1 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4 * common.opt (-static-pie): New alias.
5 (shared): Negate static-pie.
6 (-no-pie): Update help text.
8 (static-pie): New option.
9 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12 (LINK_EH_SPEC): Likewise.
13 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16 * gcc.c (LINK_COMMAND_SPEC): Likewise.
17 (init_gcc_specs): Likewise.
18 (init_spec): Likewise.
19 (display_help): Update help message for -pie.
20 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
23 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
25 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
26 (movdi_aarch64): Likewise.
27 (movti_aarch64): Likewise.
29 2017-09-12 Simon Wright <simon@pushface.org>
32 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
33 calculation of the minor version, always output as 0.
35 2017-09-12 Jakub Jelinek <jakub@redhat.com>
38 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
39 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
40 on it early, rather than manual conversion late. For
41 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
42 instead of performing manual conversion.
44 2017-09-12 Carl Love <cel@us.ibm.com>
46 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
47 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
49 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
50 VMULOSW): Add definitions.
51 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
52 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
53 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
54 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
55 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
57 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
59 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
61 (movti_aarch64): Likewise.
62 (movdf_aarch64): Likewise.
63 (movtf_aarch64): Likewise.
64 (load_pairdi): Likewise.
65 (store_pairdi): Likewise.
66 (load_pairdf): Likewise.
67 (store_pairdf): Likewise.
68 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
69 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
70 (ldr_got_small_<mode>): Likewise.
71 (ldr_got_small_28k_<mode>): Likewise.
72 (ldr_got_tiny): Likewise.
73 * config/aarch64/iterators.md (ldst_sz): New.
74 (ldpstp_sz): Likewise.
75 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
77 (thunderx_load): Split load_8 to load_16.
78 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
80 (thunderx2t99_storepair_basic): Split store_8 to store_16.
81 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
82 (xgene1_store_pair): Split store_8 to store_16.
83 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
84 (falkor_st_0_st_sd): Split store_8 to store_16.
86 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
88 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
89 and store1/2/3/4 to store_4/8/12/16.
90 * config/aarch64/aarch64.md: Update for rename.
91 * config/arm/arm.md: Likewise.: Likewise.
92 * config/arm/arm.c: Likewise.
93 * config/arm/thumb1.md: Likewise.
94 * config/arm/thumb2.md: Likewise.
95 * config/arm/vfp.md: Likewise.
96 * config/arm/arm-generic.md: Likewise.
97 * config/arm/arm1020e.md: Likewise.
98 * config/arm/arm1026ejs.md: Likewise.
99 * config/arm/arm1136jfs.md: Likewise.
100 * config/arm/arm926ejs.md: Likewise.
101 * config/arm/cortex-a15.md: Likewise.
102 * config/arm/cortex-a17.md: Likewise.
103 * config/arm/cortex-a5.md: Likewise.
104 * config/arm/cortex-a53.md: Likewise.
105 * config/arm/cortex-a57.md: Likewise.
106 * config/arm/cortex-a7.md: Likewise.
107 * config/arm/cortex-a8.md: Likewise.
108 * config/arm/cortex-a9.md: Likewise.
109 * config/arm/cortex-m4.md: Likewise.
110 * config/arm/cortex-m7.md: Likewise.
111 * config/arm/cortex-r4.md: Likewise.
112 * config/arm/exynos-m1.md: Likewise.
113 * config/arm/fa526.md: Likewise.
114 * config/arm/fa606te.md: Likewise.
115 * config/arm/fa626te.md: Likewise.
116 * config/arm/fa726te.md: Likewise.
117 * config/arm/fmp626.md: Likewise.
118 * config/arm/iwmmxt.md: Likewise.
119 * config/arm/ldmstm.md: Likewise.
120 * config/arm/marvell-pj4.md: Likewise.
121 * config/arm/xgene1.md: Likewise.
122 * config/aarch64/thunderx.md: Likewise.
123 * config/aarch64/thunderx2t99.md: Likewise.
124 * config/aarch64/falkor.md: Likewise.
126 2017-09-12 Martin Liska <mliska@suse.cz>
128 * attribs.c (private_lookup_attribute): New function.
129 * attribs.h (private_lookup_attribute): Declared here.
130 (lookup_attribute): Called from this place.
132 2017-09-12 Richard Biener <rguenther@suse.de>
134 PR tree-optimization/82157
135 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
136 stmts with side-effects.
138 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
139 Alan Hayward <alan.hayward@arm.com>
140 David Sherwood <david.sherwood@arm.com>
142 * target.def (hard_regno_nregs): New hook.
143 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
144 * targhooks.h (default_hard_regno_nregs): Declare.
145 * targhooks.c (default_hard_regno_nregs): New function.
146 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
147 (TARGET_HARD_REGNO_NREGS): ...this hook.
148 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
149 (CLASS_MAX_NREGS): Likewise.
150 * doc/tm.texi: Regenerate.
151 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
152 instead of HARD_REGNO_NREGS.
153 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
154 HARD_REGNO_NREGS in the comment.
155 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
156 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
157 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
158 Return an unsigned int.
159 (TARGET_HARD_REGNO_NREGS): Redefine.
160 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
161 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
162 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
163 (arc_hard_regno_nregs): New function.
164 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
165 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
166 (arm_hard_regno_nregs): New function.
167 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
168 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
169 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
170 (TARGET_HARD_REGNO_NREGS): Redefine.
171 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
172 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
173 (HARD_REGNO_NREGS): Delete.
174 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
175 (cr16_hard_regno_nregs): New function.
176 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
177 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
178 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
179 (cris_hard_regno_nregs): New function.
180 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
181 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
182 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
183 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
184 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
185 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
186 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
187 (frv_hard_regno_nregs): Make static. Take and return an
189 (frv_class_max_nregs): Remove outdated copy of documentation.
190 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
191 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
192 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
193 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
194 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
195 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
196 (TARGET_HARD_REGNO_NREGS): Redefine.
197 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
198 (CLASS_MAX_NREGS): Update comment.
199 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
200 (ia64_hard_regno_nregs): New function.
201 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
202 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
203 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
204 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
205 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
207 (m32c_hard_regno_nregs): Likewise. Make static.
208 (TARGET_HARD_REGNO_NREGS): Redefine.
209 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
210 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
211 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
212 (m68k_hard_regno_nregs): New function.
213 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
214 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
215 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
216 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
217 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
218 Take and return an unsigned int.
219 (TARGET_HARD_REGNO_NREGS): Redefine.
220 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
221 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
222 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
223 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
224 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
225 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
226 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
227 (msp430_hard_regno_nregs): Make static. Take and return an
229 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
230 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
231 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
232 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
233 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
234 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
235 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
236 (TARGET_HARD_REGNO_NREGS): Redefine.
237 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
238 (PA_HARD_REGNO_NREGS): ...this.
239 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
240 (PA_HARD_REGNO_NREGS): ...this.
241 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
242 (pa_hard_regno_nregs): New function.
243 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
244 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
245 (pdp11_hard_regno_nregs): New function.
246 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
247 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
248 (rs6000_hard_regno_nregs_hook): New function.
249 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
250 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
251 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
252 Take and return an unsigned int. Move earlier in file.
253 (TARGET_HARD_REGNO_NREGS): Redefine.
254 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
255 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
256 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
257 (rl78_hard_regno_nregs): Make static. Take and return an
259 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
260 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
261 (rs6000_hard_regno_nregs_hook): New function.
262 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
263 * config/rx/rx.c (rx_hard_regno_nregs): New function.
264 (TARGET_HARD_REGNO_NREGS): Redefine.
265 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
266 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
267 instead of HARD_REGNO_NREGS.
268 (s390_hard_regno_nregs): New function.
269 (s390_hard_regno_mode_ok): Add comment from s390.h.
270 (TARGET_HARD_REGNO_NREGS): Redefine.
271 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
272 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
273 (sh_hard_regno_nregs): New function.
274 (sh_pass_in_reg_p): Use it.
275 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
276 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
277 (sparc_hard_regno_nregs): New function.
278 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
279 * config/spu/spu.c (spu_hard_regno_nregs): New function.
280 (spu_function_arg_advance): Use it, supplying a valid register number.
281 (TARGET_HARD_REGNO_NREGS): Redefine.
282 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
283 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
284 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
285 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
286 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
287 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
288 (CLASS_MAX_NREGS): Remove copy of old documentation.
289 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
290 (visium_hard_regno_nregs): New function.
291 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
292 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
293 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
294 xtensa_hard_regno_nregs): New function.
295 * system.h (HARD_REGNO_NREGS): Poison.
297 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
299 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
300 hard_regno_nregs instead of HARD_REGNO_NREGS.
301 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
302 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
303 (c6x_expand_epilogue): Likewise.
304 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
305 (frv_read_iacc_argument): Likewise.
306 * config/sh/sh.c: Include regs.h.
307 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
308 (regs_used): Likewise.
309 (output_stack_adjust): Likewise.
310 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
311 * expmed.c: Include regs.h.
312 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
313 * ree.c: Include regs.h.
314 (combine_reaching_defs): Use hard_regno_nregs instead of
316 (add_removable_extension): Likewise.
318 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
320 * regs.h (hard_regno_nregs): Turn into a function.
321 (end_hard_regno): Update accordingly.
322 * caller-save.c (setup_save_areas): Likewise.
323 (save_call_clobbered_regs): Likewise.
324 (replace_reg_with_saved_mem): Likewise.
325 (insert_restore): Likewise.
326 (insert_save): Likewise.
327 * combine.c (can_change_dest_mode): Likewise.
328 (move_deaths): Likewise.
329 (distribute_notes): Likewise.
330 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
331 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
332 (rs6000_split_multireg_move): Likewise.
333 (rs6000_register_move_cost): Likewise.
334 (rs6000_memory_move_cost): Likewise.
335 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
336 (rs6000_split_multireg_move): Likewise.
337 (rs6000_register_move_cost): Likewise.
338 (rs6000_memory_move_cost): Likewise.
339 * cselib.c (cselib_reset_table): Likewise.
340 (cselib_lookup_1): Likewise.
341 * emit-rtl.c (set_mode_and_regno): Likewise.
342 * function.c (aggregate_value_p): Likewise.
343 * ira-color.c (setup_profitable_hard_regs): Likewise.
344 (check_hard_reg_p): Likewise.
345 (calculate_saved_nregs): Likewise.
346 (assign_hard_reg): Likewise.
347 (improve_allocation): Likewise.
348 (calculate_spill_cost): Likewise.
349 * ira-emit.c (modify_move_list): Likewise.
350 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
351 (ira_hard_reg_in_set_p): Likewise.
352 * ira.c (setup_reg_mode_hard_regset): Likewise.
353 (clarify_prohibited_class_mode_regs): Likewise.
354 (check_allocation): Likewise.
355 * lra-assigns.c (find_hard_regno_for_1): Likewise.
356 (lra_setup_reg_renumber): Likewise.
357 (setup_try_hard_regno_pseudos): Likewise.
358 (spill_for): Likewise.
359 (assign_hard_regno): Likewise.
360 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
361 * lra-constraints.c (in_class_p): Likewise.
362 (lra_constraint_offset): Likewise.
363 (simplify_operand_subreg): Likewise.
364 (lra_constraints): Likewise.
365 (split_reg): Likewise.
366 (split_if_necessary): Likewise.
367 (invariant_p): Likewise.
368 (inherit_in_ebb): Likewise.
369 * lra-lives.c (process_bb_lives): Likewise.
370 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
371 (get_hard_regs): Likewise.
372 (do_remat): Likewise.
373 * lra-spills.c (assign_spill_hard_regs): Likewise.
374 * mode-switching.c (create_pre_exit): Likewise.
375 * postreload.c (reload_combine_recognize_pattern): Likewise.
376 * recog.c (peep2_find_free_register): Likewise.
377 * regcprop.c (kill_value_regno): Likewise.
378 (set_value_regno): Likewise.
379 (copy_value): Likewise.
380 (maybe_mode_change): Likewise.
381 (find_oldest_value_reg): Likewise.
382 (copyprop_hardreg_forward_1): Likewise.
383 * regrename.c (check_new_reg_p): Likewise.
384 (regrename_do_replace): Likewise.
385 * reload.c (push_reload): Likewise.
386 (combine_reloads): Likewise.
387 (find_dummy_reload): Likewise.
388 (operands_match_p): Likewise.
389 (find_reloads): Likewise.
390 (find_equiv_reg): Likewise.
391 (reload_adjust_reg_for_mode): Likewise.
392 * reload1.c (count_pseudo): Likewise.
393 (count_spilled_pseudo): Likewise.
394 (find_reg): Likewise.
395 (clear_reload_reg_in_use): Likewise.
396 (free_for_value_p): Likewise.
397 (allocate_reload_reg): Likewise.
398 (choose_reload_regs): Likewise.
399 (reload_adjust_reg_for_temp): Likewise.
400 (emit_reload_insns): Likewise.
401 (delete_output_reload): Likewise.
402 * rtlanal.c (subreg_get_info): Likewise.
403 * sched-deps.c (sched_analyze_reg): Likewise.
404 * sel-sched.c (init_regs_for_mode): Likewise.
405 (mark_unavailable_hard_regs): Likewise.
406 (choose_best_reg_1): Likewise.
407 (verify_target_availability): Likewise.
408 * valtrack.c (dead_debug_insert_temp): Likewise.
409 * var-tracking.c (track_loc_p): Likewise.
410 (emit_note_insn_var_location): Likewise.
411 * varasm.c (make_decl_rtl): Likewise.
412 * reginfo.c (choose_hard_reg_mode): Likewise.
413 (init_reg_modes_target): Refer directly to
414 this_target_regs->x_hard_regno_nregs.
416 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
418 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
419 instead of hard_regno_nregs.
421 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
423 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
424 end_hard_regno instead of hard_regno_nregs.
425 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
426 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
427 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
428 * ira-color.c (improve_allocation): Likewise.
429 * lra-assigns.c (find_hard_regno_for_1): Likewise.
430 * lra-lives.c (mark_regno_live): Likewise.
431 (mark_regno_dead): Likewise.
432 * lra-remat.c (operand_to_remat): Likewise.
433 * lra.c (collect_non_operand_hard_regs): Likewise.
434 * postreload.c (reload_combine_note_store): Likewise.
435 (move2add_valid_value_p): Likewise.
436 * reload.c (regno_clobbered_p): Likewise.
438 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
440 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
442 * config/v850/v850.c (v850_reorg): Likewise.
443 * reload.c (refers_to_regno_for_reload_p): Likewise.
444 (find_equiv_reg): Likewise.
445 * reload1.c (reload_reg_reaches_end_p): Likewise.
447 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
449 * caller-save.c (add_used_regs): Use REG_NREGS instead of
451 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
452 * config/arm/arm.c (output_move_neon): Likewise.
453 (arm_attr_length_move_neon): Likewise.
454 (neon_split_vcombine): Likewise.
455 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
456 (c6x_mark_reg_written): Likewise.
457 (c6x_dwarf_register_span): Likewise.
458 * config/i386/i386.c (ix86_save_reg): Likewise.
459 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
460 (rws_access_reg): Likewise.
461 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
462 * mode-switching.c (create_pre_exit): Likewise.
463 * ree.c (combine_reaching_defs): Likewise.
464 (add_removable_extension): Likewise.
465 * regcprop.c (find_oldest_value_reg): Likewise.
466 (copyprop_hardreg_forward_1): Likewise.
467 * reload.c (reload_inner_reg_of_subreg): Likewise.
468 (push_reload): Likewise.
469 (combine_reloads): Likewise.
470 (find_dummy_reload): Likewise.
471 (reload_adjust_reg_for_mode): Likewise.
472 * reload1.c (find_reload_regs): Likewise.
473 (forget_old_reloads_1): Likewise.
474 (reload_reg_free_for_value_p): Likewise.
475 (reload_adjust_reg_for_temp): Likewise.
476 (emit_reload_insns): Likewise.
477 (delete_output_reload): Likewise.
478 * sel-sched.c (choose_best_reg_1): Likewise.
479 (choose_best_pseudo_reg): Likewise.
481 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
482 Alan Hayward <alan.hayward@arm.com>
483 David Sherwood <david.sherwood@arm.com>
485 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
486 * target.def (slow_unaligned_access): New hook.
487 * targhooks.h (default_slow_unaligned_access): Declare.
488 * targhooks.c (default_slow_unaligned_access): New function.
489 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
490 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
491 * doc/tm.texi: Regenerate.
492 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
493 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
494 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
496 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
497 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
499 (rs6000_slow_unaligned_access): New function.
500 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
501 (expand_block_compare): Likewise.
502 (expand_strn_compare): Likewise.
503 (rs6000_rtx_costs): Likewise.
504 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
505 (riscv_slow_unaligned_access): Likewise.
506 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
507 (riscv_slow_unaligned_access_p): ...this and make static.
508 (riscv_option_override): Update accordingly.
509 (riscv_slow_unaligned_access): New function.
510 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
511 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
512 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
513 (rs6000_slow_unaligned_access): New function.
514 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
515 (rs6000_rtx_costs): Likewise.
516 * config/rs6000/rs6000-string.c (expand_block_compare)
517 (expand_strn_compare): Use targetm.slow_unaligned_access instead
518 of SLOW_UNALIGNED_ACCESS.
519 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
520 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
521 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
522 of SLOW_UNALIGNED_ACCESS.
523 * expmed.c (simple_mem_bitfield_p): Likewise.
524 * expr.c (alignment_for_piecewise_move): Likewise.
525 (emit_group_load_1): Likewise.
526 (emit_group_store): Likewise.
527 (copy_blkmode_from_reg): Likewise.
528 (emit_push_insn): Likewise.
529 (expand_assignment): Likewise.
530 (store_field): Likewise.
531 (expand_expr_real_1): Likewise.
532 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
533 * lra-constraints.c (simplify_operand_subreg): Likewise.
534 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
535 * gimple-ssa-store-merging.c: Likewise in block comment at start
537 * tree-ssa-strlen.c: Include target.h.
538 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
539 of SLOW_UNALIGNED_ACCESS.
540 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
542 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
544 PR rtl-optimization/82185
545 * expmed.c (emit_store_flag_int): Only test tem if it has been
548 2017-09-12 Richard Biener <rguenther@suse.de>
551 * match.pd ((FTYPE) N CMP CST): Fix typo.
553 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
555 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
557 (mips_near_type_p): Add 'short_call' attribute as a synonym
559 * doc/extend.texi (short_call): Document new function attribute.
561 2017-09-12 Jakub Jelinek <jakub@redhat.com>
564 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
565 assertion check that in the condition.
566 (get_atomic_generic_size): Likewise. Before testing if parameter
567 has pointer type, if it has array type, call for C++
568 default_conversion to perform array-to-pointer conversion.
570 2017-09-12 Richard Biener <rguenther@suse.de>
572 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
573 for operations we cannot scalarize.
575 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
577 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
578 vectors heap vectors. Clean up comments.
579 Make visited_bbs a reference.
580 (profitable_jump_thread_path): Make GC
581 vectors heap vectors. Clean up comments.
583 (convert_and_register_jump_thread_path): Make GC vectors heap
585 (check_subpath_and_update_thread_path): Same. Clean up comments.
586 Make visited_bbs a reference.
587 (handle_phi): Abstract common code to to
588 register_jump_thread_path_if_profitable.
589 Rename VAR_BB to DEF_BB.
591 Make GC vectors heap vectors.
592 Make visited_bbs a reference.
593 (handle_assignment): Same.
594 (register_jump_thread_path_if_profitable): New.
595 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
597 Make GC vectors heap vectors. Clean up comments.
598 Make visited_bbs a reference.
599 (find_jump_threads_backwards): Make visited_bbs live in the stack.
600 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
603 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
606 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
607 words of E_DImode object are reachable by xtensa_uimm8x4 access.
609 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
611 Revert r251800 and r251799.
613 2017-09-11 Martin Jambor <mjambor@suse.cz>
616 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
617 arguments in advance.
618 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
619 use it to find predecessor edges.
620 (naive_outof_ssa): Collect vector of predecessors.
622 2017-09-08 Jason Merrill <jason@redhat.com>
624 PR c++/70029 - ICE with ref-qualifier and -flto
625 * langhooks.h (struct lang_hooks_for_types): Add
626 copy_lang_qualifiers.
627 * attribs.c (build_type_attribute_qual_variant): Use it.
628 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
630 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
631 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
633 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
636 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
637 (*mulsi3_sp64): New instruction.
638 (mulsi3): New expander.
640 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
642 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
644 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
646 * sancov.c: Include memmodel.h.
648 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
651 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
654 2017-09-07 Carl Love <cel@us.ibm.com>
656 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
657 the sldi instruction.
659 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
661 * sancov.c: Include tm_p.h.
663 2017-09-07 Jakub Jelinek <jakub@redhat.com>
666 * output.h (switch_to_other_text_partition): New declaration.
667 * varasm.c (switch_to_other_text_partition): New function.
668 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
669 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
670 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
671 to the other text partition before emitting LCL label and switch back
672 after emitting the word after it.
674 2017-09-07 Richard Biener <rguenther@suse.de>
676 * passes.def (pass_split_crit_edges): Remove instance before PRE.
677 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
678 critical edges here, after loop init.
679 (pass_data_pre): Remove PROP_no_crit_edges flags.
680 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
681 for valueization of call args to avoid leaking VN_TOP.
682 (visit_use): Assert we do not visit default defs.
683 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
684 Use error_mark_node to more easily detect leaking VN_TOP.
685 All default-defs are varying, not VN_TOP. Mark them visited.
686 (run_scc_vn): Make code match comment.
688 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
690 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
691 OPTION_MASK_FLOAT128_KEYWORD.
692 (POWERPC_MASKS): Likewise.
693 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
694 support for the -mfloat128-type option, and make -mfloat128
695 default on PowerPC Linux systems. Define or undefine
696 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
697 Define __float128 to be __ieee128 if IEEE 128-bit support is
698 enabled, or undefine it.
699 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
700 Delete defining __FLOAT128_TYPE__.
701 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
702 -mfloat128-type option and make -mfloat128 default on PowerPC
704 (TARGET_FLOAT128_TYPE): Likewise.
705 (-mfloat128-type): Likewise.
706 * config/rs6000/rs6000.c (rs6000_option_override_internal):
707 Delete the -mfloat128-type option and make -mfloat128 default on
708 PowerPC Linux systems. Always use __ieee128 to be the keyword for
709 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
710 128-bit floating point is enabled. Change tests from using
711 -mfloat128-type to -mfloat128.
