/79725
18201 * tree-ssa-sink.c (statement_sink_location): Return whether
18202 failure reason was zero uses. Move that check later.
18203 (sink_code_in_bb): Deal with zero uses by removing the stmt
18206 2017-04-24 Richard Biener <rguenther@suse.de>
18209 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18210 pointer-based references.
18212 2017-04-24 Richard Biener <rguenther@suse.de>
18215 * pass_manager.h (pass_manager::operator new): Remove.
18216 (pass_manager::operator delete): Likewise.
18217 * passes.c (pass_manager::operator new): Remove.
18218 (pass_manager::operator delete): Likewise.
18219 (pass_manager::pass_manager): Zero individual pass members.
18221 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18224 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18225 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18226 Check "XEXP (src, 1)" operand here.
18227 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18228 Check "XEXP (src, 1)" operand here.
18229 (dimode_scalar_chain::make_vector_copies): Detect count register
18230 of a shift instruction. Zero extend count register from QImode
18231 to DImode to satisfy vector shift pattern count operand predicate.
18232 Substitute vector shift count operand with a DImode copy.
18233 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18236 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18238 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18239 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18240 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18241 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18242 (UNSPEC_NOREX_MEM): Remove definition.
18244 2017-04-21 Richard Biener <rguenther@suse.de>
18246 PR tree-optimization/79547
18247 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18248 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18249 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18250 without any constraints.
18252 2017-04-21 Richard Biener <rguenther@suse.de>
18254 PR tree-optimization/78847
18255 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18257 2017-04-21 Richard Biener <rguenther@suse.de>
18259 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18260 (build_distinct_type_copy): Likewise.
18261 (build_variant_type_copy): Likewise.
18262 * tree.c (build_qualified_type): Pass down mem-stat info.
18263 (build_distinct_type_copy): Likewise.
18264 (build_variant_type_copy): Likewise.
18266 2017-04-21 Richard Biener <rguenther@suse.de>
18268 PR tree-optimization/80237
18269 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18271 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18272 for a simplified result.
18274 2016-04-21 Richard Biener <rguenther@suse.de>
18276 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18277 sth as strict as a simple_iv but a chrec without symbols and an
18278 operand defined in the loop we are peeling (and not some subloop).
18279 (propagate_constants_for_unrolling): Propagate all constants.
18281 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18284 * config/i386/i386.c (print_reg): Remove assert for disalowed
18285 regno values, call output_operand_lossage instead.
18287 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18290 * config/i386/constraints.md (Yc): New register constraint.
18291 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18292 Use Yc constraint for alternative 2 of operand 0. Remove
18293 preferred_for_speed attribute.
18295 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18297 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18298 lastprivate clauses in SIMT case.
18300 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18302 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18304 2017-04-20 Richard Biener <rguenther@suse.de>
18306 PR tree-optimization/57796
18307 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18308 as N scalar stores.
18309 (vect_model_load_cost): Cost gathers as N scalar loads.
18311 2017-04-20 Richard Biener <rguenther@suse.de>
18313 * ggc-page.c (ggc_allocated_p): Rename to ...
18314 (safe_lookup_page_table_entry): ... this and return the lookup
18316 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18318 2017-04-20 Richard Biener <rguenther@suse.de>
18320 PR tree-optimization/80453
18321 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18322 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18323 from the conditions.
18324 (vn_phi_eq): Pass them down.
18325 (vn_phi_lookup): Record them.
18326 (vn_phi_insert): Likewise.
18328 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18330 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18331 uninitialized variable warning to avoid buffer overrun.
18333 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18336 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18337 is suppressed for '{ 0 }' in C.
18339 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18341 * BASE-VER: Set to 8.0.0.
18343 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18345 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18346 priority .init_array and .fini_array section with SECTION_NOTYPE
18349 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18351 PR middle-end/80423
18352 * tree.h (build_array_type): Add typeless_storage default argument.
18353 * tree.c (type_cache_hasher::equal): Also compare
18354 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18355 (build_array_type): Add typeless_storage argument, set
18356 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18358 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18359 (build_array_type): Likewise. Add typeless_storage argument.
18361 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18362 Jakub Jelinek <jakub@redhat.com>
18364 PR tree-optimization/80426
18365 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18366 operation on symbolic operands, also compute the overflow for the
18367 invariant part when the operation degenerates into a negation.
18369 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18372 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18373 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18376 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18378 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18381 * config/avr/avr.c (tree.h): Include it.
18382 (cgraph.h): Include it.
18383 (avr_encode_section_info): Don't warn for uninitialized progmem
18384 variable if it's just an alias.
18386 2017-04-19 Richard Biener <rguenther@suse.de>
18389 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18390 when needed by AutoPGO.
18392 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18395 * doc/lto.texi: Remove an extra 'that'.
18397 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18399 PR rtl-optimization/80429
18400 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18401 are only used in debug insns.
18403 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18404 Vladimir Makarov <vmakarov@redhat.com>
18406 * config/sparc/predicates.md (input_operand): Add comment. Return
18407 true for any memory operand when LRA is in progress.
18408 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18410 2017-04-18 Jeff Law <law@redhat.com>
18413 * mips.md ({return,simple_return}_internal): Do not overwrite
18416 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18418 PR tree-optimization/80443
18419 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18420 instead of adding 1, subtract -1 and similarly instead of subtracting
18423 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18425 PR rtl-optimization/80357
18426 * haifa-sched.c (tmp_bitmap): New variable.
18427 (model_recompute): Handle duplicate use records.
18428 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18429 (free_global_sched_pressure_data): Free it.
18431 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18434 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18435 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18436 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18437 instead of SYSTEM_HEADER_DIR.
18439 2017-04-18 Jeff Law <law@redhat.com>
18441 PR middle-end/80422
18442 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18443 predecessors after walking up the insn chain.
18445 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18448 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18449 sizetype type into debug info.
18451 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18454 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18455 unneeded test for TARGET_UPPER_REGS_SF.
18456 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18458 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18461 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18462 instead of gsi_after_labels.
18464 2017-04-18 Jeff Law <law@redhat.com>
18466 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18470 2017-04-13 Jeff Law <law@redhat.com>
18471 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18472 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18474 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18477 * config/avr/avr.c (intl.h): Include it.
18478 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18480 2017-04-18 Martin Liska <mliska@suse.cz>
18482 PR gcov-profile/78783
18483 * gcov-tool.c (gcov_output_files): Validate that destination
18484 file is either removed by the tool or by a user.
18486 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18487 Guy Benyei <guybe@mellanox.com>
18489 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18490 block, and do not negate it, the stored id is already negative.
18492 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18494 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18496 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18499 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18500 masks of options that should be turned off if the VSX vector
18501 options are turned off.
18502 (OTHER_P8_VECTOR_MASKS): Likewise.
18503 (OTHER_VSX_VECTOR_MASKS): Likewise.
18504 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18505 rs6000_disable_incompatible_switches to validate no type switches
18507 (rs6000_incompatible_switch): New function to disallow turning on
18508 other vector options if -mno-vsx, -mno-power8-vector, or
18509 -mno-power9-vector are specified.
18511 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18513 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18515 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18517 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18518 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18519 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18520 (ARG_POINTER_CFA_OFFSET): Likewise.
18522 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18524 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18525 conditions to take advantage of various optimizations.
18527 2017-04-13 Jeff Law <law@redhat.com>
18529 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18530 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18531 (zero_extendsidi2_dext): Likewise.
18533 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18536 * fold-const.c (fold_ternary_loc): Revert
18537 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18540 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18542 PR rtl-optimization/80343
18543 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18544 new scratch pseudo.
18546 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18549 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18550 to ubsan_encode_value.
18552 2017-04-13 Jeff Law <law@redhat.com>
18554 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18555 appearing in DEBUG_INSNs.
18557 2017-04-13 Martin Liska <mliska@suse.cz>
18559 PR gcov-profile/80413
18560 * gcov-io.c (gcov_write_string): Copy to buffer just when
18561 allocated size is greater than zero.
18563 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18566 * dwarf2out.c (decls_for_scope): Ignore declarations of
18567 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18569 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18572 * ipa-visibility.c (non_local_p): Fix typos.
18573 (localize_node): When localizing symbol in same comdat group,
18574 dissolve the group only when we know external symbols are going
18576 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18578 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18580 PR tree-optimization/79390
18581 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18582 order does not result in usable sequence, retry with reversed operand
18588 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18589 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18590 op0 instead of fold_convert_loc (loc, type, arg0).
18592 2017-04-12 Jeff Law <law@redhat.com>
18594 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18595 has a delay slot in the generated code.
18597 * config/cris/cris.md (cris_preferred_reload_class): Return
18598 GENNONACR_REGS rather than GENERAL_REGS.
18600 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18603 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18604 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18605 signedness of the result type.
18607 2017-04-12 Richard Biener <rguenther@suse.de>
18608 Jeff Law <law@redhat.com>
18610 PR tree-optimization/80359
18611 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18612 trim stores to TARGET_MEM_REFs.
18614 2017-04-12 Richard Biener <rguenther@suse.de>
18616 PR tree-optimization/79390
18617 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18618 threading case even more.
18620 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18623 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18624 for quad_address_p for TImode, instead of just not indexed_address.
18626 2017-04-12 Richard Biener <rguenther@suse.de>
18627 Bernd Edlinger <bernd.edlinger@hotmail.de>
18629 PR middle-end/79671
18630 * alias.c (component_uses_parent_alias_set_from): Handle
18631 TYPE_TYPELESS_STORAGE.
18632 (get_alias_set): Likewise.
18633 * tree-core.h (tree_type_common): Add typeless_storage flag.
18634 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18635 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18636 for types containing members with TYPE_TYPELESS_STORAGE.
18637 (place_field): Likewise.
18638 (layout_type): Likewise for ARRAY_TYPE.
18639 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18640 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18641 TYPE_TYPELESS_STORAGE.
18642 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18644 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18647 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18648 first argument to type.
18650 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18654 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18655 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18656 (rs6000_expand_binop_builtin): Likewise.
18657 (rs6000_expand_ternop_builtin): Likewise; also add missing
18658 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18659 vshasigma built-ins.
18660 * doc/extend.texi: Document that vec_xxpermdi's third argument
18661 must be a constant.
18663 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18665 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18666 Use shift_const cost parameter when calculating gain of STV shifts.
18668 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18670 PR rtl-optimization/70478
18671 * lra-constraints.c (process_alt_operands): Check memory for
18672 disfavoring memory insn operand.
18674 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18676 PR middle-end/80100
18677 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18678 left shift in unsigned HOST_WIDE_INT type.
18680 PR rtl-optimization/80385
18681 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18682 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18685 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18686 if they have any depend clauses.
18688 2017-04-11 Martin Liska <mliska@suse.cz>
18691 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18692 * ipa-split.c (split_function): Create a local comdat symbol
18693 if caller is in a comdat group.
18695 2017-04-11 Martin Liska <mliska@suse.cz>
18698 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18701 2017-04-11 Martin Sebor <msebor@redhat.com>
18703 PR middle-end/80364
18704 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18705 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18707 (directive::set_width, directive::set_precision, format_character):
18709 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18712 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18715 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18716 conflict, set target->arch_name instead of target->cpu_name.
18718 2017-04-11 Richard Biener <rguenther@suse.de>
18720 PR tree-optimization/80374
18721 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18722 build_zero_cst, remove fold_convertible_p check again.
18724 2017-04-11 Martin Liska <mliska@suse.cz>
18727 * ubsan.c (instrument_object_size): Do not instrument register
18730 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18733 * config/i386/i386-builtin-types.def
18734 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18735 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18736 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18737 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18738 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18739 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18740 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18741 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18742 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18743 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18744 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18745 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18746 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18747 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18748 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18749 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18750 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18751 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18752 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18753 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18754 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18755 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18756 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18757 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18758 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18759 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18760 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18761 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18762 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18763 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18764 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18765 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18766 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18767 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18768 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18769 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18770 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18771 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18772 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18773 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18774 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18775 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18776 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18777 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18778 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18779 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18781 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18782 flag to second_arg_count, handle 4 argument function type _COUNT
18783 aliases, handle second_arg_count on second argument rather than last.
18785 2017-04-10 Jeff Law <law@redhat.com>
18787 PR tree-optimization/80374
18788 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18789 record anything if we can not convert integer_zero_node to the
18792 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18795 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18796 Enhance special handling given to the TARGET_P9_MINMAX option in
18797 relation to certain other options.
18799 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18801 PR tree-optimization/80153
18802 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18803 remove POINTER_PLUS_EXPR's base part directly, rather than through
18806 2017-04-10 Richard Biener <rguenther@suse.de>
18807 Bin Cheng <bin.cheng@arm.com>
18809 PR tree-optimization/80153
18810 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18811 the first element of pointer type aff_tree. Build result expr in
18813 (add_elt_to_tree): Convert to type unconditionally. Remove other
18814 fold_convert calls.
18815 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18816 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18818 2017-04-10 Richard Biener <rguenther@suse.de>
18820 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18823 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18825 PR rtl-optimization/70478
18826 * lra-constraints.c (curr_small_class_check): New.
18827 (update_and_check_small_class_inputs): New.
18828 (process_alt_operands): Update curr_small_class_check. Disfavor
18829 alternative insn memory operands. Check available regs for small
18832 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18835 * config/mips/mips.opt (-mvirt): Update description.
18836 * doc/invoke.texi (-mvirt): Likewise.
18838 2017-04-10 Richard Biener <rguenther@suse.de>
18840 PR middle-end/80362
18841 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18842 looking for NEGATE_EXPR in -A / -B to A / B folding.
18844 2017-04-10 Martin Liska <mliska@suse.cz>
18846 PR gcov-profile/80224
18847 * gcov.c (print_usage): Fix usage string.
18848 (get_gcov_intermediate_filename): Remove.
18849 (output_gcov_file): Use both for normal and intermediate format.
18850 (generate_results): Do not initialize special file for
18851 intermediate format.
18853 2017-04-10 Richard Biener <rguenther@suse.de>
18855 PR tree-optimization/80304
18856 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18859 2017-04-10 Nathan Sidwell <nathan@acm.org>
18862 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18863 (rs6000_init_builtins): Use it.
18865 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18867 * config/arm/arm.md (<mrc>): Add mode to SET source.
18868 (<mrrc>): Likewise.
18870 2017-04-10 Richard Biener <rguenther@suse.de>
18872 PR middle-end/80344
18873 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18875 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18878 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18879 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18880 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18881 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18882 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18883 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18884 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18885 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18886 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18887 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18888 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18889 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18890 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18891 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18892 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18893 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18894 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18895 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18896 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18897 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18898 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18899 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18900 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18902 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18904 PR rtl-optimization/70478
18905 * lra-constraints.c: Reverse the last patch.
18907 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18909 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18910 Add comment for WCHAR_T.
18912 2017-04-08 Martin Liska <mliska@suse.cz>
18915 2017-04-07 Martin Liska <mliska@suse.cz>
18918 * ipa-split.c (split_function): Add function part to a same comdat
18921 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18924 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18926 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18928 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18929 (vec_pairing): ...to this.
18930 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18931 (rs6000_sched_init): Adjust for name change.
18932 (struct rs6000_sched_context): Likewise.
18933 (rs6000_init_sched_context): Likewise.
18934 (rs6000_set_sched_context): Likewise.
18936 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18942 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18944 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18945 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18946 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18948 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18950 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18952 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18954 PR rtl-optimization/70703
18955 * ira-color.c (update_conflict_hard_regno_costs): Use
18956 int64_t instead of HOST_WIDE_INT.
18958 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18960 PR rtl-optimization/70478
18961 * lra-constraints.c (process_alt_operands): Disfavor alternative
18962 insn memory operands.
18964 2017-04-07 Jeff Law <law@redhat.com>
18966 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18967 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18969 2017-04-07 Martin Liska <mliska@suse.cz>
18972 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18973 Show error message instead of an ICE.
18975 2017-04-07 Martin Liska <mliska@suse.cz>
18978 * ipa-split.c (split_function): Add function part to a same comdat
18981 2017-04-07 Richard Biener <rguenther@suse.de>
18983 PR middle-end/80341
18984 * tree.c (get_unwidened): Also handle ! for_type case for
18986 * convert.c (do_narrow): Split out from ...
18987 (convert_to_integer_1): ... here. Do not pass final truncation
18988 type to get_unwidened for TRUNC_DIV_EXPR.
18990 2017-04-07 Richard Biener <rguenther@suse.de>
18992 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18994 (aff_combination_const): Adjust.
18995 (aff_combination_scale): Likewise.
18996 (aff_combination_add_elt): Likewise.
18997 (aff_combination_add_cst): Likewise.
18998 (aff_combination_convert): Likewise.
18999 (add_elt_to_tree): Likewise. Remove unused argument.
19000 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19002 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19004 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19006 * config/arm/arm.c (arm_default_short_enums): Use
19007 ARM_DEFAULT_SHORT_ENUMS.
19008 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19010 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19013 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19014 members with redundant out-of-class redeclaration.
19016 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19019 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19020 * config/i386/i386.md (*zero_extendsidi2):
19021 Add (?*x,*x) and (?*v,*v) alternatives.
19023 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19026 * config/i386/i386.c (ix86_expand_builtin)
19027 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19028 mode from insn data. Convert operands to insn operand mode.
19029 Copy operands that don't satisfy insn predicate to a register.
19031 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19033 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19036 2017-04-06 Richard Biener <rguenther@suse.de>
19038 PR tree-optimization/80334
19039 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19040 preserve alignment of accesses.
19042 2017-04-06 Richard Biener <rguenther@suse.de>
19044 PR tree-optimization/80262
19045 * tree-sra.c (build_ref_for_offset): Preserve address-space
19047 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19048 Drop useless address-space information on MEM_REF offsets.
19050 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19052 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19054 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19056 PR rtl-optimization/70703
19057 * ira-color.c (update_conflict_hard_regno_costs): Use
19058 HOST_WIDE_INT instead of long.
19060 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19063 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19064 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19066 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19067 for x86_64 target. Handle -m3dnowa option.
19069 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19071 PR rtl-optimization/70703
19072 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19073 (update_conflict_hard_regno_costs): Use long instead of unsigned
19074 arithmetic for cost calculation.
19076 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19077 Bernd Edlinger <bernd.edlinger@hotmail.de>
19080 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19083 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19086 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19087 ptr_mode with Pmode throughout.
19088 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19089 into probe_stack_range and use DImode.
19091 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19094 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19095 call_eh_return is true.
19097 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19099 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19100 Initialize last_match_fntype_index.
19102 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19105 * tree-nvr.c: Include internal-fn.h.
19106 (pass_return_slot::execute): Ignore internal calls without
19109 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19110 Richard Biener <rguenther@suse.de>
19113 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19114 captures used multiple times, except for the last use.
19115 * generic-match-head.c: Include gimplify.h.
19117 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19119 PR tree-optimization/79390
19120 * target.h (struct noce_if_info): Declare.
19121 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19122 * target.def (noce_conversion_profitable_p): New target hook.
19123 * ifcvt.h (struct noce_if_info): New type, moved from ...
19124 * ifcvt.c (struct noce_if_info): ... here.
19125 (noce_conversion_profitable_p): Renamed to ...
19126 (default_noce_conversion_profitable_p): ... this. No longer
19128 (noce_try_store_flag_constants, noce_try_addcc,
19129 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19130 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19131 instead of noce_conversion_profitable_p.
19132 * config/i386/i386.c: Include ifcvt.h.
19133 (ix86_option_override_internal): Don't override
19134 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19135 (ix86_noce_conversion_profitable_p): New function.
19136 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19137 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19138 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19139 * doc/tm.texi: Regenerated.
19141 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19143 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19146 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19149 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19150 instructions for small multiply cores.
19152 2017-04-04 Jeff Law <law@redhat.com>
19154 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19156 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19157 that are not set by the loop over the elements.
19159 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19162 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19163 int mode, convert_modes it to mode as unsigned, otherwise use
19164 lowpart_subreg to mode rather than SImode.
19165 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19166 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19167 Use DImode instead of SImode for the shift count operand.
19168 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19171 2017-04-04 Richard Biener <rguenther@suse.de>
19173 PR middle-end/80281
19174 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19175 arithmetic done for the negate or the plus. Simplify.
19176 (A - (-B) -> A + B): Likewise.
19177 * fold-const.c (split_tree): Make sure to not negate pointers.
19179 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19181 PR rtl-optimization/60818
19182 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19183 a compare of comparisons with the thing compared if this results
19184 in a different machine mode.
19186 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19188 * alias.c (base_alias_check): Fix typo in comment.
19189 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19190 * cgraphunit.c (symbol_table::compile): Likewise.
19191 * collect2.c (maybe_run_lto_and_relink): Likewise.
19192 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19193 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19194 * config/avr/avr.c (avr_map_op_t): Likewise.
19195 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19196 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19197 * config/epiphany/epiphany.md (movcc): Likewise.
19198 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19199 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19201 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19202 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19203 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19204 * config/sh/sh.c (sh_rtx_costs): Likewise.
19205 * fold-const.c (fold_truth_andor): Likewise.
19206 * genautomata.c (collapse_flag): Likewise.
19207 * gengtype.h (struct type::u::s): Likewise.
19208 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19209 * input.c (FORMAT_AMOUNT): Likewise.
19210 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19211 (known_aggs_to_agg_replacement_list): Likewise.
19212 * ipa-inline-analysis.c: Likewise.
19213 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19214 * ipa-polymorphic-call.c
19215 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19216 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19217 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19219 * modulo-sched.c (apply_reg_moves): Likewise.
19220 * omp-expand.c (build_omp_regions_1): Likewise.
19221 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19222 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19223 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19224 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19225 * value-prof.c: Likewise.
19226 * var-tracking.c (val_reset): Likewise.
19228 2017-04-03 Richard Biener <rguenther@suse.de>
19230 PR tree-optimization/80275
19231 * fold-const.c (split_address_to_core_and_offset): Handle
19234 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19236 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19237 descriptors is at least equal to that of functions.
19239 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19241 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19243 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19246 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19247 (mov<IMOD4:mode>): New expander.
19248 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19250 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19252 PR rtl-optimization/79405
19253 * fwprop.c (propagations_left): New variable.
19254 (forward_propagate_into): Decrement it.
19255 (fwprop_init): Initialize it.
19256 (fw_prop): If the variable has reached zero, stop propagating.
19257 (fwprop_addr): Ditto.
19259 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19262 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19263 a FUNCTION_DECL, pass it as decl instead of origin to
19266 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19268 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19271 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19274 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19275 TARGET_VSX_SMALL_INTEGER.
19277 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19279 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19280 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19282 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19284 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19285 extraction from odd-numbered MSA register.
19287 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19289 PR middle-end/80173
19290 * expmed.c (store_bit_field_1): Don't attempt to create
19291 a word subreg out of hard registers wider than word if they
19292 have HARD_REGNO_NREGS of 1 for their mode.
19294 PR middle-end/80163
19295 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19296 conversions to integer types wider than word and pointer.
19299 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19300 (rtx_equal_for_cselib_p): Pass 0 to it.
19301 * cselib.c (cselib_hasher::equal): Likewise.
19302 (rtx_equal_for_cselib_1): Add depth argument. If depth
19303 is 128, don't look up VALUE locs and punt. Increment
19304 depth in recursive calls when walking VALUE locs.
19306 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19308 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19309 (make_gcov_file_name): Use the canonical path name for generating
19311 (read_line): Fix handling of files with ascii null bytes.
19313 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19315 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19316 to initialise a vector register instead
19317 of using a const_int.
19319 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19321 PR translation/80189
19322 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19323 diagnostic messages.
19325 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19328 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19329 (dfp_diex_<mode>): Update mode of operand 1.
19330 * doc/extend.texi (dxex, dxexq): Document change to return type.
19331 (diex, diexq): Document change to argument type.
19333 2017-03-30 Martin Jambor <mjambor@suse.cz>
19336 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19337 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19338 it reflects the signature changes performed at the callee side.
19339 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19340 to cgraph_build_function_type_skip_args.
19341 (build_function_decl_skip_args): Adjust call to the above function.
19343 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19346 * config/i386/sse.md
19347 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19348 register as dest whenever it is a MEM not rtx_equal_p to the
19349 corresponding dup operand, and when forcing into reg move the
19350 reg into the memory afterwards.
19351 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19352 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19353 for the force_reg mode.
19354 (avx512vl_vextractf128<mode>): Use register as dest either
19355 always when a MEM, or when it is a MEM not rtx_equal_p to the
19356 corresponding dup operand, or even not when it is a CONST_VECTOR
19357 depending on the mode and lo vs. hi.
19358 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19360 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19361 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19362 Likewise. Require that operands[2] is even.
19363 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19364 Remove extraneous parens. Require that operands[2] is a multiple
19366 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19367 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19368 disallow memory then.
19370 2017-03-30 Richard Biener <rguenther@suse.de>
19372 PR tree-optimization/77498
19373 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19374 to non-constants over backedges.
19376 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19378 PR rtl-optimization/80233
19379 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19380 as last_combined_insn. Do not test for BARRIER_P separately.
19382 2017-03-29 Andreas Schwab <schwab@suse.de>
19385 * calls.c (prepare_call_address): Convert funexp to Pmode before
19386 copying to temp reg.
19388 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19390 PR tree-optimization/80158
19391 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19392 Handle possible future case of more than one alternate
19394 (replace_rhs_if_not_dup): Likewise.
19395 (replace_one_candidate): Likewise.
19397 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19399 PR rtl-optimization/80193
19400 * ira.c (ira): Do not check allocation for LRA.
19402 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19404 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19405 (nvptx_output_simt_exit): Declare.
19406 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19407 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19408 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19409 (nvptx_declare_function_name): ...here. Emit declaration of local
19410 memory space buffer for omp_simt_enter insn.
19411 (nvptx_output_unisimt_switch): New.
19412 (nvptx_output_softstack_switch): New.
19413 (nvptx_output_simt_enter): New.
19414 (nvptx_output_simt_exit): New.