712 (rs6000_mangle_type): Use the correct mangling for the __float128
713 type even if normal long double is restricted to 64-bits.
714 (floatn_mode): Enable the _Float128 type by default on VSX Linux
716 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
717 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
718 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
720 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
721 documentation for -mfloat128.
723 2017-09-06 Olivier Hainque <hainque@adacore.com>
725 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
727 2017-09-06 Wish Wu <wishwu007@gmail.com>
728 Jakub Jelinek <jakub@redhat.com>
730 * asan.c (initialize_sanitizer_builtins): Add
731 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
732 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
733 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
734 BT_FN_VOID_UINT64_PTR variables.
735 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
736 (BT_FN_VOID_UINT16_UINT16): Likewise.
737 (BT_FN_VOID_UINT32_UINT32): Likewise.
738 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
739 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
740 (BT_FN_VOID_UINT64_PTR): Likewise.
741 * common.opt (flag_sanitize_coverage): New variable.
742 (fsanitize-coverage=trace-pc): Remove.
743 (fsanitize-coverage=): Add.
744 * flag-types.h (enum sanitize_coverage_code): New enum.
745 * fold-const.c (fold_range_test): Disable non-short-circuit
746 optimization if flag_sanitize_coverage.
747 (fold_truth_andor): Likewise.
748 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
749 * opts.c (COVERAGE_SANITIZER_OPT): Define.
750 (coverage_sanitizer_opts): New array.
751 (get_closest_sanitizer_option): Add OPTS argument, handle also
752 OPT_fsanitize_coverage_.
753 (parse_sanitizer_options): Adjusted to also handle
754 OPT_fsanitize_coverage_.
755 (common_handle_option): Add OPT_fsanitize_coverage_.
756 * sancov.c (instrument_comparison, instrument_switch): New function.
757 (sancov_pass): Add trace-cmp support.
758 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
759 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
760 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
761 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
762 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
763 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
764 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
765 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
766 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
767 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
769 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
771 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
772 error. Only quit immediately if parsing is complete.
773 (BEGIN): Initialize fatal_err and parse_done.
774 (begin fpu, end fpu): Check number of arguments.
775 (begin arch, end arch): Likewise.
776 (begin cpu, end cpu): Likewise.
777 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
778 (optalias): Likewise.
780 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
782 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
783 * config/arm/arm-isa.h: Delete. Move definitions to ...
784 * arm-cpus.in: ... here. Use new feature and fgroup values.
785 * config/arm/arm.c (arm_option_override): Use lower case for feature
787 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
788 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
789 * config/arm/parsecpu.awk (END): Add new command 'isa'.
791 (print_isa_bits_for): New function.
792 (gen_isa): New function.
793 (gen_comm_data): Use print_isa_bits_for.
794 (define feature): New keyword.
795 (define fgroup): New keyword.
796 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
797 (arm-isa.h): Add rule to generate file.
798 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
799 case for feature bit names.
801 2017-09-06 Richard Biener <rguenther@suse.de>
803 * tree-ssa-pre.c (NECESSARY): Remove.
804 (create_expression_by_pieces): Do not touch pass-local flags.
805 (insert_into_preds_of_block): Likewise.
806 (do_pre_regular_insertion): Likewise.
807 (eliminate_insert): Likewise.
808 (eliminate_dom_walker::before_dom_children): Likewise.
809 (fini_eliminate): Do not look at inserted_exprs.
810 (mark_operand_necessary): Remove.
811 (remove_dead_inserted_code): Replace with simple work-list
812 algorithm based on inserted_exprs and SSA uses.
813 (pass_pre::execute): Re-order fini_eliminate and
814 remove_dead_inserted_code.
816 2017-09-06 Olivier Hainque <hainque@adacore.com>
818 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
819 for VxWorks 7. Adjust surrounding comments.
821 2017-09-06 Richard Biener <rguenther@suse.de>
823 * gimple-ssa-strength-reduction.c
824 (find_candidates_dom_walker::before_dom_children): Also allow
827 2017-09-06 Richard Biener <rguenther@suse.de>
829 PR tree-optimization/82108
830 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
831 for gap in the non-permutation SLP case.
833 2017-09-06 Martin Jambor <mjambor@suse.cz>
835 PR tree-optimization/82078
836 * tree-sra.c (sort_and_splice_var_accesses): Move call to
837 add_access_to_work_queue...
838 (build_accesses_from_assign): ...here.
839 (propagate_all_subaccesses): Make sure racc is the group
840 representative, if there is one.
842 2017-09-06 Jakub Jelinek <jakub@redhat.com>
845 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
848 2017-09-06 Richard Biener <rguenther@suse.de>
850 * gimple-ssa-strength-reduction.c
851 (find_candidates_dom_walker::before_doom_children): Use a
852 type and not a mode check.
854 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
857 * config/arm/predicates.md (arm_general_adddi_operand): Create new
859 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
861 2017-09-05 Jeff Law <law@redhat.com>
863 PR tree-optimization/64910
864 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
865 cases where we have 3 or more operands.
867 2017-09-05 Jakub Jelinek <jakub@redhat.com>
870 * omp-low.c (lower_omp_for): Recompute tree invariant if
871 gimple_omp_for_initial/final is ADDR_EXPR.
874 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
875 into gimple val before gimplification fo the COND_EXPR.
877 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
879 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
880 REGION_COPY argument.
881 (thread_through_all_blocks): Remove unused argument to
882 duplicate_thread_path.
884 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
885 Alan Hayward <alan.hayward@arm.com>
886 David Sherwood <david.sherwood@arm.com>
888 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
889 Take a scalar_mode rather than a machine_mode.
890 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
891 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
892 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
893 (aarch64_gen_adjusted_ldpstp): Likewise.
894 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
896 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
897 Alan Hayward <alan.hayward@arm.com>
898 David Sherwood <david.sherwood@arm.com>
900 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
901 Take a scalar_int_mode instead of a machine_mode.
902 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
903 (aarch64_output_scalar_simd_mov_immediate): Likewise.
904 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
905 (aarch64_simd_attr_length_rglist): Delete.
906 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
907 a scalar_int_mode instead of a machine_mode.
908 (aarch64_add_offset): Likewise.
909 (aarch64_internal_mov_immediate): Likewise
910 (aarch64_add_constant_internal): Likewise.
911 (aarch64_add_constant): Likewise.
912 (aarch64_movw_imm): Likewise.
913 (aarch64_rtx_arith_op_extract_p): Likewise.
914 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
915 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
916 Remove assert that the mode isn't a vector.
917 (aarch64_output_scalar_simd_mov_immediate): Likewise.
918 (aarch64_expand_mov_immediate): Update calls after above changes.
919 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
920 (aarch64_and_bitmask_imm): Check for scalar integer modes.
921 (aarch64_move_imm): Likewise.
922 (aarch64_can_const_movi_rtx_p): Likewise.
923 (aarch64_strip_extend): Likewise.
924 (aarch64_extr_rtx_p): Likewise.
925 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
926 a CONST_INT when the mode parameter is VOIDmode.
927 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
929 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
931 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
932 * stor-layout.c (bitwise_mode_for_mode): Likewise.
933 (bitwise_type_for_mode): Update accordingly.
935 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
937 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
938 * stor-layout.c (mode_for_size_tree): Likewise.
939 (mode_for_array): Update accordingly.
940 (layout_decl): Likewise.
941 (compute_record_mode): Likewise. Only set the mode once.
943 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
945 * target.def (get_mask_mode): Change return type to opt_mode.
947 * doc/tm.texi: Regenerate.
948 * targhooks.h (default_get_mask_mode): Return an opt_mode.
949 * targhooks.c (default_get_mask_mode): Likewise.
950 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
951 * optabs-query.c (can_vec_mask_load_store_p): Update use of
952 targetm.get_mask_mode.
953 * tree.c (build_truth_vector_type): Likewise.
955 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
957 * machmode.h (mode_for_vector): Return an opt_mode.
958 * stor-layout.c (mode_for_vector): Likewise.
959 (mode_for_int_vector): Update accordingly.
960 (layout_type): Likewise.
961 * config/i386/i386.c (emit_memmov): Likewise.
962 (ix86_expand_set_or_movmem): Likewise.
963 (ix86_expand_vector_init): Likewise.
964 (ix86_get_mask_mode): Likewise.
965 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
967 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
968 * expmed.c (extract_bit_field_1): Likewise.
969 * expr.c (expand_expr_real_2): Likewise.
970 * optabs-query.c (can_vec_perm_p): Likewise.
971 (can_vec_mask_load_store_p): Likewise.
972 * optabs.c (expand_vec_perm): Likewise.
973 * targhooks.c (default_get_mask_mode): Likewise.
974 * tree-vect-stmts.c (vectorizable_store): Likewise.
975 (vectorizable_load): Likewise.
976 (get_vectype_for_scalar_type_and_size): Likewise.
978 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
980 * machmode.h (mode_for_int_vector): New function.
981 * stor-layout.c (mode_for_int_vector): Likewise.
982 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
983 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
984 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
985 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
986 (s390_expand_vcond): Likewise.
988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
990 * machmode.h (opt_machine_mode): New type.
991 (opt_mode<T>): Allow construction from anything that can be
993 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
994 (mode_for_size): Return an opt_machine_mode.
995 * stor-layout.c (mode_for_size): Likewise.
996 (mode_for_size_tree): Update call accordingly.
997 (bitwise_mode_for_mode): Likewise.
998 (make_fract_type): Likewise.
999 (make_accum_type): Likewise.
1000 * caller-save.c (replace_reg_with_saved_mem): Update call
1002 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1003 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1004 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1005 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1006 * expmed.c (extract_bit_field_1): Likewise.
1007 * reload.c (get_secondary_mem): Likewise.
1008 * varasm.c (assemble_integer): Likewise.
1009 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1012 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1014 * machmode.h (decimal_float_mode_for_size): New function.
1015 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1016 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1017 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1018 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1019 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1020 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1022 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1024 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1025 (get_builtin_sync_mode): Likewise.
1026 (expand_ifn_atomic_compare_exchange): Likewise.
1027 (expand_builtin_atomic_clear): Likewise.
1028 (expand_builtin_atomic_test_and_set): Likewise.
1029 (fold_builtin_atomic_always_lock_free): Likewise.
1030 * calls.c (compute_argument_addresses): Likewise.
1031 (emit_library_call_value_1): Likewise.
1032 (store_one_arg): Likewise.
1033 * combine.c (combine_instructions): Likewise.
1034 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1035 * config/arm/arm.c (arm_function_value): Likewise.
1036 (aapcs_allocate_return_reg): Likewise.
1037 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1038 * config/i386/i386.c (construct_container): Likewise.
1039 (ix86_gimplify_va_arg): Likewise.
1040 (ix86_expand_sse_cmp): Likewise.
1041 (emit_memmov): Likewise.
1042 (emit_memset): Likewise.
1043 (expand_small_movmem_or_setmem): Likewise.
1044 (ix86_expand_pextr): Likewise.
1045 (ix86_expand_pinsr): Likewise.
1046 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1047 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1049 * config/mips/mips.c (mips_function_value_1) Likewise.
1050 (mips_block_move_straight): Likewise.
1051 (mips_expand_ins_as_unaligned_store): Likewise.
1052 * config/powerpcspe/powerpcspe.c
1053 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1054 (rs6000_darwin64_record_arg_flush): Likewise.
1055 * config/rs6000/rs6000.c
1056 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1057 (rs6000_darwin64_record_arg_flush): Likewise.
1058 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1059 (sparc_function_value_1): Likewise.
1060 * config/spu/spu.c (adjust_operand): Likewise.
1061 (spu_emit_branch_or_set): Likewise.
1062 (arith_immediate_p): Likewise.
1063 * emit-rtl.c (gen_lowpart_common): Likewise.
1064 * expr.c (expand_expr_real_1): Likewise.
1065 * function.c (assign_parm_setup_block): Likewise.
1066 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1067 * reload1.c (alter_reg): Likewise.
1068 * stor-layout.c (mode_for_vector): Likewise.
1069 (layout_type): Likewise.
1071 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1073 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1074 (spu_convert_move): Likewise.
1075 * lower-subreg.c (resolve_simple_move): Likewise.
1077 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1080 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1081 define_insn to a define_expand.
1082 (altivec_vsum2sws_direct): New define_insn.
1083 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1085 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1087 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1090 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1095 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1096 realign stack if stack alignment needed is less than incoming
1099 2017-09-05 Marek Polacek <polacek@redhat.com>
1102 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1105 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1107 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1109 2017-09-05 Richard Biener <rguenther@suse.de>
1111 PR tree-optimization/82084
1112 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1114 2017-09-05 Richard Biener <rguenther@suse.de>
1116 PR tree-optimization/82102
1117 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1119 2017-09-05 Martin Liska <mliska@suse.cz>
1121 PR tree-optimization/82032
1122 * tree-cfg.c (generate_range_test): New function.
1123 * tree-cfg.h (generate_range_test): Declared here.
1124 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1125 (cleanup_control_expr_graph): Use it.
1126 * tree-switch-conversion.c (try_switch_expansion): Remove
1128 (emit_case_nodes): Use generate_range_test.
1130 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1133 * config/i386/i386.md (*<btsc><mode>_mask): Add
1134 TARGET_USE_BT to insn constraint.
1135 (*btr<mode>_mask): Ditto.
1137 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1139 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1140 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1142 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1145 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1146 TARGET_NEON and TARGET_IWMMXT.
1147 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1148 TARGET_NEON and TARGET_IWMMXT.
1149 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1151 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1153 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1154 (ix86_rewrite_tls_address): Ditto.
1155 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1156 (ix86_rewrite_tls_address_1): Ditto.
1157 (ix86_rewrite_tls_address): Ditto.
1158 * config/i386/predicates.md (tls_address_pattern): New predicate.
1159 * config/i386/i386.md (TLS address splitter): New splitter.
1161 2017-09-04 Richard Biener <rguenther@suse.de>
1163 PR tree-optimization/82084
1164 * fold-const.h (can_native_encode_string_p): Declare.
1165 * fold-const.c (can_native_encode_string_p): Factor out from ...
1166 (native_encode_string): ... here.
1167 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1168 vectorizing stores from constants we later cannot handle.
1170 2017-09-04 Marek Polacek <polacek@redhat.com>
1173 * doc/invoke.texi: Update -Wtautological-compare documentation.
1175 2017-09-04 Jeff Law <law@redhat.com>
1177 PR tree-optimization/64910
1178 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1179 swap the first and last operand if the last is a constant.
1181 2017-09-04 Marek Polacek <polacek@redhat.com>
1184 * convert.c (do_narrow): When sanitizing signed integer overflows,
1185 bail out for signed types.
1186 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1188 2017-09-04 Richard Biener <rguenther@suse.de>
1190 PR tree-optimization/82060
1191 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1192 Move devirtualization after stmt folding and before EH/AB/noreturn
1193 cleanup to get the stmt refs canonicalized. Use a bool instead
1194 of gimple_modified_p since that doesn't work for NOPs. Schedule
1195 NOPs generated by folding for removal.
1197 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1198 Alan Hayward <alan.hayward@arm.com>
1199 David Sherwood <david.sherwood@arm.com>
1201 * coretypes.h (pad_direction): New enum.
1202 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1203 (FUNCTION_ARG_PADDING): Likewise.
1204 * target.def (function_arg_padding): New hook.
1205 * targhooks.h (default_function_arg_padding): Declare.
1206 * targhooks.c (default_function_arg_padding): New function.
1207 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1208 (TARGET_FUNCTION_ARG_PADDING): ...this.
1209 * doc/tm.texi: Regenerate.
1210 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1211 instead of direction.
1212 (compute_argument_addresses): Likewise.
1213 (load_register_parameters): Likewise.
1214 (emit_library_call_value_1): Likewise.
1215 (store_one_arg): Use targetm.calls.function_arg_padding instead
1216 of FUNCTION_ARG_PADDING.
1217 (must_pass_in_stack_var_size_or_pad): Likewise.
1218 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1219 (emit_group_store): Likewise.
1220 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1221 instead of FUNCTION_ARG_PADDING.
1222 (emit_push_insn): Likewise, and propagate enum change throughout
1224 * function.h (direction): Delete.
1225 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1227 * function.c (assign_parm_find_stack_rtl): Likewise.
1228 (assign_parm_setup_block_p): Likewise.
1229 (assign_parm_setup_block): Likewise.
1230 (gimplify_parameters): Likewise.
1231 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1232 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1234 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1235 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1236 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1237 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1238 (aarch64_function_arg_padding): ...this new function.
1239 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1240 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1241 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1242 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1243 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1244 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1245 (arm_pad_arg_upward): Replace with...
1246 (arm_function_arg_padding): ...this new function.
1247 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1249 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1250 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1251 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1252 (ia64_hpux_function_arg_padding): Replace with...
1253 (ia64_function_arg_padding): ...this new function. Use pad_direction
1254 instead of direction. Check for TARGET_HPUX.
1255 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1256 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1257 (iq2000_function_arg_padding): New function.
1258 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1259 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1260 (mips_function_arg_padding): ...this new function.
1261 (mips_pad_reg_upward): Update accordingly.
1262 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1263 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1264 targetm.calls.function_arg_padding.
1265 (FUNCTION_ARG_PADDING): Delete.
1266 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1267 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1268 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1269 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1270 (nios2_block_reg_padding): Return pad_direction instead of direction.
1271 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1272 instead of direction.
1273 (nios2_function_arg_padding): Likewise. Make static.
1274 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1275 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1276 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1277 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1278 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1279 (pa_function_arg_padding): Make static. Return pad_direction instead
1281 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1282 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1283 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1284 instead of direction. Use targetm.calls.function_arg_padding.
1285 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1286 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1287 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1288 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1289 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1291 (function_arg_padding): Rename to...
1292 (rs6000_function_arg_padding): ...this. Make static. Return
1293 pad_direction instead of direction.
1294 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1295 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1296 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1297 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1298 instead of direction. Use targetm.calls.function_arg_padding.
1299 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1300 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1301 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1302 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1303 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1304 (function_arg_padding): Rename to...
1305 (rs6000_function_arg_padding): ...this. Make static. Return
1306 pad_direction instead of direction.
1307 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1308 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1309 * config/s390/s390.c (s390_function_arg_padding): New function.
1310 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1311 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1312 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1313 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1314 (function_arg_padding): Rename to...
1315 (sparc_function_arg_padding): ...this. Make static. Return
1316 pad_direction instead of direction.
1317 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1318 * config/spu/spu.c (spu_function_arg_padding): New function.
1319 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1320 * system.h (FUNCTION_ARG_PADDING): Poison.
1322 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1323 Alan Hayward <alan.hayward@arm.com>
1324 David Sherwood <david.sherwood@arm.com>
1326 * target.def (modes_tieable_p): New hook.
1327 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1328 (TARGET_MODES_TIEABLE_P): ...this.
1329 * doc/tm.texi.in: Regenerate.
1330 * hooks.h (hook_bool_mode_mode_true): Declare.
1331 * hooks.c (hook_bool_mode_mode_true): New function.
1332 * combine.c (subst): Use targetm.modes_tieable_p instead of
1334 * dse.c (find_shift_sequence): Likewise.
1335 * expmed.c (extract_low_bits): Likewise.
1336 * lower-subreg.c: Include target.h.
1337 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1339 * rtlanal.c (rtx_cost): Likewise.
1340 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1341 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1342 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1343 (TARGET_MODES_TIEABLE_P): Redefine.
1344 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1345 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1346 (TARGET_MODES_TIEABLE_P): Redefine.
1347 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1348 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1349 (arc_modes_tieable_p): New function.
1350 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1351 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1352 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1353 (arm_modes_tieable_p): Make static.
1354 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1355 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1356 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1357 (TARGET_MODES_TIEABLE_P): Redefine.
1358 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1359 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1360 (TARGET_MODES_TIEABLE_P): Redefine.
1361 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1362 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1363 (cr16_modes_tieable_p): New function.
1364 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1365 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1366 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1367 (TRULY_NOOP_TRUNCATION): Update comment.
1368 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1369 (TRULY_NOOP_TRUNCATION): Update comment.
1370 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1371 (frv_modes_tieable_p): New function.
1372 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1373 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1374 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1375 (TARGET_MODES_TIEABLE_P): Redefine.
1376 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1377 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1378 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1379 (TARGET_MODES_TIEABLE_P): Redefine.
1380 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1381 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1382 (ia64_modes_tieable_p): New function.
1383 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1384 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1385 (iq2000_modes_tieable_p): New function.
1386 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1387 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1388 (lm32_modes_tieable_p): New function.
1389 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1390 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1391 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1392 (TARGET_MODES_TIEABLE_P): Redefine.
1393 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1394 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1395 (m32r_modes_tieable_p): New function.
1396 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1397 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1398 (m68k_modes_tieable_p): New function.
1399 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1400 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1401 (mcore_modes_tieable_p): New function.
1402 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1403 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1405 (TARGET_MODES_TIEABLE_P): Redefine.
1406 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1407 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1408 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1409 (TARGET_MODES_TIEABLE_P): Redefine.
1410 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1411 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1412 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1413 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1414 (mn10300_modes_tieable_p): ...this and make static.
1415 (TARGET_MODES_TIEABLE_P): Redefine.
1416 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1417 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1418 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1419 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1420 (msp430_modes_tieable_p): Make static.
1421 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1422 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1423 (TARGET_MODES_TIEABLE_P): Redefine.
1424 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1425 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1426 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1427 (TARGET_MODES_TIEABLE_P): Redefine.
1428 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1429 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1430 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1431 (TARGET_MODES_TIEABLE_P): Redefine.
1432 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1433 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1434 (pdp11_modes_tieable_p): New function.
1435 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1436 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1437 (rs6000_modes_tieable_p): New function.
1438 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1439 * config/powerpcspe/powerpcspe.md: Update comment.
1440 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1441 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1442 (TARGET_MODES_TIEABLE_P): Redefine.
1443 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1444 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1445 (rl78_modes_tieable_p): New function.
1446 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1447 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1448 (rs6000_modes_tieable_p): New function.
1449 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1450 * config/rs6000/rs6000.md: Update comment.
1451 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1452 * config/rx/rx.c (rx_modes_tieable_p): New function.
1453 (TARGET_MODES_TIEABLE_P): Redefine.
1454 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1455 * config/s390/s390.c (s390_modes_tieable_p): New function.
1456 (TARGET_MODES_TIEABLE_P): Redefine.
1457 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1458 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1459 (sh_modes_tieable_p): New function.
1460 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1461 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1462 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1463 (sparc_modes_tieable_p): Make static.
1464 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1465 * config/spu/spu.c (spu_modes_tieable_p): New function.
1466 (TARGET_MODES_TIEABLE_P): Redefine.
1467 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1468 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1469 (TARGET_MODES_TIEABLE_P): Redefine.
1470 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1471 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1472 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1473 * config/v850/v850.c (v850_modes_tieable_p): New function.
1474 (TARGET_MODES_TIEABLE_P): Redefine.
1475 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1476 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1477 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1478 (visium_modes_tieable_p): New function.
1479 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1480 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1481 (xtensa_modes_tieable_p): New function.
1482 * system.h (MODES_TIEABLE_P): Poison.
1484 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1485 Alan Hayward <alan.hayward@arm.com>
1486 David Sherwood <david.sherwood@arm.com>
1488 * target.def (hard_regno_mode_ok): New hook.
1489 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1490 (TARGET_HARD_REGNO_MODE_OK): ...this.
1491 * doc/tm.texi.in: Regenerate.
1492 * hooks.h (hook_bool_uint_mode_true): Declare.
1493 * hooks.c (hook_bool_uint_mode_true): New function.
1494 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1496 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1497 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1498 instead of HARD_REGNO_MODE_OK.
1499 * caller-save.c: Include target.h.
1500 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1502 * combine.c (can_combine_p): Likewise.
1503 (combinable_i3pat): Likewise.
1504 (can_change_dest_mode): Likewise.
1505 * expr.c (init_expr_target): Likewise.
1506 (convert_move): Likewise.
1507 (convert_modes): Likewise.
1508 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1509 (setup_prohibited_mode_move_regs): Likewise.
1510 * ira.h (target_ira): Likewise.
1511 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1512 * lra-constraints.c (process_alt_operands): Likewise.
1513 (split_reg): Likewise.
1514 * recog.c (peep2_find_free_register): Likewise.
1515 * ree.c (combine_reaching_defs): Likewise.
1516 * regcprop.c (maybe_mode_change): Likewise.
1517 * reginfo.c (init_reg_sets_1): Likewise.
1518 (choose_hard_reg_mode): Likewise.
1519 (simplifiable_subregs): Likewise.
1520 * regrename.c (check_new_reg_p): Likewise.
1521 * reload.c (find_valid_class): Likewise.
1522 (find_valid_class_1): Likewise.
1523 (reload_inner_reg_of_subreg): Likewise.
1524 (push_reload): Likewise.
1525 (combine_reloads): Likewise.
1526 (find_dummy_reload): Likewise.
1527 (find_reloads): Likewise.
1528 * reload1.c (find_reg): Likewise.
1529 (set_reload_reg): Likewise.
1530 (allocate_reload_reg): Likewise.
1531 (choose_reload_regs): Likewise.
1532 (reload_adjust_reg_for_temp): Likewise.
1533 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1534 (simplify_subreg_regno): Likewise.
1535 * sel-sched.c (init_regs_for_mode): Likewise.
1536 * varasm.c (make_decl_rtl): Likewise.
1537 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1538 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1540 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1541 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1542 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1543 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1544 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1545 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1546 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1547 (arc_mode_class): Delete.
1548 (HARD_REGNO_MODE_OK): Delete.
1549 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1550 (arc_hard_regno_mode_ok): Rename old array to...
1551 (arc_hard_regno_mode_ok_modes): ...this.
1552 (arc_conditional_register_usage): Update accordingly.
1553 (arc_mode_class): Make static.
1554 (arc_hard_regno_mode_ok): New function.
1555 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1556 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1557 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1558 (arm_hard_regno_mode_ok): Make static.
1559 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1561 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1562 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1563 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1565 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1566 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1567 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1568 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1569 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1570 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1571 * config/bfin/predicates.md (valid_reg_operand): Use
1572 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1573 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1574 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1576 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1577 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1578 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1579 (cr16_hard_regno_mode_ok): Make static and return a bool.
1580 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1581 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1582 (cris_hard_regno_mode_ok): New function.
1583 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1584 (epiphany_mode_class): Delete.
1585 (HARD_REGNO_MODE_OK): Delete.
1586 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1587 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1588 (hard_regno_mode_ok): Rename to...
1589 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1590 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1591 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1593 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1594 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1595 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1596 (frv_hard_regno_mode_ok): Make static and return a bool.
1597 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1599 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1600 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1601 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1602 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1604 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1605 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1606 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1607 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1609 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1610 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1611 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1612 (ia64_hard_regno_mode_ok): New function.
1613 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1614 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1615 (iq2000_hard_regno_mode_ok): New function.
1616 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1617 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1618 (lm32_hard_regno_mode_ok): New function.
1619 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1620 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1621 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1622 instead of HARD_REGNO_MODE_OK.
1623 (m32c_hard_regno_ok): Rename to...
1624 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1625 (m32c_cannot_change_mode_class): Update accordingly.
1626 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1627 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1628 (m32r_mode_class): Delete.
1629 (HARD_REGNO_MODE_OK): Delete.
1630 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1631 (m32r_hard_regno_mode_ok): Rename to...
1632 (m32r_hard_regno_modes): ...this.
1633 (m32r_mode_class): Make static.
1634 (m32r_hard_regno_mode_ok): New function.
1635 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1636 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1637 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1638 (m68k_hard_regno_mode_ok): Make static.
1639 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1640 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1641 (mcore_hard_regno_mode_ok): New function.
1642 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1643 (HARD_REGNO_MODE_OK): Delete.
1644 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1646 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1647 (microblaze_hard_regno_mode_ok): New function.
1648 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1649 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1650 (mips_hard_regno_mode_ok): Delete.
1651 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1652 (mips_hard_regno_mode_ok_p): ...this and make static.
1653 (mips_hard_regno_mode_ok_p): Rename to...
1654 (mips_hard_regno_mode_ok_uncached): ...this.
1655 (mips_hard_regno_mode_ok): New function.
1656 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1657 of HARD_REGNO_MODE_OK.
1658 (mips_option_override): Update after above name changes.
1659 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1660 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1661 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1662 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1663 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1664 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1665 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1666 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1667 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1668 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1669 (msp430_hard_regno_mode_ok): Make static and return a bool.
1670 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1671 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1672 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1674 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1675 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1676 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1677 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1678 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1679 (PA_HARD_REGNO_MODE_OK): ...this
1680 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1681 (PA_HARD_REGNO_MODE_OK): ...this.
1682 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1683 (pa_hard_regno_mode_ok): New function.
1684 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1685 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1686 (pdp11_hard_regno_mode_ok): New function.
1687 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1688 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1690 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1692 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1693 (rs6000_hard_regno_mode_ok): Rename to...
1694 (rs6000_hard_regno_mode_ok_uncached): ...this.
1695 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1696 (rs6000_hard_regno_mode_ok): New function.
1697 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1698 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1699 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1700 (riscv_hard_regno_mode_ok): ...this and make static.
1701 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1702 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1703 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1704 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1705 (rl78_hard_regno_mode_ok): Make static and return bool.
1706 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1707 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1709 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1711 (rs6000_hard_regno_mode_ok): Rename to...
1712 (rs6000_hard_regno_mode_ok_uncached): ...this.
1713 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1714 (rs6000_hard_regno_mode_ok): New function.
1715 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1716 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1717 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1718 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1719 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1720 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1721 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1722 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1723 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1724 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1725 (sh_hard_regno_mode_ok): Make static.
1726 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1727 instead of HARD_REGNO_MODE_OK.
1728 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1729 (sparc_mode_class): Delete.
1730 (HARD_REGNO_MODE_OK): Delete.
1731 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1732 (hard_regno_mode_classes): Make static.
1733 (sparc_mode_class): Likewise.
1734 (sparc_hard_regno_mode_ok): New function.
1735 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1736 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1737 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1739 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1740 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1741 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1742 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1743 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1745 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1746 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1747 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1748 (visium_hard_regno_mode_ok): New function.
1749 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1750 instead of HARD_REGNO_MODE_OK.
1751 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1752 (HARD_REGNO_MODE_OK): Delete.
1753 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1754 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1755 (xtensa_option_override): Update accordingly.
1756 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1757 (xtensa_hard_regno_mode_ok): New function.
1758 * system.h (HARD_REGNO_MODE_OK): Poison.
1760 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1761 Alan Hayward <alan.hayward@arm.com>
1762 David Sherwood <david.sherwood@arm.com>
1764 * target.def (hard_regno_call_part_clobbered): New hook.
1765 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1766 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1767 * doc/tm.texi: Regenerate.
1768 * hooks.h (hook_bool_uint_mode_false): Declare.
1769 * hooks.c (hook_bool_uint_mode_false): New function.
1770 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1771 * cselib.c (cselib_process_insn): Use
1772 targetm.hard_regno_call_part_clobbered instead of
1773 HARD_REGNO_CALL_PART_CLOBBERED.
1774 * ira-conflicts.c (ira_build_conflicts): Likewise.
1775 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1776 * lra-constraints.c (need_for_call_save_p): Likewise.
1777 * lra-lives.c: Include target.h.
1778 (check_pseudos_live_through_calls): Use
1779 targetm.hard_regno_call_part_clobbered instead of
1780 HARD_REGNO_CALL_PART_CLOBBERED.
1781 * regcprop.c: Include target.h.
1782 (copyprop_hardreg_forward_1): Use
1783 targetm.hard_regno_call_part_clobbered instead of
1784 HARD_REGNO_CALL_PART_CLOBBERED.
1785 * reginfo.c (choose_hard_reg_mode): Likewise.
1786 * regrename.c (check_new_reg_p): Likewise.
1787 * reload.c (find_equiv_reg): Likewise.
1788 * reload1.c (emit_reload_insns): Likewise.
1789 * sched-deps.c (deps_analyze_insn): Likewise.
1790 * sel-sched.c (init_regs_for_mode): Likewise.
1791 (mark_unavailable_hard_regs): Likewise.
1792 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1793 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1794 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1796 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1797 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1798 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1800 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1802 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1803 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1804 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1806 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1807 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1808 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1810 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1811 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1813 * config/powerpcspe/powerpcspe.c
1814 (rs6000_hard_regno_call_part_clobbered): New function.
1815 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1816 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1817 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1819 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1820 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1821 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1823 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1824 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1825 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1827 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1828 Alan Hayward <alan.hayward@arm.com>
1829 David Sherwood <david.sherwood@arm.com>
1831 * rtl.h (subreg_memory_offset): Declare.
1832 * emit-rtl.c (subreg_memory_offset): New function.
1833 * expmed.c (store_bit_field_1): Use it.
1834 * expr.c (undefined_operand_subword_p): Likewise.
1835 * simplify-rtx.c (simplify_subreg): Likewise.
1837 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1839 PR rtl-optimization/57448
1842 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1843 using atomic_load pattern.
1844 (expand_atomic_store): Likewise.
1846 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1849 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1850 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1851 handling. Use replace_call_with_value with NULL instead of
1852 gsi_replace, unlink_stmt_vdef and release_defs.
1854 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1857 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1859 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1862 * rtl.h (emit_library_call_value_1): Declare.
1863 (emit_library_call): Replace declaration with a series of overloads.
1864 Remove the parameter count argument.
1865 (emit_library_call_value): Likewise.
1866 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1867 with an "rtx_mode_t *".
1868 (emit_library_call_value): Delete.
1869 (emit_library_call): Likewise.
1870 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1871 (asan_emit_allocas_unpoison): Likewise.
1872 * builtins.c (expand_builtin_powi): Likewise.
1873 (expand_asan_emit_allocas_unpoison): Likewise.
1874 * cfgexpand.c (expand_main_function): Likewise.
1875 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1876 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1877 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1878 * config/arm/arm.c (arm_trampoline_init): Likewise.
1879 (arm_call_tls_get_addr): Likewise.
1880 (arm_expand_divmod_libfunc): Likewise.
1881 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1882 (smulsi3_highpart): Likewise.
1883 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1884 (c6x_expand_compare): Likewise.
1885 (c6x_expand_movmem): Likewise.
1886 * config/frv/frv.c (frv_trampoline_init): Likewise.
1887 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1888 (ix86_expand_divmod_libfunc): Likewise.
1889 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1890 (ia64_expand_compare): Likewise.
1891 (ia64_profile_hook): Likewise.
1892 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1893 (nonlocal_goto): Likewise.
1894 (restore_stack_nonlocal): Likewise.
1895 * config/m32r/m32r.c (block_move_call): Likewise.
1896 (m32r_trampoline_init): Likewise.
1897 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1898 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1899 (m68k_call_m68k_read_tp): Likewise.
1900 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1901 (microblaze_expand_divide): Likewise.
1902 * config/mips/mips.h (mips_args): Likewise.
1903 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1904 (MIPS_ICACHE_SYNC): Likewise.
1905 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1906 (nios2_trampoline_init): Likewise.
1907 * config/pa/pa.c (hppa_tls_call): Likewise.
1908 (pa_trampoline_init): Likewise.
1909 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1910 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1911 (expand_strn_compare): Likewise.
1912 (rs6000_generate_compare): Likewise.
1913 (rs6000_expand_float128_convert): Likewise.
1914 (output_profile_hook): Likewise.
1915 (rs6000_trampoline_init): Likewise.
1916 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1917 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1918 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1919 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1920 (rs6000_generate_compare): Likewise.
1921 (rs6000_expand_float128_convert): Likewise.
1922 (output_profile_hook): Likewise.
1923 (rs6000_trampoline_init): Likewise.
1924 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1925 * config/sh/sh.c (sh_trampoline_init): Likewise.
1926 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1927 (sparc_emit_float_lib_cmp): Likewise.
1928 (sparc32_initialize_trampoline): Likewise.
1929 (sparc64_initialize_trampoline): Likewise.
1930 (sparc_profile_hook): Likewise.
1931 * config/spu/spu.c (ea_load_store): Likewise.
1932 * config/spu/spu.md (floatunssidf2): Likewise.
1933 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1934 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1935 * config/visium/visium.c (expand_block_move_4): Likewise.
1936 (expand_block_move_2): Likewise.
1937 (expand_block_move_1): Likewise.
1938 (expand_block_set_4): Likewise.
1939 (expand_block_set_2): Likewise.
1940 (expand_block_set_1): Likewise.
1941 (visium_trampoline_init): Likewise.
1942 (visium_profile_hook): Likewise.
1943 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1944 (xtensa_setup_frame_addresses): Likewise.
1945 (xtensa_trampoline_init): Likewise.
1946 * except.c (sjlj_emit_function_enter): Likewise.
1947 (sjlj_emit_function_exit): Likewise.
1948 * explow.c (allocate_dynamic_stack_space): Likewise.
1949 (probe_stack_range): Likewise.
1950 * expr.c (convert_mode_scalar): Likewise.
1951 * optabs.c (expand_binop): Likewise.
1952 (expand_twoval_binop_libfunc): Likewise.
1953 (expand_unop): Likewise.
1954 (prepare_cmp_insn): Likewise.
1955 (prepare_float_lib_cmp): Likewise.
1956 (expand_float): Likewise.
1957 (expand_fix): Likewise.
1958 (expand_fixed_convert): Likewise.
1959 (maybe_emit_sync_lock_test_and_set): Likewise.
1960 (expand_atomic_compare_and_swap): Likewise.
1961 (expand_mem_thread_fence): Likewise.
1962 (expand_atomic_fetch_op): Likewise.
1964 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1966 * doc/generic.texi (OpenACC): Adjust URL.
1967 * doc/invoke.texi (C Dialect Options): Ditto.
1969 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1971 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1972 predicate for operand 1. Add (m,<S>) constraint.
1973 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1974 Prevent memory operand 1 with register operand 2.
1976 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1978 PR rtl-optimization/82024
1979 * combine.c (try_combine): If the combination result is a PARALLEL,
1980 and we only need to retain the SET in there that would be placed
1981 at I2, check that we can place that at I3 instead, before doing so.
1983 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1986 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1988 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1989 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1992 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1993 Jeff Law <law@redhat.com>
1995 * varasm.c (bss_initializer_p): Do not put constants into .bss
1996 (categorize_decl_for_section): Handle bss_initializer_p returning
1997 false when DECL_INITIAL is NULL.
1999 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2002 * config/s390/s390.c (s390_can_inline_p): New function.
2004 2017-09-01 Jeff Law <law@redhat.com>
2006 PR tree-optimization/82052
2007 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2008 Always initialize the returned slot after a hash table miss
2009 when INSERT is true.
2011 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2013 * config/s390/s390.md (mem_signal_fence): Remove.
2014 * doc/md.texi (mem_signal_fence): Remove.
2015 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2017 * target-insns.def (mem_signal_fence): Remove.
2019 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2022 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2025 * asan.c (create_odr_indicator): Strip name encoding from assembler
2026 name before appending it after __odr_asan_.
2028 2017-09-01 Martin Liska <mliska@suse.cz>
2030 PR tree-optimization/82059
2031 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2032 frequency only when an edge is redirected.
2034 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2036 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2037 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2038 (arc_conditional_register_usage): Remove ARC600 lp_count
2040 (arc_file_start): Emit Tag_ARC_CPU_variation.
2041 (arc_can_use_doloop_p): New conditions to use ZOLs.
2042 (hwloop_fail): New function.
2043 (hwloop_optimize): Likewise.
2044 (hwloop_pattern_reg): Likewise.
2045 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2046 (arc_reorg_loops): New function, calls reorg_loops.
2047 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2048 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2050 (arc_loop_hazard): Remove function, functionality moved into
2052 (arc_hazard): Remove arc_loop_hazard call.
2053 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2054 into hwloop_optimize.
2055 (arc_label_align): Remove ZOL handling.
2056 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2057 * config/arc/arc.md (doloop_begin): Remove pattern.
2058 (doloop_begin_i): Likewise.
2059 (doloop_end_i): Likewise.
2060 (doloop_fallback): Likewise.
2061 (doloop_fallback_m): Likewise.
2062 (doloop_end): Reimplement expand.
2063 (arc_lp): New pattern for LP instruction.
2064 (loop_end): New pattern.
2065 (loop_fail): Likewise.
2066 (decrement_and_branch_until_zero): Likewise.
2067 * config/arc/arc.opt (mlpc-width): New option.
2068 * doc/invoke.texi (mlpc-width): Document option.
2070 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2072 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2073 (arc_ccfsm_advance): Fix checking for delay slots.
2074 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2076 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2078 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2080 (movhi_insn): Update store instruction constraint which are saving
2081 6-bit short immediates.
2082 (movsi_insn): Consider also short scaled load operations.
2083 (zero_extendhisi2_i): Use Usd constraint instead of T.
2084 (extendhisi2_i): Add q constraint.
2085 (arc_clzsi2): Add type and length attributes.
2086 (arc_ctzsi2): Likewise.
2087 * config/arc/constraints.md (Usc): Update constraint, the
2088 assembler can parse two relocations for a single instruction.
2090 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2092 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2093 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2095 2017-08-31 Olivier Hainque <hainque@adacore.com>
2097 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2098 match as powerpc-wrs-vxworks*.
2100 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2102 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2103 register constraint for by-element operand.
2104 (aarch64_mls_elt_merge<mode>): Likewise.
2106 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2108 * config/arc/arc.c (arc_can_follow_jump): Check for short
2111 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2113 * config.gcc: Use g.opt for arc.
2114 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2115 functionality moved to ...
2116 (legitimate_scaled_address_p): New function, ...here.
2117 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2118 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2119 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2121 (arc_override_options): Handle G option.
2122 (arc_output_pic_addr_const): Correct function definition.
2123 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2124 (arc_decl_anon_ns_mem_p): Delete.
2125 (arc_in_small_data_p): Overhaul this function to take into
2126 consideration the value given via G option.
2127 (arc_rewrite_small_data_1): Renamed and corrected old
2128 arc_rewrite_small_data function.
2129 (arc_rewrite_small_data): New function.
2130 (small_data_pattern): Don't use pic_offset_table_rtx.
2131 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2132 * config/arc/simdext.md (movmisalignv2hi): Use
2133 prepare_move_operands function.
2135 (movmisalign*): Likewise.
2136 * doc/invoke.texi (ARC options): Document -G option.
2138 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2140 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2142 * config/arc/arc.c (arc_print_operand): Output scalled address for
2143 sdata whenever is possible.
2144 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2145 load/stores are available.
2146 (compact_sda_memory_operand): Check for the alignment required by
2147 code density instructions.
2148 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2150 * config/arc/constraints.md (Usd): Update constraint.
2151 (Us0): New constraint.
2152 (Usc): Update constraint.
2154 2017-08-31 Richard Biener <rguenther@suse.de>
2157 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2160 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2162 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2163 Resize type_signature.
2165 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2166 Alan Hayward <alan.hayward@arm.com>
2167 David Sherwood <david.sherwood@arm.com>
2169 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2170 subregs whose inner modes can be stored in GPRs.
2171 (aarch64_classify_index): Likewise.
2173 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2174 Alan Hayward <alan.hayward@arm.com>
2175 David Sherwood <david.sherwood@arm.com>
2177 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2178 (V_INT_EQUIV): ...this.
2179 (v_cmp_result): Rename to...
2180 (v_int_equiv): ...this.
2181 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2182 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2183 (copysign<mode>3): Likewise.
2184 (aarch64_simd_bsl<mode>_internal): Likewise.
2185 (aarch64_simd_bsl<mode>): Likewise.
2186 (vec_cmp<mode><mode>): Likewise.
2187 (vcond<mode><mode>): Likewise.
2188 (vcond<v_cmp_mixed><mode>): Likewise.
2189 (vcondu<mode><v_cmp_mixed>): Likewise.
2190 (aarch64_cm<optab><mode>): Likewise.
2191 (aarch64_cmtst<mode>): Likewise.
2192 (aarch64_fac<optab><mode>): Likewise.
2193 (vec_perm_const<mode>): Likewise.
2194 (vcond_mask_<mode><v_cmp_result>): Rename to...
2195 (vcond_mask_<mode><v_int_equiv>): ...this.
2196 (vec_cmp<mode><v_cmp_result>): Rename to...
2197 (vec_cmp<mode><v_int_equiv>): ...this.