19415 * config/nvptx/nvptx.h (struct machine_function): New fields
19416 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19417 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19418 (UNSPECV_SIMT_EXIT): Ditto.
19419 (omp_simt_enter_insn): New insn.
19420 (omp_simt_enter): New expansion.
19421 (omp_simt_exit): New insn.
19422 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19424 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19425 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19426 (expand_GOMP_SIMT_EXIT): New.
19427 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19428 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19429 (GOMP_SIMT_EXIT): Ditto.
19430 * target-insns.def (omp_simt_enter): New insn.
19431 (omp_simt_exit): Ditto.
19432 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19434 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19435 (lower_rec_input_clauses): Likewise.
19436 (lower_lastprivate_clauses): Handle SIMT privatization.
19438 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19439 (ompdevlow_adjust_simt_enter): New.
19440 (find_simtpriv_var_op): New.
19441 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19442 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19444 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19445 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19446 (copy_decl_for_dup_finish): Ditto.
19448 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19450 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19453 * config/i386/i386.c (ix86_option_override_internal): Always
19454 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19456 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19458 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19460 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19462 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19463 mark new edge's irreducible flag accordign to it.
19464 (vect_do_peeling): Check loop preheader edge's irreducible flag
19465 and pass it to function slpeel_add_loop_guard.
19467 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19469 PR tree-optimization/80218
19470 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19471 Update block frequencies and counts.
19473 2017-03-28 Richard Biener <rguenther@suse.de>
19475 PR tree-optimization/78644
19476 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19477 of a simplification result we may not use it at all.
19479 2017-03-28 Richard Biener <rguenther@suse.de>
19482 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19483 without arguments, generate default definition of a SSA name.
19485 2017-03-28 Richard Biener <rguenther@suse.de>
19487 PR middle-end/80222
19488 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19489 TYPE_REF_CAN_ALIAS_ALL references.
19490 * fold-const.c (fold_indirect_ref_1): Likewise.
19492 2017-03-28 Martin Liska <mliska@suse.cz>
19495 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19496 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19498 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19499 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19501 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19502 (EXTRA_SPECS): Define.
19503 (SUBTARGET_EXTRA_SPECS): Likewise.
19504 (SUBTARGET_CPP_SPEC): Likewise.
19505 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19506 SUBTARGET_EXTRA_SPECS.
19507 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19509 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19511 * config/arc/simdext.md (vst64_insn): Update pattern.
19512 (vld32wh_insn): Likewise.
19513 (vld32wl_insn): Likewise.
19514 (vld64_insn): Likewise.
19515 (vld32_insn): Likewise.
19517 2017-03-28 Marek Polacek <polacek@redhat.com>
19520 * fold-const.c (fold_comparison): Use protected_set_expr_location
19521 instead of SET_EXPR_LOCATION.
19523 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19525 * tree.c (add_expr): Avoid name lookup warning.
19527 2017-03-27 Jeff Law <law@redhat.com>
19529 PR tree-optimization/80216
19530 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19531 function name. Limit recursion depth.
19532 (record_temporary_equivalences): Corresponding changes.
19534 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19536 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19539 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19542 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19544 * cfgcleanup.c (reg_note_cfa_p): New array.
19545 (insns_have_identical_cfa_notes): New function.
19546 (old_insns_match_p): Don't cross-jump in between /f
19547 and non-/f instructions. If both i1 and i2 are frame related,
19548 verify all CFA notes, their order and content.
19550 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19553 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19554 HImode and SImode with zero extend to DImode to one insn.
19555 (bswap<mode>2_extenddi): Likewise.
19556 (bswapsi2_extenddi): Likewise.
19557 (bswaphi2_extendsi): Likewise.
19558 (bswaphi2): Combine bswap HImode and SImode into one insn.
19559 Separate memory insns from swapping register.
19560 (bswapsi2): Likewise.
19561 (bswap<mode>2): Likewise.
19562 (bswaphi2_internal): Delete, no longer used.
19563 (bswapsi2_internal): Likewise.
19564 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19565 store, and gpr<-gpr swap insns.
19566 (bswap<mode>2_store): Likewise.
19567 (bswaphi2_reg): Register only splitter, combine with the splitter.
19568 (bswaphi2 splitter): Likewise.
19569 (bswapsi2_reg): Likewise.
19570 (bswapsi2 splitter): Likewise.
19571 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19572 the insns into load, store, and register/register insns.
19573 (bswapdi2_ldbrx): Likewise.
19574 (bswapdi2_load): Likewise.
19575 (bswapdi2_store): Likewise.
19576 (bswapdi2_reg): Likewise.
19578 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19580 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19581 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19583 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19586 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19588 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19589 special handling for target option conflicts between dform
19590 options (-mpower9-dform, -mpower9-dform-vector,
19591 -mpower9-dform-scalar) and -mno-direct-move.
19593 2017-03-27 Richard Biener <rguenther@suse.de>
19595 PR tree-optimization/80181
19596 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19598 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19600 * config/arc/predicates.md (move_double_src_operand): Replace the
19601 call to move_double_src_operand with a call to address_operand.
19603 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19605 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19606 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19607 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19609 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19611 * config/arc/predicates.md (long_immediate_loadstore_operand):
19612 Consider scaled addresses cases.
19614 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19616 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19617 restored when in interrupt.
19618 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19619 doesn't have delay slot.
19621 2017-03-27 Richard Biener <rguenther@suse.de>
19624 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19625 inlined thunk clones.
19627 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19630 * asan.c (instrument_derefs): Copy over last operand from
19631 original COMPONENT_REF to the new COMPONENT_REF with
19632 DECL_BIT_FIELD_REPRESENTATIVE.
19633 * ubsan.c (instrument_object_size): Likewise.
19635 2017-03-27 Richard Biener <rguenther@suse.de>
19637 PR tree-optimization/80170
19638 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19639 sure DR/SCEV didnt fold in constants we do not see when looking
19640 at the reference base alignment.
19642 2017-03-27 Richard Biener <rguenther@suse.de>
19644 PR middle-end/80171
19645 * gimple-fold.c (fold_ctor_reference): Properly guard against
19646 NULL return value from canonicalize_constructor_val.
19648 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19651 * config/i386/i386.c (ix86_expand_builtin)
19652 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19653 flags reg setting and flags reg using instructions.
19654 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19655 clobbering instructions to zero extend op2.
19657 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19659 * doc/install.texi (Configuration) <--with-aix-soname>:
19660 Update link to AIX ld.
19662 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19664 PR rtl-optimization/80160
19665 PR rtl-optimization/80159
19666 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19667 reg_alternate_class into account.
19669 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19672 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19673 to consider in curr_insn_transform.
19675 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19677 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19678 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19679 and emit_mode_inner.
19681 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19683 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19684 argument to the overloaded builtin variants. Use the new flag to
19685 deprecate certain builtin variants.
19686 * config/s390/s390-builtin-types.def: Add new builtin types.
19687 * config/s390/s390-builtins.h: Support new flags field for
19688 overloaded builtins.
19689 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19690 (s390_macro_to_expand): Enable vector float data type.
19691 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19692 builtins by incrementing the __VEC__ version number.
19693 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19695 (s390_resolve_overloaded_builtin): Emit error messages depending
19696 on the builtin flags.
19697 * config/s390/s390.c (s390_expand_builtin): Support additional
19698 flags argument. Change error message to match the messages
19699 emitted in s390-c.c.
19700 * config/s390/s390.md: New UNSPEC_* constants.
19701 (op_type): Add new instruction types.
19702 * config/s390/vecintrin.h: Add new builtins and test data class
19704 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19705 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19706 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19707 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19708 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19709 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19711 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19712 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19713 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19714 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19716 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19717 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19718 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19719 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19721 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19722 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19723 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19724 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19725 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19726 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19727 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19729 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19730 ("vec_scatter_element<V_HW_4:mode>_DI")
19731 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19732 ("vec_fpint<mode>", "vflls")
19733 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19734 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19735 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19736 ("*vec_cmphe<mode>_cc"): ... these.
19738 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19739 mode constant instead of magic value.
19741 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19743 * config/s390/s390.c (s390_expand_vec_compare): Support other
19744 vector floating point modes than just V2DF.
19745 (s390_expand_vcond): Likewise.
19746 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19747 (s390_cannot_change_mode_class): Prevent mode changes between TF
19748 and V1TF in vector registers.
19749 * config/s390/s390.md (DF, SF): New mode attributes.
19750 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19751 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19752 SFmode support for VRs.
19753 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19755 (VFT, VF_HW): New mode iterators.
19756 (vw, sdx): New mode attributes.
19757 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19758 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19759 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19760 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19761 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19762 also the new vector floating point modes. Renaming to ...
19764 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19765 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19766 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19767 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19768 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19769 ("vec_unordered<mode>"): ... these.
19771 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19772 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19773 ("*vec_extendv2df"): New insn definitions.
19775 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19777 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19778 ("mulditi3_2", "*muldi3_sign"): New patterns.
19779 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19780 rename the pattern definition.
19782 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19784 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19786 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19790 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19791 instruction if possible.
19792 * config/s390/vector.md (vec_halfnumelts): New mode
19794 ("*vec_vllezlf<mode>"): New pattern.
19796 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19798 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19799 ("popcountv4si2", "popcountv2di2"): Rename to ...
19800 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19801 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19803 ("popcount<mode>2_vxe"): New pattern.
19805 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19807 * common/config/s390/s390-common.c (processor_flags_table): Add
19809 * config.gcc: Add arch12.
19810 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19811 Default to arch12 for unknown CPU model numbers.
19812 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19813 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19814 PROCESSOR_max sanity check.
19815 * config/s390/s390-opts.h (enum processor_type): Add
19817 * config/s390/s390.c (processor_table): Add arch12.
19818 (s390_expand_builtin): Add check for B_VXE flag.
19819 (s390_issue_rate): Add PROCESSOR_ARCH12.
19820 (s390_get_sched_attrmask): Likewise.
19821 (s390_get_unit_mask): Likewise.
19822 (s390_sched_score): Enable z13 scheduling for arch12.
19823 (s390_sched_reorder): Likewise.
19824 (s390_sched_variable_issue): Likewise.
19825 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19827 (s390_tune_attr): Use z13 scheduling also for arch12.
19828 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19829 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19830 (TARGET_VXE_P): New macros.
19831 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19832 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19833 * config/s390/s390.opt: Add arch12 as processor_type.
19835 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19837 * config/s390/s390.md
19838 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19839 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19840 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19842 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19843 Rename expanders to ...
19845 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19846 ("fixuns_truncdddi2_emu"): ... these.
19848 ("fixuns_trunc<mode>si2_emu"): New expander.
19850 ("*fixuns_truncdfdi2_z13"): Rename to ...
19851 ("*fixuns_truncdfdi2_vx"): ... this.
19853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19855 * config/s390/2964.md: Remove the single element vector compare
19856 instructions which are no longer used.
19857 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19859 (s390_canonicalize_comparison): Remove handling of DFmode
19861 (s390_expand_vec_compare_scalar): Remove function.
19862 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19863 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19865 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19867 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19869 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19871 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19872 will anyway by matched by mov<mode>_64dfp.
19874 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19876 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19877 vlef/vstef. Add missing operand to vleif.
19879 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19881 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19882 pair for all vector types with 64 bit elements.
19883 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19884 * config/s390/vector.md (V_HW_64): ... here.
19885 (V_128_NOSINGLE): New mode iterator.
19886 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19887 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19888 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19889 ("*vec_load_pairv2di"): Change to ...
19890 ("*vec_load_pair<mode>"): ... this one.
19892 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19894 * config/s390/constraints.md: Add comments.
19895 (jKK): Reject element sizes > 8 bytes.
19896 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19898 * config/s390/s390.md: Add the s_operand checks formerly in
19899 s390_split_ok_p to various splitters where they are still
19901 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19902 for 128 bit vectors. Plus two splitters.
19904 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19906 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19909 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19912 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19913 error if the boundary argument is not constant.
19915 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19917 PR rtl-optimization/80112
19918 * loop-doloop.c (doloop_condition_get): Don't check condition
19919 if cmp isn't SET with IF_THEN_ELSE src.
19921 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19923 PR tree-optimization/80158
19924 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19925 replacing a candidate statement, also replace it for the
19926 candidate's alternate interpretation.
19927 (replace_rhs_if_not_dup): Likewise.
19928 (replace_one_candidate): Likewise.
19930 2017-03-24 Richard Biener <rguenther@suse.de>
19932 PR tree-optimization/80167
19933 * graphite-isl-ast-to-gimple.c
19934 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19936 (translate_isl_ast_to_gimple::get_rename): Likewise.
19938 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19940 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19941 handling of certain combinations of target options, including the
19942 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19943 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19945 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19948 * config/arm/arm.md (*load_multiple): Add reload_completed to
19949 matching condition.
19951 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19952 Richard Biener <rguenth@suse.de>
19954 PR tree-optimization/79908
19955 PR tree-optimization/80136
19956 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19957 been cast away, gimplify_and_add suffices.
19959 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19961 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19963 2017-03-23 Richard Biener <rguenther@suse.de>
19965 PR tree-optimization/80032
19966 * gimplify.c (gimple_push_cleanup): Forced unconditional
19967 cleanups still have to go to the conditional_cleanups
19970 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19972 PR tree-optimization/80072
19973 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19975 (next_operand_entry_id): Change type to unsigned int.
19976 (sort_by_operand_rank): Make sure to return the right return value
19977 even if unsigned fields are bigger than INT_MAX.
19978 (struct oecount): Change cnt and id type to unsigned int.
19979 (oecount_hasher::equal): Formatting fix.
19980 (oecount_cmp): Make sure to return the right return value
19981 even if unsigned fields are bigger than INT_MAX.
19982 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19985 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19986 TREE_READONLY on result if writing it more than once.
19989 * doc/invoke.texi (-fsanitize=thread): Document that with
19990 -fnon-call-exceptions atomics are not able to throw
19994 * tsan.c: Include tree-eh.h.
19995 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19996 maybe_clean_or_replace_eh_stmt where needed.
19997 (instrument_memory_accesses): Add cfg_changed argument.
19998 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19999 if it returned true.
20000 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20002 PR rtl-optimization/63191
20003 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20004 wrapper function, moved the whole old content into ...
20005 (ix86_delegitimize_address_1): ... this. New inline function.
20006 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20007 true as last argument instead of ix86_delegitimize_address.
20009 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20011 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20012 cortexa57_branch_cost.
20014 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20016 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20018 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20021 * doc/md.texi (Constraints): Document wA constraint.
20022 * config/rs6000/constraints.md (wA): New.
20023 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20024 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20025 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20026 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20028 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20031 * gimplify.c (is_oacc_declared): New function.
20032 (oacc_default_clause): Use it to set default flags for acc declared
20033 variables inside parallel regions.
20034 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20035 declared variables.
20036 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20037 declare attribute to any decl as necessary.
20039 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20042 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20043 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20044 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20045 (arm_arch_lpae): This.
20046 * config/arm/arm.c (arm_arch7ve): Rename into ...
20047 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20048 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20051 2017-03-22 Martin Liska <mliska@suse.cz>
20054 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20055 error message instead of an ICE.
20057 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20059 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20061 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20063 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20068 2017-03-21 Martin Sebor <msebor@redhat.com>
20070 * doc/extend.texi: Use "cannot" instead of "can't."
20071 * doc/hostconfig.texi: Same.
20072 * doc/install.texi: Same.
20073 * doc/invoke.texi: Same.
20074 * doc/loop.texi: Same.
20075 * doc/md.texi: Same.
20076 * doc/objc.texi: Same.
20077 * doc/rtl.texi: Same.
20078 * doc/tm.texi: Same.
20079 * doc/tm.texi.in: Same.
20080 * doc/trouble.texi: Same.
20082 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20085 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20086 (collect_checksum_attributes): Set it.
20087 (die_checksum_ordered): Use it.
20089 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20091 PR tree-optimization/79908
20092 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20093 change: For a VA_ARG whose LHS has been cast away, use
20094 force_gimple_operand to construct the side effects.
20096 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20098 PR translation/80001
20099 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20100 more amenable to translation.
20101 (oacc_loop_auto_partitions): Likewise.
20103 2017-03-21 Marek Polacek <polacek@redhat.com>
20104 Martin Sebor <msebor@redhat.com>
20106 PR tree-optimization/80109
20107 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20108 on INTEGRAL_TYPE_P.
20110 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20111 Segher Boessenkool <segher@kernel.crashing.org>
20114 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20115 check reg_used_between_p between insn and one of succ or succ2
20116 depending on if succ is artificial insn not inserted into insn
20119 2017-03-21 Martin Liska <mliska@suse.cz>
20121 PR gcov-profile/80081
20122 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20123 * doc/gcc.texi: Include gcov-dump stuff.
20124 * doc/gcov-dump.texi: New file.
20126 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20128 PR rtl-optimization/79150
20129 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20130 conditional jump, if the jump is the last insn of the loop.
20132 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20133 Richard Biener <rguenth@suse.de>
20135 PR tree-optimization/79908
20136 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20137 been cast away, use force_gimple_operand to construct the side
20140 2017-03-21 Martin Liska <mliska@suse.cz>
20142 PR libfortran/79956
20143 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20146 2017-03-21 Brad Spengler <spender@grsecurity.net>
20149 * plugin.c (htab_hash_plugin): New function.
20150 (add_new_plugin): Use it and adjust.
20151 (parse_plugin_arg_opt): Adjust.
20152 (init_one_plugin): Likewise.
20154 2017-03-21 Richard Biener <rguenther@suse.de>
20156 PR tree-optimization/80032
20157 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20158 if set force the cleanup to happen unconditionally.
20159 (gimplify_target_expr): Push inserted clobbers with force_uncond
20160 to avoid them being removed by control-dependent DCE.
20162 2017-03-21 Richard Biener <rguenther@suse.de>
20164 PR tree-optimization/80122
20165 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20166 va_arg_pack_len when the inlined call stmt requires pack
20168 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20170 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20173 * tsan.c (instrument_builtin_call): If the memory model argument
20174 is not a constant, assume it is valid.
20177 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20180 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20182 PR rtl-optimization/79910
20183 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20184 if its dest is used by an insn before I2 (other than the combined
20185 insns themselves, which are properly handled already).
20187 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20190 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20192 * combine.c (record_used_regs): New static function.
20193 (try_combine): Handle situations where there is an additional
20194 instruction between I2 and I3 which needs to have a LOG_LINK
20198 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20200 * combine.c (try_combine): Delete redundant i1 test. Call
20201 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20203 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20206 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20207 alternatives 13/14.
20209 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20211 PR tree-optimization/80054
20212 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20213 the optimization if a PHI or any of its arguments is not dominated
20214 by the candidate's basis. Use gphi* rather than gimple* as
20216 (replace_profitable_candidates): Clean up a gimple* variable that
20217 should be a gphi* variable.
20219 2017-03-20 Martin Sebor <msebor@redhat.com>
20222 * doc/extend.texi (attribute constructor): Document present limitation.
20224 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20227 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20228 __POWER9_VECTOR__ #ifdef control, change template definition to
20229 use Power9-specific built-in function.
20230 (vec_any_eq): Likewise.
20231 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20232 to control outcomes from this test.
20233 (vector_ae_<mode>p): For VEC_F modes, likewise.
20235 2017-03-20 Ian Lance Taylor <iant@google.com>
20237 * config/i386/i386.c (ix86_function_regparm): Save an extra
20238 register for -fsplit-stack with DECL_STATIC_CHAIN.
20240 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20243 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20244 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20246 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20248 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20250 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20253 2017-03-20 Marek Polacek <polacek@redhat.com>
20256 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20258 2017-03-20 Richard Biener <rguenther@suse.de>
20260 PR tree-optimization/80113
20261 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20262 allocate extra SSA name for PHI def.
20263 (add_close_phis_to_outer_loops): Likewise.
20264 (add_close_phis_to_merge_points): Likewise.
20265 (copy_loop_close_phi_args): Likewise.
20266 (copy_cond_phi_nodes): Likewise.
20268 2017-03-20 Martin Liska <mliska@suse.cz>
20270 PR middle-end/79753
20271 * tree-chkp.c (chkp_build_returned_bound): Do not build
20272 returned bounds for a LHS that's not a BOUNDED_P type.
20274 2017-03-20 Martin Liska <mliska@suse.cz>
20278 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20279 COMPLEX_CST and VECTOR_CST.
20281 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20284 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20285 target operand. A new splitter adds the clobber statement in case
20286 the target operand is dead anyway.
20288 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20290 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20291 to age-old versions of binutils and glibc.
20293 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20295 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20297 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20299 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20301 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20303 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20304 requirement for binutils 2.13.
20306 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20308 * combine.c (try_combine): Delete redundant i1 test. Call
20309 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20311 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20313 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20314 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20316 <riscv64-*-elf>: Re-arrange section
20317 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20318 <riscv32-*-linux>: Likewise.
20319 <riscv64-*-elf>: Likewise
20320 <riscv64-*-linux>: Likewise.
20322 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20325 * aarch64.opt(verbose-cost-dump): Fix typo.
20327 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20330 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20331 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20333 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20335 * reload.c (find_reloads): When reloading a nonoffsettable address,
20336 use RELOAD_OTHER for it and its address reloads.
20338 PR rtl-optimization/79910
20339 * combine.c (record_used_regs): New static function.
20340 (try_combine): Handle situations where there is an additional
20341 instruction between I2 and I3 which needs to have a LOG_LINK
20344 2017-03-17 Jeff Law <law@redhat.com>
20346 PR tree-optimization/71437
20347 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20348 conditional in the hash table first.
20349 (vrp_dom_walker::before_dom_children): Extract condition from
20350 ASSERT_EXPR. Record condition, its inverion and any implied
20351 conditions as well.
20353 2017-03-17 Marek Polacek <polacek@redhat.com>
20354 Markus Trippelsdorf <markus@trippelsdorf.de>
20356 PR tree-optimization/80079
20357 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20360 2017-03-17 Richard Biener <rguenther@suse.de>
20362 PR middle-end/80075
20363 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20364 Properly verify the LHS before the RHS possibly claims to be
20366 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20369 2017-03-17 Martin Jambor <mjambor@suse.cz>
20371 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20372 (List of -O2 options): Likewise.
20373 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20376 2017-03-17 Tom de Vries <tom@codesourcery.com>
20378 * gcov-dump.c (print_usage): Print bug_report_url.
20380 2017-03-17 Richard Biener <rguenther@suse.de>
20382 PR middle-end/80050
20383 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20384 (parser::peek): Likewise.
20386 2017-03-17 Richard Biener <rguenther@suse.de>
20388 PR tree-optimization/80048
20389 * sese.c (free_sese_info): Properly release rename_map and
20390 copied_bb_map elements.
20392 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20394 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20395 Add linked-list forward and backlinks. Insert on
20396 construction, remove on destruction.
20397 (class pass_store_merging): Add m_stores_head field.
20398 (pass_store_merging::terminate_and_process_all_chains):
20399 Iterate over m_stores_head list.
20400 (pass_store_merging::terminate_all_aliasing_chains):
20402 (pass_store_merging::execute): Check for debug stmts first.
20403 Push new chains onto the m_stores_head stack.
20405 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20408 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20409 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20410 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20412 2017-03-16 Jeff Law <law@redhat.com>
20414 PR tree-optimization/71437
20415 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20416 member function. Implementation moved into after_dom_children
20417 member function and into the threader's thread_outgoing_edges
20419 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20420 some code into new thread_outgoing_edges.
20421 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20422 definition. Simplify marker handling (do it here). Assume we always
20423 have the available expression and the const/copies tables.
20424 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20426 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20427 * tree-vrp.c (equiv_stack): No longer file scoped.
20428 (vrp_dom_walker): New class.
20429 (vrp_dom_walker::before_dom_children): New member function.
20430 (vrp_dom_walker::after_dom_children): Likewise.
20431 (identify_jump_threads): Setup domwalker. Use it rather than
20432 walking edges in a random order by hand. Simplify setup/finalization.
20433 (finalize_jump_threads): Remove.
20434 (vrp_finalize): Do not call identify_jump_threads here.
20435 (execute_vrp): Do it here instead and call thread_through_all_blocks
20438 PR tree-optimization/71437
20439 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20441 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20443 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20444 (dom_opt_dom_walker::thread_across_edge): Remove
20445 handle_dominating_asserts argument. All callers changed.
20446 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20447 changes. Remove calls to lhs_of_dominating_assert. Other
20448 uses of handle_dominating_asserts turn into unconditional code
20449 (simplify_control_stmt_condition_1): Likewise.
20450 (simplify_control_stmt_condition): Likewise.
20451 (thread_through_normal_block, thread_across_edge): Likewise.
20452 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20453 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20454 object if it is not an SSA_NAME.
20455 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20456 before calling into the VRP specific simplifiers.
20457 (identify_jump_threads): Remove handle_dominating_asserts
20460 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20463 * tree-diagnostic.c (default_tree_printer): No longer static.
20464 * tree-diagnostic.h (default_tree_printer): New prototype.
20466 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20468 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20469 Change ins into fmov.
20471 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20473 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20474 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20475 Use h_con constraint for operand 1.
20476 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20477 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20479 2017-03-15 Jeff Law <law@redhat.com>
20481 PR tree-optimization/71437
20482 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20483 (record_temporary_equivalences): Use it.
20485 PR tree-optimization/71437
20486 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20487 tree-ssa-scopedtables.
20488 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20489 (record_conditions, record_cond, vuse_eq): Likewise.
20490 (record_edge_info): Adjust to API tweak of record_conditions.
20491 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20492 (record_temporary_equivalences, optimize_stmt): Likewise.
20493 (eliminate_redundant_computations): Likewise.
20494 (record_equivalences_from_stmt): Likewise.
20495 * tree-ssa-scopedtables.c: Include options.h and params.h.
20496 (vuse_eq): New function, moved from tree-ssa-dom.c
20497 (build_and_record_new_cond): Likewise.