2199 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2200 Alan Hayward <alan.hayward@arm.com>
2201 David Sherwood <david.sherwood@arm.com>
2203 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2205 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2206 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2207 (UNSPEC_LD4_DREG): New unspecs.
2208 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2209 (aarch64_ld2<mode>_dreg_be): Replace with...
2210 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2212 (aarch64_ld3<mode>_dreg_le)
2213 (aarch64_ld3<mode>_dreg_be): Replace with...
2214 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2216 (aarch64_ld4<mode>_dreg_le)
2217 (aarch64_ld4<mode>_dreg_be): Replace with...
2218 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2221 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2223 PR tree-optimization/81987
2224 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2225 insert an initializer in a location not dominated by the stride
2228 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2230 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2231 on the entire header of the finally block in the fallthru case.
2233 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2235 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2237 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2239 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2240 rs6000_emit_move_from_cr and call renamed function.
2241 (rs6000_emit_prologue): Call renamed functions.
2242 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2243 movesi_from_cr, remove volatile CRs.
2245 2017-08-30 Jon Beniston <jon@beniston.com>
2246 Richard Biener <rguenther@suse.de>
2248 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2249 of VECTOR_MODE_P check.
2250 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2251 element vector types.
2253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2255 * df.h (df_read_modify_subreg_p): Remove in favor of...
2256 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2257 const_rtx instead of an rtx.
2258 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2259 * df-problems.c (df_word_lr_mark_ref): Likewise.
2260 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2261 (mark_pseudo_reg_dead): Likewise.
2262 (mark_ref_dead): Likewise.
2263 * reginfo.c (init_subregs_of_mode): Likewise.
2264 * sched-deps.c (sched_analyze_1): Likewise.
2265 * df-scan.c (df_def_record_1): Likewise.
2266 (df_uses_record): Likewise.
2267 (df_read_modify_subreg_p): Remove in favor of...
2268 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2269 const_rtx instead of an rtx.
2271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * rtl.h (partial_subreg_p): New function.
2276 * caller-save.c (save_call_clobbered_regs): Use it.
2277 * calls.c (expand_call): Likewise.
2278 * combine.c (combinable_i3pat): Likewise.
2279 (simplify_set): Likewise.
2280 (make_extraction): Likewise.
2281 (make_compound_operation_int): Likewise.
2282 (gen_lowpart_or_truncate): Likewise.
2283 (force_to_mode): Likewise.
2284 (make_field_assignment): Likewise.
2285 (reg_truncated_to_mode): Likewise.
2286 (record_truncated_value): Likewise.
2287 (move_deaths): Likewise.
2288 * cse.c (record_jump_cond): Likewise.
2289 (cse_insn): Likewise.
2290 * cselib.c (cselib_lookup_1): Likewise.
2291 * expmed.c (extract_bit_field_using_extv): Likewise.
2292 * function.c (assign_parm_setup_reg): Likewise.
2293 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2294 * ira-build.c (create_insn_allocnos): Likewise.
2295 * lra-coalesce.c (merge_pseudos): Likewise.
2296 * lra-constraints.c (match_reload): Likewise.
2297 (simplify_operand_subreg): Likewise.
2298 (curr_insn_transform): Likewise.
2299 * lra-lives.c (process_bb_lives): Likewise.
2300 * lra.c (new_insn_reg): Likewise.
2301 (lra_substitute_pseudo): Likewise.
2302 * regcprop.c (mode_change_ok): Likewise.
2303 (maybe_mode_change): Likewise.
2304 (copyprop_hardreg_forward_1): Likewise.
2305 * reload.c (push_reload): Likewise.
2306 (find_reloads): Likewise.
2307 (find_reloads_subreg_address): Likewise.
2308 * reload1.c (alter_reg): Likewise.
2309 (eliminate_regs_1): Likewise.
2310 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2312 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2314 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2315 back to if statements, including unpack.
2317 2017-08-30 Martin Liska <mliska@suse.cz>
2320 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2322 (func_checker::compare_asm_inputs_outputs): ... this function.
2323 (func_checker::compare_gimple_asm): Use the function to compare
2324 also ASM constrains.
2325 * ipa-icf-gimple.h: Rename the function.
2327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2328 Alan Hayward <alan.hayward@arm.com>
2329 David Sherwood <david.sherwood@arm.com>
2331 * coretypes.h (complex_mode): New type.
2332 * gdbhooks.py (build_pretty_printer): Handle it.
2333 * machmode.h (complex_mode): New class.
2334 (complex_mode::includes_p): New function.
2335 (is_complex_int_mode): Likewise.
2336 (is_complex_float_mode): Likewise.
2337 * genmodes.c (get_mode_class): Handle complex mode classes.
2338 * function.c (expand_function_end): Use is_complex_int_mode.
2340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2341 Alan Hayward <alan.hayward@arm.com>
2342 David Sherwood <david.sherwood@arm.com>
2344 * coretypes.h (scalar_mode_pod): New typedef.
2345 * gdbhooks.py (build_pretty_printer): Handle it.
2346 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2347 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2348 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2349 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2353 Alan Hayward <alan.hayward@arm.com>
2354 David Sherwood <david.sherwood@arm.com>
2356 * machmode.h (mode_for_vector): Take a scalar_mode instead
2358 * stor-layout.c (mode_for_vector): Likewise.
2359 * explow.c (promote_mode): Use as_a <scalar_mode>.
2360 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2363 Alan Hayward <alan.hayward@arm.com>
2364 David Sherwood <david.sherwood@arm.com>
2366 * target.def (preferred_simd_mode): Take a scalar_mode
2367 instead of a machine_mode.
2368 * targhooks.h (default_preferred_simd_mode): Likewise.
2369 * targhooks.c (default_preferred_simd_mode): Likewise.
2370 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2371 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2372 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2373 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2374 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2375 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2376 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2377 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2379 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2380 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2381 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2382 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2383 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2384 * doc/tm.texi: Regenerate.
2385 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2389 Alan Hayward <alan.hayward@arm.com>
2390 David Sherwood <david.sherwood@arm.com>
2392 * target.def (scalar_mode_supported_p): Take a scalar_mode
2393 instead of a machine_mode.
2394 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2395 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2396 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2397 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2398 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2399 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2400 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2401 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2402 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2403 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2404 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2405 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2406 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2407 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2409 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2410 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2411 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2412 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2413 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2415 * doc/tm.texi: Regenerate.
2417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2418 Alan Hayward <alan.hayward@arm.com>
2419 David Sherwood <david.sherwood@arm.com>
2421 * coretypes.h (opt_scalar_mode): New typedef.
2422 * gdbhooks.py (build_pretty_printers): Handle it.
2423 * machmode.h (mode_iterator::get_2xwider): Add overload for
2425 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2427 * expr.c (convert_mode_scalar): Likewise.
2428 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2429 * optabs.c (expand_float): Likewise.
2430 (expand_fix): Likewise.
2431 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2434 Alan Hayward <alan.hayward@arm.com>
2435 David Sherwood <david.sherwood@arm.com>
2437 * optabs.c (expand_float): Explicitly check for scalars before
2438 using a branching expansion.
2439 (expand_fix): Likewise.
2441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2442 Alan Hayward <alan.hayward@arm.com>
2443 David Sherwood <david.sherwood@arm.com>
2445 * expr.c (convert_mode): Split scalar handling out into...
2446 (convert_mode_scalar): ...this new function. Treat the modes
2449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2450 Alan Hayward <alan.hayward@arm.com>
2451 David Sherwood <david.sherwood@arm.com>
2453 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2455 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2458 Alan Hayward <alan.hayward@arm.com>
2459 David Sherwood <david.sherwood@arm.com>
2461 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2462 rather than a machine_mode.
2463 (fixed_from_string): Likewise.
2464 (fixed_convert): Likewise.
2465 (fixed_convert_from_int): Likewise.
2466 (fixed_convert_from_real): Likewise.
2467 (real_convert_from_fixed): Likewise.
2468 * fixed-value.c (fixed_from_double_int): Likewise.
2469 (fixed_from_string): Likewise.
2470 (fixed_convert): Likewise.
2471 (fixed_convert_from_int): Likewise.
2472 (fixed_convert_from_real): Likewise.
2473 (real_convert_from_fixed): Likewise.
2474 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2477 Alan Hayward <alan.hayward@arm.com>
2478 David Sherwood <david.sherwood@arm.com>
2480 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2481 of separate mode class checks. Do not allow vector modes here.
2482 (immed_wide_int_const): Use as_a <scalar_mode>.
2483 * explow.c (trunc_int_for_mode): Likewise.
2484 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2485 (wi::shwi): Likewise.
2486 (wi::min_value): Likewise.
2487 (wi::max_value): Likewise.
2488 * dwarf2out.c (loc_descriptor): Likewise.
2489 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2493 Alan Hayward <alan.hayward@arm.com>
2494 David Sherwood <david.sherwood@arm.com>
2496 * tree.h (SCALAR_TYPE_MODE): New macro.
2497 * expr.c (expand_expr_addr_expr_1): Use it.
2498 (expand_expr_real_2): Likewise.
2499 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2500 (fold_convert_const_fixed_from_int): Likewise.
2501 (fold_convert_const_fixed_from_real): Likewise.
2502 (native_encode_fixed): Likewise
2503 (native_encode_complex): Likewise
2504 (native_encode_vector): Likewise.
2505 (native_interpret_fixed): Likewise.
2506 (native_interpret_real): Likewise.
2507 (native_interpret_complex): Likewise.
2508 (native_interpret_vector): Likewise.
2509 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2510 (simd_clone_adjust_argument_types): Likewise.
2511 (simd_clone_init_simd_arrays): Likewise.
2512 (simd_clone_adjust): Likewise.
2513 * stor-layout.c (layout_type): Likewise.
2514 * tree.c (build_minus_one_cst): Likewise.
2515 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2516 * tree-inline.c (estimate_move_cost): Likewise.
2517 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2518 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2519 (vectorizable_reduction): Likewise.
2520 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2521 (vect_recog_mixed_size_cond_pattern): Likewise.
2522 (check_bool_pattern): Likewise.
2523 (adjust_bool_pattern): Likewise.
2524 (search_type_for_mask_1): Likewise.
2525 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2526 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2527 (vectorizable_load): Likewise.
2528 (vectorizable_store): Likewise.
2529 * ubsan.c (ubsan_encode_value): Likewise.
2530 * varasm.c (output_constant): Likewise.
2532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2533 Alan Hayward <alan.hayward@arm.com>
2534 David Sherwood <david.sherwood@arm.com>
2536 * coretypes.h (scalar_mode): New class.
2537 * machmode.h (scalar_mode): Likewise.
2538 (scalar_mode::includes_p): New function.
2539 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2540 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2541 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2542 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2543 * expmed.c (store_bit_field_1): Likewise.
2544 (extract_bit_field_1): Likewise.
2545 * expr.c (write_complex_part): Likewise.
2546 (read_complex_part): Likewise.
2547 (emit_move_complex_push): Likewise.
2548 (expand_expr_real_2): Likewise.
2549 * function.c (assign_parm_setup_reg): Likewise.
2550 (assign_parms_unsplit_complex): Likewise.
2551 * optabs.c (expand_binop): Likewise.
2552 * rtlanal.c (subreg_get_info): Likewise.
2553 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2554 * varasm.c (output_constant_pool_2): Likewise.
2556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2557 Alan Hayward <alan.hayward@arm.com>
2558 David Sherwood <david.sherwood@arm.com>
2560 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2562 (expmed_mult_highpart_optab): Likewise.
2563 (expmed_mult_highpart): Likewise.
2565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2566 Alan Hayward <alan.hayward@arm.com>
2567 David Sherwood <david.sherwood@arm.com>
2569 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2570 instead of a machine_mode.
2571 (builtin_memset_read_str): Likewise.
2572 * builtins.c (c_readstr): Likewise.
2573 (builtin_memcpy_read_str): Likewise.
2574 (builtin_strncpy_read_str): Likewise.
2575 (builtin_memset_read_str): Likewise.
2576 (builtin_memset_gen_str): Likewise.
2577 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2578 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2579 instead of a machine_mode.
2580 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2582 (make_extraction): Likewise.
2583 (try_widen_shift_mode): Take and return scalar_int_modes instead
2585 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2586 a scalar_int_mode instead of a machine_mode.
2587 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2588 (avr_addr_space_pointer_mode): Likewise.
2589 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2590 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2591 (msp430_unwind_word_mode): Likewise.
2592 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2593 (spu_addr_space_pointer_mode): Likewise.
2594 (spu_addr_space_address_mode): Likewise.
2595 (spu_libgcc_cmp_return_mode): Likewise.
2596 (spu_libgcc_shift_count_mode): Likewise.
2597 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2598 (rl78_addr_space_pointer_mode): Likewise.
2599 (fl78_unwind_word_mode): Likewise.
2600 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2602 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2603 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2604 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2605 (mips_valid_pointer_mode): Likewise.
2606 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2607 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2608 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2610 (ft32_addr_space_address_mode): Likewise.
2611 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2612 scalar_int_mode instead of a machine_mode.
2613 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2615 (m32c_addr_space_address_mode): Likewise.
2616 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2617 (rs6000_eh_return_filter_mode): Likewise.
2618 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2619 (rs6000_eh_return_filter_mode): Likewise.
2620 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2621 (s390_libgcc_shift_count_mode): Likewise.
2622 (s390_unwind_word_mode): Likewise.
2623 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2625 * target.def (mode_rep_extended): Likewise.
2626 (valid_pointer_mode): Likewise.
2627 (addr_space.valid_pointer_mode): Likewise.
2628 (eh_return_filter_mode): Return a scalar_int_mode rather than
2630 (libgcc_cmp_return_mode): Likewise.
2631 (libgcc_shift_count_mode): Likewise.
2632 (unwind_word_mode): Likewise.
2633 (addr_space.pointer_mode): Likewise.
2634 (addr_space.address_mode): Likewise.
2635 * doc/tm.texi: Regenerate.
2636 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2638 (do_jump): Use scalar_int_mode for local variables.
2639 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2640 rather than a machine_mode.
2641 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2642 (scompare_loc_descriptor_wide): Likewise.
2643 (scompare_loc_descriptor_narrow): Likewise.
2644 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2646 * except.c (sjlj_emit_dispatch_table): Likewise.
2647 (expand_builtin_eh_copy_values): Likewise.
2648 * explow.c (convert_memory_address_addr_space_1): Likewise.
2649 Take a scalar_int_mode rather than a machine_mode.
2650 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2651 than a machine_mode.
2652 (memory_address_addr_space): Use scalar_int_mode for local variables.
2653 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2654 rather than a machine_mode.
2655 * expmed.c (mask_rtx): Likewise.
2656 (init_expmed_one_conv): Likewise.
2657 (expand_mult_highpart_adjust): Likewise.
2658 (extract_high_half): Likewise.
2659 (expmed_mult_highpart_optab): Likewise.
2660 (expmed_mult_highpart): Likewise.
2661 (expand_smod_pow2): Likewise.
2662 (expand_sdiv_pow2): Likewise.
2663 (emit_store_flag_int): Likewise.
2664 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2666 (extract_low_bits): Likewise.
2667 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2669 * expr.c (pieces_addr::adjust): Likewise.
2670 (can_store_by_pieces): Likewise.
2671 (store_by_pieces): Likewise.
2672 (clear_by_pieces_1): Likewise.
2673 (expand_expr_addr_expr_1): Likewise.
2674 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2675 (expand_expr_real_1): Likewise.
2676 (try_casesi): Likewise.
2677 * final.c (shorten_branches): Likewise.
2678 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2679 type of "mode" to machine_mode.
2680 * internal-fn.c (expand_arith_overflow_result_store): Take a
2681 scalar_int_mode rather than a machine_mode.
2682 (expand_mul_overflow): Use scalar_int_mode for local variables.
2683 * loop-doloop.c (doloop_modify): Likewise.
2684 (doloop_optimize): Likewise.
2685 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2686 than a machine_mode.
2687 (expand_doubleword_shift_condmove): Likewise.
2688 (expand_doubleword_shift): Likewise.
2689 (expand_doubleword_clz): Likewise.
2690 (expand_doubleword_popcount): Likewise.
2691 (expand_doubleword_parity): Likewise.
2692 (expand_absneg_bit): Use scalar_int_mode for local variables.
2693 (prepare_float_lib_cmp): Likewise.
2694 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2695 rather than a machine_mode.
2696 (convert_memory_address_addr_space): Likewise.
2697 (get_mode_bounds): Likewise.
2698 (get_address_mode): Return a scalar_int_mode rather than a
2700 * rtlanal.c (get_address_mode): Likewise.
2701 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2702 than a machine_mode.
2703 * targhooks.c (default_mode_rep_extended): Likewise.
2704 (default_valid_pointer_mode): Likewise.
2705 (default_addr_space_valid_pointer_mode): Likewise.
2706 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2707 than a machine_mode.
2708 (default_libgcc_cmp_return_mode): Likewise.
2709 (default_libgcc_shift_count_mode): Likewise.
2710 (default_unwind_word_mode): Likewise.
2711 (default_addr_space_pointer_mode): Likewise.
2712 (default_addr_space_address_mode): Likewise.
2713 * targhooks.h (default_eh_return_filter_mode): Likewise.
2714 (default_libgcc_cmp_return_mode): Likewise.
2715 (default_libgcc_shift_count_mode): Likewise.
2716 (default_unwind_word_mode): Likewise.
2717 (default_addr_space_pointer_mode): Likewise.
2718 (default_addr_space_address_mode): Likewise.
2719 (default_mode_rep_extended): Take a scalar_int_mode rather than
2721 (default_valid_pointer_mode): Likewise.
2722 (default_addr_space_valid_pointer_mode): Likewise.
2723 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2725 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2726 rather than a machine_mode.
2727 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2728 for local variables.
2729 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2730 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2731 than a machine_mode.
2733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2734 Alan Hayward <alan.hayward@arm.com>
2735 David Sherwood <david.sherwood@arm.com>
2737 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2738 the mode argument to scalar_int_mode.
2739 (do_jump_by_parts_zero_rtx): Likewise.
2740 (do_jump_by_parts_equality_rtx): Likewise.
2741 (do_jump_by_parts_greater): Take a mode argument.
2742 (do_jump_by_parts_equality): Likewise.
2743 (do_jump_1): Update calls accordingly.
2745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2746 Alan Hayward <alan.hayward@arm.com>
2747 David Sherwood <david.sherwood@arm.com>
2749 * is-a.h (safe_dyn_cast): New function.
2750 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2751 (jump_table_for_label): Likewise.
2752 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2753 instead of an rtx_insn *.
2754 (shorten_branches): Use dyn_cast instead of LABEL_P and
2755 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2756 rtx_jump_table_data::get_data_mode.
2757 (final_scan_insn): Likewise.
2759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2760 Alan Hayward <alan.hayward@arm.com>
2761 David Sherwood <david.sherwood@arm.com>
2763 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2764 trying to combine a full-register integer set with a subreg
2767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2768 Alan Hayward <alan.hayward@arm.com>
2769 David Sherwood <david.sherwood@arm.com>
2771 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2772 that is always either address_mode or pointer_mode.
2774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2775 Alan Hayward <alan.hayward@arm.com>
2776 David Sherwood <david.sherwood@arm.com>
2778 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2779 when the two are known to be equal.
2781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2782 Alan Hayward <alan.hayward@arm.com>
2783 David Sherwood <david.sherwood@arm.com>
2785 * simplify-rtx.c (simplify_const_unary_operation): Use
2786 is_a <scalar_int_mode> instead of checking for a nonzero
2787 precision. Forcibly convert op_mode to a scalar_int_mode
2788 in that case. More clearly differentiate the operand and
2789 result modes and use the former when deciding what the value
2790 of a count-bits operation should be. Use is_int_mode instead
2791 of checking for a MODE_INT. Remove redundant check for whether
2792 this mode has a zero precision.
2794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2795 Alan Hayward <alan.hayward@arm.com>
2796 David Sherwood <david.sherwood@arm.com>
2798 * optabs.c (widen_leading): Change the type of the mode argument
2799 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2800 (widen_bswap): Likewise.
2801 (expand_parity): Likewise.
2802 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2803 (expand_ffs): Likewise.
2804 (epand_unop): Check for scalar integer modes before calling the
2807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2808 Alan Hayward <alan.hayward@arm.com>
2809 David Sherwood <david.sherwood@arm.com>
2811 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2813 (expand_expr_real_1): Update call accordingly.
2815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2816 Alan Hayward <alan.hayward@arm.com>
2817 David Sherwood <david.sherwood@arm.com>
2819 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2820 value_mode arguments. Use scalar_int_mode internally.
2821 (store_bit_field_1): Rename the new integer mode from imode
2822 to op0_mode and use it instead of GET_MODE (op0). Update calls
2823 to store_split_bit_field, store_bit_field_using_insv and
2824 store_fixed_bit_field.
2825 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2826 Use scalar_int_mode internally. Use a bit count rather than a mode
2827 when calculating the largest bit size for get_best_mode.
2828 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2829 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2830 Remove assertion that OP0 has a scalar integer mode.
2831 (store_split_bit_field): Add op0_mode and value_mode arguments.
2832 Update calls to extract_fixed_bit_field.
2833 (extract_bit_field_using_extv): Add an op0_mode argument.
2834 Use scalar_int_mode internally.
2835 (extract_bit_field_1): Rename the new integer mode from imode to
2836 op0_mode and use it instead of GET_MODE (op0). Update calls to
2837 extract_split_bit_field, extract_bit_field_using_extv and
2838 extract_fixed_bit_field.
2839 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2840 to extract_split_bit_field and extract_fixed_bit_field_1.
2841 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2842 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2844 (extract_split_bit_field): Add an op0_mode argument. Update call
2845 to extract_fixed_bit_field.
2847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2848 Alan Hayward <alan.hayward@arm.com>
2849 David Sherwood <david.sherwood@arm.com>
2851 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2852 * explow.c (hard_function_value): Likewise.
2853 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2854 convert_to_mode call outside the loop.
2855 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2856 for the mode iterator. Require the mode specified by max_pieces
2858 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2860 (copy_blkmode_to_reg): Likewise.
2861 (set_storage_via_setmem): Likewise.
2862 * optabs.c (prepare_cmp_insn): Likewise.
2863 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2864 * stor-layout.c (finish_bitfield_representative): Likewise.
2866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2867 Alan Hayward <alan.hayward@arm.com>
2868 David Sherwood <david.sherwood@arm.com>
2870 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2871 * expr.c (convert_move): Use them.