20498 (record_conditions): Likewise. Accept vector of conditions rather
20499 than edge_equivalence structure for first argument.
20500 for the first argument.
20501 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20502 from tree-ssa-dom.c.
20503 (avail_exprs_stack::record_cond): Likewise.
20504 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20505 from tree-ssa-dom.c.
20506 (avail_exprs_stack): Add new member functions lookup_avail_expr
20508 (record_conditions): Declare.
20510 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20513 * lra-constraints.c (process_alt_operands): Increase reject for
20514 reloading an input/output operand.
20516 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20519 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20520 insns to convert from signed/unsigned char/short to IEEE 128-bit
20522 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20524 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20527 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20528 subreg of inner mode for values already in registers.
20530 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20532 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20533 iteration reg is used after the loop.
20535 2017-03-14 Martin Sebor <msebor@redhat.com>
20537 PR tree-optimization/79800
20538 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20539 precision in negative-positive range.
20540 (format_floating): Call non-const overload with adjusted precision.
20542 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20545 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20548 2017-03-14 Martin Sebor <msebor@redhat.com>
20550 PR middle-end/80020
20551 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20552 * builtins.def (aligned_alloc): Use it.
20555 * Makefile.in (GTFILES): Add calls.c.
20556 * calls.c: Include "gt-calls.h".
20558 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20560 PR rtl-optimization/79728
20561 * regs.h (struct target_regs): New field
20562 x_contains_allocatable_regs_of_mode.
20563 (contains_allocatable_regs_of_mode): New macro.
20564 * reginfo.c (init_reg_sets_1): Initialize it, and change
20565 contains_reg_of_mode so it includes global regs as well.
20566 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20567 rather than contains_regs_of_mode.
20569 2017-03-14 Martin Liska <mliska@suse.cz>
20571 * doc/invoke.texi: Document options that can't be combined with
20572 -fcheck-pointer-bounds.
20574 2017-03-14 Martin Liska <mliska@suse.cz>
20576 PR middle-end/79831
20577 * doc/invoke.texi (-Wchkp): Document the option.
20579 2017-03-14 Martin Liska <mliska@suse.cz>
20581 * Makefile.in: Install gcov-dump.
20583 2017-03-14 Martin Liska <mliska@suse.cz>
20585 * multiple_target.c (expand_target_clones): Bail out for
20586 an invalid attribute.
20588 2017-03-14 Richard Biener <rguenther@suse.de>
20590 * alias.c (struct alias_set_entry): Pack properly.
20591 * cfgloop.h (struct loop): Likewise.
20592 * cse.c (struct set): Likewise.
20593 * ipa-utils.c (struct searchc_env): Likewise.
20594 * loop-invariant.c (struct invariant): Likewise.
20595 * lra-remat.c (struct cand): Likewise.
20596 * recog.c (struct change_t): Likewise.
20597 * rtl.h (struct address_info): Likewise.
20598 * symbol-summary.h (function_summary): Likewise.
20599 * tree-loop-distribution.c (struct partition): Likewise.
20600 * tree-object-size.c (struct object_size_info): Likewise.
20601 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20602 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20603 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20604 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20605 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20606 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20607 (struct _stmt_vec_info): Likewise.
20609 2017-03-14 Martin Liska <mliska@suse.cz>
20612 * multiple_target.c (create_dispatcher_calls): Check that
20613 a target can create a function dispatcher.
20615 2017-03-14 Martin Liska <mliska@suse.cz>
20618 * multiple_target.c (expand_target_clones): Drop local.local
20619 flag for default implementation.
20621 2017-03-14 Richard Biener <rguenther@suse.de>
20623 PR tree-optimization/80030
20624 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20626 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20628 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20629 gcc_fallthrough() instead of __attribute__((fallthrough));
20631 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20633 * doc/gcc.texi: Remove "up" link to (DIR).
20634 * doc/gccint.texi: Ditto.
20636 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20638 * doc/install.texi (Specific) <avr>: Remove reference to
20641 2017-03-13 Jeff Law <law@redhat.com>
20643 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20644 attribute rather than comments.
20646 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20647 match_scratch operand is highest.
20649 2017-03-13 Martin Liska <mliska@suse.cz>
20651 PR middle-end/78339
20652 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20653 is a CHKP clone, use original declaration.
20655 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20657 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20658 (arc_conditional_register_usage): Use a different allocation order
20659 when optimizing for size.
20660 * common/config/arc/arc-common.c (arc_option_optimization_table):
20661 Section anchors default on when optimizing for size.
20663 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20665 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20667 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20669 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20670 * config/arc/arc.md (cpu_facility): Add cd variant.
20671 (*movqi_insn): Add code density variant.
20672 (*movhi_insn): Likewise.
20673 (*movqi_insn): Likewise.
20674 (*addsi3_mixed): Likewise.
20675 (subsi3_insn): Likewise.
20677 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20679 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20681 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20683 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20684 expressions with MINUS and UNARY ops.
20686 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20689 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20691 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20692 between vec_select and vector argument.
20693 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20694 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20695 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20696 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20697 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20698 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20700 2017-03-13 Richard Biener <rguenther@suse.de>
20703 * params.def (vect-max-peeling-for-alignment): Fix typo.
20705 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20707 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20708 issue that only occurred with binutils below 2.18.
20710 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20712 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20713 refer to binutils 2.11/2.12 minimum.
20715 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20717 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20718 ftp.kernel.org and simplify binutils requirement.
20720 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20722 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20724 (Optimize Options): Ditto. Also remove redundancy.
20726 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20728 PR translation/79848
20729 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20731 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20732 to G_ to avoid double translation.
20734 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20736 PR translation/79923
20737 * auto-profile.c (get_combined_location): Convert leading
20738 character of diagnostics to lower case and remove trailing period.
20739 (read_profile): Likewise for various diagnostics.
20740 * config/arm/arm.c (arm_option_override): Remove trailing period
20741 from various diagnostics.
20742 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20743 (msp430_expand_delay_cycles): Likewise.
20745 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20748 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20749 full command-line argument, rather than just "str".
20750 (aarch64_validate_march): Likewise.
20751 (aarch64_validate_mtune): Likewise.
20753 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20755 PR rtl-optimization/78911
20756 * lra-assigns.c (must_not_spill_p): New function.
20757 (spill_for): Use it.
20759 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20761 PR tree-optimization/79981
20762 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20763 ATOMIC_COMPARE_EXCHANGE ifn result.
20764 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20765 IFN_ATOMIC_COMPARE_EXCHANGE.
20767 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20770 * opts.c (parse_sanitizer_options): Add missing question mark to
20771 "did you mean" message.
20773 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20775 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20777 (VMULEUH_UNS): Likewise.
20778 (VMULOUB_UNS): Likewise.
20779 (VMULOUH_UNS): Likewise.
20780 * config/rs6000/rs6000.c (builtin_function_type): Remove
20781 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20783 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20786 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20787 x with result of extra_parsing_for_operand_code_0.
20788 (function_reader::extra_parsing_for_operand_code_0): Convert
20789 return type from void to rtx, returning x. When reading
20790 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20791 larger size containing struct block_symbol.
20793 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20795 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20796 -mfloat128-hardware without -m64.
20798 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20801 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20802 entries to the case statement that marks unsigned arguments to
20803 overloaded functions.
20805 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20808 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20810 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20813 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20814 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20816 2017-03-10 Martin Liska <mliska@suse.cz>
20820 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20821 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20824 2017-03-10 Olivier Hainque <hainque@adacore.com>
20826 * tree-switch-conversion (array_value_type): Start by resetting
20827 candidate type to it's main variant.
20829 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20831 PR rtl-optimization/79909
20832 * combine.c (try_combine): Use simplify_replace_rtx on individual
20833 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20834 of the whole CALL_INSN_FUNCTION_USAGE.
20836 PR tree-optimization/79972
20837 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20838 get_range_info on SSA_NAMEs. Formatting fixes.
20840 2017-03-10 Richard Biener <rguenther@suse.de>
20841 Jakub Jelinek <jakub@redhat.com>
20843 PR tree-optimization/77975
20844 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20845 edge to be constant.
20846 (get_val_for): For constant x return it. Formatting fix.
20847 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20848 would use the same bases as the current one.
20850 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20852 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20853 instead of vec_select for V1TImode.
20854 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20856 (VSX_LE_128): Add V1TI to this mode iterator.
20857 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20858 (*vsx_le_perm_store_<mode>): Likewise.
20859 (pre-reload splitter for VSX stores): Likewise.
20860 (post-reload splitter for VSX stores): Likewise.
20861 (*vsx_xxpermdi2_le_<mode>): Likewise.
20862 (*vsx_lxvd2x2_le_<mode>): Likewise.
20863 (*vsx_stxvd2x2_le_<mode>): Likewise.
20865 2017-03-09 Michael Eager <eager@eagercon.com>
20867 Correct failures with --enable-checking=yes,rtl.
20869 * config/microblaze/microblaze.c (microblaze_expand_shift):
20870 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20871 test for const0_rtx.
20872 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20873 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20875 2017-03-09 Richard Biener <rguenther@suse.de>
20877 PR tree-optimization/79977
20878 * graphite-scop-detection.c (scop_detection::merge_sese):
20879 Handle the case of extra exits to blocks dominating the entry.
20881 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20883 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20886 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20888 PR rtl-optimization/79949
20889 * lra-constraints.c (process_alt_operands): Check memory when
20890 trying to predict a cycle. Print about the overall increase.
20892 2017-03-09 Richard Biener <rguenther@suse.de>
20894 PR middle-end/79971
20895 * gimple-expr.c (useless_type_conversion_p): Preserve
20896 TYPE_SATURATING for fixed-point types.
20898 2017-03-09 Richard Biener <rguenther@suse.de>
20901 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20902 alignment of BLKmode params.
20904 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20907 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20908 (VALL_NO_V2Q): Likewise.
20910 * config/aarch64/aarch64-simd.md
20911 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20913 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20914 VALL_NO_V2Q mode iterator.
20915 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20917 2017-03-09 Martin Liska <mliska@suse.cz>
20919 PR tree-optimization/79631
20920 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20921 tree_int_cst_sign_bit just for INTEGER constants.
20923 2017-03-09 Martin Liska <mliska@suse.cz>
20927 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20930 2017-03-09 Marek Polacek <polacek@redhat.com>
20933 * tree.c (inchash::add_expr): Handle TREE_VEC.
20935 2017-03-09 Martin Liska <mliska@suse.cz>
20938 (chkp_narrow_size_and_offset): New function.
20939 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20940 (void chkp_parse_bit_field_ref): New function.
20941 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20942 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20944 2017-03-09 Martin Liska <mliska@suse.cz>
20947 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20948 (chkp_find_bounds_1): Remove gcc_unreachable.
20950 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20953 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20954 BUILT_IN_SYNC*, determine the access type from the size suffix and
20955 always build a MEM_REF with that type. Handle forgotten
20956 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20959 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20960 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20961 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20962 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20963 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20964 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20965 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20966 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20967 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20968 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20969 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20970 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20971 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20972 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20973 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20974 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20975 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20976 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20977 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20978 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20979 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20980 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20981 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20982 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20983 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20984 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20985 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20986 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20987 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20988 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20989 definitions outside of __OPTIMIZE__ guarded section.
20992 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20993 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20994 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20995 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20998 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21000 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21001 ("vfenez<mode>"): Add missing constraints.
21003 2017-03-08 Martin Sebor <msebor@redhat.com>
21006 * config/nds32/nds32.c (nds32_option_override):
21007 Fix misspelled diagnostic.
21009 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21012 * gimplify.c (gimplify_omp_for): Replace index var in outer
21013 taskloop statement with an artificial variable and add
21014 OMP_CLAUSE_PRIVATE clause for it.
21016 2017-03-08 Richard Biener <rguenther@suse.de>
21018 PR tree-optimization/79955
21019 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21020 for accesses that are completely outside of the variable.
21022 2017-03-08 Andrew Haley <aph@redhat.com>
21024 PR tree-optimization/79943
21025 * tree-ssa-loop-split.c (compute_new_first_bound): When
21026 calculating the new upper bound, (END-BEG) should be added, not
21029 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21031 * config/avr/avr.md (setmemhi): Make sure match_dup
21032 operand number comes before match_scratch.
21034 2017-03-08 Richard Biener <rguenther@suse.de>
21036 PR tree-optimization/79920
21037 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21038 with ncopies == 1 to ...
21039 (vect_transform_slp_perm_load): ... here. Properly compute
21040 all element loads by iterating VF times over the group. Do
21041 not handle ncopies (computed in a broken way) in
21042 vect_create_mask_and_perm.
21044 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21047 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21048 is a uniform vector, use uniform_vector_p return value instead of
21049 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21051 2017-03-07 Marek Polacek <polacek@redhat.com>
21053 PR middle-end/79809
21054 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21055 (alloca_call_type): Likewise.
21057 2017-03-07 Martin Liska <mliska@suse.cz>
21059 * gcov.c (process_args): Put comment to correct location.
21061 2017-03-07 Martin Liska <mliska@suse.cz>
21063 PR middle-end/68270
21064 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21065 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21066 (chkp_narrow_bounds_for_field): Likewise.
21067 (chkp_parse_array_and_component_ref): Pass one more argument to
21070 2017-03-07 Richard Biener <rguenther@suse.de>
21072 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21075 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21077 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21078 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21080 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21083 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21084 to end of description.
21085 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21087 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21089 PR rtl-optimization/79901
21090 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21092 (*avx512f_<code><mode>3<mask_name>): ... this.
21093 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21094 iterator instead of VI8_AVX2_AVX512BW.
21096 PR rtl-optimization/79901
21097 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21098 min/max expander, expand it using expand_vec_cond_expr.
21101 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21104 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21107 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21108 to void * for PCH reasons.
21109 * dwarf2out.c (output_loc_operands, output_die): Cast
21110 v.val_vec.array to unsigned char *.
21112 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21115 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21118 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21120 PR rtl-optimization/79571
21121 * lra-constraints.c (process_alt_operands): Calculate static
21122 reject and subtract it from overall when only addresses will be
21125 2017-03-06 Julia Koval <julia.koval@intel.com>
21128 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21129 incoming stack boundary to 128 for 64-bit targets.
21131 2017-03-06 Richard Biener <rguenther@suse.de>
21133 PR tree-optimization/79894
21134 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21135 to NULL after folding it.
21137 2017-03-06 Richard Biener <rguenther@suse.de>
21139 PR tree-optimization/79824
21140 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21141 check disabling peeling for gaps.
21143 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21145 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21146 attributes): Document gettimeofday.
21148 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21150 * config/s390/s390.c (s390_option_override_internal): Set
21151 PARAM_MIN_VECT_LOOP_BOUND
21153 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21155 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21156 * config/s390/s390.md: Likewise.
21158 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21161 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21162 (<avx2_avx512>_perm<mode>): Rename to ...
21163 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21165 (<avx512>_perm<mode>_mask): Rename to ...
21166 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21168 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21169 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21170 instead of VI8F_256_512.
21171 (avx512f_perm<mode>): New define_expand.
21172 (avx512f_perm<mode>_mask): Likewise.
21173 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21174 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21176 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21178 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21179 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21181 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21183 2017-03-06 Martin Liska <mliska@suse.cz>
21186 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21187 when having a SSA NAME w/o VAR_DECL assigned to it.
21189 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21191 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21192 msa_dpsub_<su>_d): Fix MODE for vec_select.
21194 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21196 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21198 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21200 2017-03-06 Richard Biener <rguenther@suse.de>
21202 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21203 * plugin.c (register_plugin_info): Likewise.
21204 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21206 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21208 * config/i386/sse.md (sse_storehps, sse_storelps,
21209 avx_<castmode><avxsizesuffix>_<castmode>,
21210 avx512f_<castmode><avxsizesuffix>_<castmode>,
21211 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21212 in condition that at least one operand is not a MEM.
21214 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21216 PR middle-end/79805
21217 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21218 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21220 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21221 gimple_call_nothrow_p flag based on whether original builtin can throw.
21222 If it can, emit following stmts on the fallthrough edge.
21223 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21224 don't create new bb if inserting just debug stmts on the edge, try to
21225 insert them on the fallthru bb or just reset debug stmts.
21227 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21230 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21231 restore recog_data (including the operand rtxes inside it) around
21232 the call to get_insn_template.
21234 2017-03-03 Martin Sebor <msebor@redhat.com>
21236 PR tree-optimization/79699
21237 * context.c (context::~context): Free MPFR caches to avoid
21238 a memory leak on program exit.
21240 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21242 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21243 Use wide_int::ulow () instead of .elt (0).
21245 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21247 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21248 (*pushxf): Limit oF constraint to 32bit targets and add oC
21249 constraint for 64bit targets.
21250 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21251 (*pushdf): Change rmF constraint to rmC.
21253 2017-03-03 Martin Liska <mliska@suse.cz>
21255 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21256 Remove unused variable.
21258 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21261 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21262 is a memory operand, increase num_memory.
21263 (ix86_expand_args_builtin): Likewise.
21265 2017-03-03 Jan Hubicka <jh@suse.cz>
21268 * ipa-devirt.c (maybe_record_node): Properly handle
21269 __cxa_pure_virtual visibility.
21271 2017-03-03 Martin Liska <mliska@suse.cz>
21273 PR tree-optimization/79803
21274 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21276 (pass_loop_prefetch::execute): Disabled optimization if an
21277 assumption about L1 cache size is not met.
21279 2017-03-03 Martin Liska <mliska@suse.cz>
21281 PR rtl-optimization/79574
21282 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21283 (hash_scan_set): Likewise.
21284 (dump_hash_table): Likewise.
21285 (hoist_code): Likewise.
21287 2017-03-03 Richard Biener <rguenther@suse.de>
21289 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21290 in place of uhigh ().
21291 (fixed_convert_from_real): Likewise.
21293 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21296 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21298 2017-03-03 Richard Biener <rguenther@suse.de>
21300 PR middle-end/79818
21301 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21302 TYPE_OVERFLOW_UNDEFINED check.
21304 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21306 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21308 (vector_ae_<mode>_p): Likewise.
21309 (vector_nez_<mode>_p): Likewise.
21310 (vector_ne_v2di_p): Likewise.
21311 (vector_ae_v2di_p): Likewise.
21312 (vector_ne_<mode>_p): Likewise.
21313 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21315 (vsx_tsqrt<mode>2_fe): Likewise.
21317 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21320 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21322 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21324 PR rtl-optimization/79780
21325 * cprop.c (one_cprop_pass): When second and further conditional trap
21326 in a single basic block is turned into an unconditional trap, turn it
21327 into a deleted note to avoid RTL verification failures.
21329 2017-03-02 Richard Biener <rguenther@suse.de>
21331 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21333 2017-03-02 Richard Biener <rguenther@suse.de>
21335 PR tree-optimization/79345
21337 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21338 param and abort the walk, returning -1 if it is hit.
21339 (walk_aliased_vdefs): Take a limit param and pass it on.
21340 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21341 defaulting to 0 and return a signed int.
21342 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21343 (check_defs): New helper.
21344 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21345 about uninitialized memory.
21346 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21347 bogus uninitialized warning.
21348 (fixed_convert_from_real): Likewise.
21350 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21352 PR tree-optimization/66768
21353 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21354 iv_use if base object can't be determined.
21356 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21358 PR tree-optimization/79345
21359 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21360 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21361 (get_pattern_stats): Initialize it.
21362 * genemit.c (gen_expand): Verify match_scratch numbers come after
21363 match_operand/match_dup numbers.
21364 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21365 match_scratch numbers.
21366 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21368 * config/s390/s390.md (trunctdsd2): Likewise.
21370 2017-03-02 Richard Biener <rguenther@suse.de>
21372 * wide-int.h (wide_int_storage::operator=): Implement in terms
21375 2017-03-02 Richard Biener <rguenther@suse.de>
21377 PR tree-optimization/79777
21378 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21379 the to insert expression to sth existing.
21381 2017-03-01 Martin Sebor <msebor@redhat.com>
21383 PR middle-end/79692
21384 * gimple-ssa-sprintf.c
21385 (directive::known_width_and_precision): New function.
21386 (format_integer): Use it.
21387 (get_mpfr_format_length): Consider the full range of precision
21388 when computing %g output with the # flag. Set the likely byte
21389 count to 3 rather than 1 when precision is indeterminate.
21390 (format_floating): Correct the lower bound of precision.
21392 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21394 * doc/invoke.texi: Document default code model for 64-bit Linux.
21396 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21399 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21400 udiv rather than div since input pattern is unsigned.
21402 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21404 * config/i386/i386.c (print_reg): Warn for values of
21405 unsupported size in integer register.
21407 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21410 * config/rs6000/predicates.md (current_file_function_operand): Do
21411 not allow self calls to be local if the function is replaceable.
21413 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21416 * config/rs6000/altivec.h (vec_ctz and others): Change the
21417 preprocessor macro that controls conditional compilation from
21418 _ARCH_PWR9 to __POWER9_VECTOR__.
21419 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21420 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21421 control (instead of _ARCH_PWR9 control) so that template
21422 definition uses power9-specific function.
21423 (vec_any_eq): Likewise.
21424 (vec_all_ne): Change macro definition to use a power9-specific
21425 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21426 _ARCH_PWR9 control).
21427 (vec_any_eq) Likewise.
21428 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21429 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21430 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21431 support for xvcmpnedp instruction.
21432 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21433 macro expansion so that Power9 implementation of vec_all_ne does
21434 not use the AltiVec predicate framework.
21435 (VCMPNEH_P): Likewise.
21436 (VCMPNEW_P): Likewise.
21437 (VCMPNED_P): Likewise.
21438 (VCMPNEFP_P): Likewise.
21439 (VCMPNEDP_P): Likewise.
21440 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21441 implementation of vec_any_eq to not use AltiVec predicate
21443 (VCMPAEH_P): Likewise.
21444 (VCMPAEW_P): Likewise.
21445 (VCMPAED_P): Likewise.
21446 (VCMPAEFP_P): Likewise.
21447 (VCMPAEDP_P): Likewise.
21448 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21449 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21450 not use the AltiVec predicate framework.
21451 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21452 of vec_any_eq to not use AltiVec predicate framework.
21453 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21454 support for predefined __POWER9_VECTOR__ macro to indicate that
21455 Power9 instruction selection is enabled.
21456 (altivec_overloaded_builtins): Remove extraneous
21457 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21458 function argument types RS6000_BTI_bool_V16QI and
21459 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21460 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21461 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21462 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21463 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21464 Power9 for implementations of vec_cmpne. Change the signature for
21465 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21466 (representing vec_all_ne) to remove the previously described first
21467 argument of type RS6000_BTI_INTSI, as this was an artifact of
21468 reliance on the AltiVec predicate framework, which is no longer
21469 used in the implementation of these functions. Add
21470 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21471 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21472 since, unlike the AltiVec predicate framework implementation, we
21473 do not share function descriptors between vec_alle and vec_anyeq.
21474 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21475 set of modes that receive special treatment even when
21476 TARGET_P9_VECTOR is true. The special treatment emits code that
21477 does not depend on Power9 instructions.
21478 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21479 define_expand to not rely on AltiVec predicate framework.
21480 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21482 (vector_ne_v2di_p): Change this define_expand to not rely on
21483 AltiVec predicate framework.
21484 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21486 (vector_ne_<mode>_p): Change this define_expand to not rely on
21487 AltiVec predicate framework.
21488 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21490 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21491 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21492 define_insn pattern.
21493 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21494 define_insn pattern because the xvcmpne<VSs>. instruction is not
21496 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21497 instruction is not supported.
21499 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21501 * config/nvptx/nvptx.c: Include intl.h.
21503 2017-03-01 Martin Jambor <mjambor@suse.cz>
21506 * ipa-prop.h (ipa_bits): Removed field known.
21507 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21508 to pointers. Adjusted their comments to warn about their sharing.
21509 (ipcp_transformation_summary): Change bits to a vector of pointers.
21510 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21511 (ipa_get_ipa_bits_for_value): Declare.
21512 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21513 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21514 (ipa_bits_hash_table): Likewise.
21515 (ipa_vr_ggc_hash_traits): Likewise.
21516 (ipa_vr_hash_table): Likewise.
21517 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21518 being pointers and vr_known being removed.
21519 (ipa_set_jf_unknown): Likewise.
21520 (ipa_get_ipa_bits_for_value): New function.
21521 (ipa_set_jfunc_bits): Likewise.
21522 (ipa_get_value_range): New overloaded functions.
21523 (ipa_set_jfunc_vr): Likewise.
21524 (ipa_compute_jump_functions_for_edge): Use the above functions to
21525 construct bits and vr parts of jump functions.
21526 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21527 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21529 (ipcp_grow_transformations_if_necessary): Also allocate
21530 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21532 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21533 them. Fix too long lines.
21534 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21535 vr_known being removed.
21536 (ipa_read_jump_function): Use new setter functions to construct bits
21537 and vr parts of jump functions or set them to NULL.
21538 (write_ipcp_transformation_info): Adjust for bits being pointers.
21539 (read_ipcp_transformation_info): Likewise.
21540 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21542 Include gt-ipa-prop.h.
21543 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21545 (ipcp_store_bits_results): Likewise.
21546 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21547 Do not write to existing jump functions but use a temporary instead.
21549 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21552 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21553 attempt to use its first operand as BIT_FIELD_REF base.
21555 2017-03-01 Richard Biener <rguenther@suse.de>
21557 PR middle-end/79721
21558 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21559 interpolating formula in wrapping arithmetic.
21560 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21562 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21564 PR tree-optimization/79734
21565 * tree-vect-generic.c (expand_vector_condition): Optimize
21566 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21567 Handle VEC_COND_EXPR where comparison has different inner width from
21568 type's inner width.
21570 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21572 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21573 markup, and similar issues. Remove @opindex entries for things
21574 that aren't options. Add missing -mmpy-option entries.
21576 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21578 PR tree-optimization/79737
21579 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21580 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21581 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21582 instead of byte_size. Formatting fix.
21583 (shift_bytes_in_array_right): Formatting fix.
21585 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21588 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21589 condition on optimize for the leaf function test.
21591 2017-02-28 Martin Liska <mliska@suse.cz>
21594 * read-rtl-function.c (function_reader::handle_unknown_directive):
21595 Bail out when one uses -flto.
21597 2017-02-28 Martin Liska <mliska@suse.cz>
21599 * common.opt: Replace space with tabular for options of <number>
21601 * config/i386/i386.opt: Show <number> value for
21602 -mlarge-data-threshold.
21603 * opts.c (print_filtered_help): Do not display number in hexadecimal
21606 2017-02-28 Martin Liska <mliska@suse.cz>
21608 * common.opt: Fix --help=option -Q for options which are of
21611 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21613 * config/i386/i386.c (print_reg): Error out for values
21614 of 8-bit size in invalid integer register.
21616 2017-02-28 Martin Sebor <msebor@redhat.com>
21618 PR tree-optimization/79691
21619 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21621 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21624 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21625 gcc_unreachable with output_operand_lossage.
21627 2017-02-28 Richard Biener <rguenther@suse.de>
21629 PR tree-optimization/79740
21630 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21632 (visit_nary_op): Insert the nary into the hashtable if we
21633 pattern-matched sth.
21634 * tree-ssa-pre.c (eliminate_insert): Robustify.
21636 2017-02-28 Richard Biener <rguenther@suse.de>
21638 PR middle-end/79731
21639 * fold-const.c (decode_field_reference): Reject out-of-bound
21642 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21644 * config/i386/i386.c: Include intl.h.
21645 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21646 instead of just cond ? "..." : "...".
21647 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21648 * coverage.c (read_counts_file): Likewise.
21649 * omp-offload.c: Include intl.h.
21650 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21651 of just cond ? "..." : "...".
21652 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21653 of just cond ? "..." : "...".
21655 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21658 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21660 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21661 'tune for' CPU name.
21662 * config/arm/arm-cpu-data.h: Regenerated.
21664 2017-02-28 Richard Biener <rguenther@suse.de>
21666 PR tree-optimization/79732
21667 * tree-inline.c (expand_call_inline): Do not shadow var.
21669 2017-02-28 Richard Biener <rguenther@suse.de>
21671 PR tree-optimization/79723
21672 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21673 address-space properly.
21675 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21677 * doc/optinfo.texi (Optimization groups): Fix option used for
21679 * doc/invoke.texi (-fopt-info): Document "omp".
21680 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21681 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21682 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21683 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21684 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21686 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21688 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21689 OPTGROUP_OMP with "omp".
21691 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21694 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21695 for arithmetic shift of unsigned V2DI.
21697 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21699 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21700 arc/linux.h headers.
21701 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21702 (LINK_SPEC): Likewise.
21703 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21704 (EXTRA_SPECS): Likewise.
21705 (STARTFILE_SPEC): Likewise.
21706 (ENDFILE_SPEC): Likewise.
21707 (LIB_SPEC): Likewise.
21708 (TARGET_SDATA_DEFAULT): Likewise.
21709 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21710 (MULTILIB_DEFAULTS): Likewise.
21711 (DWARF2_UNWIND_INFO): Likewise.
21712 * config/arc/big.h: New file.
21713 * config/arc/elf.h: Likewise.
21714 * config/arc/linux.h: Likewise.
21715 * config/arc/t-uClibc: Remove.
21717 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21719 PR tree-optimization/77536
21720 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21721 (tree_transform_and_unroll_loop): Use above function to compute the
21722 estimated niter of unrolled loop and use it when scaling profile.
21723 Also use count info rather than frequency if it's non-zero.
21724 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21725 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21726 (vect_transform_loop): Call above function.
21728 2017-02-27 Richard Biener <rguenther@suse.de>
21730 PR tree-optimization/45397
21731 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21732 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21733 (visit_nary_op): Add pattern matching for CSEing sign-changed
21734 or truncated operations with wider ones.
21736 2017-02-27 Richard Biener <rguenther@suse.de>
21738 PR tree-optimization/79690
21739 * tree-vect-stmts.c (vectorizable_store): Use vector type
21740 built from the DR with address-space.
21742 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21744 * doc/invoke.texi (Optimize Options): Refine the description
21745 of asan-use-after-return.
21747 2017-02-25 Alan Modra <amodra@gmail.com>
21749 PR rtl-optimization/79584
21750 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21751 base, not ad->base_term, the reg within base. Remove assertion
21752 that ad->base == ad->base_term. Replace gen_int_mode using
21753 bogus mode with const0_rtx.
21755 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21757 PR middle-end/79396
21758 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21759 FMA_EXPR like tcc_binary or tcc_unary.
21761 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21764 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21766 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21767 (output_loc_operands): Handle DW_OP_call_ref and
21768 DW_OP_GNU_variable_value.
21769 (struct variable_value_struct): New type.
21770 (struct variable_value_hasher): Likewise.
21771 (variable_value_hash): New variable.
21772 (string_types): Remove.
21773 (copy_loc_descr): New function.
21774 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21775 (prepend_loc_descr_to_each): New function.
21776 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21777 instead of add_loc_descr_to_each if the first argument is single
21778 location list and the second has multiple.
21779 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21780 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21781 when looking for variable value which doesn't have other location info.
21782 (loc_list_from_tree): Formatting fix.
21783 (gen_array_type_die): Simplify DW_AT_string_length handling.
21784 (adjust_string_types): Remove.
21785 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21786 string_types. Call resolve_variable_values.
21787 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21788 (resolve_addr_in_expr): Likewise. Add A argument.
21789 (copy_deref_exprloc): Remove deref argument. Adjust for the
21790 original expression being DW_OP_GNU_variable_value with optionally
21791 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21792 optionally after it.
21793 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21794 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21795 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21796 (variable_value_hasher::hash, variable_value_hasher::equal): New
21798 (resolve_variable_value_in_expr, resolve_variable_value,
21799 resolve_variable_values, note_variable_value_in_expr,
21800 note_variable_value): New functions.
21801 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21804 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21807 * opts.h (handle_generated_option): Add GENERATED_P argument.
21808 * opts-common.c (handle_option): Adjust function comment.
21809 (handle_generated_option): Add GENERATED_P argument, pass it to
21811 (control_warning_option): Pass false to handle_generated_option
21813 * opts.c (maybe_default_option): Pass true to handle_generated_option
21815 * optc-gen.awk: Likewise.
21817 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21819 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21820 a REG, look at the REG it is a SUBREG of.
21821 (splitter for cmpeqsi_t): Ditto.
21823 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21825 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21826 the special USEs with the pattern of the insn, not the insn itself.
21828 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21831 * doc/invoke.texi: Document -mload-store-pairs.
21833 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21834 Sandra Loosemore <sandra@codesourcery.com>
21836 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21837 argument isn't a CONST_INT.
21838 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21839 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21840 (nios2_validate_compare): Bypass alternate compare logic if *op2
21841 is not a CONST_INT.
21842 (ldstwm_operation_p): Return false if first_base is not a REG or
21843 if first_offset is not a CONST_INT.
21845 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21847 * config/cris/cris.md: Use correct operand in a define_peephole2.
21849 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21851 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21853 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21855 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21856 this_insn if it is an INSN or JUMP_INSN.
21857 (force_offsettable): Look at base, not at addr.
21858 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21859 on things that aren't necessarily CONST_INTs.
21861 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21863 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21864 -mfpmath=sse is the default also for x86-32 targets with SSE2
21865 instruction set when @option{-ffast-math} is enabled
21867 2017-02-24 Jeff Law <law@redhat.com>
21869 PR rtl-optimizatoin/79286
21870 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21873 2017-02-24 Richard Biener <rguenther@suse.de>
21875 PR tree-optimization/79389
21876 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21879 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21881 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21882 function comment to reflect reality.
21883 (loop_exits_before_overflow): Fix typo in function description.
21885 2017-02-24 Richard Biener <rguenther@suse.de>
21887 PR tree-optimization/79389
21888 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21889 properly that a threading opportunity exists. Detect conditional
21890 copy/constant propagation opportunities.
21892 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21894 * config/visium/visium.md (type): Add trap.
21895 (b): New mode attribute.
21896 (*btst): Rename into...
21897 (*btst<mode>): ...this and adjust.
21898 (*cbranchsi4_btst_insn): Rename into...
21899 (*cbranch<mode>4_btst_insn): ...this and adjust.
21900 (trap): New define_insn.
21902 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21904 PR tree-optimization/79389
21905 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21906 (noce_reversed_cond_code): New function.
21907 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21908 reversed_comparison_code. Formatting fix.
21909 (noce_try_store_flag): Test rev_cond != NULL in addition to
21910 reversed_comparison_code.
21911 (noce_try_store_flag_constants): Likewise.
21912 (noce_try_store_flag_mask): Likewise.
21913 (noce_try_addcc): Use rev_cond if non-NULL instead of
21914 reversed_comparison_code.
21915 (noce_try_cmove_arith): Likewise. Formatting fixes.
21916 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21917 (noce_find_if_block): Initialize rev_cond.
21918 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21919 instead of false as last argument never attempt to reverse it
21922 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21924 PR tree-optimization/79663
21925 * tree-predcom.c (combine_chains): Process refs in reverse order
21926 only for ZERO length chains, and add explaining comment.
21928 2017-02-23 Jeff Law <law@redhat.com>
21930 PR tree-optimization/79578
21931 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21932 in call to operand_equal_p.
21934 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21937 * config/i386/cpuid.h: Fix another undefined behavior.
21939 2017-02-23 Richard Biener <rguenther@suse.de>
21941 PR tree-optimization/79683
21942 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21943 vector types for data-refs.
21945 2017-02-23 Martin Liska <mliska@suse.cz>
21947 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21949 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21951 PR middle-end/79665
21952 * internal-fn.c (get_range_pos_neg): Moved to ...
21953 * tree.c (get_range_pos_neg): ... here. No longer static.
21954 * tree.h (get_range_pos_neg): New prototype.
21955 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21956 are known to be in between 0 and signed maximum inclusive, try to
21957 expand both unsigned and signed divmod and use the cheaper one from
21960 2017-02-22 Jeff Law <law@redhat.com>
21962 PR tree-optimization/79578
21963 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21964 to compare base operands.
21966 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21969 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21970 gpc_reg_operand instead of fpr_reg_operand.
21972 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21974 * config/mips/mips.c (mips_return_in_memory): Force FP
21975 vector types to be returned in memory for o32 ABI.
21977 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21979 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21980 instead of DW_TAG_member for static data member declarations and don't
21981 set no_linkage_name for static inline data members.
21982 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21985 2017-02-22 Martin Liska <mliska@suse.cz>
21987 * doc/invoke.texi: Replace inequality signs with square brackets
21990 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21992 PR tree-optimization/68644
21993 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21995 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21998 * lra-constraints.c (simplify_operand_subreg): Handle
21999 WORD_REGISTER_OPERATIONS targets.
22001 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22004 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22005 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22006 elimination by swapping fld*.
22008 2017-02-22 Richard Biener <rguenther@suse.de>
22010 PR tree-optimization/79673
22011 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22012 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22013 irrelevant address-space qualifiers and avoiding a
22014 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22016 2017-02-22 Richard Biener <rguenther@suse.de>
22018 PR tree-optimization/79666
22019 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22020 to not symbolically negate if that may introduce undefined
22023 2017-02-22 Martin Liska <mliska@suse.cz>
22026 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22027 * data-streamer-out.c (streamer_write_gcov_count_stream):
22029 * value-prof.c (stream_out_histogram_value): Make assert more
22030 precise based on type of counter.
22032 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22035 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22036 Use nonimmediate_operand instead of memory_operand for operand 1.
22037 (float-extend standard_x87sse_constant_load splitter): Ditto.
22039 2017-02-21 Jeff Law <law@redhat.com>
22041 PR tree-optimization/79621
22042 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22043 blocks with edges to themselves.
22045 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22048 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22049 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22050 Use gimple_call_builtin_p.
22053 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22054 on temporarily removed DEBUG_INSNs.
22056 PR tree-optimization/79649
22057 * tree-loop-distribution.c (classify_partition): Give up on
22058 non-generic address space loads/stores.
22060 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22062 * doc/loop.texi (Loop manipulation): Remove nonexistent
22063 tree_ssa_loop_version from the documentation.
22064 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22066 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22069 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22070 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22071 * config/rs6000/rs6000.c: Include except.h.
22072 (rs6000_expand_split_stack_prologue): Call
22073 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22075 2017-02-21 Martin Jambor <mjambor@suse.cz>
22078 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22079 have been analyzed.
22081 2017-02-21 Martin Jambor <mjambor@suse.cz>
22083 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22084 for backward compatibility only.
22085 * doc/invoke.texi (Option Summary): Remove all references to
22086 -fipa-cp-alignment.
22088 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22092 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22094 * lra-constraints.c (curr_insn_transform): Handle
22095 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22097 2017-02-21 Martin Liska <mliska@suse.cz>
22099 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22101 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22103 PR translation/79638
22104 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22106 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22109 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22110 (arm_function_ok_for_sibcall): Return false for an indirect call by
22111 descriptor if all the argument registers are used.
22112 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22113 alignment of the function.
22115 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22117 PR tree-optimization/61441
22118 * simplify-rtx.c (simplify_const_unary_operation): For
22119 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22120 the sNaN unmodified.
22122 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22124 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22125 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22126 instead of SYSTEM_HEADER_DIR.
22128 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22129 Martin Liška <mliska@suse.cz>
22131 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22132 Fix typos and grammar, use active voice, and clarify.
22134 2017-02-20 Marek Polacek <polacek@redhat.com>
22136 PR middle-end/79537
22137 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22140 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22142 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22145 * config/i386/i386.c (ix86_expand_builtin): Handle
22146 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22147 ix86_builtins_isa[fcode].isa as a requirement of those
22148 flags and any other flag in the bitmask.
22149 (ix86_init_mmx_sse_builtins): Use 0 instead of
22150 ~OPTION_MASK_ISA_64BIT as mask.
22151 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22152 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22153 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22154 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22156 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22159 * lra-constraints.c (split_reg): Check requested split mode
22160 is supported by the register.
22162 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22164 * lra-constraints.c (simplify_operand_subreg): Remove early
22167 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22170 * lra-constraints.c (curr_insn_transform): Tighten condition
22171 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22173 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22176 * lra-constraints.c (curr_insn_transform): Handle
22177 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22179 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22182 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22184 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22186 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22189 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22192 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22194 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22197 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22199 * final.c (last_columnnum, override_columnnum): New variables.
22200 (final_start_function): Set last_columnnum, pass it to begin_prologue
22201 hook and pass 0 to dwarf2out_begin_prologue.
22202 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22203 to source_line debug hook.
22204 (notice_source_line): Compute last_columnnum and for debug_column_info
22205 return true on column changes.
22206 * debug.h (struct gcc_debug_hooks): Add column argument to
22207 source_line and begin_prologue hooks.
22208 (debug_nothing_int_charstar_int_bool): Remove prototype.
22209 (debug_nothing_int_int_charstar,
22210 debug_nothing_int_int_charstar_int_bool): New prototypes.
22211 (dwarf2out_begin_prologue): Add column argument.
22212 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22213 begin_prologue hooks.
22214 (debug_nothing_int_charstar_int_bool): Remove.
22215 (debug_nothing_int_int_charstar,
22216 debug_nothing_int_int_charstar_int_bool): New functions.
22217 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22218 through to dwarf2out_source_line.
22219 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22220 (dwarf2out_source_line): Add column argument, emit it if requested.
22221 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22223 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22224 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22225 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22226 through to dwarf2out_begin_prologue.
22227 (vmsdbgout_source_line): Add column argument, pass it through to
22228 dwarf2out_source_line.
22229 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22230 dbxout_source_line caller.
22231 (dbxout_source_line): Add column argument.
22233 * common.opt (gno-column-info, gcolumn-info): New options.
22234 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22235 (check_die): Also test for multiple DW_AT_decl_column attributes.
22236 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22237 DW_AT_decl_column if requested.
22238 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22240 (gen_variable_die): Likewise.
22241 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22242 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22245 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22246 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22247 (ix86_handle_option): Handle OPT_m3dnowa.
22248 * doc/invoke.texi (-m3dnowa): Document.
22249 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22250 -m3dnowa instead of -m3dnow -march=athlon.
22253 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22254 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22256 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22259 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22260 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22261 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22262 generator for vsx_xxpermdi_<mode>_be.
22263 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22264 force big-endian semantics.
22265 (vsx_xxpermdi_<mode>_be): New define_expand with same
22266 implementation as previous version of vsx_xxpermdi_<mode>.
22268 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22270 PR tree-optimization/79327
22271 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22272 variable, its initialization and use.
22274 2017-02-17 Julia Koval <julia.koval@intel.com>
22276 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22277 (OPTION_MASK_ISA_PKU_UNSET): New.
22278 (ix86_handle_option): Handle -mrdpid.
22279 * config/i386/cpuid.h (bit_RDPID): New.
22280 * config/i386/driver-i386.c (host_detect_local_cpu):
22281 Detect RDPID feature.
22282 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22283 * config/i386/i386-c.c (ix86_target_macros_internal):
22285 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22286 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22287 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22288 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22289 * config/i386/i386.md (define_insn "rdpid"): New.
22290 * config/i386/i386.opt Add -mrdpid.
22291 * config/i386/immintrin.h (_rdpid_u32): New.
22293 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22295 PR rtl-optimization/79541
22296 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22297 instead of transforming it into USE.
22299 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22301 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22302 If HONOR_SNANS (SFmode) force the input to a register.
22303 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22304 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22305 an frsp or similar insn.
22307 2017-02-17 Martin Liska <mliska@suse.cz>
22309 PR rtl-optimization/79577
22310 * params.def (selsched-max-sched-times): Increase minimum to 1.
22312 2017-02-17 Martin Liska <mliska@suse.cz>
22314 PR rtl-optimization/79574
22315 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22317 2017-02-17 Martin Liska <mliska@suse.cz>
22319 PR tree-optimization/79529
22320 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22321 ssa_defined_default_def_p to handle cases which are implicitly
22323 * tree-ssa.c (ssa_defined_default_def_p): New function.
22324 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22325 which are implicitly defined.
22326 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22328 2017-02-17 Richard Biener <rguenther@suse.de>
22330 PR middle-end/79576
22331 * params.def (max-ssa-name-query-depth): Limit to 10.
22333 2017-02-17 Richard Biener <rguenther@suse.de>
22335 PR tree-optimization/79552
22336 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22339 2017-02-17 Richard Biener <rguenther@suse.de>
22342 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22344 2017-02-17 Marek Polacek <polacek@redhat.com>
22346 PR middle-end/79536
22347 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22348 (fold_negate_expr): New wrapper.
22350 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22352 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22353 Correct terminology and de-emphasize pre-standard behavior.
22355 2017-02-16 Alan Modra <amodra@gmail.com>
22357 PR rtl-optimization/79286
22358 * ira.c (def_dominates_uses): New function.
22359 (update_equiv_regs): Don't create an equivalence for insns that
22360 may trap where the register def does not dominate the use.
22362 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22364 PR rtl-optimization/78127
22365 * lra.c (lra): Call lra_eliminate before finish the loop after
22368 2017-02-16 Richard Biener <rguenther@suse.de>
22370 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22372 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22373 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22374 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22375 (isl_val_int_from_wi): New function.
22376 (extract_affine_gmp): Rename to ...
22377 (extract_affine_wi): ... this, take a widest_int.
22378 (extract_affine_int): Just wrap extract_affine_wi.
22379 (add_param_constraints): Use isl_val_int_from_wi.
22380 (add_loop_constraints): Likewise, and extract_affine_wi.
22382 2017-02-15 Jeff Law <law@redhat.com>
22384 PR middle-end/79521
22385 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22386 ira_init_register_move_cost_if_necessary.
22388 2017-02-15 Martin Sebor <msebor@redhat.com>
22390 PR middle-end/32003
22391 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22392 removed in a prior commit.
22394 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22396 PR tree-optimization/79347
22397 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22398 counters during peeling.
22400 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22402 * Makefile.in (site.exp): Remove "set ISLVER".
22404 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22407 * real.c (real_from_integer): Call real_convert even for decimal.
22409 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22412 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22414 2017-02-14 Andrew Pinski <apinski@cavium.com>
22416 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22417 cores and change the partno/implementer to be correct.
22418 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22419 the 'B" as the implementer.
22420 * config/aarch64/aarch64-tune.md: Regenerate.
22422 2017-02-14 Carl Love <cel@us.ibm.com>
22424 * config/rs6000/rs6000.c: Add case statement entry to make the
22425 xvcvuxdsp built-in argument unsigned.
22426 * config/rs6000/vsx.md: Fix the source and return operand types so they
22427 match the instruction definitions from the ISA document. Fix typo
22428 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22431 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22434 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22435 member early_clobber_alts.
22436 * lra-lives.c (reg_early_clobber_p): New.
22437 (process_bb_lives): Use it.
22438 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22439 (debug_operand_data): Initialize early_clobber_alts.
22440 (setup_operand_alternative): Set up early_clobber_alts.
22441 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22442 alternatives to new_insn_reg.
22443 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22445 (lra_update_insn_regno_info): Pass the new arg.
22447 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22449 PR middle-end/79505
22450 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22451 (new_oacc_loop_raw): Don't clear already cleared fields.
22454 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22455 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22456 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22458 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22461 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22463 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22466 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22467 the extra instruction to the right place to store 128-bit constant
22470 2017-02-14 Martin Sebor <msebor@redhat.com>
22472 PR middle-end/79448
22473 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22474 warning for strings of unknown length.
22476 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22478 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22480 2017-02-14 Jeff Law <law@redhat.com>
22483 * ira-costs.c (scan_one_insn): Initialize register move costs
22484 for pseudos seen in USE/CLOBBER insns.
22486 PR tree-optimization/79095
22487 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22488 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22489 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22490 if the operands are known to be not equal, then the resulting range
22492 (intersect_ranges): If the new range is ~[0,0] and the old range is
22493 wide, then prefer ~[0,0].
22494 * tree-vrp.c (overflow_comparison_p_1): New function.
22495 (overflow_comparison_p): New function.
22496 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22497 if NAME is used in an overflow test.
22498 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22499 overflow check that can be expressed as an equality test, then adjust
22500 ops to be that equality test.
22502 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22504 * config/s390/s390-builtin-types.def: Remove flags argument.
22505 * config/s390/s390.c (s390_init_builtins): Likewise.
22507 2017-02-14 Martin Liska <mliska@suse.cz>
22509 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22510 vector. Fix trailing white spaces.
22512 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22514 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22517 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22519 PR rtl-optimization/68664
22520 * config/arm/arm.c (arm_sched_can_speculate_insn):
22521 New function. Declare prototype.
22522 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22524 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22526 PR rtl-optimization/68664
22527 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22529 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22531 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22533 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22534 max skip bytes for function, loop and jump.
22536 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22538 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22539 ABS_EXPR for gimple dump.
22541 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22544 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22546 PR tree-optimization/79408
22547 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22548 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22549 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22550 also if rhs1 is INTEGER_CST.
22552 2017-02-14 Richard Biener <rguenther@suse.de>
22554 PR middle-end/79432
22555 * tree-into-ssa.c (insert_phi_nodes): When the function can
22556 have abnormal edges rewrite SSA names with broken use-def
22557 dominance out of SSA and register them for PHI insertion.
22559 2017-02-13 Martin Sebor <msebor@redhat.com>
22561 PR middle-end/79496
22562 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22563 clearing info.nowrite flag when snprintf size argument is a range.
22565 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22567 * cprop.c (cprop_jump): Add missing space in string literal.
22568 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22569 (get_constraint_for_component_ref): Likewise.
22570 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22571 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22572 * lra-constraints.c (process_alt_operands): Likewise.
22573 * ipa-inline.c (inline_small_functions): Likewise.
22574 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22575 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22576 * trans-mem.c (diagnose_tm_1_op): Likewise.
22577 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22578 (grid_parallel_clauses_gridifiable): Likewise.
22580 * config/nvptx/mkoffload.c (process): Add space in between
22583 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22584 "MOD4_SSE_REGS" and "ALL_REGS".
22586 * spellcheck.c (test_data): Add , in between "foo" and "food".
22588 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22591 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22592 boundary crossing check and subsequent code generation agree.
22594 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22596 * config/aarch64/aarch64.c (has_memory_op): Delete.
22597 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22600 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22602 PR rtl-optimization/79388
22603 PR rtl-optimization/79450
22604 * combine.c (distribute_notes): When removing TEM_INSN for which
22605 corresponding dest has last value recorded, invalidate that last
22608 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22610 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22611 of explicit '@'. Add missing assembly comment marker on branch costs
22614 2017-02-13 Nathan Sidwell <nathan@acm.org>
22616 * gengtype-lex.l (<in_struct>): Add '/'.
22618 2017-02-13 Martin Liska <mliska@suse.cz>
22621 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22623 2017-02-13 Richard Biener <rguenther@suse.de>
22625 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22627 * configure: Re-generate.
22628 * config.in: Likewise.
22629 * graphite-dependences.c: Simplify as if
22630 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22631 * graphite-isl-ast-to-gimple.c: Likewise.
22632 * graphite-optimize-isl.c: Likewise.
22633 * graphite-poly.c: Likewise.
22634 * graphite-sese-to-poly.c: Likewise.
22635 * graphite.h: Likewise.
22636 * toplev.c: Include isl/version.h and use isl_version () for
22637 printing the ISL version.
22638 * doc/install.texi: Update ISL requirement.
22640 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22642 * doc/standards.texi (Standards): Update reference to
22645 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22647 * doc/extend.texi (Named Address Spaces): sourceware.org now
22649 * doc/install.texi (Binaries): Ditto.
22652 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22654 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22655 standard terminology "stringize"/"stringizing" throughout.
22656 * doc/cppinternals.texi: Likewise.
22658 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22660 * doc/extend.texi: Fix some spelling mistakes and typos.