2872 (convert_modes): Likewise.
2873 (store_expr_with_bounds): Likewise.
2875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2876 Alan Hayward <alan.hayward@arm.com>
2877 David Sherwood <david.sherwood@arm.com>
2879 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2880 parameter for the mode of "x". Remove the "known_x", "known_mode"
2881 and "known_ret" arguments. Change the type of the mode argument
2883 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2884 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2885 (reg_num_sign_bit_copies_for_combine): Likewise.
2886 * rtlanal.c (nonzero_bits1): Likewise.
2887 (num_sign_bit_copies1): Likewise.
2888 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2889 (reg_num_sign_bit_copies_general): Likewise.
2890 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2891 (reg_nonzero_bits_general): Likewise.
2893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2894 Alan Hayward <alan.hayward@arm.com>
2895 David Sherwood <david.sherwood@arm.com>
2897 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2898 than in subroutines. Return 1 for non-integer modes.
2899 (cached_num_sign_bit_copies): Change the type of the mode parameter
2901 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2902 classes. Handle CONST_INT_P first and then check whether X also
2903 has a scalar integer mode. Check the same thing for inner registers
2904 of a SUBREG and for values that are being extended or truncated.
2906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2907 Alan Hayward <alan.hayward@arm.com>
2908 David Sherwood <david.sherwood@arm.com>
2910 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2911 in subroutines. Return the mode mask for non-integer modes.
2912 (cached_nonzero_bits): Change the type of the mode parameter
2914 (nonzero_bits1): Likewise. Remove early exit for other mode
2915 classes. Handle CONST_INT_P first and then check whether X
2916 also has a scalar integer mode.
2918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2919 Alan Hayward <alan.hayward@arm.com>
2920 David Sherwood <david.sherwood@arm.com>
2922 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2923 Return a scalar_int_mode and assert that the size is greater than
2925 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2926 (op_by_pieces_d::op_by_pieces_d): Likewise.
2927 (op_by_pieces_d::run): Likewise.
2928 (can_store_by_pieces): Likewise.
2930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2931 Alan Hayward <alan.hayward@arm.com>
2932 David Sherwood <david.sherwood@arm.com>
2934 * combine.c (extract_left_shift): Add a mode argument and update
2936 (make_compound_operation_int): Change the type of the mode parameter
2937 to scalar_int_mode and update the call to extract_left_shift.
2939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2940 Alan Hayward <alan.hayward@arm.com>
2941 David Sherwood <david.sherwood@arm.com>
2943 * combine.c (simplify_and_const_int): Change the type of the mode
2944 parameter to scalar_int_mode.
2945 (simplify_and_const_int_1): Likewise. Update recursive call.
2947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2948 Alan Hayward <alan.hayward@arm.com>
2949 David Sherwood <david.sherwood@arm.com>
2951 * combine.c (simplify_compare_const): Check that the mode is a
2952 scalar_int_mode (rather than VOIDmode) before testing its
2954 (simplify_comparison): Move COMPARISON_P handling out of the
2955 loop and restrict the latter part of the loop to scalar_int_modes.
2956 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2957 and when considering SUBREG_REGs. Use is_int_mode instead of
2958 checking GET_MODE_CLASS against MODE_INT.
2960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2961 Alan Hayward <alan.hayward@arm.com>
2962 David Sherwood <david.sherwood@arm.com>
2964 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2965 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2966 shift_unit_mode and for modes involved in scalar shifts.
2968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2969 Alan Hayward <alan.hayward@arm.com>
2970 David Sherwood <david.sherwood@arm.com>
2972 * combine.c (force_int_to_mode): New function, split out from...
2973 (force_to_mode): ...here. Keep xmode up-to-date and use it
2974 instead of GET_MODE (x).
2976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2977 Alan Hayward <alan.hayward@arm.com>
2978 David Sherwood <david.sherwood@arm.com>
2980 * optabs-query.h (extraction_insn::struct_mode): Change type to
2981 opt_scalar_int_mode and update comment.
2982 (extraction_insn::field_mode): Change type to scalar_int_mode.
2983 (extraction_insn::pos_mode): Likewise.
2984 * combine.c (make_extraction): Update accordingly.
2985 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2986 (get_optab_extraction_insn): Likewise.
2987 * recog.c (simplify_while_replacing): Likewise.
2988 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2989 parameter to opt_scalar_int_mode.
2991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2992 Alan Hayward <alan.hayward@arm.com>
2993 David Sherwood <david.sherwood@arm.com>
2995 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2996 to a scalar_int_mode instead of a machine_mode.
2997 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2998 (get_best_mode): Return a boolean and use a pointer argument to store
2999 the selected mode. Replace the limit mode parameter with a bit limit.
3000 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3001 for the values returned by bit_field_mode_iterator::next_mode.
3002 (store_bit_field): Update call to get_best_mode.
3003 (store_fixed_bit_field): Likewise.
3004 (extract_fixed_bit_field): Likewise.
3005 * expr.c (optimize_bitfield_assignment_op): Likewise.
3006 * fold-const.c (optimize_bit_field_compare): Likewise.
3007 (fold_truth_andor_1): Likewise.
3008 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3009 Update for new type of m_mode.
3010 (get_best_mode): As above.
3012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3013 Alan Hayward <alan.hayward@arm.com>
3014 David Sherwood <david.sherwood@arm.com>
3016 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3017 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3018 (store_bit_field): Check is_a <scalar_int_mode> before calling
3019 strict_volatile_bitfield_p.
3020 (extract_bit_field): Likewise.
3022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3023 Alan Hayward <alan.hayward@arm.com>
3024 David Sherwood <david.sherwood@arm.com>
3026 * target.def (cstore_mode): Return a scalar_int_mode.
3027 * doc/tm.texi: Regenerate.
3028 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3029 * targhooks.h (default_cstore_mode): Likewise.
3030 * targhooks.c (default_cstore_mode): Likewise, using a forced
3032 * expmed.c (emit_cstore): Expect the target of the cstore to be
3035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3036 Alan Hayward <alan.hayward@arm.com>
3037 David Sherwood <david.sherwood@arm.com>
3039 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3041 (niter_desc): Likewise mode.
3042 (iv_analyze): Add a mode parameter.
3044 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3045 and change its type to scalar_int_mode.
3046 * loop-iv.c: Update commentary at head of file.
3047 (iv_constant): Pass the mode paraeter before the rtx it describes
3048 and change its type to scalar_int_mode. Remove VOIDmode handling.
3049 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3050 (iv_extend): Likewise.
3051 (shorten_into_mode): Likewise.
3052 (iv_add): Use scalar_int_mode.
3053 (iv_mult): Likewise.
3054 (iv_shift): Likewise.
3055 (canonicalize_iv_subregs): Likewise.
3056 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3057 it describes and change its mode to scalar_int_mode. Also change
3058 the type of the returned inner_mode to scalar_int_mode.
3059 (get_biv_step): Likewise, turning outer_mode from a pointer
3060 into a direct parameter. Update call to get_biv_step_1.
3061 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3062 iv_constant and get_biv_step.
3063 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3064 and change its type to scalar_int_mode. Don't initialise iv->mode
3065 to VOIDmode and remove later checks for its still being VOIDmode.
3066 Update calls to iv_analyze_op and iv_analyze_expr. Check
3067 is_a <scalar_int_mode> when changing the mode under consideration.
3068 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3069 Update call to iv_analyze_expr.
3070 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3071 inner register is not also a scalar_int_mode. Update call to
3073 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3074 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3075 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3076 separate mode class checks. Update calls to iv_analyze. Remove
3077 fix-up of VOIDmodes after iv_analyze_biv.
3078 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3079 don't have a scalar_int_mode. Update call to biv_p.
3081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3082 Alan Hayward <alan.hayward@arm.com>
3083 David Sherwood <david.sherwood@arm.com>
3085 * cfgexpand.c (convert_debug_memory_address): Use
3086 as_a <scalar_int_mode>.
3087 * combine.c (expand_compound_operation): Likewise.
3088 (make_extraction): Likewise.
3089 (change_zero_ext): Likewise.
3090 (simplify_comparison): Likewise.
3091 * cse.c (cse_insn): Likewise.
3092 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3093 (mem_loc_descriptor): Likewise.
3094 (loc_descriptor): Likewise.
3095 * expmed.c (init_expmed_one_mode): Likewise.
3096 (synth_mult): Likewise.
3097 (emit_store_flag_1): Likewise.
3098 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3099 of a comparison with size.
3100 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3101 (reduce_to_bit_field_precision): Likewise.
3102 * function.c (expand_function_end): Likewise.
3103 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3104 * loop-doloop.c (doloop_modify): Likewise.
3105 * optabs.c (expand_binop): Likewise.
3106 (expand_unop): Likewise.
3107 (expand_copysign_absneg): Likewise.
3108 (prepare_cmp_insn): Likewise.
3109 (maybe_legitimize_operand): Likewise.
3110 * recog.c (const_scalar_int_operand): Likewise.
3111 * rtlanal.c (get_address_mode): Likewise.
3112 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3113 (simplify_cond_clz_ctz): Likewise.
3114 * tree-nested.c (get_nl_goto_field): Likewise.
3115 * tree.c (build_vector_type_for_mode): Likewise.
3116 * var-tracking.c (use_narrower_mode): Likewise.
3118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3119 Alan Hayward <alan.hayward@arm.com>
3120 David Sherwood <david.sherwood@arm.com>
3122 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3123 * builtins.c (expand_builtin_signbit): Use it.
3124 * cfgexpand.c (expand_debug_expr): Likewise.
3125 * dojump.c (do_jump): Likewise.
3126 (do_compare_and_jump): Likewise.
3127 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3128 * expmed.c (make_tree): Likewise.
3129 * expr.c (expand_expr_real_2): Likewise.
3130 (expand_expr_real_1): Likewise.
3131 (try_casesi): Likewise.
3132 * fold-const-call.c (fold_const_call_ss): Likewise.
3133 * fold-const.c (unextend): Likewise.
3134 (extract_muldiv_1): Likewise.
3135 (fold_single_bit_test): Likewise.
3136 (native_encode_int): Likewise.
3137 (native_encode_string): Likewise.
3138 (native_interpret_int): Likewise.
3139 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3140 * internal-fn.c (expand_addsub_overflow): Likewise.
3141 (expand_neg_overflow): Likewise.
3142 (expand_mul_overflow): Likewise.
3143 (expand_arith_overflow): Likewise.
3144 * match.pd: Likewise.
3145 * stor-layout.c (layout_type): Likewise.
3146 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3147 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3148 * tree-ssanames.c (get_range_info): Likewise.
3149 * tree-switch-conversion.c (array_value_type) Likewise.
3150 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3151 (vect_recog_divmod_pattern): Likewise.
3152 (vect_recog_mixed_size_cond_pattern): Likewise.
3153 * tree-vrp.c (extract_range_basic): Likewise.
3154 (simplify_float_conversion_using_ranges): Likewise.
3155 * tree.c (int_fits_type_p): Likewise.
3156 * ubsan.c (instrument_bool_enum_load): Likewise.
3157 * varasm.c (mergeable_string_section): Likewise.
3158 (narrowing_initializer_constant_valid_p): Likewise.
3159 (output_constant): Likewise.
3161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3162 Alan Hayward <alan.hayward@arm.com>
3163 David Sherwood <david.sherwood@arm.com>
3165 * machmode.h (NARROWEST_INT_MODE): New macro.
3166 * expr.c (alignment_for_piecewise_move): Use it instead of
3167 GET_CLASS_NARROWEST_MODE (MODE_INT).
3168 (push_block): Likewise.
3169 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3171 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3174 Alan Hayward <alan.hayward@arm.com>
3175 David Sherwood <david.sherwood@arm.com>
3177 * postreload.c (move2add_valid_value_p): Change the type of the
3178 mode parameter to scalar_int_mode.
3179 (move2add_use_add2_insn): Add a mode parameter and use it instead
3181 (move2add_use_add3_insn): Likewise.
3182 (reload_cse_move2add): Update accordingly.
3184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3185 Alan Hayward <alan.hayward@arm.com>
3186 David Sherwood <david.sherwood@arm.com>
3188 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3190 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3191 * optabs.c (expand_unop): Likewise.
3193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3194 Alan Hayward <alan.hayward@arm.com>
3195 David Sherwood <david.sherwood@arm.com>
3197 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3198 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3199 (popcount_loc_descriptor): Likewise.
3200 (bswap_loc_descriptor): Likewise.
3201 (rotate_loc_descriptor): Likewise.
3202 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3203 calling the functions above.
3205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3206 Alan Hayward <alan.hayward@arm.com>
3207 David Sherwood <david.sherwood@arm.com>
3209 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3211 (try_combine): Likewise.
3212 (simplify_if_then_else): Likewise.
3213 * cse.c (cse_insn): Likewise.
3214 * dwarf2out.c (mem_loc_descriptor): Likewise.
3215 * emit-rtl.c (gen_lowpart_common): Likewise.
3216 * simplify-rtx.c (simplify_truncation): Likewise.
3217 (simplify_binary_operation_1): Likewise.
3218 (simplify_const_relational_operation): Likewise.
3219 (simplify_ternary_operation): Likewise.
3220 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3223 Alan Hayward <alan.hayward@arm.com>
3224 David Sherwood <david.sherwood@arm.com>
3226 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3227 * reload.c (push_reload): Likewise.
3228 (find_reloads): Likewise.
3230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3231 Alan Hayward <alan.hayward@arm.com>
3232 David Sherwood <david.sherwood@arm.com>
3234 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3235 (make_compound_operation_int): Likewise.
3236 (change_zero_ext): Likewise.
3237 * expr.c (convert_move): Likewise.
3238 (convert_modes): Likewise.
3239 * fwprop.c (forward_propagate_subreg): Likewise.
3240 * loop-iv.c (get_biv_step_1): Likewise.
3241 * optabs.c (widen_operand): Likewise.
3242 * postreload.c (move2add_valid_value_p): Likewise.
3243 * recog.c (simplify_while_replacing): Likewise.
3244 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3245 (simplify_binary_operation_1): Likewise. Remove redundant
3246 mode equality check.
3248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3249 Alan Hayward <alan.hayward@arm.com>
3250 David Sherwood <david.sherwood@arm.com>
3252 * combine.c (combine_simplify_rtx): Add checks for
3253 is_a <scalar_int_mode>.
3254 (simplify_if_then_else): Likewise.
3255 (make_field_assignment): Likewise.
3256 (simplify_comparison): Likewise.
3257 * ifcvt.c (noce_try_bitop): Likewise.
3258 * loop-invariant.c (canonicalize_address_mult): Likewise.
3259 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3262 Alan Hayward <alan.hayward@arm.com>
3263 David Sherwood <david.sherwood@arm.com>
3265 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3266 is_a <scalar_int_mode> instead of != BLKmode.
3268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3269 Alan Hayward <alan.hayward@arm.com>
3270 David Sherwood <david.sherwood@arm.com>
3272 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3273 instead of != VOIDmode.
3274 * combine.c (if_then_else_cond): Likewise.
3275 (change_zero_ext): Likewise.
3276 * dwarf2out.c (mem_loc_descriptor): Likewise.
3277 (loc_descriptor): Likewise.
3278 * rtlanal.c (canonicalize_condition): Likewise.
3279 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3282 Alan Hayward <alan.hayward@arm.com>
3283 David Sherwood <david.sherwood@arm.com>
3285 * simplify-rtx.c (simplify_binary_operation_1): Use
3286 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3289 Alan Hayward <alan.hayward@arm.com>
3290 David Sherwood <david.sherwood@arm.com>
3292 * wide-int.h (int_traits<unsigned char>) New class.
3293 (int_traits<unsigned short>) Likewise.
3294 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3295 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3297 * combine.c (set_nonzero_bits_and_sign_copies): Use
3298 is_a <scalar_int_mode>.
3299 (find_split_point): Likewise.
3300 (combine_simplify_rtx): Likewise.
3301 (simplify_logical): Likewise.
3302 (expand_compound_operation): Likewise.
3303 (expand_field_assignment): Likewise.
3304 (make_compound_operation): Likewise.
3305 (extended_count): Likewise.
3306 (change_zero_ext): Likewise.
3307 (simplify_comparison): Likewise.
3308 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3309 (ucompare_loc_descriptor): Likewise.
3310 (minmax_loc_descriptor): Likewise.
3311 (mem_loc_descriptor): Likewise.
3312 (loc_descriptor): Likewise.
3313 * expmed.c (init_expmed_one_mode): Likewise.
3314 * lra-constraints.c (lra_constraint_offset): Likewise.
3315 * optabs.c (prepare_libcall_arg): Likewise.
3316 * postreload.c (move2add_note_store): Likewise.
3317 * reload.c (operands_match_p): Likewise.
3318 * rtl.h (load_extend_op): Likewise.
3319 * rtlhooks.c (gen_lowpart_general): Likewise.
3320 * simplify-rtx.c (simplify_truncation): Likewise.
3321 (simplify_unary_operation_1): Likewise.
3322 (simplify_binary_operation_1): Likewise.
3323 (simplify_const_binary_operation): Likewise.
3324 (simplify_const_relational_operation): Likewise.
3325 (simplify_subreg): Likewise.
3326 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3327 * var-tracking.c (adjust_mems): Likewise.
3328 (prepare_call_arguments): Likewise.
3330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3331 Alan Hayward <alan.hayward@arm.com>
3332 David Sherwood <david.sherwood@arm.com>
3334 * machmode.h (is_int_mode): New fuction.
3335 * combine.c (find_split_point): Use it.
3336 (combine_simplify_rtx): Likewise.
3337 (simplify_if_then_else): Likewise.
3338 (simplify_set): Likewise.
3339 (simplify_shift_const_1): Likewise.
3340 (simplify_comparison): Likewise.
3341 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3342 * cse.c (notreg_cost): Likewise.
3343 (cse_insn): Likewise.
3344 * cselib.c (cselib_lookup_1): Likewise.
3345 * dojump.c (do_jump_1): Likewise.
3346 (do_compare_rtx_and_jump): Likewise.
3347 * dse.c (get_call_args): Likewise.
3348 * dwarf2out.c (rtl_for_decl_init): Likewise.
3349 (native_encode_initializer): Likewise.
3350 * expmed.c (emit_store_flag_1): Likewise.
3351 (emit_store_flag): Likewise.
3352 * expr.c (convert_modes): Likewise.
3353 (store_field): Likewise.
3354 (expand_expr_real_1): Likewise.
3355 * fold-const.c (fold_read_from_constant_string): Likewise.
3356 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3357 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3358 * optabs.c (expand_binop): Likewise.
3359 (expand_unop): Likewise.
3360 (expand_abs_nojump): Likewise.
3361 (expand_one_cmpl_abs_nojump): Likewise.
3362 * simplify-rtx.c (mode_signbit_p): Likewise.
3363 (val_signbit_p): Likewise.
3364 (val_signbit_known_set_p): Likewise.
3365 (val_signbit_known_clear_p): Likewise.
3366 (simplify_relational_operation_1): Likewise.
3367 * tree.c (vector_type_mode): Likewise.
3369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3370 Alan Hayward <alan.hayward@arm.com>
3371 David Sherwood <david.sherwood@arm.com>
3373 * machmode.h (smallest_mode_for_size): Fix formatting.
3374 (smallest_int_mode_for_size): New function.
3375 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3376 instead of smallest_mode_for_size.
3377 * combine.c (make_extraction): Likewise.
3378 * config/arc/arc.c (arc_expand_movmem): Likewise.
3379 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3380 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3381 * config/s390/s390.c (s390_expand_insv): Likewise.
3382 * config/sparc/sparc.c (assign_int_registers): Likewise.
3383 * config/spu/spu.c (spu_function_value): Likewise.
3384 (spu_function_arg): Likewise.
3385 * coverage.c (get_gcov_type): Likewise.
3386 (get_gcov_unsigned_t): Likewise.
3387 * dse.c (find_shift_sequence): Likewise.
3388 * expmed.c (store_bit_field_1): Likewise.
3389 * expr.c (convert_move): Likewise.
3390 (store_field): Likewise.
3391 * internal-fn.c (expand_arith_overflow): Likewise.
3392 * optabs-query.c (get_best_extraction_insn): Likewise.
3393 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3394 * stor-layout.c (layout_type): Likewise.
3395 (initialize_sizetypes): Likewise.
3396 * targhooks.c (default_get_mask_mode): Likewise.
3397 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3400 Alan Hayward <alan.hayward@arm.com>
3401 David Sherwood <david.sherwood@arm.com>
3403 * machmode.h (opt_mode::else_blk): New function.
3404 (int_mode_for_mode): Declare.
3405 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3406 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3408 * cfgexpand.c (expand_debug_expr): Likewise.
3409 * combine.c (gen_lowpart_or_truncate): Likewise.
3410 (gen_lowpart_for_combine): Likewise.
3411 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3412 * config/avr/avr.c (avr_to_int_mode): Likewise.
3413 (avr_out_plus_1): Likewise.
3414 (avr_out_plus): Likewise.
3415 (avr_out_round): Likewise.
3416 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3417 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3418 (s390_expand_vcond): Likewise.
3419 * config/spu/spu.c (spu_split_immediate): Likewise.
3420 (spu_expand_mov): Likewise.
3421 * dse.c (get_stored_val): Likewise.
3422 * expmed.c (store_bit_field_1): Likewise.
3423 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3425 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3426 (extract_low_bits): Likewise.
3427 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3428 handling rather than repeating the check.
3429 (emit_group_store): Likewise.
3430 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3431 * optabs.c (expand_absneg_bit): Likewise.
3432 (expand_copysign_absneg): Likewise.
3433 (expand_copysign_bit): Likewise.
3434 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3435 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3436 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3437 * var-tracking.c (prepare_call_arguments): Likewise.
3438 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3439 int_mode_for_mode instead of mode_for_size.
3440 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3443 Alan Hayward <alan.hayward@arm.com>
3444 David Sherwood <david.sherwood@arm.com>
3446 * machmode.h (int_mode_for_size): New function.
3447 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3448 instead of mode_for_size.
3449 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3451 * combine.c (expand_field_assignment): Use int_mode_for_size
3452 instead of mode_for_size.
3453 (make_extraction): Likewise.
3454 (simplify_shift_const_1): Likewise.
3455 (simplify_comparison): Likewise.
3456 * dojump.c (do_jump): Likewise.
3457 * dwarf2out.c (mem_loc_descriptor): Likewise.
3458 * emit-rtl.c (init_derived_machine_modes): Likewise.
3459 * expmed.c (flip_storage_order): Likewise.
3460 (convert_extracted_bit_field): Likewise.
3461 * expr.c (copy_blkmode_from_reg): Likewise.
3462 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3463 * internal-fn.c (expand_mul_overflow): Likewise.
3464 * lower-subreg.c (simple_move): Likewise.
3465 * optabs-libfuncs.c (init_optabs): Likewise.
3466 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3467 * tree.c (vector_type_mode): Likewise.
3468 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3469 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3470 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3471 * tree-vect-stmts.c (vectorizable_load): Likewise.
3472 (vectorizable_store): Likewise.
3474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3475 Alan Hayward <alan.hayward@arm.com>
3476 David Sherwood <david.sherwood@arm.com>
3478 * coretypes.h (pod_mode): New type.
3479 (scalar_int_mode_pod): New typedef.
3480 * machmode.h (pod_mode): New class.
3481 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3482 * genmodes.c (emit_mode_int_n): Update accordingly.
3483 * lower-subreg.h (target_lower_subreg): Change type to
3484 scalar_int_mode_pod.
3485 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3486 scalar_int_mode_pod.
3488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3489 Alan Hayward <alan.hayward@arm.com>
3490 David Sherwood <david.sherwood@arm.com>
3492 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3493 machine_mode to scalar_int_mode.
3494 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3495 (rs6000_option_override_internal): Remove cast to int.
3496 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3497 machine_mode to scalar_int_mode.
3498 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3499 (rs6000_option_override_internal): Remove cast to int.
3500 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3501 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3503 * config/s390/s390.c (s390_expand_builtin): Likewise.
3504 * coretypes.h (scalar_int_mode): New type.
3505 (opt_scalar_int_mode): New typedef.
3506 * machmode.h (scalar_int_mode): New class.
3507 (scalar_int_mode::includes_p): New function.
3508 (byte_mode): Change type to scalar_int_mode.
3509 (word_mode): Likewise.
3510 (ptr_mode): Likewise.
3511 * emit-rtl.c (byte_mode): Likewise.
3512 (word_mode): Likewise.
3513 (ptr_mode): Likewise.
3514 (init_derived_machine_modes): Update accordingly.
3515 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3516 and MODE_PARTIAL_INT.
3517 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3518 opt_scalar_int_mode.
3520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3521 Alan Hayward <alan.hayward@arm.com>
3522 David Sherwood <david.sherwood@arm.com>
3524 * target.def (libgcc_floating_mode_supported_p): Take a
3526 * doc/tm.texi: Regenerate.
3527 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3529 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3530 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3534 Alan Hayward <alan.hayward@arm.com>
3535 David Sherwood <david.sherwood@arm.com>
3537 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3538 * doc/tm.texi: Regenerate.
3539 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3540 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3541 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3542 * targhooks.h (default_floatn_mode): Likewise.
3543 * targhooks.c (default_floatn_mode): Likewise.
3544 * tree.c (build_common_tree_nodes): Update accordingly.
3546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3547 Alan Hayward <alan.hayward@arm.com>
3548 David Sherwood <david.sherwood@arm.com>
3550 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3551 (mode_iterator::iterate_p): Likewise.
3552 (mode_iterator::get_wider): Likewise.
3553 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3556 Alan Hayward <alan.hayward@arm.com>
3557 David Sherwood <david.sherwood@arm.com>
3559 * coretypes.h (opt_scalar_float_mode): New typedef.
3560 * machmode.h (float_mode_for_size): New function.
3561 * emit-rtl.c (double_mode): Delete.
3562 (init_emit_once): Use float_mode_for_size.
3563 * stor-layout.c (layout_type): Likewise.
3564 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3567 Alan Hayward <alan.hayward@arm.com>
3568 David Sherwood <david.sherwood@arm.com>
3570 * output.h (assemble_real): Take a scalar_float_mode.
3571 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3572 * config/arm/arm.md (consttable_4): Likewise.
3573 (consttable_8): Likewise.
3574 (consttable_16): Likewise.
3575 * config/mips/mips.md (consttable_float): Likewise.
3576 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3577 * varasm.c (assemble_real): Take a scalar_float_mode.
3578 (output_constant_pool_2): Update accordingly.
3579 (output_constant): Likewise.
3581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3582 Alan Hayward <alan.hayward@arm.com>
3583 David Sherwood <david.sherwood@arm.com>
3585 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3586 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3587 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3588 (native_encode_real): Likewise.
3589 (native_interpret_real): Likewise.
3590 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3591 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3594 Alan Hayward <alan.hayward@arm.com>
3595 David Sherwood <david.sherwood@arm.com>
3597 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3598 <scalar_float_mode>. Simplify.
3599 (gen_extend_conv_libfunc): Likewise.
3601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3602 Alan Hayward <alan.hayward@arm.com>
3603 David Sherwood <david.sherwood@arm.com>
3605 * coretypes.h (scalar_float_mode): New type.
3606 * machmode.h (mode_traits::from_int): Use machine_mode if
3607 USE_ENUM_MODES is defined.
3608 (is_a): New function.
3610 (dyn_cast): Likewise.
3611 (scalar_float_mode): New class.
3612 (scalar_float_mode::includes_p): New function.
3613 (is_float_mode): Likewise.
3614 * gdbhooks.py (MachineModePrinter): New class.
3615 (build_pretty_printer): Use it for scalar_float_mode.
3616 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3617 (format_helper::format_helper): Turn into a template.
3618 * genmodes.c (get_mode_class): New function.
3619 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3620 or machine_mode if none.
3621 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3622 as_a <scalar_float_mode>.
3623 * dwarf2out.c (mem_loc_descriptor): Likewise.
3624 (insert_float): Likewise.
3625 (add_const_value_attribute): Likewise.
3626 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3627 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3628 (expand_unop): Update accordingly.
3629 (expand_abs_nojump): Likewise.
3630 (expand_copysign_absneg): Take a scalar_float_mode.
3631 (expand_copysign_bit): Likewise.
3632 (expand_copysign): Update accordingly.
3634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3635 Alan Hayward <alan.hayward@arm.com>
3636 David Sherwood <david.sherwood@arm.com>
3638 * coretypes.h (opt_mode): New class.
3639 * machmode.h (opt_mode): Likewise.
3640 (opt_mode::else_void): New function.
3641 (opt_mode::require): Likewise.
3642 (opt_mode::exists): Likewise.
3643 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3644 (GET_MODE_2XWIDER_MODE): Likewise.
3645 (mode_iterator::get_wider): Update accordingly.
3646 (mode_iterator::get_2xwider): Likewise.
3647 (mode_iterator::get_known_wider): Likewise, turning into a template.
3648 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3649 forcing a wider mode to exist.
3650 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3651 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3652 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3653 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3654 * lower-subreg.c (init_lower_subreg): Likewise.
3655 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3656 on the final iteration.
3657 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3658 a wider mode exists before asking for a move pattern.
3659 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3660 forcing a wider mode to exist.
3661 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3662 returning false if no such mode exists.
3663 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3664 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3665 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3666 Avoid checking for a MODE_INT if we already know the mode is not a
3668 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3669 forcing a wider mode to exist.
3670 (expmed_mult_highpart_optab): Likewise.
3671 (expmed_mult_highpart): Likewise.
3672 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3674 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3675 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3676 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3677 * internal-fn.c (expand_mul_overflow): Update use of
3678 GET_MODE_2XWIDER_MODE.
3679 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3680 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3681 GET_MODE_WIDER_MODE.
3682 (convert_plusminus_to_widen): Likewise.
3683 * tree-switch-conversion.c (array_value_type): Likewise.
3684 * var-tracking.c (emit_note_insn_var_location): Likewise.
3685 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3686 Return false inside rather than outside the loop if no wider mode
3688 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3689 and GET_MODE_2XWIDER_MODE
3690 (can_compare_p): Use else_void.
3691 * gdbhooks.py (OptMachineModePrinter): New class.
3692 (build_pretty_printer): Use it for opt_mode.
3694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3695 Alan Hayward <alan.hayward@arm.com>
3696 David Sherwood <david.sherwood@arm.com>
3698 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3699 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3702 Alan Hayward <alan.hayward@arm.com>
3703 David Sherwood <david.sherwood@arm.com>
3705 * machmode.h (mode_traits): New structure.
3706 (get_narrowest_mode): New function.
3707 (mode_iterator::start): Likewise.
3708 (mode_iterator::iterate_p): Likewise.
3709 (mode_iterator::get_wider): Likewise.
3710 (mode_iterator::get_known_wider): Likewise.
3711 (mode_iterator::get_2xwider): Likewise.
3712 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3713 (FOR_EACH_MODE): Likewise.
3714 (FOR_EACH_MODE_FROM): Likewise.
3715 (FOR_EACH_MODE_UNTIL): Likewise.
3716 (FOR_EACH_WIDER_MODE): Likewise.
3717 (FOR_EACH_2XWIDER_MODE): Likewise.
3718 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3719 * combine.c (simplify_comparison): Likewise
3720 * config/i386/i386.c (type_natural_mode): Likewise.
3721 * cse.c (cse_insn): Likewise.
3722 * dse.c (find_shift_sequence): Likewise.
3723 * emit-rtl.c (init_derived_machine_modes): Likewise.
3724 (init_emit_once): Likewise.
3725 * explow.c (hard_function_value): Likewise.
3726 * expmed.c (extract_fixed_bit_field_1): Likewise.
3727 (extract_bit_field_1): Likewise.
3728 (expand_divmod): Likewise.
3729 (emit_store_flag_1): Likewise.
3730 * expr.c (init_expr_target): Likewise.
3731 (convert_move): Likewise.
3732 (alignment_for_piecewise_move): Likewise.
3733 (widest_int_mode_for_size): Likewise.
3734 (emit_block_move_via_movmem): Likewise.
3735 (copy_blkmode_to_reg): Likewise.
3736 (set_storage_via_setmem): Likewise.
3737 (compress_float_constant): Likewise.
3738 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3739 * optabs-query.c (get_best_extraction_insn): Likewise.
3740 * optabs.c (expand_binop): Likewise.
3741 (expand_twoval_unop): Likewise.
3742 (expand_twoval_binop): Likewise.
3743 (widen_leading): Likewise.
3744 (widen_bswap): Likewise.
3745 (expand_parity): Likewise.
3746 (expand_unop): Likewise.
3747 (prepare_cmp_insn): Likewise.
3748 (prepare_float_lib_cmp): Likewise.
3749 (expand_float): Likewise.
3750 (expand_fix): Likewise.
3751 (expand_sfix_optab): Likewise.
3752 * postreload.c (move2add_use_add2_insn): Likewise.
3753 * reg-stack.c (reg_to_stack): Likewise.
3754 * reginfo.c (choose_hard_reg_mode): Likewise.
3755 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3756 * stor-layout.c (mode_for_size): Likewise.
3757 (smallest_mode_for_size): Likewise.
3758 (mode_for_vector): Likewise.
3759 (finish_bitfield_representative): Likewise.
3760 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3761 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3762 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3763 * var-tracking.c (prepare_call_arguments): Likewise.
3765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3766 Alan Hayward <alan.hayward@arm.com>
3767 David Sherwood <david.sherwood@arm.com>
3769 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3770 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3771 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3772 * machmode.h (mode_size): Move earlier in file.
3773 (mode_precision): Likewise.
3774 (mode_inner): Likewise.
3775 (mode_nunits): Likewise.
3776 (mode_unit_size): Likewise.
3777 (unit_unit_precision): Likewise.
3778 (mode_wider): Likewise.
3779 (mode_2xwider): Likewise.
3780 (machine_mode): New class.
3781 (mode_to_bytes): New function.
3782 (mode_to_bits): Likewise.
3783 (mode_to_precision): Likewise.
3784 (mode_to_inner): Likewise.
3785 (mode_to_unit_size): Likewise.
3786 (mode_to_unit_precision): Likewise.
3787 (mode_to_nunits): Likewise.
3788 (GET_MODE_SIZE): Use mode_to_bytes.
3789 (GET_MODE_BITSIZE): Use mode_to_bits.
3790 (GET_MODE_PRECISION): Use mode_to_precision.
3791 (GET_MODE_INNER): Use mode_to_inner.
3792 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3793 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3794 (GET_MODE_NUNITS): Use mode_to_nunits.
3795 * system.h (ALWAYS_INLINE): New macro.
3796 * config/powerpcspe/powerpcspe-c.c
3797 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3798 int for arg1_mode and arg2_mode.
3800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3801 Alan Hayward <alan.hayward@arm.com>
3802 David Sherwood <david.sherwood@arm.com>
3804 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3805 Prefix mode names with E_ in case statements.
3806 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3807 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3808 (aarch64_split_simd_move): Likewise.
3809 (aarch64_gen_storewb_pair): Likewise.
3810 (aarch64_gen_loadwb_pair): Likewise.
3811 (aarch64_gen_store_pair): Likewise.
3812 (aarch64_gen_load_pair): Likewise.
3813 (aarch64_get_condition_code_1): Likewise.
3814 (aarch64_constant_pool_reload_icode): Likewise.
3815 (get_rsqrte_type): Likewise.
3816 (get_rsqrts_type): Likewise.
3817 (get_recpe_type): Likewise.
3818 (get_recps_type): Likewise.
3819 (aarch64_gimplify_va_arg_expr): Likewise.
3820 (aarch64_simd_container_mode): Likewise.
3821 (aarch64_emit_load_exclusive): Likewise.
3822 (aarch64_emit_store_exclusive): Likewise.
3823 (aarch64_expand_compare_and_swap): Likewise.
3824 (aarch64_gen_atomic_cas): Likewise.
3825 (aarch64_emit_bic): Likewise.
3826 (aarch64_emit_atomic_swap): Likewise.
3827 (aarch64_emit_atomic_load_op): Likewise.
3828 (aarch64_evpc_trn): Likewise.
3829 (aarch64_evpc_uzp): Likewise.
3830 (aarch64_evpc_zip): Likewise.
3831 (aarch64_evpc_ext): Likewise.
3832 (aarch64_evpc_rev): Likewise.
3833 (aarch64_evpc_dup): Likewise.
3834 (aarch64_gen_ccmp_first): Likewise.
3835 (aarch64_gen_ccmp_next): Likewise.
3836 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3837 (alpha_emit_xfloating_libcall): Likewise.
3838 (emit_insxl): Likewise.
3839 (alpha_arg_type): Likewise.
3840 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3841 (arc_preferred_simd_mode): Likewise.
3842 (arc_secondary_reload): Likewise.
3843 (get_arc_condition_code): Likewise.
3844 (arc_print_operand): Likewise.
3845 (arc_legitimate_constant_p): Likewise.
3846 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3847 * config/arc/arc.md (casesi_load): Likewise.
3848 (casesi_compact_jump): Likewise.
3849 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3850 (cc_use_register): Likewise.
3851 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3852 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3853 (arm_init_iwmmxt_builtins): Likewise.
3854 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3855 (neon_expand_vector_init): Likewise.
3856 (arm_attr_length_move_neon): Likewise.
3857 (maybe_get_arm_condition_code): Likewise.
3858 (arm_emit_vector_const): Likewise.
3859 (arm_preferred_simd_mode): Likewise.
3860 (arm_output_iwmmxt_tinsr): Likewise.
3861 (thumb1_output_casesi): Likewise.
3862 (thumb2_output_casesi): Likewise.
3863 (arm_emit_load_exclusive): Likewise.
3864 (arm_emit_store_exclusive): Likewise.
3865 (arm_expand_compare_and_swap): Likewise.
3866 (arm_evpc_neon_vuzp): Likewise.
3867 (arm_evpc_neon_vzip): Likewise.
3868 (arm_evpc_neon_vrev): Likewise.
3869 (arm_evpc_neon_vtrn): Likewise.
3870 (arm_evpc_neon_vext): Likewise.
3871 (arm_validize_comparison): Likewise.
3872 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3873 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3874 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3875 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3876 (c6x_preferred_simd_mode): Likewise.
3877 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3878 (epiphany_rtx_costs): Likewise.
3879 * config/epiphany/predicates.md (proper_comparison_operator):
3881 * config/frv/frv.c (condexec_memory_operand): Likewise.
3882 (frv_emit_move): Likewise.
3883 (output_move_single): Likewise.
3884 (output_condmove_single): Likewise.
3885 (frv_hard_regno_mode_ok): Likewise.
3886 (frv_matching_accg_mode): Likewise.
3887 * config/h8300/h8300.c (split_adds_subs): Likewise.
3888 (h8300_rtx_costs): Likewise.
3889 (h8300_print_operand): Likewise.
3890 (compute_mov_length): Likewise.
3891 (output_logical_op): Likewise.
3892 (compute_logical_op_length): Likewise.
3893 (compute_logical_op_cc): Likewise.
3894 (h8300_shift_needs_scratch_p): Likewise.
3895 (output_a_shift): Likewise.
3896 (compute_a_shift_length): Likewise.
3897 (compute_a_shift_cc): Likewise.
3898 (expand_a_rotate): Likewise.
3899 (output_a_rotate): Likewise.
3900 * config/i386/i386.c (classify_argument): Likewise.
3901 (function_arg_advance_32): Likewise.
3902 (function_arg_32): Likewise.
3903 (function_arg_64): Likewise.
3904 (function_value_64): Likewise.
3905 (ix86_gimplify_va_arg): Likewise.
3906 (ix86_legitimate_constant_p): Likewise.
3907 (put_condition_code): Likewise.
3908 (split_double_mode): Likewise.
3909 (ix86_avx256_split_vector_move_misalign): Likewise.
3910 (ix86_expand_vector_logical_operator): Likewise.
3911 (ix86_split_idivmod): Likewise.
3912 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3913 (ix86_build_const_vector): Likewise.
3914 (ix86_build_signbit_mask): Likewise.
3915 (ix86_match_ccmode): Likewise.
3916 (ix86_cc_modes_compatible): Likewise.
3917 (ix86_expand_branch): Likewise.
3918 (ix86_expand_sse_cmp): Likewise.
3919 (ix86_expand_sse_movcc): Likewise.
3920 (ix86_expand_int_sse_cmp): Likewise.
3921 (ix86_expand_vec_perm_vpermi2): Likewise.
3922 (ix86_expand_vec_perm): Likewise.
3923 (ix86_expand_sse_unpack): Likewise.
3924 (ix86_expand_int_addcc): Likewise.
3925 (ix86_split_to_parts): Likewise.
3926 (ix86_vectorize_builtin_gather): Likewise.
3927 (ix86_vectorize_builtin_scatter): Likewise.
3928 (avx_vpermilp_parallel): Likewise.
3929 (inline_memory_move_cost): Likewise.
3930 (ix86_tieable_integer_mode_p): Likewise.
3931 (x86_maybe_negate_const_int): Likewise.
3932 (ix86_expand_vector_init_duplicate): Likewise.
3933 (ix86_expand_vector_init_one_nonzero): Likewise.
3934 (ix86_expand_vector_init_one_var): Likewise.
3935 (ix86_expand_vector_init_concat): Likewise.
3936 (ix86_expand_vector_init_interleave): Likewise.
3937 (ix86_expand_vector_init_general): Likewise.
3938 (ix86_expand_vector_set): Likewise.
3939 (ix86_expand_vector_extract): Likewise.
3940 (emit_reduc_half): Likewise.
3941 (ix86_emit_i387_round): Likewise.
3942 (ix86_mangle_type): Likewise.
3943 (ix86_expand_round_sse4): Likewise.
3944 (expand_vec_perm_blend): Likewise.
3945 (canonicalize_vector_int_perm): Likewise.
3946 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3947 (expand_vec_perm_1): Likewise.
3948 (expand_vec_perm_interleave3): Likewise.
3949 (expand_vec_perm_even_odd_pack): Likewise.
3950 (expand_vec_perm_even_odd_1): Likewise.
3951 (expand_vec_perm_broadcast_1): Likewise.
3952 (ix86_vectorize_vec_perm_const_ok): Likewise.
3953 (ix86_expand_vecop_qihi): Likewise.
3954 (ix86_expand_mul_widen_hilo): Likewise.
3955 (ix86_expand_sse2_abs): Likewise.
3956 (ix86_expand_pextr): Likewise.
3957 (ix86_expand_pinsr): Likewise.
3958 (ix86_preferred_simd_mode): Likewise.
3959 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3960 * config/i386/sse.md (*andnot<mode>3): Likewise.
3961 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3962 (*<code><mode>3): Likewise.
3963 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3964 (ia64_expand_atomic_op): Likewise.
3965 (ia64_arg_type): Likewise.
3966 (ia64_mode_to_int): Likewise.
3967 (ia64_scalar_mode_supported_p): Likewise.
3968 (ia64_vector_mode_supported_p): Likewise.
3969 (expand_vec_perm_broadcast): Likewise.
3970 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3971 (iq2000_function_arg_advance): Likewise.
3972 (iq2000_function_arg): Likewise.
3973 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3974 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3975 (m68k_libcall_value): Likewise.
3976 (m68k_function_value): Likewise.
3977 (sched_attr_op_type): Likewise.
3978 * config/mcore/mcore.c (mcore_output_move): Likewise.
3979 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3981 (microblaze_function_arg): Likewise.
3982 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3983 (mips_print_operand): Likewise.
3984 (mips_mode_ok_for_mov_fmt_p): Likewise.
3985 (mips_vector_mode_supported_p): Likewise.
3986 (mips_preferred_simd_mode): Likewise.
3987 (mips_expand_vpc_loongson_even_odd): Likewise.
3988 (mips_expand_vec_unpack): Likewise.
3989 (mips_expand_vi_broadcast): Likewise.
3990 (mips_expand_vector_init): Likewise.
3991 (mips_expand_vec_reduc): Likewise.
3992 (mips_expand_msa_cmp): Likewise.
3993 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3994 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3995 (cc_flags_for_mode): Likewise.
3996 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3997 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3998 (nds32_output_casesi_pc_relative): Likewise.
3999 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4000 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4001 (nvptx_gen_unpack): Likewise.
4002 (nvptx_gen_pack): Likewise.
4003 (nvptx_gen_shuffle): Likewise.
4004 (nvptx_gen_wcast): Likewise.
4005 (nvptx_preferred_simd_mode): Likewise.
4006 * config/pa/pa.c (pa_secondary_reload): Likewise.
4007 * config/pa/predicates.md (base14_operand): Likewise.