22661 * doc/invoke.texi: Likewise.
22663 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22666 * params.def (inline-min-speedup) Change from 10 to 8.
22668 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22670 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22673 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22676 * ipa-inline-analysis.c (get_minimal_bb): New function.
22677 (record_modified): Use it.
22678 (remap_edge_change_prob): Handle also ancestor functions.
22680 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22682 * doc/contrib.texi (Contributors): Remove broken link into
22683 the Mauve CVS repository.
22685 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22687 PR middle-end/79454
22688 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22689 result computation whenever lhs doesn't have vector mode, not
22690 just when it has BLKmode.
22692 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22694 * doc/makefile.texi (profiledbootstrap): Refer to the
22695 installation instructions only in textual form.
22697 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22700 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22702 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22704 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22705 (Specific): Update mingw-w64 reference.
22707 (Specific): Remove broken link to Renesas RX processor.
22709 2017-02-10 Richard Biener <rguenther@suse.de>
22711 * toplev.c (process_options): Do not mention obsolete graphite
22712 options when printing sorry message about missing graphite support.
22713 Mention -floop-nest-optimize.
22715 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22717 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22718 (vtst_p16): Likewise.
22719 (vtstq_p8): Likewise.
22720 (vtstq_p16): Likewise.
22722 (vtstq_p64): Likewise.
22723 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22724 (vset_lane_p64): New.
22725 (vsetq_lane_p64): New.
22727 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22729 PR tree-optimization/79411
22730 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22731 stmt operands are SSA_NAMEs used in abnormal phis.
22732 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22735 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22738 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22741 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22743 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22745 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22747 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22751 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22752 not arbitrary TREE_CONSTANT.
22755 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22756 "omp declare target link" attribute unless is_global_var.
22757 * omp-offload.c (find_link_var_op): Likewise.
22759 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22760 Chung-Lin Tang <cltang@codesourcery.com>
22762 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22764 (gimplify_adjust_omp_clauses): Don't delete TILE.
22765 (gimplify_omp_for): Deal with TILE.
22766 * internal-fn.c (expand_GOACC_TILE): New function.
22767 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22769 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22770 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22772 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22773 avoid DIV for outermost collapse var.
22774 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22775 Remove out of date comments, fix whitespace.
22776 * omp-general.c (omp_extract_for_data): Deal with tiling.
22777 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22778 adjust OLF_DIM_BASE value.
22779 (struct omp_for_data): Add tiling field.
22780 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22781 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22782 for auto loops. Remove default auto determining, moved to
22783 oacc_loop_fixed_partitions.
22784 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22785 stmts, add e_mask field.
22786 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22787 (oacc_thread_numbers): Use oacc_dim_call.
22788 (oacc_xform_tile): New.
22789 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22790 (finish_oacc_loop): Adjust for ifns vector.
22791 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22792 add case for GOACC_TILE fns.
22793 (oacc_loop_xform_loop): Delete.
22794 (oacc_loop_process): Iterate over call list directly, and add
22795 handling for GOACC_TILE fns.
22796 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22798 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22799 vector partitioning to outer loops. Assign 2 partitions to loops
22800 when available. Add TILE handling.
22801 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22802 (execite_oacc_device_lower): Process GOACC_TILE fns,
22803 ignore unknown specs.
22804 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22805 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22806 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22808 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22810 * configure.ac (ACX_BUGURL): Update.
22811 * configure: Regenerate.
22813 2017-02-09 Richard Biener <rguenther@suse.de>
22815 PR tree-optimization/69823
22816 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22817 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22819 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22821 * config/arc/arc-c.def: Add __NPS400__ definition.
22822 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22823 (TARGET_NPS400): Define.
22825 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22827 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22829 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22830 pointer, arch_info.
22831 (arc_cpu_types): Fill the arch_info field with a pointer into the
22832 arc_arch_types table.
22833 (arc_selected_cpu): Declare.
22834 * config/arc/arc.c (arc_selected_cpu): Make global.
22835 (arc_selected_arch): Delete.
22836 (arc_base_cpu): Delete.
22837 (arc_override_options): Remove references to deleted variables,
22838 update access to arch information.
22839 (ARC_OPT): Update access to arch information.
22840 (ARC_OPTX): Likewise.
22841 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22842 (TARGET_ARC600): Update access to arch information.
22843 (TARGET_ARC601): Likewise.
22844 (TARGET_ARC700): Likewise.
22845 (TARGET_EM): Likewise.
22846 (TARGET_HS): Likewise.
22847 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22850 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22853 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22854 condition/operands for integer GE/LE/GEU/LEU operations.
22856 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22858 PR translation/79397
22859 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22862 2017-02-08 Martin Jambor <mjambor@suse.cz>
22865 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22866 whether allocation happened.
22867 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22868 nothing was allocated.
22870 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22872 PR tree-optimization/79408
22873 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22874 constant, but SSA_NAME with a known integer range, use the minimum
22875 of that range instead of op1 to determine if modulo can be replaced
22876 with its first operand.
22878 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22880 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22882 2017-02-08 Richard Biener <rguenther@suse.de>
22884 PR tree-optimization/71824
22885 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22886 Check all loops contained in the merged region.
22888 2017-02-07 Andrew Pinski <apinski@cavium.com>
22890 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22892 2017-02-07 Andrew Pinski <apinski@cavium.com>
22894 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22895 (thunderxt88): Likewise.
22896 (thunderxt81): Disable LSE and change v8.1 to v8.
22897 (thunderxt83): Likewise.
22899 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22900 Richard Biener <rguenther@suse.de>
22902 PR middle-end/79399
22903 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22904 type from int to size_t.
22905 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22907 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22909 PR rtl-optimization/79386
22910 * cprop.c (bypass_conditional_jumps): Initialize
22911 bypass_last_basic_block already before splitting bbs after
22912 unconditional traps...
22913 (bypass_conditional_jumps): ... rather than here.
22916 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22917 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22918 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22919 fix -masm=intel patterns.
22921 2017-02-07 Richard Biener <rguenther@suse.de>
22923 PR tree-optimization/79256
22924 PR middle-end/79278
22925 * builtins.c (get_object_alignment_2): Use min_align_of_type
22926 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22927 and ADJUST_FIELD_ALIGN.
22929 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22931 * doc/tm.texi: Regenerate.
22932 * stor-layout.c (layout_decl): Adjust.
22933 (update_alignment_for_field): Likewise.
22934 (place_field): Likewise.
22935 (min_align_of_type): Likewise.
22936 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22937 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22938 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22939 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22940 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22941 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22942 * config/i386/i386.c (x86_field_alignment): Likewise.
22943 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22944 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22945 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22946 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22947 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22948 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22952 2017-01-30 Richard Biener <rguenther@suse.de>
22954 PR tree-optimization/79256
22955 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22956 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22959 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22961 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22962 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22963 builtins to SImode and emit a zero-extend, if necessary.
22965 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22967 * docs/invoke.texi (RISC-V Options): Alphabetize.
22969 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22971 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22974 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22976 * config/riscv/riscv.c: New file.
22977 * common/config/riscv/riscv-common.c: Likewise.
22978 * config.gcc: Likewise.
22979 * config/riscv/constraints.md: Likewise.
22980 * config/riscv/elf.h: Likewise.
22981 * config/riscv/generic.md: Likewise.
22982 * config/riscv/linux.h: Likewise.
22983 * config/riscv/multilib-generator: Likewise.
22984 * config/riscv/peephole.md: Likewise.
22985 * config/riscv/pic.md: Likewise.
22986 * config/riscv/predicates.md: Likewise.
22987 * config/riscv/riscv-builtins.c: Likewise.
22988 * config/riscv/riscv-c.c: Likewise.
22989 * config/riscv/riscv-ftypes.def: Likewise.
22990 * config/riscv/riscv-modes.def: Likewise.
22991 * config/riscv/riscv-opts.h: Likewise.
22992 * config/riscv/riscv-protos.h: Likewise.
22993 * config/riscv/riscv.h: Likewise.
22994 * config/riscv/riscv.md: Likewise.
22995 * config/riscv/riscv.opt: Likewise.
22996 * config/riscv/sync.md: Likewise.
22997 * config/riscv/t-elf-multilib: Likewise.
22998 * config/riscv/t-linux: Likewise.
22999 * config/riscv/t-linux-multilib: Likewise.
23000 * config/riscv/t-riscv: Likewise.
23001 * configure.ac: Likewise.
23002 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23003 Waterman as RISC-V maintainers.
23004 * doc/install.texi: Add RISC-V entries.
23005 * doc/invoke.texi: Add RISC-V options section.
23006 * doc/md.texi: Add RISC-V constraints section.
23007 * configure: Regenerated.
23009 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23012 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23013 false values to be constant vectors with all 0 or all 1 bits set.
23014 (vcondu<mode><mode>): Likewise.
23015 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23017 (fpmask_comparison_operator): Update comment.
23018 (vecint_comparison_operator): New predicate.
23019 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23020 vector conditionals when the true and false values are constant
23021 vectors with all 0 bits or all 1 bits set.
23023 2017-02-06 Martin Sebor <msebor@redhat.com>
23025 PR tree-optimization/79376
23026 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23028 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23030 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23031 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23032 to simplify split condition.
23034 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23036 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23037 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23040 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23042 PR rtl-optimization/68664
23043 * target.def (can_speculate_insn): New hook.
23044 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23045 * doc/tm.texi: Regenerate.
23046 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23047 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23048 (rs6000_sched_can_speculate_insn): New function.
23050 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23052 PR tree-optimization/79284
23053 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23054 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23055 vectorizable_mask_load_store, vectorizable_operation,
23056 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23057 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23058 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23059 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23060 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23061 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23062 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23063 is_gimple_assign (stmt). Replace another such test with
23064 is_gimple_assign (stmt).
23066 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23069 * config/avr/avr.c (rtl-iter.h): Include it.
23070 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23071 (avr_legitimate_combined_insn): ...and implementation.
23073 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23075 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23076 * config/s390/s390.c (s390_const_operand_ok)
23077 (s390_canonicalize_comparison, s390_extract_part)
23078 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23079 (s390_contiguous_bitmask_p, s390_rtx_costs)
23080 (legitimize_pic_address): Likewise.
23081 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23082 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23083 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23084 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23085 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23087 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23089 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23090 REGNO($0) == REGNO($1).
23092 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23094 * config/s390/linux.h(SIZE_TYPE): Add comment.
23096 2017-02-06 Julian Brown <julian@codesourcery.com>
23097 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23098 Virendra Pathak <virendra.pathak@broadcom.com>
23100 * config/aarch64/aarch64-cores.def: Change the scheduler
23102 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23103 * config/aarch64/thunderx2t99.md: New file.
23105 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23107 * doc/standards.texi (Go Language): Update link to language
23110 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23112 * tree-eh.c (lower_resx): Sanitize profile.
23113 (cleanup_empty_eh_move_lp): Likewise.
23115 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23118 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23120 * cfgloopmanip.h (loop_version): Update prototype.
23121 * modulo-sched.c (sms_schedule): Update call of loop_version.
23122 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23123 * tree-parloops.c (gen_parallel_loop): Likewise.
23124 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23125 * tree-ssa-loop-split.c (split_loop): Likewise.
23126 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23127 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23129 2017-02-05 Martin Liska <mliska@suse.cz>
23132 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23134 (print_operand_address): Initialize a struct to zero.
23136 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23138 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23139 garbage collector only in textual form.
23141 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23143 * doc/extend.texi (x86 specific memory model extensions for
23144 transactional memory): Simplify a phrase.
23146 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23149 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23150 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23151 (atomic_storedi_1): Likewise.
23153 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23155 PR tree-optimization/79338
23156 * tree-parloops.c (gather_scalar_reductions): Don't call
23157 vect_analyze_loop_form for loop->inner before destroying loop's
23160 2017-02-03 Martin Sebor <msebor@redhat.com>
23162 PR tree-optimization/79327
23163 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23164 when precision has resulted in leading zeros.
23165 (format_integer): Adjust the likely counter to assume an unknown
23166 argument that may be zero is non-zero.
23168 2017-02-03 Jason Merrill <jason@redhat.com>
23171 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23172 avoid copying non-taken branch.
23174 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23176 PR tree-optimization/79340
23177 * tree-vect-loop.c (vectorizable_reduction): Release
23178 vec_defs elements after safe_splicing them into other vectors.
23181 PR tree-optimization/79327
23182 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23183 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23185 (format_integer): Use wide_int_to_tree instead of build_int_cst
23186 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23187 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23188 of shortest and longest sequence.
23190 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23192 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23193 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23195 2017-02-03 Walter Lee <walt@tilera.com>
23198 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23199 after initial stackframe link reg save.
23200 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23202 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23205 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23206 wu for stxssp alternative.
23208 2017-02-03 Martin Sebor <msebor@redhat.com>
23210 PR tree-optimization/79352
23211 * gimple-fold.c (get_range_strlen): Add argument.
23212 (get_range_strlen): Change return type to bool.
23213 (get_maxval_strlen): Pass in a dummy argument.
23214 * gimple-fold.h (get_range_strlen): Change return type to bool.
23215 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23216 * tree.h (array_at_struct_end_p): Add argument.
23217 * tree.c (array_at_struct_end_p): Handle it.
23219 2017-02-03 Martin Liska <mliska@suse.cz>
23222 * multiple_target.c (create_dispatcher_calls): Redirect edge
23223 from a caller of a dispatcher.
23224 (expand_target_clones): Make the clones local.
23225 (ipa_target_clone): Do both target clones and resolvers.
23226 (ipa_dispatcher_calls): Remove the pass.
23227 (pass_dispatcher_calls::gate): Likewise.
23228 (make_pass_dispatcher_calls): Likewise.
23229 * passes.def (pass_target_clone): Put as very first IPA early
23232 2017-02-03 Martin Liska <mliska@suse.cz>
23234 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23235 in case of a function with ifunc attribute.
23237 2017-02-03 Martin Liska <mliska@suse.cz>
23239 * cgraph.c (cgraph_node::dump): Dump function version info.
23240 * symtab.c (symtab_node::dump_base): Add missing new line.
23242 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23244 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23245 (ifcombine_ifandif): Use it.
23247 2017-02-03 Martin Liska <mliska@suse.cz>
23249 * doc/invoke.texi: Document default value for
23250 use-after-scope-direct-emission-threshold.
23252 2017-02-03 Martin Liska <mliska@suse.cz>
23254 PR tree-optimization/79339
23255 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23256 (format_floating): Likewise.
23258 2017-02-03 Martin Liska <mliska@suse.cz>
23261 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23263 (ipa_node_params_t::remove): Likewise.
23264 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23265 initialization from removed ipa_node_params_t::insert.
23266 (ipa_node_params::~ipa_node_params): Move from removed
23267 ipa_node_params_t::release.
23268 * symbol-summary.h (symbol_summary::m_released): New member.
23269 Do not release a summary twice. Do not allow to call finalizer
23270 for types of a summary that live in GGC memory.
23272 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23274 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23277 2017-02-02 Martin Sebor <msebor@redhat.com>
23279 PR middle-end/79275
23280 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23281 (format_string): Tighten up the range of output for non-constant
23282 strings and correct the expected range for wide non-constant strings.
23284 2017-02-02 Martin Sebor <msebor@redhat.com>
23286 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23288 PR middle-end/32003
23289 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23291 (-fdump-tree-@var): Add to index and document how to come up
23292 with pass-specific option and dump file names.
23293 (-fdump-passes): Clarify where to look for output.
23295 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23297 PR middle-end/77445
23298 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23299 statistics of the analyzed path; allow threading for speed when
23300 any of BBs along the path are optimized for speed.
23302 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23304 PR middle-end/78468
23305 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23306 settings of the virtual registers.
23309 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23311 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23312 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23315 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23317 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23318 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23320 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23322 * config/s390/s390.md: Add missing comments with the expanded
23324 * config/s390/vector.md: Likewise.
23325 * config/s390/vx-builtins.md: Likewise.
23327 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23330 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23331 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23332 conditions on a single line.
23334 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23336 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23337 __S390_VX__ to __VX__.
23339 2017-02-01 Andrew Pinski <apinski@cavium.com>
23341 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23342 stmt_info to record_stmt_cost.
23343 (vect_get_known_peeling_cost): Pass stmt_info if known to
23345 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23346 cpu_vector_cost field into
23347 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23348 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23349 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23350 splitting of scalar_stmt_cost and vec_stmt_cost.
23351 (thunderx_vector_cost): Likewise.
23352 (cortexa57_vector_cost): LIkewise.
23353 (exynosm1_vector_cost): Likewise.
23354 (xgene1_vector_cost): Likewise.
23355 (thunderx2t99_vector_cost): Improve after the splitting of the two
23357 (aarch64_builtin_vectorization_cost): Update for the splitting of
23358 scalar_stmt_cost and vec_stmt_cost.
23360 2017-02-01 Torvald Riegel <triegel@redhat.com>
23361 Richard Henderson <rth@redhat.com>
23363 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23364 conditional on existance of a fast atomic load.
23365 * optabs-query.c (can_atomic_load_p): New function.
23366 * optabs-query.h (can_atomic_load_p): Declare it.
23367 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23368 no fast atomic load is available for the particular size of access.
23369 (expand_atomic_compare_and_swap): Likewise.
23370 (expand_atomic_load): Likewise.
23371 (expand_atomic_store): Likewise.
23372 (expand_atomic_fetch_op): Likewise.
23373 * testsuite/lib/target-supports.exp
23374 (check_effective_target_sync_int_128): Remove x86 because it provides
23375 no fast atomic load.
23376 (check_effective_target_sync_int_128_runtime): Likewise.
23378 2017-02-01 Richard Biener <rguenther@suse.de>
23380 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23381 (graphite_transform_loops): Provide opt-info for optimized nests.
23382 * tree-parloop.c (parallelize_loops): Provide opt-info for
23383 parallelized loops.
23385 2017-02-01 Richard Biener <rguenther@suse.de>
23387 PR middle-end/79315
23388 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23389 was not set before.
23391 2017-02-01 Richard Biener <rguenther@suse.de>
23393 PR tree-optimization/71824
23394 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23395 Verify the loops are valid in the merged SESE region.
23396 (scop_detection::can_represent_loop_1): Check analyzing the
23397 evolution of the number of iterations in the region succeeds.
23399 2017-01-31 Ian Lance Taylor <iant@golang.org>
23401 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23402 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23404 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23406 PR preprocessor/79210
23407 * input.c (get_substring_ranges_for_loc): Replace line_width
23408 assertion with error-handling.
23410 2017-01-31 Richard Biener <rguenther@suse.de>
23412 PR tree-optimization/77318
23413 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23414 (create_pw_aff_from_tree): Take loop parameter.
23415 (add_condition_to_pbb): Pass loop of the condition to
23416 create_pw_aff_from_tree.
23418 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23420 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23421 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23423 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23427 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23429 (convert_int_to_float128): Likewise.
23430 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23431 (convert_int_to_float128): Likewise.
23432 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23433 (UNSPEC_IEEE128_CONVERT): Likewise.
23434 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23435 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23436 Use local variables for IBM extended format.
23437 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23438 (fix_trunc<mode>si2_fprs): Likewise.
23439 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23440 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23441 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23442 to know that we can now have integers of all sizes in vector
23444 (fix<uns>_<mode>di2_hw): Likewise.
23445 (float<uns>_<mode>si2_hw): Likewise.
23446 (fix_<mode>si2_hw): Likewise.
23447 (fixuns_<mode>si2_hw): Likewise.
23448 (float<uns>_<mode>di2_hw): Likewise.
23449 (float_<mode>di2_hw): Likewise.
23450 (float_<mode>si2_hw): Likewise.
23451 (floatuns_<mode>di2_hw): Likewise.
23452 (floatuns_<mode>si2_hw): Likewise.
23453 (xscvqp<su>wz_<mode>): Delete, no longer used.
23454 (xscvqp<su>dz_<mode>): Likewise.
23455 (xscv<su>dqp_<mode>): Likewise.
23456 (ieee128_mfvsrd_64bit): Likewise.
23457 (ieee128_mfvsrd_32bit): Likewise.
23458 (ieee128_mfvsrwz): Likewise.
23459 (ieee128_mtvsrw): Likewise.
23460 (ieee128_mtvsrd_64bit): Likewise.
23461 (ieee128_mtvsrd_32bit): Likewise.
23463 2017-01-31 Martin Liska <mliska@suse.cz>
23466 * ipa-prop.c (ipa_free_all_node_params): Call release method
23467 instead of ~sumbol_summary to not to trigger double times
23470 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23472 PR tree-optimization/71691
23473 * bitmap.h (class auto_bitmap): New.
23474 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23475 is_maybe_undefined instead of ssa_undefined_value_p.
23477 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23479 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23480 __S390_ARCH_LEVEL__ to __ARCH__.
23482 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23484 PR tree-optimization/79267
23485 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23486 if should_remove_lhs_p is true.
23488 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23491 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23492 (add_alignment_attribute): New.
23493 (base_type_die): Add alignment attribute.
23494 (subrange_type_die): Likewise.
23495 (modified_type_die): Likewise.
23496 (gen_array_type_die): Likewise.
23497 (gen_descr_array_type_die: Likewise.
23498 (gen_enumeration_type_die): Likewise.
23499 (gen_subprogram_die): Likewise.
23500 (gen_variable_die): Likewise.
23501 (gen_field_die): Likewise.
23502 (gen_ptr_to_mbr_type_die): Likewise.
23503 (gen_struct_or_union_type_die): Likewise.
23504 (gen_subroutine_type_die): Likewise.
23505 (gen_typedef_die): Likewise.
23506 (base_type_cmp): Compare alignment attribute.
23508 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23511 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23512 (setb_unsigned) New pattern for setb with CCUNS.
23513 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23514 subfc./subfe sequence to avoid overflow problems. Generate a
23515 shorter sequence with cmpld/setb for power9.
23516 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23517 for generating subfc. instruction.
23518 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23519 now uses this instruction.
23521 2017-01-30 Ian Lance Taylor <iant@google.com>
23524 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23525 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23527 2017-01-30 Martin Sebor <msebor@redhat.com>
23529 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23530 Move constant to the right of a relational operator.
23531 (get_mpfr_format_length, format_character, format_string): Ditto.
23532 (should_warn_p, maybe_warn): Same.
23534 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23536 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23539 * asan.c (get_translation_unit_decl): Remove function.
23540 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23542 2017-01-30 Martin Liska <mliska@suse.cz>
23544 PR gcov-profile/79259
23545 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23546 -fprofile-generate.
23548 2017-01-30 Martin Liska <mliska@suse.cz>
23551 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23552 Initialize variables with NULL value.
23554 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23557 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23559 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23561 2017-01-30 Richard Biener <rguenther@suse.de>
23563 PR tree-optimization/79276
23564 * tree-vrp.c (process_assert_insertions): Properly adjust common
23565 when removing a duplicate.
23567 * gcc.dg/torture/pr79276.c: New testcase.
23569 2017-01-30 Richard Biener <rguenther@suse.de>
23571 PR tree-optimization/79256
23572 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23573 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23575 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23577 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23580 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23581 ("*r<noxa>sbg_<mode>_sll_bitmask")
23582 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23583 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23584 Use contiguous_bitmask_nowrap_operand.
23586 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23589 * config/rs6000/altivec.h (vec_xl): Revise #define.
23590 (vec_xst): Likewise.
23592 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23594 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23596 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23598 PR rtl-optimization/79194
23599 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23600 traps before call to bypass_conditional_jumps.
23602 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23604 PR tree-optimization/71374
23605 * lra-constraints.c (check_conflict_input_operands): New.
23606 (match_reload): Use it.
23608 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23611 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23612 account to calculate conflict_set.
23614 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23616 PR rtl-optimization/78559
23617 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23618 other_insn in combine.
23620 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23622 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23623 uint16_type_node for BT_UINT16.
23625 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23627 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23628 "RTL Tests" to menu.
23629 (GIMPLE Tests): New node.
23630 (RTL Tests): New node.
23632 2017-01-27 Richard Biener <rguenther@suse.de>
23634 PR tree-optimization/79245
23635 * tree-loop-distribution.c (distribute_loop): Apply cost
23636 modeling also to detected patterns.
23638 2017-01-27 Richard Biener <rguenther@suse.de>
23640 PR tree-optimization/71433
23641 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23642 (compare_assert_loc): New function.
23643 (process_assert_insertions): Sort and optimize assert locations
23644 to remove duplicates and push down identical assertions on
23645 edges to their destination block.
23647 2017-01-27 Richard Biener <rguenther@suse.de>
23649 PR tree-optimization/79244
23650 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23651 out SSA names even if abnormal.
23653 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23655 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23656 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23657 instead of MPFR_RNDN.
23659 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23662 * arm.c (arm_option_override): Don't call build_target_option_node
23663 until after doing all option overrides.
23664 (arm_valid_target_attribute_tree): Likewise.
23666 2017-01-27 Martin Liska <mliska@suse.cz>
23668 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23669 for {cd}tors and C++ {cd}tors.
23671 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23673 * config/s390/s390.md ("*setmem_long_and")
23674 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23676 2017-01-26 Martin Sebor <msebor@redhat.com>
23678 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23681 2017-01-26 Martin Sebor <msebor@redhat.com>
23683 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23684 HAVE_DFmode before using XFmode or DFmode.
23685 (parse_directive): Avoid using the z length modifier to avoid
23686 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23688 PR middle-end/78703
23689 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23690 to accept adjustment as an array.
23691 (get_int_range): New function.
23692 (struct directive): Make width and prec arrays.
23693 (directive::set_width, directive::set_precision): Call get_int_range.
23694 (format_integer, format_floating): Handle width and precision ranges.
23695 (format_string, parse_directive): Same.
23697 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23700 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23701 comdat_type_p set, just clone them, but keep the children in the
23705 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23706 which have direct callers with -fvar-tracking-assignments enabled
23708 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23709 inside of type units.