4008 * config/powerpcspe/powerpcspe-c.c
4009 (altivec_resolve_overloaded_builtin): Likewise.
4010 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4012 (rs6000_preferred_simd_mode): Likewise.
4013 (output_vec_const_move): Likewise.
4014 (rs6000_expand_vector_extract): Likewise.
4015 (rs6000_split_vec_extract_var): Likewise.
4016 (reg_offset_addressing_ok_p): Likewise.
4017 (rs6000_legitimate_offset_address_p): Likewise.
4018 (rs6000_legitimize_address): Likewise.
4019 (rs6000_emit_set_const): Likewise.
4020 (rs6000_const_vec): Likewise.
4021 (rs6000_emit_move): Likewise.
4022 (spe_build_register_parallel): Likewise.
4023 (rs6000_darwin64_record_arg_recurse): Likewise.
4024 (swap_selector_for_mode): Likewise.
4025 (spe_init_builtins): Likewise.
4026 (paired_init_builtins): Likewise.
4027 (altivec_init_builtins): Likewise.
4028 (do_load_for_compare): Likewise.
4029 (rs6000_generate_compare): Likewise.
4030 (rs6000_expand_float128_convert): Likewise.
4031 (emit_load_locked): Likewise.
4032 (emit_store_conditional): Likewise.
4033 (rs6000_output_function_epilogue): Likewise.
4034 (rs6000_handle_altivec_attribute): Likewise.
4035 (rs6000_function_value): Likewise.
4036 (emit_fusion_gpr_load): Likewise.
4037 (emit_fusion_p9_load): Likewise.
4038 (emit_fusion_p9_store): Likewise.
4039 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4040 (fusion_gpr_mem_load): Likewise.
4041 (fusion_addis_mem_combo_load): Likewise.
4042 (fusion_addis_mem_combo_store): Likewise.
4043 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4044 (fusion_gpr_mem_load): Likewise.
4045 (fusion_addis_mem_combo_load): Likewise.
4046 (fusion_addis_mem_combo_store): Likewise.
4047 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4049 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4050 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4051 (rs6000_preferred_simd_mode): Likewise.
4052 (output_vec_const_move): Likewise.
4053 (rs6000_expand_vector_extract): Likewise.
4054 (rs6000_split_vec_extract_var): Likewise.
4055 (reg_offset_addressing_ok_p): Likewise.
4056 (rs6000_legitimate_offset_address_p): Likewise.
4057 (rs6000_legitimize_address): Likewise.
4058 (rs6000_emit_set_const): Likewise.
4059 (rs6000_const_vec): Likewise.
4060 (rs6000_emit_move): Likewise.
4061 (rs6000_darwin64_record_arg_recurse): Likewise.
4062 (swap_selector_for_mode): Likewise.
4063 (paired_init_builtins): Likewise.
4064 (altivec_init_builtins): Likewise.
4065 (rs6000_expand_float128_convert): Likewise.
4066 (emit_load_locked): Likewise.
4067 (emit_store_conditional): Likewise.
4068 (rs6000_output_function_epilogue): Likewise.
4069 (rs6000_handle_altivec_attribute): Likewise.
4070 (rs6000_function_value): Likewise.
4071 (emit_fusion_gpr_load): Likewise.
4072 (emit_fusion_p9_load): Likewise.
4073 (emit_fusion_p9_store): Likewise.
4074 * config/rx/rx.c (rx_gen_move_template): Likewise.
4075 (flags_from_mode): Likewise.
4076 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4077 (s390_slb_comparison): Likewise.
4078 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4079 (s390_vector_mode_supported_p): Likewise.
4080 (s390_cc_modes_compatible): Likewise.
4081 (s390_match_ccmode_set): Likewise.
4082 (s390_canonicalize_comparison): Likewise.
4083 (s390_emit_compare_and_swap): Likewise.
4084 (s390_branch_condition_mask): Likewise.
4085 (s390_rtx_costs): Likewise.
4086 (s390_secondary_reload): Likewise.
4087 (__SECONDARY_RELOAD_CASE): Likewise.
4088 (s390_expand_cs): Likewise.
4089 (s390_preferred_simd_mode): Likewise.
4090 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4091 * config/sh/sh.c (sh_print_operand): Likewise.
4092 (dump_table): Likewise.
4093 (sh_secondary_reload): Likewise.
4094 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4095 * config/sh/sh.md (casesi_worker_1): Likewise.
4096 (casesi_worker_2): Likewise.
4097 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4098 (fcc_comparison_operator): Likewise.
4099 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4100 (emit_soft_tfmode_cvt): Likewise.
4101 (sparc_preferred_simd_mode): Likewise.
4102 (output_cbranch): Likewise.
4103 (sparc_print_operand): Likewise.
4104 (sparc_expand_vec_perm_bmask): Likewise.
4105 (vector_init_bshuffle): Likewise.
4106 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4107 (spu_vector_mode_supported_p): Likewise.
4108 (spu_expand_insv): Likewise.
4109 (spu_emit_branch_or_set): Likewise.
4110 (spu_handle_vector_attribute): Likewise.
4111 (spu_builtin_splats): Likewise.
4112 (spu_builtin_extract): Likewise.
4113 (spu_builtin_promote): Likewise.
4114 (spu_expand_sign_extend): Likewise.
4115 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4116 (tilegx_simd_int): Likewise.
4117 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4118 (tilepro_simd_int): Likewise.
4119 * config/v850/v850.c (const_double_split): Likewise.
4120 (v850_print_operand): Likewise.
4121 (ep_memory_offset): Likewise.
4122 * config/vax/vax.c (vax_rtx_costs): Likewise.
4123 (vax_output_int_move): Likewise.
4124 (vax_output_int_add): Likewise.
4125 (vax_output_int_subtract): Likewise.
4126 * config/visium/predicates.md (visium_branch_operator): Likewise.
4127 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4128 (visium_print_operand_address): Likewise.
4129 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4130 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4131 (xtensa_expand_conditional_branch): Likewise.
4132 (xtensa_copy_incoming_a7): Likewise.
4133 (xtensa_output_literal): Likewise.
4134 * dfp.c (decimal_real_maxval): Likewise.
4135 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4138 Alan Hayward <alan.hayward@arm.com>
4139 David Sherwood <david.sherwood@arm.com>
4141 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4142 (mode_nunits_inline): Likewise.
4143 (mode_inner_inline): Likewise.
4144 (mode_unit_size_inline): Likewise.
4145 (mode_unit_precision_inline): Likewise.
4146 (emit_insn_modes_h): Likewise. Also emit a #define of the
4148 (emit_mode_wider): Add an E_ prefix to mode names.
4149 (emit_mode_complex): Likewise.
4150 (emit_mode_inner): Likewise.
4151 (emit_mode_adjustments): Likewise.
4152 (emit_mode_int_n): Likewise.
4153 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4154 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4155 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4156 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4157 (CRC32_BUILTIN, ENTRY): Likewise.
4158 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4159 (aarch64_pop_regs): Likewise.
4160 (aarch64_process_components): Likewise.
4161 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4162 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4163 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4164 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4165 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4166 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4168 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4169 (pcmpistr): Likewise.
4170 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4171 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4172 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4174 * config/rl78/rl78.c (mduc_regs): Likewise.
4175 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4176 (htm_expand_builtin): Likewise.
4177 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4178 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4179 * config/xtensa/xtensa.c (print_operand): Likewise.
4180 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4181 (NUM_MODE_VECTOR_INT): Likewise.
4182 * genoutput.c (null_operand): Likewise.
4183 (output_operand_data): Likewise.
4184 * genrecog.c (print_parameter_value): Likewise.
4185 * lra.c (debug_operand_data): Likewise.
4187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4189 * dwarf2out.c (scompare_loc_descriptor_wide)
4190 (scompare_loc_descriptor_narrow): New functions, split out from...
4191 (scompare_loc_descriptor): ...here.
4192 * expmed.c (emit_store_flag_int): New function, split out from...
4193 (emit_store_flag): ...here.
4195 2017-08-30 Richard Biener <rguenther@suse.de>
4197 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4198 (dwarf2out_early_finish): Move setting of AT_pubnames from
4199 early debug output to early finish.
4201 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4203 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4204 and -mdata-region to the assembler.
4206 2017-08-30 Richard Biener <rguenther@suse.de>
4208 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4210 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4211 (dwarf2out_early_global_decl): Only generate a DIE for the
4212 abstract origin if it doesn't already exist or is a declaration DIE.
4213 (resolve_addr): Do not add the linkage name twice when
4214 generating a stub DIE for the DW_TAG_GNU_call_site target.
4216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4218 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4219 Use machine_mode rather than int for arg1_mode.
4221 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4224 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4225 that the second argument of the built-in functions to unpack
4226 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4227 switch statement instead a lot of if statements.
4228 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4229 Allow 64-bit values to be in Altivec registers as well as
4230 traditional floating point registers.
4231 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4233 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4235 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4236 MAX_REGS_PER_ADDRESS == 1.
4238 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4240 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4241 * config/i386/i386.c (ix86_option_override_internal): Simplify
4242 setting of opts->x_flag_entry.
4244 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4245 Jakub Jelinek <jakub@redhat.com>
4246 Richard Biener <rguenther@suse.de>
4248 PR tree-optimization/81503
4249 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4250 folded constant fits in the target type; reorder tests for clarity.
4252 2017-08-29 Martin Liska <mliska@suse.cz>
4254 * passes.def: Include pass_lower_switch.
4255 * stmt.c (dump_case_nodes): Remove and move to
4256 tree-switch-conversion.
4257 (case_values_threshold): Likewise.
4258 (expand_switch_as_decision_tree_p): Likewise.
4259 (emit_case_decision_tree): Likewise.
4260 (expand_case): Likewise.
4261 (balance_case_nodes): Likewise.
4262 (node_has_low_bound): Likewise.
4263 (node_has_high_bound): Likewise.
4264 (node_is_bounded): Likewise.
4265 (emit_case_nodes): Likewise.
4266 (struct simple_case_node): New struct.
4267 (add_case_node): Remove.
4268 (emit_case_dispatch_table): Use vector instead of case_list.
4269 (reset_out_edges_aux): Remove.
4270 (compute_cases_per_edge): Likewise.
4271 (expand_case): Build list of simple_case_node.
4272 (expand_sjlj_dispatch_table): Use it.
4273 * tree-switch-conversion.c (struct case_node): Moved from
4274 stmt.c and adjusted.
4275 (emit_case_nodes): Likewise.
4276 (node_has_low_bound): Likewise.
4277 (node_has_high_bound): Likewise.
4278 (node_is_bounded): Likewise.
4279 (case_values_threshold): Likewise.
4280 (reset_out_edges_aux): Likewise.
4281 (compute_cases_per_edge): Likewise.
4282 (add_case_node): Likewise.
4283 (dump_case_nodes): Likewise.
4284 (balance_case_nodes): Likewise.
4285 (expand_switch_as_decision_tree_p): Likewise.
4286 (emit_jump): Likewise.
4287 (emit_case_decision_tree): Likewise.
4288 (try_switch_expansion): Likewise.
4289 (do_jump_if_equal): Likewise.
4290 (emit_cmp_and_jump_insns): Likewise.
4291 (fix_phi_operands_for_edge): New function.
4292 (record_phi_operand_mapping): Likewise.
4293 (class pass_lower_switch): New pass.
4294 (pass_lower_switch::execute): New function.
4295 (make_pass_lower_switch): Likewise.
4296 (conditional_probability):
4297 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4298 * tree-pass.h: Add make_pass_lower_switch.
4300 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4303 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4306 2017-08-29 Richard Biener <rguenther@suse.de>
4308 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4309 we do not add a DW_AT_inline attribute twice.
4310 (gen_subprogram_die): Remove code setting DW_AT_inline on
4311 DECL_ABSTRACT_P nodes.
4313 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4315 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4316 calls to internal functions.
4317 (gimplify_modify_expr): Likewise.
4318 * tree-call-cdce.c (use_internal_fn): Likewise.
4319 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4320 (convert_to_divmod): Set the nothrow flag.
4321 * tree-if-conv.c (predicate_mem_writes): Likewise.
4322 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4323 (vectorizable_call): Likewise.
4324 (vectorizable_store): Likewise.
4325 (vectorizable_load): Likewise.
4326 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4327 (vect_recog_mask_conversion_pattern): Likewise.
4329 2017-08-29 Martin Liska <mliska@suse.cz>
4332 * gcc.c (driver_handle_option): Add new argument.
4333 * opts-common.c (handle_option): Pass
4334 target_option_override_hook.
4335 * opts-global.c (lang_handle_option): Add new option.
4336 (set_default_handlers): Add new argument.
4337 (decode_options): Likewise.
4338 * opts.c (target_handle_option): Likewise.
4339 (common_handle_option): Call target_option_override_hook.
4340 * opts.h (struct cl_option_handler_func): Add hook for
4341 target option override.
4342 (struct cl_option_handlers): Likewise.
4343 (set_default_handlers): Add new argument.
4344 (decode_options): Likewise.
4345 (common_handle_option): Likewise.
4346 (target_handle_option): Likewise.
4347 * toplev.c (toplev::main): Pass targetm.target_option.override
4350 2017-08-29 Richard Biener <rguenther@suse.de>
4351 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4353 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4354 life to the active subtree.
4356 2017-08-28 Jeff Law <law@redhat.com>
4358 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4359 derive_equivalences.
4360 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4362 (edge_info::derive_equivalences): New private member function
4364 * tree-ssa-dom.c (class edge_info): Changed from a struct
4365 to a class. Add ctor/dtor, methods and data members.
4366 (edge_info::edge_info): Renamed from allocate_edge_info.
4367 Initialize additional members.
4368 (edge_info::~edge_info): New.
4369 (free_dom_edge_info): Delete the edge info.
4370 (record_edge_info): Use new class & associated member functions.
4371 Tighten forms for testing for edge equivalences.
4372 (record_temporary_equivalences): Iterate over the simple
4373 equivalences rather than assuming there's only one per edge.
4374 (cprop_into_successor_phis): Iterate over the simple
4375 equivalences rather than assuming there's only one per edge.
4376 (optimize_stmt): Use operand_equal_p rather than pointer
4377 equality for mini-DSE code.
4379 2017-08-28 Nathan Sidwell <nathan@acm.org>
4381 * gcc.c (execute): Fold SIGPIPE handling into switch
4382 statement. Adjust internal error message.
4384 2017-08-28 Richard Biener <rguenther@suse.de>
4387 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4388 Do nothing for removed DIEs.
4390 2017-08-28 Richard Biener <rguenther@suse.de>
4392 PR tree-optimization/81977
4393 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4396 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4399 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4400 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4401 using targetm.gen_mem_thread_fence.
4403 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4406 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4407 predicate to register_operand. Reorder operands.
4408 (*btr<mode>): Ditto.
4409 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4410 (*btr<mode>_mask): Ditto.
4412 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4414 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4415 * config/rs6000/xmmintrin.h: New file.
4416 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4418 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4421 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4422 parameter and_insn and return it.
4423 (recombine_lvx_pattern): Insert a copy to ensure availability of
4424 the base register of the copied masking operation at the point of
4425 the instruction replacement.
4426 (recombine_stvx_pattern): Likewise.
4428 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4430 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4431 undocumented switches.
4432 (-mpower9-dform-vector): Likewise.
4433 (-mpower9-dform): Likewise.
4434 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4435 comments to delete references to -mpower9-dform* switches.
4436 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4437 Delete reference to -mpower9-dform* switches, test for
4438 -mpower9-vector instead.
4439 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4440 (OTHER_P9_VECTOR_MASKS): Likewise.
4441 (POWERPC_MASKS): Likewise.
4442 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4443 tests against -mpower9-dform* to -mpower9-vector. Delete code
4444 that checked for -mpower9-dform* consistancy with other options.
4445 Add test for -mpower9-misc to enable other power9 switches.
4446 (rs6000_init_hard_regno_mode_ok): Likewise.
4447 (rs6000_option_override_internal): Likewise.
4448 (rs6000_emit_prologue): Likewise.
4449 (rs6000_emit_epilogue): Likewise.
4450 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4451 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4452 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4454 (emit_fusion_p9_store): Likewise.
4455 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4456 resetting these macros if the assembler does not support ISA 3.0
4458 (TARGET_P9_DFORM_VECTOR): Likewise.
4459 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4460 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4462 2017-08-25 Alan Modra <amodra@gmail.com>
4464 PR rtl-optimization/81747
4465 * cse.c (cse_extended_basic_block): Don't attempt to record
4466 equivalences for degenerate conditional branches that branch
4467 to their fall-through.
4469 2017-08-24 Martin Sebor <msebor@redhat.com>
4472 * gimple-fold.c (size_must_be_zero_p): New function.
4473 (gimple_fold_builtin_memory_op): Call it.
4475 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4477 * config/rs6000/mm_malloc.h: New file.
4479 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4481 PR tree-optimization/81913
4482 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4483 analysis when either IVs in condition can wrap.
4485 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4487 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4488 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4490 2017-08-24 Richard Biener <rguenther@suse.de>
4493 * targhooks.c (default_target_can_inline_p): Properly
4494 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4495 is present and always compare.
4496 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4497 infer -mfpmath=sse from TARGET_SSE_P.
4498 (ix86_can_inline_p): Properly use target_option_default_node when
4499 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4501 2017-08-24 Richard Biener <rguenther@suse.de>
4504 * dwarf2out.c (output_die): Handle flag_generate_offload like
4506 (output_comp_unit): Likewise.
4507 (gen_array_type_die): Likewise.
4508 (dwarf2out_early_finish): Likewise.
4509 (note_variable_value_in_expr): Likewise.
4510 (dwarf2out_finish): Likewise. Adjust assert.
4511 * cgraphunit.c (symbol_table::compile): Move setting of
4512 flag_generate_offload earlier ...
4513 (symbol_table::finalize_compilation_unit): ... here, before
4514 early debug finalization.
4516 2017-08-24 Richard Biener <rguenther@suse.de>
4518 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4519 and ipa-fnsummary.h.
4520 (ix86_can_inline_p): When ix86_fpmath flags do not match
4521 check whether the callee uses FP math at all.
4523 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4526 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4527 instead of TYPE_PRECISION.
4529 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4530 Alan Hayward <alan.hayward@arm.com>
4531 David Sherwood <david.sherwood@arm.com>
4533 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4535 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4537 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4539 (ix86_frame::stack_realign_allocate): New field.
4540 (struct machine_frame_state): Modify comments.
4541 (machine_frame_state::sp_realigned_fp_end): New field.
4542 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4544 (sp_valid_at): Add assertion to assure no attempt to access invalid
4545 offset of a realigned stack.
4546 (fp_valid_at): Likewise.
4547 (choose_baseaddr): Modify comments.
4548 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4549 ix86_expand_prologue.
4550 (ix86_expand_prologue): Modify stack realignment and allocation.
4551 (ix86_expand_epilogue): Modify comments.
4552 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4553 avx2_runtime, avx512f, and avx512f_runtime.
4555 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4557 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4558 (mstackrealign): Do not init to -1.
4559 * config/i386/i386.c (ix86_option_override_internal):
4560 Check opts_set, not opts when setting default value of
4561 opts->x_ix86_force_align_arg_pointer.
4563 2017-08-23 Richard Biener <rguenther@suse.de>
4565 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4566 lang_hooks.decl_printable_name.
4567 * print-rtl-function.c (print_rtx_function): Likewise.
4568 * tree-pretty-print.c (dump_function_header): Likewise.
4570 2017-08-23 Richard Biener <rguenther@suse.de>
4573 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4574 -g0 at compile-time.
4576 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4579 * doc/sourcebuild.texi (Other hardware attributes):
4582 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4585 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4586 Fix single-use check.
4588 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4590 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4592 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4594 * doc/install.texi: Modify to add more details on running selected
4597 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4599 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4600 is combined with -mabi=ms.
4601 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4604 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4606 PR tree-optimization/81488
4607 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4608 and cached_basis fields.
4609 (MAX_SPREAD): New constant.
4610 (alloc_cand_and_find_basis): Initialize new fields.
4611 (clear_visited): New function.
4612 (create_phi_basis_1): Rename from create_phi_basis, set visited
4613 and cached_basis fields.
4614 (create_phi_basis): New wrapper function.
4615 (phi_add_costs_1): Rename from phi_add_costs, add spread
4616 parameter, set visited field, short-circuit when limits reached.
4617 (phi_add_costs): New wrapper function.
4618 (record_phi_increments_1): Rename from record_phi_increments, set
4620 (record_phi_increments): New wrapper function.
4621 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4622 (phi_incr_cost): New wrapper function.
4623 (all_phi_incrs_profitable_1): Rename from
4624 all_phi_incrs_profitable, set visited field.
4625 (all_phi_incrs_profitable): New wrapper function.
4627 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4628 Alan Hayward <alan.hayward@arm.com>
4629 David Sherwood <david.sherwood@arm.com>
4631 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4632 that takes the outer and inner modes.
4633 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4634 comparison as the canonical test for a paradoxical subreg.
4635 * combine.c (simplify_set): Use paradoxical_subreg_p.
4636 (make_extraction): Likewise.
4637 (force_to_mode): Likewise.
4638 (rtx_equal_for_field_assignment_p): Likewise.
4639 (gen_lowpart_for_combine): Likewise.
4640 (simplify_comparison): Likewise.
4641 * cse.c (equiv_constant): Likewise.
4642 * expmed.c (store_bit_field_1): Likewise.
4643 * final.c (alter_subreg): Likewise.
4644 * fwprop.c (propagate_rtx): Likewise.
4645 (forward_propagate_subreg): Likewise.
4646 * ira-conflicts.c (ira_build_conflicts): Likewise.
4647 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4648 * lra-constraints.c (curr_insn_transform): Likewise.
4649 (split_reg): Likewise.
4650 * lra-eliminations.c (move_plus_up): Likewise.
4651 (lra_eliminate_regs_1): Likewise.
4652 * recog.c (general_operand): Likewise.
4653 * ree.c (combine_reaching_defs): Likewise.
4654 * reload.c (push_reload): Likewise.
4655 (find_reloads): Likewise.
4656 * reload1.c (elimination_effects): Likewise.
4657 (compute_reload_subreg_offset): Likewise.
4658 (choose_reload_regs): Likewise.
4659 * rtlanal.c (subreg_lsb_1): Likewise.
4660 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4661 (simplify_subreg): Likewise.