23711 2017-01-26 Martin Sebor <msebor@redhat.com>
23713 PR middle-end/78703
23714 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23716 (struct format_result): Replace number_chars, number_chars_min,
23717 and number_chars_max with a single member of struct result_range.
23719 (format_result::operator+=): Adjust.
23720 (struct fmtresult): Remove bounded. Handle likely and unlikely
23722 (fmtresult::adjust_for_width_or_precision): New function.
23723 (fmtresult:type_max_digits): New function.
23724 (bytes_remaining): Handle likely and unlikely counters.
23725 (min_bytes_remaining): Remove.
23726 (format_percent): Simplify.
23727 (format_integer, format_floating): Set likely and unlikely counters.
23728 (get_string_length, format_character, format_string): Same.
23729 (format_plain, should_warn_p): New function.
23730 (maybe_warn): Call should_warn_p. Update diagnostic messages
23731 and handle those for all directives, including plain strings.
23732 (format_directive): Handle likely and unlikely counters.
23733 Remove unnecessary quoting from diagnostics. Add an informational
23735 (add_bytes): Remove.
23736 (pass_sprintf_length::compute_format_length): Simplify.
23737 (try_substitute_return_value): Handle likely and unlikely counters.
23739 2017-01-26 Carl Love <cel@us.ibm.com>
23741 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23742 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23744 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23747 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23748 endianess for subregs into account.
23749 * lra-constraints.c (lra_constraints): Do risky transformations
23750 always on the first iteration.
23751 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23752 last_call_used_reg_set.
23753 (process_bb_lives): Define and use last_call_used_reg_set.
23754 * lra.c (lra): Always continue after lra_constraints on the first
23757 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23759 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23761 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23763 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23765 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23766 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23767 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23768 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23769 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23770 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23771 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23772 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23773 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23775 2017-01-26 Marek Polacek <polacek@redhat.com>
23778 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23779 for the third operand.
23781 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23783 PR middle-end/79236
23784 * omp-low.c (struct omp_context): Add simt_stmt field.
23785 (scan_omp_for): Return omp_context *.
23786 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23787 context to the _simt_ SIMD stmt.
23788 (lower_omp_for): For combined SIMD with sibling _simt_
23789 SIMD, make sure to use the same decls in _looptemp_
23790 clauses as in the sibling.
23792 2017-01-26 David Sherwood <david.sherwood@arm.com>
23794 PR middle-end/79212
23795 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23798 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23801 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23802 emit fld b; fld a; if possible.
23804 * brig-builtins.def: Update copyright years.
23805 * config/arm/arm_acle_builtins.def: Update copyright years.
23807 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23810 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23811 constraint instead of o for the stxsd instruction.
23813 2017-01-25 Carl Love <cel@us.ibm.com>
23815 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23816 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23818 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23820 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23822 2017-01-25 Richard Biener <rguenther@suse.de>
23824 PR tree-optimization/69264
23825 * target.def (vector_alignment_reachable): Improve documentation.
23826 * doc/tm.texi: Regenerate.
23827 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23829 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23830 earlier changes with respect to TYPE_USER_ALIGN.
23831 (vector_alignment_reachable_p): Likewise. Improve dumping.
23833 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23836 * config/arm/arm.md (xordi3): Force constant operand into a register
23839 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23841 * doc/invoke.texi (-fstore-merging): Correct default optimization
23842 levels at which it is enabled.
23843 (-O): Move -fstore-merging from list to...
23846 2017-01-25 Richard Biener <rguenther@suse.de>
23849 * omp-expand.c: Include debug.h.
23850 (expand_omp_taskreg): Make sure to generate early debug before
23851 outlining anything from a function.
23852 (expand_omp_target): Likewise.
23853 (grid_expand_target_grid_body): Likewise.
23855 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23858 * asan.c (get_translation_unit_decl): New function.
23859 (asan_add_global): Extract modules file name from globals
23860 TRANSLATION_UNIT_DECL name.
23862 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23865 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23866 for long calls with APCS frame and VFP.
23868 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23870 * cfg.c (original_copy_tables_initialized_p): New function.
23871 * cfg.h (original_copy_tables_initialized_p): New decl.
23872 * cfgrtl.c (relink_block_chain): Guard the call to
23873 free_original_copy_tables with a call to
23874 original_copy_tables_initialized_p.
23875 * cgraph.h (symtab_node::native_rtl_p): New decl.
23876 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23877 (symtab_node::needed_p): Don't assert for early assembly output
23878 for __RTL functions.
23879 (cgraph_node::finalize_function): Set "force_output" for __RTL
23881 (cgraph_node::analyze): Bail out early for __RTL functions.
23882 (analyze_functions): Update assertion to support __RTL functions.
23883 (cgraph_node::expand): Bail out early for __RTL functions.
23884 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23886 * function.h (struct function): Update comment for field
23888 * gimple-expr.c: Include "tree-pass.h".
23889 (gimple_has_body_p): Return false for __RTL functions.
23890 * Makefile.in (OBJS): Add run-rtl-passes.o.
23891 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23893 (gcc::pass_manager::get_clean_slate): New accessor.
23894 * passes.c: Include "insn-addr.h".
23895 (should_skip_pass_p): Add logging. Update logic for running
23896 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23897 property-provider override so it is only done for gimple passes.
23899 (skip_pass): New function.
23900 (execute_one_pass): Call skip_pass when skipping passes.
23901 * read-md.c (md_reader::read_char): Support filtering
23902 the input to a subset of line numbers.
23903 (md_reader::md_reader): Initialize fields
23904 m_first_line and m_last_line.
23905 (md_reader::read_file_fragment): New function.
23906 * read-md.h (md_reader::read_file_fragment): New decl.
23907 (md_reader::m_first_line): New field.
23908 (md_reader::m_last_line): New field.
23909 * read-rtl-function.c (function_reader::create_function): Only
23910 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23911 curr_properties. Set DECL_INITIAL to a dummy block.
23912 (read_rtl_function_body_from_file_range): New function.
23913 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23915 * run-rtl-passes.c: New file.
23916 * run-rtl-passes.h: New file.
23918 2017-01-24 Jeff Law <law@redhat.com>
23920 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23923 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23925 PR tree-optimization/79159
23926 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23927 (record_nonwrapping_iv): Improve boundary using above function if no
23928 value range information.
23930 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23931 Martin Jambor <mjambor@suse.cz>
23933 * brig-builtins.def: New file.
23934 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23935 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23936 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23937 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23938 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23939 * builtin-types.def (BT_INT8): New.
23940 (BT_INT16): Likewise.
23941 (BT_UINT8): Likewise.
23942 (BT_UINT16): Likewise.
23943 (BT_FN_ULONG): Likewise.
23944 (BT_FN_UINT_INT): Likewise.
23945 (BT_FN_UINT_ULONG): Likewise.
23946 (BT_FN_UINT_LONG): Likewise.
23947 (BT_FN_UINT_PTR): Likewise.
23948 (BT_FN_ULONG_PTR): Likewise.
23949 (BT_FN_INT8_FLOAT): Likewise.
23950 (BT_FN_INT16_FLOAT): Likewise.
23951 (BT_FN_UINT32_FLOAT): Likewise.
23952 (BT_FN_UINT16_FLOAT): Likewise.
23953 (BT_FN_UINT8_FLOAT): Likewise.
23954 (BT_FN_UINT64_FLOAT): Likewise.
23955 (BT_FN_UINT16_UINT32): Likewise.
23956 (BT_FN_UINT32_UINT16): Likewise.
23957 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23958 (BT_FN_INT_PTR_INT): Likewise.
23959 (BT_FN_UINT_PTR_UINT): Likewise.
23960 (BT_FN_LONG_PTR_LONG): Likewise.
23961 (BT_FN_ULONG_PTR_ULONG): Likewise.
23962 (BT_FN_VOID_UINT64_UINT64): Likewise.
23963 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23964 (BT_FN_INT8_INT8_INT8): Likewise.
23965 (BT_FN_INT16_INT16_INT16): Likewise.
23966 (BT_FN_INT_INT_INT): Likewise.
23967 (BT_FN_UINT_FLOAT_UINT): Likewise.
23968 (BT_FN_FLOAT_UINT_UINT): Likewise.
23969 (BT_FN_ULONG_UINT_UINT): Likewise.
23970 (BT_FN_ULONG_UINT_PTR): Likewise.
23971 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23972 (BT_FN_UINT_UINT_UINT): Likewise.
23973 (BT_FN_VOID_UINT_PTR): Likewise.
23974 (BT_FN_UINT_UINT_PTR: Likewise.
23975 (BT_FN_UINT32_UINT64_PTR): Likewise.
23976 (BT_FN_INT_INT_UINT_UINT): Likewise.
23977 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23978 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23979 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23980 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23981 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23982 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23983 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23984 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23985 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23986 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23987 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23988 * doc/frontends.texi: List BRIG FE.
23989 * doc/install.texi (Testing): Add BRIG tesring requirements.
23990 * doc/invoke.texi (Overall Options): Mention BRIG.
23991 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23993 2017-01-24 Richard Biener <rguenther@suse.de>
23995 PR translation/79208
23996 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23998 2017-01-24 Martin Jambor <mjambor@suse.cz>
24001 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24002 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24003 and known_contexts.
24005 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24007 PR middle-end/79123
24008 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24009 casts from signed to unsigned really don't have a range.
24011 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24013 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24014 GMP_RNDx for compatiblity.
24016 2017-01-24 Martin Liska <mliska@suse.cz>
24019 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24020 that would prevent us to call alloca with -1 as argument.
24022 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24024 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24025 Avoid -Wformat-security warning.
24027 2017-01-23 Andrew Pinski <apinski@cavium.com>
24029 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24032 2017-01-23 Martin Sebor <msebor@redhat.com>
24034 PR middle-end/78703
24035 * gimple-ssa-sprintf.c (warn_level): New global.
24036 (format_integer): Use it here and throughout the rest of the file.
24037 Use the same switch to compute sign as base.
24038 (maybe_warn): New function.
24039 (format_directive): Factor out warnings into maybe_warn.
24040 Add debugging output. Use warn_level.
24041 (add_bytes): Use warn_level.
24042 (pass_sprintf_length::compute_format_length): Add debugging output.
24043 (try_substitute_return_value): Same.
24044 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24046 PR middle-end/78703
24047 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24048 (struct fmtresult, format_integer, format_floating): Adjust.
24049 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24050 (get_string_length, format_string,format_directive): Same.
24051 (pass_sprintf_length::compute_format_length): Same.
24052 (try_substitute_return_value): Simplify slightly.
24054 PR middle-end/78703
24055 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24056 (fmtresult::operator+=): Outlined.
24057 (struct fmtresult): Add ctors.
24058 (struct conversion_spec): Rename...
24059 (struct directive): ...to this. Add and remove data members.
24060 (directive::set_width, directive::set_precision): New functions.
24061 (format_percent): Use fmtresult ctor.
24062 (get_width_and_precision): Remove.
24063 (format_integer): Make naming changes. Avoid computing width and
24065 (format_floating): Same. Adjust indentation.
24066 (format_character, format_none): New functions.
24067 (format_string): Moved character handling to format_character.
24068 (format_directive): Remove arguments, change return type.
24069 (parse_directive): New function.
24070 (pass_sprintf_length::compute_format_length): Move directive
24071 parsing to parse_directive.
24073 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24075 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24076 (assign_assembler_name_if_needed): ... this.
24077 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24078 (assign_assembler_name_if_needed): ... this.
24079 (free_lang_data_in_cgraph): Adjust callers.
24080 * cgraphunit.c (cgraph_node::analyze): Likewise.
24081 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24084 2017-01-23 Richard Biener <rguenther@suse.de>
24086 PR tree-optimization/79088
24087 PR tree-optimization/79188
24088 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24089 resetting loop bounds after last path deletion. Reset loop
24090 bounds of the target loop, make code match the comments.
24091 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24092 Make sure loops need no fixups.
24094 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24096 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24097 exponent support with double type for first argument.
24098 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24099 type returned by __builtin_vec_extract_sig,
24100 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24101 functions from "vector int" to "vector unsigned int" or from
24102 "vector long long int" to "vector unsigned long long int".
24103 Changed type returned by __builtin_vec_extract_exp,
24104 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24105 functions from "vector int" to "vector unsigned int" or from
24106 "vector long long int" to "vector unsigned long long int".
24107 Changed return type of __builtin_vec_test_data_class,
24108 __builtin_vec_test_data_class_sp, and
24109 __builtin_vec_test_data_class_dp from "vector int" to
24110 "vector bool int" or from "vector long long int" to "vector bool
24111 long long int" and changed second argument type from "unsigned
24112 int" to "int". Added new overloaded function forms "vector float
24113 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24114 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24115 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24116 double, vector unsigned long long int)" and "vector double
24117 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24118 long int)". Changed return type of
24119 __builtin_scalar_test_data_class and
24120 __builtin_scalar_test_data_class_sp and
24121 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24122 int" and changed second argument from "unsigned int" to "int".
24123 Changed type returned by __builtin_scalar_test_neg,
24124 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24125 from "int" to "bool int". Added new overloaded function form
24126 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24127 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24128 exponent double-precision with floating point first argument.
24129 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24130 documentation of scalar_test_data_class, scalar_test_neg,
24131 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24132 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24133 vec_test_data_class built-in functions to reflect refinements in
24134 their type signatures.
24136 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24138 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24140 (aarch64_elf_asm_destructor): Likewise.
24142 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24144 PR rtl-optimization/78634
24145 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24146 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24147 * ifcvt.c (noce_try_cmove): Add missing cost check.
24149 PR rtl-optimization/71724
24150 * combine.c (if_then_else_cond): Look for situations where it is
24151 beneficial to undo the work of one of the recursive calls.
24153 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24155 PR tree-optimization/70754
24156 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24157 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24158 combined stmt before it if not NULL.
24159 (combine_chains): Process refs reversely and compute dominance point
24162 2017-01-23 Martin Liska <mliska@suse.cz>
24164 PR tree-optimization/79196
24165 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24166 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24168 (strlen_optimize_stmt): Call the renamed function.
24170 2017-01-23 Michael Matz <matz@suse.de>
24172 PR tree-optimization/78384
24173 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24175 2017-01-23 Richard Biener <rguenther@suse.de>
24177 PR tree-optimization/79186
24178 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24179 both incoming edges before moving an assert.
24181 2017-01-23 Martin Jambor <mjambor@suse.cz>
24183 * ipa-prop.c (load_from_param_1): Removed.
24184 (load_from_unmodified_param): Bits from load_from_param_1 put back
24186 (load_from_param): Removed.
24187 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24188 with stmt. Reverted back to use of load_from_unmodified_param.
24190 2017-01-23 Martin Jambor <mjambor@suse.cz>
24193 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24194 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24195 field a pointer to garbage collected vector, mark lattices and
24196 ipcp_orig_node with GTY((skip)).
24197 (ipa_get_param_count): Adjust to descriptors being a pointer.
24198 (ipa_get_param): Likewise.
24199 (ipa_get_type): Likewise.
24200 (ipa_get_param_move_cost): Likewise.
24201 (ipa_set_param_used): Likewise.
24202 (ipa_get_controlled_uses): Likewise.
24203 (ipa_set_controlled_uses): Likewise.
24204 (ipa_is_param_used): Likewise.
24205 (ipa_node_params_t): Move into garbage collector. New methods insert
24207 (ipa_node_params_sum): Annotate wth GTY(()).
24208 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24210 (ipa_load_from_parm_agg): Adjust declaration.
24211 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24212 * ipa-profile.c (ipa_profile): Likewise.
24213 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24214 (ipa_populate_param_decls): Make descriptors parameter garbage
24216 (ipa_dump_param): Adjust to descriptors being a pointer.
24217 (ipa_alloc_node_params): Likewise.
24218 (ipa_initialize_node_params): Likewise.
24219 (load_from_param_1): Make descriptors parameter garbage collected.
24220 (load_from_unmodified_param): Likewise.
24221 (load_from_param): Likewise.
24222 (ipa_load_from_parm_agg): Likewise.
24223 (ipa_node_params::~ipa_node_params): Removed.
24224 (ipa_free_all_node_params): Remove call to delete operator.
24225 (ipa_node_params_t::insert): New.
24226 (ipa_node_params_t::remove): Likewise.
24227 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24228 copy known_csts and known_contexts vectors.
24229 (ipa_read_node_info): Adjust to descriptors being a pointer.
24230 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24232 (ipcp_transform_function): Make descriptors variable garbage
24235 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24237 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24238 * config/i386/avx512dqintrin.h: Ditto.
24239 * config/i386/avx512fintrin.h: Ditto.
24240 * config/i386/i386.c: Handle new builtins.
24241 * config/i386/i386-builtin.def: Add new builtins.
24242 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24243 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24245 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24246 Martin Liska <mliska@suse.cz>
24248 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24249 * asan.c (asan_expand_poison_ifn): Support stores and use
24250 appropriate ASAN report function.
24251 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24252 * internal-fn.def (ASAN_POISON_USE): Declare.
24253 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24254 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24255 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24256 ASAN_POISON calls w/o LHS.
24257 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24258 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24259 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24260 * gimplify.c (asan_poison_variables): Add attribute
24261 use_after_scope_memory to variables that really needs to live
24263 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24264 having the attribute.
24266 2017-01-23 Martin Liska <mliska@suse.cz>
24268 * asan.c (create_asan_shadow_var): New function.
24269 (asan_expand_poison_ifn): Likewise.
24270 * asan.h (asan_expand_poison_ifn): New declaration.
24271 * internal-fn.c (expand_ASAN_POISON): Likewise.
24272 * internal-fn.def (ASAN_POISON): New builtin.
24273 * sanopt.c (pass_sanopt::execute): Expand
24274 asan_expand_poison_ifn.
24275 * tree-inline.c (copy_decl_for_dup_finish): Make function
24277 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24278 * tree-ssa.c (is_asan_mark_p): New function.
24279 (execute_update_addresses_taken): Rewrite local variables
24280 (identified just by use-after-scope as addressable) into SSA.
24282 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24284 * doc/install.texi (Specific): opensource.apple.com uses https
24285 now. Remove trailing slash.
24287 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24289 * README.Portability: Remove note on an Irix compatibility issue.
24291 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24293 * gcov.c (INCLUDE_ALGORITHM): Define.
24294 (INCLUDE_VECTOR): Define.
24295 No longer include <vector> and <algorithm> directly.
24297 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24299 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24301 * doc/invoke.texi (Code Gen Options): Ditto.
24303 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24306 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24308 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24310 rtl-optimization/79125
24311 * cprop.c (local_cprop_pass): Handle cases where we make an
24312 unconditional trap.
24314 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24318 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24319 read from, for big endian.
24321 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24323 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24324 register pauth builtins for LP64 only.
24326 2017-01-20 Marek Polacek <polacek@redhat.com>
24329 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24332 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24334 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24336 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24337 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24338 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24340 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24343 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24344 in big-endian mode when they are not a single duplicated value.
24346 2017-01-20 Richard Biener <rguenther@suse.de>
24348 * BASE-VER: Bump to 7.0.1.
24350 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24352 * omp-low.c (omplow_simd_context): New struct. Use it...
24353 (lower_rec_simd_input_clauses): ...here and...
24354 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24355 references to idx, lane, max_vf, is_simt.
24357 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24359 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24362 2017-01-20 Martin Jambor <mjambor@suse.cz>
24364 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24365 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24367 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24368 (GTFILES): Rename hsa.c to hsa-common.c.
24369 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24370 * hsa-dump.c: Likewise.
24371 * hsa-gen.c: Likewise.
24372 * hsa-regalloc.c: Likewise.
24373 * ipa-hsa.c: Likewise.
24374 * omp-expand.c: Likewise.
24375 * omp-low.c: Likewise.
24376 * toplev.c: Likewise.
24378 2017-01-20 Marek Polacek <polacek@redhat.com>
24381 * doc/invoke.texi: Document -Wduplicated-branches.
24382 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24383 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24384 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24385 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24386 return 0 only when not OEP_LEXICOGRAPHIC.
24387 (fold_build_cleanup_point_expr): Use the expression
24388 location when building CLEANUP_POINT_EXPR.
24389 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24390 * tree.c (add_expr): Handle error_mark_node.
24392 2017-01-20 Martin Liska <mliska@suse.cz>
24395 * tree-profile.c (init_ic_make_global_vars): Do not call
24397 (gimple_init_gcov_profiler): Likewise.
24399 2017-01-20 Martin Liska <mliska@suse.cz>
24402 * cgraph.h (maybe_create_reference): Remove argument and
24404 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24406 * ipa-cp.c (create_specialized_node): Likewise.
24407 * symtab.c (symtab_node::maybe_create_reference): Handle
24408 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24410 2017-01-20 Martin Liska <mliska@suse.cz>
24412 * read-rtl-function.c (function_reader::create_function): Use
24413 build_decl instread of build_decl_stat.
24415 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24417 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24418 * config/i386/avx512dqintrin.h: Ditto.
24419 * config/i386/avx512fintrin.h: Ditto.
24420 * config/i386/i386-builtin-types.def: Add new types.
24421 * config/i386/i386.c: Handle new types.
24422 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24423 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24424 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24425 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24426 (__builtin_ia32_kshiftridi): New.
24427 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24429 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24433 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24434 define to rs6000_init_stack_protect_guard.
24435 (rs6000_init_stack_protect_guard): New function.
24437 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24438 Yunqiang Su <yunqiang.su@imgtec.com>
24440 * config.gcc (supported_defaults): Add madd4.
24441 (with_madd4): Add validation.
24442 (all_defaults): Add madd4.
24443 * config/mips/mips.opt (mmadd4): New option.
24444 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24446 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24448 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24449 (ISA_HAS_FUSED_MADD4): Likewise.
24450 * doc/invoke.texi (-mmadd4): Document the new option.
24451 * doc/install.texi (--with-madd4): Document the new option.
24453 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24455 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24456 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24457 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24458 (aarch64_init_pauth_hint_builtins): New.
24459 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24460 (aarch64_expand_builtin): Expand new builtins.
24462 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24464 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24465 * combine-stack-adj.c (no_unhandled_cfa): Handle
24466 REG_CFA_TOGGLE_RA_MANGLE.
24467 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24468 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24469 info for return address signing.
24470 (aarch64_expand_epilogue): Likewise.
24472 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24474 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24475 * config/aarch64/aarch64-protos.h
24476 (aarch64_return_address_signing_enabled): New declaration.
24477 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24479 (aarch64_expand_prologue): Sign return address before it's pushed onto
24481 (aarch64_expand_epilogue): Authenticate return address fetched from
24483 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24484 (aarch64_attributes): New function attributes for "sign-return-address".
24485 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24486 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24487 ("*do_return"): Generate combined instructions according to key index.
24488 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24489 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24491 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24492 * config/aarch64/aarch64.opt (msign-return-address=): New.
24493 * doc/extend.texi (AArch64 Function Attributes): Documents
24494 "sign-return-address=".
24495 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24497 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24499 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24500 overall option summary.
24502 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24504 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24505 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24506 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24507 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24509 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24511 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24512 -mpower9-minmax by default for -mcpu=power9.
24513 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24514 128-bit floating point.
24516 2017-01-20 Alan Modra <amodra@gmail.com>
24518 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24519 optimizing for size.
24521 2017-01-20 Alan Modra <amodra@gmail.com>
24524 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24525 for strcmp and strncmp from corresponding builtin decl.
24527 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24529 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24530 instead of i386/rtems-64.h.
24531 * config/i386/rtems-64.h: Remove.
24533 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24537 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24539 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24541 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24543 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24544 Change int to HOST_WIDE_INT.
24545 * config/aarch64/aarch64-protos.h
24546 (aarch64_simd_gen_const_vector_dup): Likewise.
24547 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24549 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24551 * langhooks-def.h (lhd_type_for_size): New decl.
24552 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24553 * langhooks.c (lhd_type_for_size): New function, taken from
24556 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24558 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24559 define_bypass for CR latency.
24560 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24561 (power9-alu2): Add define_bypass for CR latency.
24563 (power9-mul): Update insn latency.
24564 (power9-mul-compare): Update insn latency, bypass latency and remove
24567 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24569 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24571 * config/aarch64/aarch64.md
24572 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24573 aarch64_nopcrelative_literal_loads.
24574 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24576 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24578 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24579 TARGET_LOONGSON_3A.
24580 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24582 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24585 * config.gcc (supported_defaults): Add lxc1-sxc1.
24586 (with_lxc1_sxc1): Add validation.
24587 (all_defaults): Add lxc1-sxc1.
24588 * config/mips/mips.opt (mlxc1-sxc1): New option.
24589 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24591 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24592 __mips_no_lxc1_sxc1.
24593 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24594 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24595 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24597 2017-01-19 Richard Biener <rguenther@suse.de>
24599 PR tree-optimization/72488
24600 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24601 sure to restore SSA info.
24602 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24604 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24606 PR rtl-optimization/79121
24607 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24608 of the inner type when shifting an extended value.
24610 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24613 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24614 interposable aliases.
24616 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24619 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24620 Use the evmergelohi instruction.
24621 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24622 (mov_sitf_e500_subreg8_2_be): Likewise.
24623 (mov_sitf_e500_subreg12_2_le): Likewise.
24624 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24625 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24626 (mov_sitf_e500_subreg8_2_le): Likewise.
24627 (mov_sitf_e500_subreg12_2_be): Likewise.
24629 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24631 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24632 attribute from vecsimple to vecperm.
24633 (altivec_vbpermq2): Likewise.
24635 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24638 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24640 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24641 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24642 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24643 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24644 case where N arg is SIZE_MAX.
24645 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24646 (cmpstrsi): Add pattern.
24648 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24650 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24651 __builtin_vec_revb builtins.
24652 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24653 built-in functions to support generation of the ISA 3.0 XXBR<x>
24654 vector byte reverse instructions.
24655 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24656 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24657 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24658 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24659 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24660 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24661 (P9V_BUILTIN_VEC_REVB): Likewise.
24662 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24663 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24664 (p9_xxbrq_v16qi): Likewise.
24665 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24666 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24667 (p9_xxbrh_v8hi): Likewise.
24668 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24669 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24670 vec_revb built-in functions.
24672 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24674 PR rtl-optimization/78952
24675 * config/i386/i386.md (any_extract): New code iterator.
24676 (*insvqi_2): Use any_extract for source operand.
24677 (*insvqi_3): Use any_shiftrt for source operand.
24679 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24681 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24683 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24685 2017-01-18 Matthias Klose <doko@ubuntu.com>
24687 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24689 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24691 * config/rs6000/altivec.h (vec_bperm): Change #define.
24692 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24693 (altivec_vbpermq2): New define_insn.
24694 (altivec_vbpermd): Likewise.
24695 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24696 function interface.
24697 (VBPERMD): Likewise.
24698 (VBPERM): New polymorphic function interface.
24699 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24700 Add entries for P9V_BUILTIN_VEC_VBPERM.
24701 * doc/extend.texi: Add interfaces for vec_bperm.
24703 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24705 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24706 first letter of error messages.
24707 (s390_resolve_overloaded_builtin): Likewise.
24708 * config/s390/s390.c (s390_expand_builtin): Likewise.
24709 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24710 (s390_valid_target_attribute_inner_p): Likewise.
24711 * config/s390/s390.md ("tabort"): Likewise.
24713 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24715 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24716 (ISA_AVOID_DIV_HILO): New macro.
24717 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24718 (ISA_HAS_DDIV): Likewise.
24720 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24722 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24724 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24726 * doc/invoke.texi (fabi-version): Spelling fix.
24728 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24731 * doc/invoke.texi (fabi-version): Mention mangling fix for
24734 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24737 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24739 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24742 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24745 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24747 (stack_protect_test): Ditto.
24748 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24749 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24751 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24752 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24753 -mstack-protector-guard-offset=.
24754 (RS/6000 and PowerPC Options): Ditto.
24756 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24758 * config/i386/i386.h (MASK_CLASS_P): New define.
24759 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24760 there are no registers from different register sets also when
24761 mask registers are used. Update function comment.
24762 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24763 to (*k/*r) and (*k/*km) alternatives.
24765 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24767 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24768 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24769 (EH_RETURN_HANDLER_RTX): New define.
24770 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24771 Force frame pointer in EH return functions.
24772 (aarch64_expand_epilogue): Add barrier for eh_return.
24773 (aarch64_final_eh_return_addr): Remove.
24774 (aarch64_eh_return_handler_rtx): New function.
24775 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24777 (aarch64_eh_return_handler_rtx): New prototype.
24779 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24781 * config/rs6000/altivec.h (vec_rlmi): New #define.
24782 (vec_vrlnm): Likewise.
24783 (vec_rlnm): Likewise.
24784 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24785 (UNSPEC_VRLNM): Likewise.
24786 (VIlong): New mode iterator.
24787 (altivec_vrl<VI_char>mi): New define_insn.
24788 (altivec_vrl<VI_char>nm): Likewise.
24789 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24791 (VRLDNM): Likewise.
24792 (RLNM): New polymorphic function entry.
24793 (VRLWMI): New monomorphic function entry.
24794 (VRLDMI): Likewise.
24795 (RLMI): New polymorphic function entry.
24796 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24797 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24798 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24801 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24804 * dwarf2out.c (field_byte_offset): Restore the
24805 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24806 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24807 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24810 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24813 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24815 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24818 * dwarf2out.c (add_data_member_location_attribute): For constant
24819 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24820 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24821 DW_AT_byte_size attributes.
24823 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24825 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24826 after forcing to constant memory when the code model is medium.
24828 2017-01-17 Julia Koval <julia.koval@intel.com>
24831 * config/i386/avx512fintrin.h
24832 (_mm512_i32gather_ps): Change __addr type to void const*.
24833 (_mm512_mask_i32gather_ps): Ditto.
24834 (_mm512_i32gather_pd): Ditto.
24835 (_mm512_mask_i32gather_pd): Ditto.
24836 (_mm512_i64gather_ps): Ditto.
24837 (_mm512_mask_i64gather_ps): Ditto.
24838 (_mm512_i64gather_pd): Ditto.
24839 (_mm512_mask_i64gather_pd): Ditto.
24840 (_mm512_i32gather_epi32): Ditto.
24841 (_mm512_mask_i32gather_epi32): Ditto.
24842 (_mm512_i32gather_epi64): Ditto.
24843 (_mm512_mask_i32gather_epi64): Ditto.
24844 (_mm512_i64gather_epi32): Ditto.
24845 (_mm512_mask_i64gather_epi32): Ditto.
24846 (_mm512_i64gather_epi64): Ditto.
24847 (_mm512_mask_i64gather_epi64): Ditto.
24848 (_mm512_i32scatter_ps): Change __addr type to void*.
24849 (_mm512_mask_i32scatter_ps): Ditto.
24850 (_mm512_i32scatter_pd): Ditto.
24851 (_mm512_mask_i32scatter_pd): Ditto.
24852 (_mm512_i64scatter_ps): Ditto.
24853 (_mm512_mask_i64scatter_ps): Ditto.
24854 (_mm512_i64scatter_pd): Ditto.
24855 (_mm512_mask_i64scatter_pd): Ditto.
24856 (_mm512_i32scatter_epi32): Ditto.
24857 (_mm512_mask_i32scatter_epi32): Ditto.
24858 (_mm512_i32scatter_epi64): Ditto.
24859 (_mm512_mask_i32scatter_epi64): Ditto.
24860 (_mm512_i64scatter_epi32): Ditto.
24861 (_mm512_mask_i64scatter_epi32): Ditto.
24862 (_mm512_i64scatter_epi64): Ditto.
24863 (_mm512_mask_i64scatter_epi64): Ditto.
24864 * config/i386/avx512pfintrin.h
24865 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24866 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24867 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24868 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24869 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24870 (_mm512_prefetch_i32scatter_ps): Ditto.
24871 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24872 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24873 (_mm512_prefetch_i64scatter_pd): Ditto.
24874 (_mm512_prefetch_i64scatter_ps): Ditto.
24875 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24876 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24877 * config/i386/avx512vlintrin.h
24878 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24879 (_mm_mmask_i32gather_ps): Ditto.
24880 (_mm256_mmask_i32gather_pd): Ditto.
24881 (_mm_mmask_i32gather_pd): Ditto.
24882 (_mm256_mmask_i64gather_ps): Ditto.
24883 (_mm_mmask_i64gather_ps): Ditto.
24884 (_mm256_mmask_i64gather_pd): Ditto.
24885 (_mm_mmask_i64gather_pd): Ditto.
24886 (_mm256_mmask_i32gather_epi32): Ditto.
24887 (_mm_mmask_i32gather_epi32): Ditto.
24888 (_mm256_mmask_i32gather_epi64): Ditto.
24889 (_mm_mmask_i32gather_epi64): Ditto.
24890 (_mm256_mmask_i64gather_epi32): Ditto.
24891 (_mm_mmask_i64gather_epi32): Ditto.
24892 (_mm256_mmask_i64gather_epi64): Ditto.
24893 (_mm_mmask_i64gather_epi64): Ditto.
24894 (_mm256_i32scatter_ps): Change __addr type to void*.
24895 (_mm256_mask_i32scatter_ps): Ditto.
24896 (_mm_i32scatter_ps): Ditto.
24897 (_mm_mask_i32scatter_ps): Ditto.
24898 (_mm256_i32scatter_pd): Ditto.
24899 (_mm256_mask_i32scatter_pd): Ditto.
24900 (_mm_i32scatter_pd): Ditto.
24901 (_mm_mask_i32scatter_pd): Ditto.
24902 (_mm256_i64scatter_ps): Ditto.
24903 (_mm256_mask_i64scatter_ps): Ditto.
24904 (_mm_i64scatter_ps): Ditto.
24905 (_mm_mask_i64scatter_ps): Ditto.
24906 (_mm256_i64scatter_pd): Ditto.
24907 (_mm256_mask_i64scatter_pd): Ditto.
24908 (_mm_i64scatter_pd): Ditto.
24909 (_mm_mask_i64scatter_pd): Ditto.
24910 (_mm256_i32scatter_epi32): Ditto.
24911 (_mm256_mask_i32scatter_epi32): Ditto.
24912 (_mm_i32scatter_epi32): Ditto.
24913 (_mm_mask_i32scatter_epi32): Ditto.
24914 (_mm256_i32scatter_epi64): Ditto.
24915 (_mm256_mask_i32scatter_epi64): Ditto.
24916 (_mm_i32scatter_epi64): Ditto.
24917 (_mm_mask_i32scatter_epi64): Ditto.
24918 (_mm256_i64scatter_epi32): Ditto.
24919 (_mm256_mask_i64scatter_epi32): Ditto.
24920 (_mm_i64scatter_epi32): Ditto.
24921 (_mm_mask_i64scatter_epi32): Ditto.
24922 (_mm256_i64scatter_epi64): Ditto.
24923 (_mm256_mask_i64scatter_epi64): Ditto.
24924 (_mm_i64scatter_epi64): Ditto.
24925 (_mm_mask_i64scatter_epi64): Ditto.
24926 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24927 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24928 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24929 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24930 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24931 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24932 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24933 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24934 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24935 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24936 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24937 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24938 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24939 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24940 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24941 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24942 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24943 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24944 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24945 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24946 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24947 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24948 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24949 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24950 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24951 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24952 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24953 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24954 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24955 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24956 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24957 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24958 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24959 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24960 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24961 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24962 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24963 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24964 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24965 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24966 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24967 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24968 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24969 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24970 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24971 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24972 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24973 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24974 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24975 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24976 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24977 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24978 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24979 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24980 definitions accordingly.
24982 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24983 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24986 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24989 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24992 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24993 hard regs for inner regno.
24995 2017-01-17 Martin Liska <mliska@suse.cz>
24998 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24999 assumption and add comment.
25001 2017-01-17 Nathan Sidwell <nathan@acm.org>
25003 * ipa-visibility.c (localize_node): New function, broken out of ...
25004 (function_and_variable_visibility): ... here. Call it.
25006 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25008 PR middle-end/77445
25009 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25010 correctly set frequency of oudgoing edge.
25011 (duplicate_thread_path): Fix profile updating.
25013 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25016 * configure.ac: Add GCC_BASE_VER.
25017 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25018 version from BASE-VER file.
25019 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25020 (gcc.o): Depend on $(BASEVER).
25021 * common.opt (dumpfullversion): New option.
25022 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25023 * doc/invoke.texi: Document -dumpfullversion.
25024 * doc/install.texi: Document --with-gcc-major-version-only.
25025 * configure: Regenerated.
25027 2017-01-17 Richard Biener <rguenther@suse.de>
25029 PR tree-optimization/71433
25030 * tree-vrp.c (register_new_assert_for): Merge same asserts
25031 on all incoming edges.
25032 (process_assert_insertions_for): Handle insertions at the
25035 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25037 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25038 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25040 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25043 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25046 2017-01-17 Alan Modra <amodra@gmail.com>
25049 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25050 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25051 symbolic stack limit when pic.
25053 2017-01-16 Martin Sebor <msebor@redhat.com>
25055 PR tree-optimization/78608
25056 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25058 2017-01-16 Jeff Law <law@redhat.com>
25061 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25062 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25063 for several include directories that may be relative to sysroot.
25064 * config/i386/x-mingw32 (gplus_includedir): Define.
25065 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25066 (native_system_includedir): Likewise.
25067 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25068 override if TARGET_SYSTEM_ROOT is defined.
25069 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25071 PR tree-optimization/79090
25072 PR tree-optimization/33562
25073 PR tree-optimization/61912
25074 PR tree-optimization/77485
25075 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25076 and computed trims into the dump file.
25078 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25080 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25082 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25085 * gimplify.c (gimplify_init_constructor): If want_value and
25086 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25090 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25091 sequence. Formatting fixes.
25092 (doloop_optimize): Formatting fixes.
25095 * gcc.c (debug_level_greater_than_spec_func): New function.
25096 (static_spec_functions): Add debug-level-gt spec function.
25097 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25099 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25100 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25101 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25102 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25103 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25104 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25106 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25108 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25109 QImode fixups to general and mask registers only.
25111 2017-01-16 Carl Love <cel@us.ibm.com>
25113 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25114 for built-in functions
25115 vector signed char vec_nabs (vector signed char)
25116 vector signed short vec_nabs (vector signed short)
25117 vector signed int vec_nabs (vector signed int)
25118 vector signed long long vec_nabs (vector signed long long)
25119 vector float vec_nabs (vector float)
25120 vector double vec_nabs (vector double)
25121 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25123 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25124 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25125 * doc/extend.texi: Update the documentation file for the new built-in
25128 2017-01-16 Martin Sebor <msebor@redhat.com>
25130 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25133 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25135 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25136 UNSPEC_VSX__XXSPLTD to require special splat handling.
25138 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25141 * system.h: Poison strndup.
25143 2017-01-16 Alan Modra <amodra@gmail.com>
25146 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25149 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25151 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25153 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25155 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25156 call recog here. Assert that INSN_CODE (insn) is non-negative.
25158 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25161 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25163 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25164 in the currently scheduled RTL fragment.
25166 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25168 PR rtl-optimization/78751
25169 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25172 2017-01-14 Jeff Law <law@redhat.com>
25174 PR tree-optimization/79090
25175 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25176 variable length stores.
25177 (compute_trims): Delete dead assignment to *trim_tail.
25178 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25181 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25183 PR rtl-optimization/78626
25184 PR rtl-optimization/78727
25185 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25186 of a block, and split such blocks after everything else is finished.
25188 2017-01-14 Alan Modra <amodra@gmail.com>
25191 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25192 target legitimate_combined_insn.
25193 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25194 (rs6000_legitimate_combined_insn): New function.
25195 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25197 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25198 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25199 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25201 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25203 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25205 2017-01-13 Jeff Law <law@redhat.com>
25207 PR tree-optimization/33562
25208 PR tree-optimization/61912
25209 PR tree-optimization/77485
25210 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25212 (delete_dead_assignment): Likewise.
25213 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25214 statement to delete_dead_call and delete_dead_assignment.
25216 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25219 * substring-locations.c (format_warning_va): Strengthen case 1 so
25220 that both endpoints of the substring must be within the format
25221 range for just the substring to be printed.
25223 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25225 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25226 * config/i386/i386.c (ix86_target_string): Add missing options
25227 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25228 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25229 flags_other and ix86_target_other to flags2_other. Display unknown
25231 (ix86_valid_target_attribute_inner_p): Add missing options and
25232 reorder options by implied ISAs, as in ix86_target_string.
25234 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25236 * hash-table.h (hash_table::too_empty_p): New function.
25237 (hash_table::expand): Use it.
25238 (hash_table::traverse): Likewise.
25239 (hash_table::empty_slot): Use sizeof (value_type) instead of
25240 sizeof (PTR) to convert bytes to elements. Shrink the table
25241 if the current size is excessive for the current number of
25244 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25246 * ira-costs.c (record_reg_classes): Break from the inner loop
25247 early once alt_fail is known to be true. Update outer loop
25248 handling accordingly.
25250 2017-01-13 Jeff Law <law@redhat.com>
25252 * tree-ssa-dse.c (decrement_count): New function.
25253 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25254 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25255 when we know the partially dead statement is a mem* function.
25257 PR tree-optimization/61912
25258 PR tree-optimization/77485
25259 * tree-ssa-dse.c: Include expr.h.
25260 (maybe_trim_constructor_store): New function.
25261 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25263 PR tree-optimization/33562
25264 PR tree-optimization/61912
25265 PR tree-optimization/77485
25266 * doc/invoke.texi: Document new dse-max-object-size param.
25267 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25268 * tree-ssa-dse.c: Include params.h.
25269 (dse_store_status): New enum.
25270 (initialize_ao_ref_for_dse): New, partially extracted from
25272 (valid_ao_ref_for_dse, normalize_ref): New.
25273 (setup_live_bytes_from_ref, compute_trims): Likewise.
25274 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25275 (maybe_trim_partially_dead_store): Likewise.
25276 (maybe_trim_complex_store): Likewise.
25277 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25278 Track what bytes live from the original store. Return tri-state
25279 for dead, partially dead or live.
25280 (dse_dom_walker): Add constructor, destructor and new private members.
25281 (delete_dead_call, delete_dead_assignment): New extracted from
25283 (dse_optimize_stmt): Make a member of dse_dom_walker.
25284 Use initialize_ao_ref_for_dse.
25286 PR tree-optimization/33562
25287 PR tree-optimization/61912
25288 PR tree-optimization/77485
25289 * sbitmap.h (bitmap_count_bits): Prototype.
25290 (bitmap_clear_range, bitmap_set_range): Likewise.
25291 * sbitmap.c (bitmap_clear_range): New function.
25292 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25294 2017-01-13 Martin Liska <mliska@suse.cz>
25297 * function.c (set_cfun): Add new argument force.
25298 * function.h (set_cfun): Likewise.
25299 * ipa-inline-transform.c (inline_call): Use the function when
25300 strict alising from is dropped for function we inline to.
25302 2017-01-13 Richard Biener <rguenther@suse.de>
25304 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25305 for dumping GIMPLE INTEGER_CSTs.
25307 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25309 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25310 to 201112L since C++17.
25312 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25315 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25316 if -fsanitize=kernel-address is present.
25318 2017-01-13 Richard Biener <rguenther@suse.de>
25320 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25321 as _Literal ( type ) number in case usual suffixes do not
25322 preserve all information.
25324 2017-01-13 Richard Biener <rguenther@suse.de>
25326 PR tree-optimization/77283
25327 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25328 and ssa-iterators.h.
25329 (is_feasible_trace): Implement a cost model based on joiner
25332 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25335 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25336 char or short to __float128/_Float128 directly.
25338 2017-01-12 Martin Sebor <msebor@redhat.com>
25340 to -Wformat-overflow.
25341 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25342 (min_bytes_remaining): Same.
25343 (get_string_length): Same.
25344 (format_string): Same.
25345 (format_directive): Same.
25347 (pass_sprintf_length::handle_gimple_call): Same.
25349 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25351 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25352 info.nowrite calls with no lhs that can't throw. Return bool
25353 whether gsi_remove has been called or not.
25354 (pass_sprintf_length::handle_gimple_call): Return bool whether
25355 try_substitute_return_value called gsi_remove. Formatting fix.
25356 (pass_sprintf_length::execute): Don't use gsi_remove if
25357 handle_gimple_call returned true.
25360 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25361 be removed due to side-effects, don't remove following barrier nor
25362 turn the successor edge into fallthru edge.
25364 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25367 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25368 element-reversing loads and stores as not swappable.
25370 2017-01-12 Nathan Sidwell <nathan@acm.org>
25371 Nicolai Stange <nicstange@gmail.com>
25373 * combine.c (try_combine): Don't ignore result of overlap checking
25374 loop. Combine overlap & asm check into single loop.
25376 2017-01-12 Richard Biener <rguenther@suse.de>
25378 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25379 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25381 2017-01-12 Richard Biener <rguenther@suse.de>
25383 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25384 and TS_TARGET_OPTION directly derive from TS_BASE.
25385 * tree-core.h (tree_optimization_option): Derive from tree_base.
25386 (tree_target_option): Likewise.
25388 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25390 * config/i386/i386.c (memory_address_length): Increase len
25391 only when rip_relative_addr_p returns false.
25393 2017-01-11 Julia Koval <julia.koval@intel.com>
25395 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25396 (OPTION_MASK_ISA_SGX_SET): New.
25397 (ix86_handle_option): Handle OPT_msgx.
25398 * config.gcc: Added sgxintrin.h.
25399 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25400 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25401 * config/i386/i386.c (ix86_target_string): Add -msgx.
25403 (ix86_option_override_internal): Handle new options.
25404 (ix86_valid_target_attribute_inner_p): Add sgx.
25405 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25406 * config/i386/i386.opt: Add msgx.
25407 * config/i386/sgxintrin.h: New file.
25408 * config/i386/x86intrin.h: Add sgxintrin.h.
25410 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25413 * fold-const.c (maybe_nonzero_address): Return 1 for function
25415 (tree_single_nonzero_warnv_p): Don't handle function local objects
25419 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25422 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25425 * opts.c: Include "spellcheck.h"
25426 (struct string_fragment): New struct.
25427 (struct edit_distance_traits<const string_fragment &>): New
25429 (get_closest_sanitizer_option): New function.
25430 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25432 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25434 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25436 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25437 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25438 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25439 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25440 for initial die_offset if dwarf_split_debug_info.
25441 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25442 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25444 (output_skeleton_debug_sections): Formatting fix. Use
25445 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25446 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25448 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25450 * config/arm/cortex-a53.md: Add bypasses for
25451 cortex_a53_r2f_cvt.
25452 (cortex_a53_r2f): Only use for transfers.
25453 (cortex_a53_f2r): Likewise.
25454 (cortex_a53_r2f_cvt): Add reservation for conversions.
25455 (cortex_a53_f2r_cvt): Likewise.
25457 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25459 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25460 to all inlined functions, change static to extern.
25462 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25465 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25467 (arm_assemble_integer): Likewise.
25469 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25471 * config.gcc: Use new awk script to check CPU, FPU and architecture
25472 parameters for --with-... options.
25473 * config/arm/parsecpu.awk: New file
25474 * config/arm/arm-cpus.in: New file.
25475 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25477 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25479 * config/arm/t-arm: Update dependency rules.
25480 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25481 of processing .def files.
25482 * config/arm/genopt.sh: Deleted.
25483 * config/arm/gentune.sh: Deleted.
25484 * config/arm/arm-cores.def: Deleted.
25485 * config/arm/arm-arches.def: Deleted.
25486 * config/arm/arm-fpus.def: Deleted.
25487 * config/arm/arm-tune.md: Regenerated.
25488 * config/arm/arm-tables.opt: Regenerated.
25489 * config/arm/arm-cpu.h: New generated file.
25490 * config/arm/arm-cpu-data.h: New generated file.
25491 * config/arm/arm-cpu-cdata.h: New generated file.
25493 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25496 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25498 (input_varpool_node): Unpack dynamically_initialized bit.
25500 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25502 PR rtl-optimization/79032
25503 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25504 the alignment of the adjusted memory reference against that of MODE,
25505 instead of the alignment of the original memory reference.
25507 2017-01-11 Martin Jambor <mjambor@suse.cz>
25509 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25511 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25512 decorated functions.
25514 2017-01-11 Richard Biener <rguenther@suse.de>
25516 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25517 set range/nonnull info for PHI results. Do not set it on
25518 stmts marked for removal.
25520 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25522 * expr.c (store_field): In the bitfield case, fetch the return value
25523 from the registers before applying a single big-endian adjustment.
25524 Always do a final load for a BLKmode value not larger than a word.
25526 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25529 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25530 that we correctly handle column numbers greater than
25531 LINE_MAP_MAX_COLUMN_NUMBER.
25533 2017-01-10 Martin Sebor <msebor@redhat.com>
25535 PR middle-end/78245
25536 * gimple-ssa-sprintf.c (get_destination_size): Call
25537 {init,fini}object_sizes.
25538 * tree-object-size.c (addr_object_size): Adjust.
25539 (pass_through_call): Adjust.
25540 (pass_object_sizes::execute): Adjust.
25541 * tree-object-size.h (fini_object_sizes): Declare.
25543 2017-01-10 Martin Sebor <msebor@redhat.com>
25545 PR tree-optimization/78775
25546 * builtins.c (get_size_range): Move...
25547 * calls.c: ...to here.
25548 (alloc_max_size): Accept zero argument.
25549 (operand_signed_p): Remove.
25550 (maybe_warn_alloc_args_overflow): Call get_size_range.
25551 * calls.h (get_size_range): Declare.
25553 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25555 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25556 from TI's devices.csv file as of September 2016.
25557 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25559 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25561 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25562 * doc/invoke.texi: Likewise.
25563 * doc/md.texi: Likewise.
25564 * doc/objc.texi: Likewise.
25566 2017-01-10 Joshua Conner <joshconner@google.com>
25568 * config/arm/fuchsia-elf.h: New file.
25569 * config/fuchsia.h: New file.
25570 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25571 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25573 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25575 2016-01-10 Richard Biener <rguenther@suse.de>
25577 PR tree-optimization/79034
25578 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25579 Propagate out degenerate PHIs in the joiner.
25581 2017-01-10 Martin Liska <mliska@suse.cz>
25583 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25584 (sort_congruence_classes_by_decl_uid): Likewise.
25585 (sort_congruence_class_groups_by_decl_uid): Likewise.
25586 (sem_item_optimizer::merge_classes): Sort class, groups in these
25587 classes and members in the groups by DECL_UID of declarations.
25588 This would make merge operations stable.
25590 2017-01-10 Martin Liska <mliska@suse.cz>
25592 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25593 usage of m_classes_vec.
25594 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25595 (sem_item_optimizer::get_group_by_hash): Likewise.
25596 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25597 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25598 (sem_item_optimizer::verify_classes): Likewise.
25599 (sem_item_optimizer::process_cong_reduction): Likewise.
25600 (sem_item_optimizer::dump_cong_classes): Likewise.
25601 (sem_item_optimizer::merge_classes): Likewise.
25602 * ipa-icf.h (congruence_class_hash): Rename from
25603 congruence_class_group_hash. Remove declaration of m_classes_vec.
25605 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25607 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25608 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25609 * config.gcc: Add avx512vpopcntdqintrin.h.
25610 * config/i386/avx512vpopcntdqintrin.h: New.
25611 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25612 * config/i386/i386-builtin-types.def: Add new types.
25613 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25614 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25615 __builtin_ia32_vpopcountq_v8di_mask): New.
25616 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25617 __AVX512VPOPCNTDQ__.
25618 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25619 (PTA_AVX512VPOPCNTDQ): Define.
25620 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25621 TARGET_AVX512VPOPCNTDQ_P): Define.
25622 * config/i386/i386.opt: Add mavx512vpopcntdq.
25623 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25624 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25626 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25628 PR middle-end/77484
25629 * predict.def (PRED_CALL): Set to 67.
25631 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25633 * expr.c (store_field): In the bitfield case, if the value comes from
25634 a function call and is of an aggregate type returned in registers, do
25635 not modify the field mode; extract the value in all cases if the mode
25636 is BLKmode and the size is not larger than a word.
25638 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25641 * config/i386/cpuid.h: Fix undefined behavior.
25643 2017-01-04 Jeff Law <law@redhat.com>
25645 PR tree-optimization/79007
25646 PR tree-optimization/67955
25647 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25648 conservative for pt.null when flag_non_call_exceptions is on.
25650 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25652 PR translation/79019
25653 PR translation/79020
25654 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25655 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25656 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25658 * config/avr/avr.opt (maccumulate-args): Likewise.
25659 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25660 * common.opt (freport-bug): Likewise.
25661 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25662 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25663 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25664 translatable string.
25665 * config/i386/i386.c (function_value_32): Likewise.
25666 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25667 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25669 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25670 * common/config/msp430/msp430-common.c (msp430_handle_option):
25672 * symtab.c (symtab_node::verify_base): Likewise.
25673 * opts.c (set_debug_level): Likewise.
25674 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25675 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25676 missing whitespace to translatable strings.
25677 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25678 * config/sh/superh.h: Likewise.
25679 * config/i386/xopintrin.h: Likewise.
25680 * config/i386/znver1.md: Likewise.
25681 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25682 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25683 * double-int.h (struct double_int): Likewise.
25684 * double-int.c (div_and_round_double): Likewise.
25685 * wide-int.cc: Likewise.
25686 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25687 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25688 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25689 (crossjumps_occurred): ... this.
25690 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25693 PR tree-optimization/78899
25694 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25695 returning bool return struct loop *, NULL for failure and the new
25697 (versionable_outer_loop_p): Don't version outer loop if it has
25698 dont_vectorized bit set.
25699 (tree_if_conversion): When versioning outer loop, ensure
25700 tree_if_conversion is performed also on the inner loop of the
25701 non-vectorizable outer loop copy.
25702 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25703 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25704 prevent vectorization of it.
25705 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25706 the outer loop vectorization of the non-scalar version is attempted
25707 before vectorization of the inner loop in scalar version. If
25708 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25709 vectorization of its inner loop.
25710 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25711 has 2 inner loops, rename also on edges from bb whose single pred
25712 is outer_loop->header. Fix typo in function comment.
25714 2017-01-09 Martin Sebor <msebor@redhat.com>
25717 * asan.c (asan_emit_stack_protection): Increase local buffer size
25718 to avoid snprintf truncation warning.
25720 2017-01-09 Andrew Pinski <apinski@cavium.com>
25722 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25723 to reference thunderx2t99 for the tuning structure
25724 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25726 (thunderx2t99_extra_costs): This.
25727 * config/aarch64/aarch64-tune.md: Regenerate.
25728 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25729 (vulcan_addrcost_table): This.
25730 (vulcan_regmove_cost): Rename to ...
25731 (thunderx2t99_regmove_cost): This.
25732 (vulcan_vector_cost): Rename to ...
25733 (thunderx2t99_vector_cost): this.
25734 (vulcan_branch_cost): Rename to ...
25735 (thunderx2t99_branch_cost): This.
25736 (vulcan_tunings): Rename to ...
25737 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25738 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25740 2017-01-09 Martin Jambor <mjambor@suse.cz>
25744 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25745 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25746 (propagate_vr_accross_jump_function): Use the above function for all
25747 value range computations for pass-through jump functions and type
25748 converasion from explicit value range values.
25749 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25750 parameters from TYPE_ARG_TYPES.
25751 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25752 (ipa_write_node_info): Stream type of the actual argument.
25753 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25755 2017-01-09 Martin Liska <mliska@suse.cz>
25758 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25759 (lookup_compiler): Do not show error message with have_E.
25761 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25763 PR tree-optimization/78938
25764 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25765 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25766 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25767 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25770 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25772 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25775 2017-01-09 Richard Biener <rguenther@suse.de>
25777 PR tree-optimization/78997
25778 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25779 name condition properly.
25781 2017-01-09 Richard Biener <rguenther@suse.de>
25784 * dwarf2out.c (is_cxx): New overload with context.
25785 (is_naming_typedef_decl): Use it.
25787 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25789 * invoke.texi (Option Summary): Correct spacing in option lists
25790 and add line breaks to fix over-long lines.
25792 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25794 PR middle-end/17660
25796 * extend.texi (Common Variable Attributes): Add xref to GCC
25797 Internals manual to explain mode attribute keywords.
25799 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25802 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25803 and Preprocessor Options.
25804 (Options for Linking): Document -pthread here....
25805 (RS/6000 and PowerPC Options): ...not here.
25806 (Solaris 2 Options): ...or here.
25807 * doc/cppopts.texi: Document -pthread.
25809 2017-01-08 Martin Sebor <msebor@redhat.com>
25811 PR middle-end/77708
25812 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25813 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25814 New member functions.
25815 (format_directive): Used them.
25817 (pass_sprintf_length::handle_gimple_call): Same.
25818 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25819 to avoid truncation for any argument.
25820 (extract_affine_mul): Same.
25821 * tree.c (get_file_function_name): Same.
25823 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25825 PR middle-end/77484
25826 * predict.def (PRED_INDIR_CALL): Set to 86.
25828 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25830 PR preprocessor/54124
25831 * doc/cppopts.texi: Reformat -d subtable to list the full name
25832 of the options. Add cross-reference to the docs for the general
25833 compiler -d options.
25834 * doc/invoke.texi (Developer Options): Add cross-reference to the
25835 preprocessor-specific -d option documentation.
25837 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25839 PR preprocessor/13498
25840 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25841 redudant material, and reflect new command-line options.
25842 (System Headers): Likewise.
25844 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25846 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25847 -isystem, and -idirafter. Copy-edit.
25848 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25849 default for -ftrack-macro-expansion. Delete obsolete and
25850 badly-formatted implementation details about -fdebug-cpp output.
25851 * doc/cppwarnopts.texi: Copy-edit.
25853 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25856 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25857 that the transition from a max line width >= 1<<10 to narrower
25858 lines works correctly.
25860 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25862 * doc/options.texi (PerFunction): New.
25863 * opt-functions.awk (switch_flags): Map both Optimization and
25864 PerFunction to CL_OPTIMIZATION.
25865 * opth-gen.awk: Test for PerFunction flag along with
25867 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25868 it only when the latter is present. Skip those that don't in
25869 the hash function generator.
25870 * common.opt (fvar-tracking): Mark as PerFunction instead of
25872 (fvar-tracking-assignments): Likewise.
25873 (fvar-tracking-assignments-toggle): Likewise.
25874 (fvar-tracking-uninit): Likewise.
25876 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25878 PR translation/79018
25879 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25882 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25885 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25886 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25887 TARGET_LONG_JUMP_TABLE_OFFSETS.
25888 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25889 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25890 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25891 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25892 * config/m68k/m68k.md (tablejump expander): Likewise.
25893 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25894 TARGET_LONG_JUMP_TABLE_OFFSETS.
25895 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25896 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25898 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25899 David Holsgrove <david.holsgrove@xilinx.com>
25901 * common/config/microblaze/microblaze-common.c
25902 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25903 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25905 * config/microblaze/microblaze.c (microblaze_must_save_register)
25906 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25908 (microblaze_eh_return): New function.
25909 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25910 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25911 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25912 * config/microblaze/microblaze.md (eh_return): New pattern.
25914 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25916 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25917 GCC_DIAGNOSTIC_STRINGIFY): Define.
25919 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25921 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25923 * config/arm/arm.md (<mcrr>): New.
25925 * config/arm/arm.c (arm_arch5te): New.
25926 (arm_option_override): Set arm_arch5te.
25927 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25929 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25930 (arm_mcrr_qualifiers): ... this. New.
25931 (MRRC_QUALIFIERS): Define to...
25932 (arm_mrrc_qualifiers): ... this. New.
25933 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25935 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25936 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25937 (MRRCI, mrrc, MRRC): New.
25938 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25939 VUNSPEC_MRRC2): New.
25941 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25943 * config/arm/arm.md (<mcr>): New.
25945 * config/arm/arm.c (arm_coproc_builtin_available): Add
25946 support for mcr, mrc, mcr2 and mrc2.
25947 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25948 (arm_mcr_qualifiers): ... this. New.
25949 (MRC_QUALIFIERS): Define to ...
25950 (arm_mrc_qualifiers): ... this. New.
25951 (MCR_QUALIFIERS): Define to ...
25952 (arm_mcr_qualifiers): ... this. New.
25953 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25955 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25956 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25957 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25958 VUNSPEC_MRC2): New.
25960 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25962 * config/arm/arm.md (*ldc): New.
25966 * config/arm/arm.c (arm_coproc_builtin_available): Add
25967 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25968 (arm_coproc_ldc_stc_legitimate_address): New.
25969 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25970 'qualifier_const_pointer'.
25971 (LDC_QUALIFIERS): Define to...
25972 (arm_ldc_qualifiers): ... this. New.
25973 (STC_QUALIFIERS): Define to...
25974 (arm_stc_qualifiers): ... this. New.
25975 * config/arm/arm-protos.h
25976 (arm_coproc_ldc_stc_legitimate_address): New.
25977 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25978 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25979 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25980 stc2, stcl, stc2l): New.
25981 * config/arm/constraints.md (Uz): New.
25982 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25983 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25984 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25985 VUNSPEC_STC2L): New.
25987 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25989 * config/arm/arm.md (<cdp>): New.
25990 * config/arm/arm.c (neon_const_bounds): Rename this ...
25991 (arm_const_bounds): ... this.
25992 (arm_coproc_builtin_available): New.
25993 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25994 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25995 (CDP_QUALIFIERS): Define to...
25996 (arm_cdp_qualifiers): ... this. New.
25998 (arm_expand_builtin_args): Add case for 6 arguments.
25999 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26000 (arm_const_bounds): ... this.
26001 (arm_coproc_builtin_available): New.
26002 * config/arm/arm_acle.h (__arm_cdp): New.
26004 * config/arm/arm_acle_builtins.def (cdp): New.
26006 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26007 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26008 'arm_const_bounds'.
26009 * config/arm/types.md (coproc): New.
26010 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26011 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26012 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26013 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26015 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26017 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26018 (UBINOP_QUALIFIERS): New.
26020 (acle_builtin_data): New. Change comment.
26021 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26022 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26023 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26024 arm_acle_builtins.def.
26025 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26026 (arm_init_acle_builtins): New.
26027 (CRC32_BUILTIN): Remove.
26028 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26029 crc32cb, crc32ch and crc32cw.
26030 (arm_init_crc32_builtins): Remove.
26031 (arm_init_builtins): Use arm_init_acle_builtins rather
26032 than arm_init_crc32_builtins.
26033 (arm_expand_acle_builtin): New.
26034 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26035 * config/arm/arm_acle_builtins.def: New.
26037 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26039 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26040 (arm_builtin_datum): ... this.
26041 (arm_init_neon_builtin): Rename to ...
26042 (arm_init_builtin): ... this. Add a new parameters PREFIX
26043 and USE_SIG_IN_NAME.
26044 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26045 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26046 'arm_builtin_datum'.
26047 (arm_init_vfp_builtins): Likewise.
26048 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26049 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26050 (arm_expand_neon_args): Rename to ...
26051 (arm_expand_builtin_args): ... this. Rename builtin_arg
26052 enum values and differentiate between ARG_BUILTIN_MEMORY
26053 and ARG_BUILTIN_NEON_MEMORY.
26054 (arm_expand_neon_builtin_1): Rename to ...
26055 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26056 values, arm_expand_builtin_args and add bool parameter NEON.
26057 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26058 (arm_expand_vfp_builtin): Likewise.
26059 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26061 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26063 PR middle-end/77484
26064 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26065 * predict.c (tree_estimate_probability_bb): Reverse direction of
26066 polymorphic call predictor.
26068 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26070 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26071 (determine_pass_name_match): ...this new function and...
26072 (should_skip_pass_p): ...this new function.
26074 2017-01-06 Nathan Sidwell <nathan@acm.org>
26076 * ipa-visibility.c (function_and_variable_visibility): Reformat
26077 comments and long lines. Remove extrneous if.
26078 * symtab.c (symtab_node::make_decl_local): Fix code format.
26079 (symtab_node::set_section_for_node): Fix comment typo.
26081 2017-01-06 Martin Liska <mliska@suse.cz>
26084 * lra-constraints.c: Rename invariant to lra_invariant.
26085 * predict.c (set_even_probabilities): Initialize e to NULL.
26087 2017-01-05 Martin Sebor <msebor@redhat.com>
26089 PR tree-optimization/78910
26090 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26091 (format_integer): Correct off-by-one error in the handling
26092 of precision with negative numbers in signed conversions..
26094 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26096 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26098 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26100 PR tree-optimization/71016
26101 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26102 factor_out_conditional_conversion. Formatting fix.
26103 (factor_out_conditional_conversion): Add cond_stmt argument.
26104 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26105 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26108 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26110 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26111 read-rtl-function.o, and selftest-rtl.o.
26112 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26113 (selftest::aarch64_test_loading_full_dump): New function.
26114 (selftest::aarch64_run_selftests): New function.
26115 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26116 selftest::aarch64_run_selftests.
26117 * config/i386/i386.c
26118 (selftest::ix86_test_loading_dump_fragment_1): New function.
26119 (selftest::ix86_test_loading_call_insn): New function.
26120 (selftest::ix86_test_loading_full_dump): New function.
26121 (selftest::ix86_test_loading_unspec): New function.
26122 (selftest::ix86_run_selftests): Call the new functions.
26123 * emit-rtl.c (maybe_set_max_label_num): New function.
26124 * emit-rtl.h (maybe_set_max_label_num): New decl.
26125 * function.c (instantiate_decls): Guard call to
26126 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26127 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26129 * gensupport.c (gen_reader::gen_reader): Pass "false"
26130 for new "compact" param of rtx_reader.
26131 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26132 rather than an empty string for NULL strings.
26133 * read-md.c: Potentially include config.h rather than bconfig.h.
26134 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26135 (have_error): New global, copied from errors.c.
26136 (md_reader::read_name): Rename to...
26137 (md_reader::read_name_1): ...this, adding "out_loc" param,
26138 and converting "missing name or number" to returning false, rather
26140 (md_reader::read_name): Reimplement in terms of read_name_1.
26141 (md_reader::read_name_or_nil): New function.
26142 (md_reader::read_string): Handle "(nil)" by returning NULL.
26143 (md_reader::md_reader): Add new param "compact".
26144 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26145 (md_reader::read_file): New method.
26146 * read-md.h (md_reader::md_reader): Add new param "compact".
26147 (md_reader::read_file): New method.
26148 (md_reader::is_compact): New accessor.
26149 (md_reader::read_name): Convert return type from void to file_location.
26150 (md_reader::read_name_or_nil): New decl.
26151 (md_reader::read_name_1): New decl.
26152 (md_reader::m_compact): New field.
26153 (noop_reader::noop_reader): Pass "false" for new "compact" param
26155 (rtx_reader::rtx_reader): Add new "compact" param.
26156 (rtx_reader::read_rtx_operand): Make virtual and convert return
26157 type from void to rtx.
26158 (rtx_reader::read_until): New decl.
26159 (rtx_reader::handle_any_trailing_information): New virtual function.
26160 (rtx_reader::postprocess): New virtual function.
26161 (rtx_reader::finalize_string): New virtual function.
26162 (rtx_reader::m_in_call_function_usage): New field.
26163 (rtx_reader::m_reuse_rtx_by_id): New field.
26164 * read-rtl-function.c: New file.
26165 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26166 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26167 (selftest::verify_three_block_rtl_cfg): New decl.
26168 * read-rtl-function.h: New file.
26169 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26170 For host, include function.h, memmodel.h, and emit-rtl.h.
26171 (one_time_initialization): New function.
26172 (struct compact_insn_name): New struct.
26173 (compact_insn_names): New array.
26174 (find_code): Handle insn codes in compact dumps.
26175 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26176 (bind_subst_iter_and_attr): Likewise.
26177 (add_condition_to_string): Likewise.
26178 (add_condition_to_rtx): Likewise.
26179 (apply_attribute_uses): Likewise.
26180 (add_current_iterators): Likewise.
26181 (apply_iterators): Likewise.
26182 (initialize_iterators): Guard usage of apply_subst_iterator with
26183 #ifdef GENERATOR_FILE.
26184 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26185 (md_reader::read_mapping): Likewise.
26186 (add_define_attr_for_define_subst): Likewise.
26187 (add_define_subst_attr): Likewise.
26188 (read_subst_mapping): Likewise.
26189 (check_code_iterator): Likewise.
26190 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26192 (one_time_initialization): New function.
26193 (rtx_reader::read_until): New method.
26194 (read_flags): New function.
26195 (parse_reg_note_name): New function.
26196 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26197 Handle reuse_rtx ids.
26198 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26199 Add parsing support for RTL dumps, mirroring the special-cases in
26200 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26201 values, and calling handle_any_trailing_information.
26202 (rtx_reader::read_rtx_operand): Convert return type from void
26203 to rtx, returning return_rtx. Handle case 'e'. Call
26204 finalize_string on XSTR and XTMPL fields.
26205 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26206 "(nil)" values were omitted. Call the postprocess vfunc on the
26208 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26209 class ctor. Initialize m_in_call_function_usage. Call
26210 one_time_initialization.
26211 * rtl-tests.c (selftest::test_uncond_jump): Call
26212 set_new_first_and_last_insn.
26213 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26214 * selftest-rtl.c: New file.
26215 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26216 (selftest::get_insn_by_uid): New decl.
26217 * selftest-run-tests.c (selftest::run_tests): Call
26218 read_rtl_function_c_tests.
26219 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26220 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26223 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26225 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26226 operands in a special way. Assert that pos+len <= mode precision.
26228 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26230 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26231 3 argument Alias with unlimited for the negative form.
26232 (fno-vect-cost-model): Removed.
26234 2017-01-05 Martin Liska <mliska@suse.cz>
26236 * hsa-gen.c (gen_hsa_divmod): New function.
26237 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26239 2017-01-05 Martin Liska <mliska@suse.cz>
26242 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26245 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26247 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26248 small constant length operands.
26250 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26252 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26253 between loop iterations.
26255 2017-01-05 Martin Liska <mliska@suse.cz>
26258 * gimplify.c (gimplify_decl_expr): Compare to
26259 asan_poisoned_variables instread of checking flags.
26260 (gimplify_target_expr): Likewise.
26261 (gimplify_expr): Likewise.
26262 (gimplify_function_tree): Conditionally initialize
26263 asan_poisoned_variables.
26265 2017-01-04 Jeff Law <law@redhat.com>
26267 PR tree-optimizatin/78812
26268 * rtl.h (contains_mem_rtx_p): Prototype.
26269 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26270 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26271 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26272 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26273 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26275 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26277 * input.c (assert_char_at_range): Default-initialize actual_range.
26279 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26281 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26282 to match the caller.
26284 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26286 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26287 insns after final jump in test to emit dummy move.
26289 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26291 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26292 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26294 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26296 * multiple_target.c (create_dispatcher_calls): Init e_next.
26297 * tree-ssa-loop-split.c (split_loop): Init border.
26298 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26301 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26306 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26307 (altivec_register_operand): Do not return true if the operand
26308 contains a SUBREG mixing SImode and SFmode.
26309 (vsx_register_operand): Likewise.
26310 (vsx_reg_sfsubreg_ok): New predicate.
26311 (vfloat_operand): Do not return true if the operand contains a
26312 SUBREG mixing SImode and SFmode.
26313 (vint_operand): Likewise.
26314 (vlogical_operand): Likewise.
26315 (gpc_reg_operand): Likewise.
26316 (int_reg_operand): Likewise.
26317 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26318 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26319 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26321 (rs6000_emit_move_si_sf_subreg): New helper function.
26322 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26323 fixup SUBREGs involving SImode and SFmode.
26324 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26325 numbers for the new peephole2 optimization.
26326 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26327 the GLIBC math library that do AND/IOR/XOR operations on single
26328 precision floating point.
26329 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26330 target macros to say whether we need to avoid SUBREGs mixing
26332 (TARGET_ALLOW_SF_SUBREG): Likewise.
26333 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26334 (UNSPEC_SI_FROM_SF): Likewise.
26335 (iorxor): Change spacing.
26336 (and_ior_xor): New iterator for AND, IOR, and XOR.
26337 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26338 (movdi_from_sf_zero_ext): Likewise.
26339 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26340 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26341 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26342 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26343 (fms<mode>4): Likewise.
26344 (fnma<mode>4): Likewise.
26345 (fnms<mode>4): Likewise.
26346 (nfma<mode>4): Likewise.
26347 (nfms<mode>4): Likewise.
26349 2017-01-04 Marek Polacek <polacek@redhat.com>
26352 * doc/invoke.texi: Document -Wpointer-compare.
26354 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26356 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26359 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26360 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26363 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26366 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26367 documentation of the powerpc_popcntb_ok attribute.
26368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26369 code to issue warning messages if a requested CPU configuration is
26370 not supported by the binary (assembler and loader) toolchain.
26371 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26372 made to define a built-in function that has been disabled.
26373 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26374 made to define a built-in function that has been disabled.
26375 (altivec_init_builtins): Add comment explaining why definition
26376 of the DST built-in functions is not preceded by an assertion
26377 check. Add assertions to prevent ICE if attempts are made to
26378 define an altivec predicate or an abs* built-in function that has
26380 (htm_init_builtins): Add comment explaining why definition of the
26381 htm built-in functions is not preceded by an assertion check.
26383 2017-01-04 Jeff Law <law@redhat.com>
26385 PR tree-optimizatin/67955
26386 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26387 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26388 the points-to solution does not include pt_null. Use DECL_PT_UID
26391 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26393 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26394 Use gen_int_mode instead of gen_lopwart for const_int operands.
26396 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26398 PR tree-optimization/71563
26399 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26400 out of range value - has low bits known to be zero.
26402 2017-01-04 Alan Modra <amodra@gmail.com>
26404 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26405 * configure: Regenerate.
26406 * config.in: Regenerate.
26408 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26411 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26412 a substring of the message, but strcmp with the whole message. Ifdef
26413 ENABLE_NLS, translate the message first using dgettext.
26415 2017-01-03 Jeff Law <law@redhat.com>
26417 PR tree-optimizatin/78856
26418 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26419 (mark_threaded_blocks): Remove code to truncate thread paths that
26420 cross multiple loop headers. Instead invalidate the cached loop
26421 iteration information and handle case of a thread path walking
26422 into an irreducible region.
26424 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26427 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26428 assertions. Add support for doing the signbit if the IEEE 128-bit
26429 floating point value is in a GPR.
26430 * config/rs6000/rs6000.md (Fsignbit): Delete.
26431 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26432 Update the length attribute if the value is in a GPR.
26433 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26434 the sign or zero extension instruction, since the value is always 0/1.
26435 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26438 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26439 extracting SImode to a GPR register so that we can generate a
26440 store, limit the vector to be in a traditional Altivec register
26441 for the vextuwrx instruction.
26443 2017-01-03 Ian Lance Taylor <iant@google.com>
26445 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26447 2017-01-03 Martin Sebor <msebor@redhat.com>
26449 PR tree-optimization/78696
26450 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26451 precision. Use MPFR for %f for greater fidelity. Correct handling
26453 (pass_sprintf_length::compute_format_length): Set width and precision
26454 specified by asrerisk to void_node for vararg functions.
26455 (try_substitute_return_value): Adjust dump output.
26457 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26459 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26461 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26463 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26464 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26465 -mlra/-mno-lra was passed to the compiler.
26467 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26469 PR rtl-optimization/65618
26470 * emit-rtl.c (try_split): Move initialization of "before" and
26471 "after" to just before the call to emit_insn_after_setloc.
26473 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26475 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26477 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26479 * dwarf2out.c (gen_enumeration_type_die): When
26480 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26482 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26484 PR tree-optimization/78965
26485 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26486 Change first argument from const call_info & to call_info &. For %n
26487 set info.nowrite to false.
26489 PR middle-end/78901
26490 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26491 possibly throwing calls.
26493 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26494 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26495 and fns handling, rather than in a separate case SSA_NAME.
26497 2017-01-02 Jeff Law <law@redhat.com>
26499 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26500 fixes for first_period and second_period variables.
26502 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26505 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26506 (*insvqi_1): New insn pattern.
26507 (*insvqi_1_mem_rex64): Ditto.
26508 (*insvqi_2): Ditto.
26509 (*insvqi_3): Rename from *insvqi.
26511 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26513 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26515 * doc/cfg.texi (Edges): Remove reference to Java.
26516 (Maintaining the CFG): Ditto.
26518 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26520 PR middle-end/77674
26521 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26522 transparent aliases.
26524 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26526 PR middle-end/77484
26527 * predict.def (PRED_CALL): Update hitrate.
26528 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26529 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26530 into direct/indirect/polymorphic variants.
26532 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26534 Update copyright years.
26536 * gcc.c (process_command): Update copyright notice dates.
26537 * gcov-dump.c (print_version): Ditto.
26538 * gcov.c (print_version): Ditto.
26539 * gcov-tool.c (print_version): Ditto.
26540 * gengtype.c (create_file): Ditto.
26541 * doc/cpp.texi: Bump @copying's copyright year.
26542 * doc/cppinternals.texi: Ditto.
26543 * doc/gcc.texi: Ditto.
26544 * doc/gccint.texi: Ditto.
26545 * doc/gcov.texi: Ditto.
26546 * doc/install.texi: Ditto.
26547 * doc/invoke.texi: Ditto.
26549 Copyright (C) 2017 Free Software Foundation, Inc.
26551 Copying and distribution of this file, with or without modification,
26552 are permitted in any medium without royalty provided the copyright
26553 notice and this notice are preserved.