4662 * var-tracking.c (track_loc_p): Likewise.
4663 * emit-rtl.c (byte_lowpart_offset): Likewise.
4664 (paradoxical_subreg_p): Delete out-of-line definition.
4666 2017-08-22 Jeff Law <law@redhat.com>
4668 PR tree-optimization/81741
4669 PR tree-optimization/71947
4670 * tree-ssa-dom.c: Include tree-inline.h.
4671 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4672 equivalences if one is more expensive to compute than the other.
4673 * tree-ssa-scopedtables.h (class const_or_copies): Make
4674 record_const_or_copy_raw method private.
4675 (class avail_exprs_stack): New method simplify_binary_operation.
4676 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4677 avail_exprs_stack::simplify_binary_operation as needed.
4678 (avail_exprs_stack::simplify_binary_operation): New function.
4680 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4682 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4683 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4684 (DOT_SYMBOLS): Likewise.
4685 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4686 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4687 (RS6000_ABI_NAME): Likewise.
4688 (TARGET_CMODEL): Likewise.
4689 (TOC_SECTION_ASM_OP): Likewise.
4690 (SET_CMODEL): New macro.
4691 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4693 2017-08-22 Richard Biener <rguenther@suse.de>
4695 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4696 to allow for free-lang-data replacements similar to verify_type_variant.
4698 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4699 Alan Hayward <alan.hayward@arm.com>
4700 David Sherwood <david.sherwood@arm.com>
4702 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4703 VOIDmode for the LABEL_REF.
4705 2017-08-22 Richard Biener <rguenther@suse.de>
4707 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4709 2017-08-22 Richard Biener <rguenther@suse.de>
4711 * common.opt (feliminate-dwarf2-dups): Ignore.
4712 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4713 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4714 same_die_p_wrap, compute_section_prefix,
4715 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4716 (comdat_symbol_id, comdat_symbol_number): Likewise.
4717 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4719 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4720 (output_die): Mark unreachable path unreachable.
4721 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4722 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4723 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4724 (dwarf2out_early_finish): Likewise.
4726 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4728 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4730 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4733 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4734 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4735 (avr_attribute_table) <io, io_low, address>: Initialize
4736 .decl_required with true.
4738 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4740 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4741 undocumented debugging options.
4742 (-mvsx-scalar-double): Likewise.
4743 (-mallow-df-permute): Likewise.
4744 (-mvectorize-builtins): Likewise.
4745 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4746 (rs6000_builtin_vectorized_function): Likewise.
4747 (rs6000_builtin_md_vectorized_function): Likewise.
4748 (rs6000_opt_vars): Likewise.
4750 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4753 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4754 (*btrq_imm): Rename from *btrq.
4755 (*btcq_imm): Rename from *btcq.
4756 (btsc): New code attribute.
4757 (*<btsc><mode>): New insn pattern.
4758 (*btr<mode>): Ditto.
4759 (*<btsc><mode>_mask): New insn_and_split pattern.
4760 (*btr<mode>_mask): Ditto.
4762 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4763 Alan Hayward <alan.hayward@arm.com>
4764 David Sherwood <david.sherwood@arm.com>
4766 * function.c (pad_below): Simplify padding calculation.
4768 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4769 Alan Hayward <alan.hayward@arm.com>
4770 David Sherwood <david.sherwood@arm.com>
4772 * target.def (function_prologue): Remove frame size argument.
4773 (function_epilogue): Likewise.
4774 * doc/tm.texi: Regenerate.
4775 * final.c (final_start_function): Update call to function_prologue.
4776 (final_end_function): Update call to function_epilogue.
4777 (default_function_pro_epilogue): Remove frame size argument.
4778 * output.h (default_function_pro_epilogue): Likewise.
4779 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4780 (arm_output_function_prologue): Likewise.
4781 * config/frv/frv.c (frv_function_prologue): Likewise.
4782 (frv_function_epilogue): Likewise.
4783 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4784 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4785 (ia64_output_function_epilogue): Likewise.
4786 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4787 (m32r_output_function_epilogue): Likewise.
4788 * config/microblaze/microblaze.c (microblaze_function_prologue)
4789 (microblaze_function_epilogue): Likewise.
4790 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4791 (mips_output_function_epilogue): Likewise.
4792 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4793 (mmix_target_asm_function_epilogue): Likewise.
4794 * config/msp430/msp430.c (msp430_start_function): Likewise.
4795 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4796 (nds32_asm_function_epilogue): Likewise.
4797 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4798 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4799 (pa_output_function_epilogue): Likewise.
4800 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4801 (rs6000_output_function_epilogue): Likewise.
4802 * config/rl78/rl78.c (rl78_start_function): Likewise.
4803 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4804 (rs6000_output_function_epilogue): Likewise.
4805 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4806 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4807 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4808 (sparc_asm_function_epilogue): Likewise.
4810 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4812 * tree.h (type_has_mode_precision_p): New function.
4813 * convert.c (convert_to_integer_1): Use it.
4814 * expr.c (expand_expr_real_2): Likewise.
4815 (expand_expr_real_1): Likewise.
4816 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4817 * match.pd: Likewise.
4818 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4819 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4820 * tree-tailcall.c (process_assignment): Likewise.
4821 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4822 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4823 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4824 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4825 (vectorizable_assignment): Likewise.
4826 (vectorizable_shift): Likewise.
4827 (vectorizable_operation): Likewise.
4828 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4830 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4832 * match.pd: Add pow (C, x) simplification.
4834 2017-08-21 Richard Biener <rguenther@suse.de>
4836 PR tree-optimization/81900
4837 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4838 for blocks with abnormal predecessors.
4839 (compute_antic): Do not set visited flag prematurely.
4841 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4844 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4846 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4848 * stor-layout.h (vector_type_mode): Move to...
4849 * tree.h (vector_type_mode): ...here.
4850 * stor-layout.c (vector_type_mode): Move to...
4851 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4853 2017-08-21 Richard Biener <rguenther@suse.de>
4855 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4856 register_external_die hooks.
4857 (debug_false_tree_charstarstar_uhwistar): Declare.
4858 (debug_nothing_tree_charstar_uhwi): Likewise.
4859 * debug.c (do_nothing_debug_hooks): Adjust.
4860 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4861 (debug_nothing_tree_charstar_uhwi): Likewise.
4862 * dbxout.c (dbx_debug_hooks): Adjust.
4863 (xcoff_debug_hooks): Likewise.
4864 * sdbout.c (sdb_debug_hooks): Likewise.
4865 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4866 * dwarf2out.c (macinfo_label_base): New global.
4867 (dwarf2out_register_external_die): New function for the
4868 register_external_die hook.
4869 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4870 (dwarf2_debug_hooks): Use them.
4871 (dwarf2_lineno_debug_hooks): Adjust.
4872 (struct die_struct): Add with_offset flag.
4873 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4874 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4875 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4876 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4877 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4878 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4879 defining section names for the early LTO debug variants.
4880 (reset_indirect_string): New helper.
4881 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4882 (print_dw_val): Add support for offsetted symbol references.
4883 (get_ultimate_context): Split out from is_cxx.
4884 (is_cxx): Use get_ultimate_context.
4885 (is_fortran): Add decl overload.
4886 (compute_comp_unit_symbol): Split out worker from
4887 compute_section_prefix.
4888 (compute_section_prefix): Call compute_comp_unit_symbol and
4889 set comdat_type_p here.
4890 (output_die): Skip DIE symbol output for the LTO added one.
4891 Handle DIE symbol references with offset.
4892 (output_comp_unit): Guard section name mangling properly.
4893 For LTO debug sections emit a symbol at the section beginning
4894 which we use to refer to its DIEs.
4895 (add_abstract_origin_attribute): For DIEs registered via
4896 dwarf2out_register_external_die directly refer to the early
4897 DIE rather than indirectly through the shadow one we created.
4898 Remove obsolete call to dwarf2out_abstract_function for
4899 non-function/block origins.
4900 (gen_array_type_die): When generating early LTO debug do
4901 not emit DW_AT_string_length.
4902 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4903 late when in LTO. As suggested place a gcc_unreachable for
4904 the DECL_ABSTRACT_P case.
4905 (gen_subprogram_die): Avoid another specification DIE
4906 for early built declarations/definitions for the late LTO case.
4907 (gen_variable_die): Add type references for late duplicated VLA dies
4909 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4910 we have the abstract instance already.
4911 (process_scope_var): Adjust decl DIE contexts in LTO which
4912 first puts them in limbo.
4913 (gen_decl_die): Do not generate type DIEs late apart from
4914 types for VLAs or for decls we do not yet have a DIE. Do not
4915 call dwarf2out_abstract_function late.
4916 (dwarf2out_early_global_decl): Make sure to create DIEs
4917 for abstract instances of a decl first.
4918 (dwarf2out_late_global_decl): Adjust comment.
4919 (output_macinfo_op): With multiple macro sections use
4920 macinfo_label_base to distinguish labels.
4921 (output_macinfo): Likewise. Update macinfo_label_base.
4922 Pass in the line info label.
4923 (note_variable_value_in_expr): When generating LTO resolve
4924 all variable values here by generating DIEs as needed.
4925 (init_sections_and_labels): Add early LTO debug flag parameter
4926 and generate different sections and names if set. Add generation
4927 counter for the labels so we can have multiple of them.
4928 (reset_dies): Helper to allow DIEs to be output multiple times.
4929 (dwarf2out_finish): When outputting DIEs to the fat part of an
4930 LTO object first reset DIEs.
4931 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4932 (modified_type_die): Check for decl_ultimate_origin being self
4934 (gen_type_die_with_usage): Likewise.
4935 (gen_typedef_die): Allow decl_ultimate_origin being self.
4936 (set_decl_abstract_flags): Remove.
4937 (set_block_abstract_flags): Likewise.
4938 (dwarf2out_abstract_function): Treat the early generated DIEs
4939 as the abstract copy and only add DW_AT_inline and
4940 DW_AT_artificial here and call set_decl_origin_self.
4941 If the DIE has an abstract origin don't do anything.
4942 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4943 if we have none yet (Go fails to build one, PR78628).
4944 (variably_modified_type_p): Prevent endless recursion for Ada
4945 cyclic pointer types.
4946 * lto-streamer-in.c: Include debug.h.
4947 (dref_queue): New global.
4948 (lto_read_tree_1): Stream in DIE references.
4949 (lto_input_tree): Register DIE references.
4950 (input_function): Stream DECL_DEBUG_ARGS.
4951 * lto-streamer-out.c: Include debug.h.
4952 (lto_write_tree_1): Output DIE references.
4953 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4954 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4955 (output_function): Stream DECL_DEBUG_ARGS.
4956 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4957 Stream DECL_ABSTRACT_ORIGIN.
4958 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4959 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4960 DECL_CONTEXT for file-scope decls.
4961 * lto-streamer.h (struct dref_entry): Declare.
4962 (dref_queue): Likewise.
4963 * cfgexpand.c (pass_expand::execute): Do not call the
4964 outlining_inline_function hook here.
4965 * lto-wrapper.c (debug_obj): New global.
4966 (tool_cleanup): Unlink it if required.
4967 (debug_objcopy): New function.
4968 (run_gcc): Handle early debug sections in the IL files by
4969 extracting them to separate files, partially linkin them and
4970 feeding the result back as result to the linker.
4971 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4972 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4973 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4974 sections into a separate segment.
4975 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4977 (darwin_asm_dwarf_section): Likewise.
4978 (darwin_asm_output_dwarf_offset): Likewise.
4979 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4981 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4982 Alan Hayward <alan.hayward@arm.com>
4983 David Sherwood <david.sherwood@arm.com>
4985 * read-md.h (md_reader::record_potential_iterator_use): Replace
4986 pointer argument with an rtx and an index.
4987 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4988 (apply_mode_iterator): Likewise.
4989 (apply_code_iterator): Likewise.
4990 (apply_int_iterator): Likewise.
4991 (apply_subst_iterator): Likewise.
4992 (record_iterator_use): Likewise.
4993 (record_attribute_use): Likewise.
4994 (md_reader::record_potential_iterator_use): Likewise. Update calls
4995 to record_iterator_use and apply_iterator.
4996 (iterator_use): Replace ptr with x and index.
4997 (attribute_use): Likewise.
4998 (apply_attribute_uses): Update calls to apply_iterator.
4999 (apply_iterators): Likewise. Update initialization of iterator_use.
5000 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5001 and record_potential_iterator_use.
5002 (rtx_reader::read_rtx_operand): Likewise.
5004 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5005 Alan Hayward <alan.hayward@arm.com>
5006 David Sherwood <david.sherwood@arm.com>
5008 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5011 2017-08-21 Richard Biener <rguenther@suse.de>
5014 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5015 at struct end conservatively when comparing common bases.
5017 2017-08-21 Richard Biener <rguenther@suse.de>
5019 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5020 (mem_ref_in_stmt): Remove.
5021 (determine_max_movement): Use ref index to get at the reference.
5022 (invariantness_dom_walker::before_dom_children): Deal with
5023 lim data already initialized.
5024 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5026 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5028 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5029 (TARGET_ISA_ROUND): Ditto.
5030 (TARGET_ROUND): Ditto.
5031 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5032 * config/i386/i386.md: Ditto.
5033 * config/i386/sse.md: Ditto.
5034 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5035 with OPTION_MASK_ISA_SSE4_1.
5037 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5040 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5041 __builtin_ia32_lzcnt_u16.
5043 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5046 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5047 (rs6000_set_current_function): Rewrite function to use it.
5049 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5052 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5053 and TYPE_WARN_IF_NOT_ALIGN.
5054 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5055 (handle_warn_if_not_align): New.
5056 (place_union_field): Call handle_warn_if_not_align.
5057 (place_field): Call handle_warn_if_not_align.
5058 Copy TYPE_WARN_IF_NOT_ALIGN.
5059 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5060 (layout_type): Likewise.
5061 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5063 (tree_decl_common): Add warn_if_not_align.
5064 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5065 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5066 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5067 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5068 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5069 * doc/extend.texi: Document warn_if_not_aligned attribute.
5070 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5072 2017-08-17 Martin Liska <mliska@suse.cz>
5075 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5076 (get_data_dependence): Use it as pointer type.
5077 (distribute_loop): Likewise.
5079 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5081 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5082 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5083 (p8_vmrgow_<mode>_direct): New define_insn.
5084 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5085 handle endianness for vmrgew and vmrgow permute patterns.
5087 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5089 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5090 * config/rs6000/rs6000-cpus.def: Remove comment.
5091 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5092 (POWERPC_MASKS): Likewise.
5093 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5094 use of TARGET_VSX_TIMODE.
5095 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5096 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5097 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5098 (rs6000_option_override_internal): Remove dead code.
5099 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5100 (rs6000_legitimize_reload_address): Likewise.
5101 (rs6000_legitimate_address_p): Likewise.
5102 (rs6000_opt_masks): Delete "vsx-timode".
5103 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5104 from function comment.
5105 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5106 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5107 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5109 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5110 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5111 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5114 (define_peephole2): Likewise.
5116 2017-08-17 Martin Sebor <msebor@redhat.com>
5119 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5120 past the end of an array.
5121 (test_pp_format): Add test cases.
5123 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5125 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5126 missing ECF_NOTHROW flags.
5128 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5131 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5132 operands residing in integer registers.
5133 (*vsx_le_perm_load_<mode>): Likewise.
5134 (*vsx_le_perm_store_<mode>): Likewise.
5135 (define_peephole2): Add peepholes to optimize the above.
5137 2017-08-17 Marek Polacek <polacek@redhat.com>
5140 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5141 to mimic what shorten_compare did. Change the return type to bool.
5142 (fold_cond_expr_with_comparison): Update call to
5143 operand_equal_for_comparison_p.
5144 (fold_ternary_loc): Likewise.
5146 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5148 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5150 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5152 2017-08-17 Richard Biener <rguenther@suse.de>
5154 * tree-ssa-structalias.c (solve_graph): When propagating
5155 to successors update the graphs succ edges and avoid duplicate work.
5157 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5160 * config/i386/i386.c (ix86_option_override_internal): Save target
5161 specific options after ix86_stack_protector_guard_reg was changed.
5163 2017-08-17 Richard Biener <rguenther@suse.de>
5165 PR tree-optimization/81827
5166 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5167 (new_var_info): Initialize it conservatively.
5168 (get_call_vi): Mark register vars.
5169 (new_scalar_tmp_constraint_exp): Likewise.
5170 (handle_rhs_call): Likewise.
5171 (handle_const_call): Likewise.
5172 (create_function_info_for): Likewise.
5173 (solve_constraints): Sort varinfos to separate register from
5174 non-register vars to pack points-to solution bitmaps during
5177 2017-08-17 Marek Polacek <polacek@redhat.com>
5179 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5181 2017-08-17 Richard Biener <rguenther@suse.de>
5183 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5184 to true when overflow is undefined and we saturated the result.
5186 2017-08-17 Alan Modra <amodra@gmail.com>
5189 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5190 Don't use store multiple if only one reg needs saving.
5191 (interesting_frame_related_regno): New function.
5192 (rs6000_frame_related): Don't emit frame info for regs that
5194 (rs6000_emit_epilogue): Likewise.
5196 2017-08-16 Nathan Sidwell <nathan@acm.org>
5198 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5199 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5200 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5201 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5202 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5203 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5204 (verify_type): Adjust for TYPE_BINFO move.
5205 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5206 process TYPE_BINFO directly.
5207 (hash_tree): Likewise.
5208 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5210 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5213 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5215 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5217 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5220 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5221 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5222 Add ix86_binary_operator_ok to insn constraint.
5223 (*iordi_1_bts): Ditto.
5224 (*xordi_1_btc): Ditto.
5225 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5226 Update corresponding peephole2 pattern.
5230 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5232 PR tree-optimization/81832
5233 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5234 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5236 2017-08-16 Marek Polacek <polacek@redhat.com>
5239 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5240 regarding size_zero_node.
5242 2017-08-16 Martin Liska <mliska@suse.cz>
5245 * config.gcc: Respect previously set extra_objs in case
5248 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5250 PR tree-optimization/81835
5251 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5252 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5253 not depend on the phi.
5255 2017-08-16 Alan Modra <amodra@gmail.com>
5257 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5260 2017-08-16 Alan Modra <amodra@gmail.com>
5262 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5263 (save_reg_p): ..into this. Update all callers.
5264 (first_reg_to_save): Simplify.
5266 2017-08-16 Alan Modra <amodra@gmail.com>
5268 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5271 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5275 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5276 constant count if that count is less than 32.
5278 2017-08-15 Nathan Sidwell <nathan@acm.org>
5280 * gcc.c (execute): Emit friendlier message if inferior is killed
5281 by an external cause.
5283 2017-08-15 Richard Biener <rguenther@suse.de>
5285 PR tree-optimization/81790
5286 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5287 CONSTRUCTORs from simplifying and VN.
5289 2017-08-14 Martin Sebor <msebor@redhat.com>
5291 * builtin-attrs.def: Add comments.
5293 2017-08-14 Martin Sebor <msebor@redhat.com>
5296 * doc/extend.texi (attribute nonstring): Document new attribute.
5298 2017-08-14 Martin Sebor <msebor@redhat.com>
5301 * tree-diagnostic.c (default_tree_printer): Handle %G.
5302 * gimple-pretty-print.h (percent_G_format): Declare new function.
5303 * gimple-pretty-print.c (percent_G_format): Define.
5304 * tree-pretty-print.c (percent_K_format): Add argument.
5306 2017-08-14 Martin Sebor <msebor@redhat.com>
5308 PR translation/79998
5309 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5310 Remove a stray space.
5312 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5315 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5316 (*iordi_1_bts): Ditto.
5317 (*xordi_1_btc): Ditto.
5319 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5322 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5323 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5325 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5326 quoted strings, and make more translator-friendly.
5327 (darwin_rs6000_override_options): Likewise.
5328 (rs6000_option_override_internal): Likewise.
5329 (rs6000_return_in_memory): Fix overlong line.
5330 (init_cmulative_args): Use quoted strings, and make more
5331 translator-friendly.
5332 (rs6000_pass_by_reference): Fix overlong line.
5333 (def_builtin): Use quoted strings.
5334 (altivec_expand_predicate_builtin): Use quoted strings, and make
5335 more translator-friendly.
5336 (htm_expand_builtin): Use quoted strings.
5337 (cpu_expand_builtin): Use quoted strings, and make more
5338 translator-friendly.
5339 (altivec_expand_builtin): Likewise.
5340 (paired_expand_predicate_builtin): Likewise.
5341 (rs6000_invalid_builtin): Likewise.
5342 (builtin_function_type): Use quoted strings.
5343 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5344 more translator-friendly.
5345 (rs6000_trampoline_init): Likewise.
5346 (rs6000_handle_altivec_attribute): Likewise.
5347 (rs6000_inner_target_options): Use quoted strings.
5348 (rs6000_disable_incompatible_switches): Likewise.
5349 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5350 strings, and make more translator-friendly.
5351 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5353 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5355 PR tree-optimization/81799
5356 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5357 cond_expr to simple gimple operand.
5359 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5362 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5364 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5368 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5369 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5370 TARGET_GASISR_PROLOGUES.
5371 * config/avr/avr.c (avr_option_override): Same.
5372 (avr_pass_pre_proep::execute): Same.
5374 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5377 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5378 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5380 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5382 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5383 define_insn_and_split. Split to a memory load from 0 in
5384 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5385 using PTR mode iterator.
5386 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5387 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5388 (*add_tp_<mode>): Redefine as define_insn_and_split.
5389 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5390 address space. Merge with *add_tp_x32 using PTR mode iterator.
5391 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5392 Split to an add with a memory load from 0 in
5393 DEFAULT_TLS_SEG_REG address space.
5395 2017-08-12 Andrew Pinski <apinski@cavium.com>
5397 * config/aarch64/aarch64-option-extensions.def (rdma):
5398 Fix feature string to what Linux prints out in /proc/cpuinfo.
5400 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5403 * dwarf2out.c (modified_type_die): For C typedef types that have
5404 an ultimate origin, process the ultimate origin instead of the
5406 (gen_typedef_die): Assert that input DECLs have no ultimate
5408 (gen_type_die_with_usage): For typedef variants that have an
5409 ultimate origin, just call gen_decl_die on the original DECL.
5410 (process_scope_var): Avoid creating DIEs for local typedefs and
5411 concrete static variables.
5413 2017-08-12 Alan Modra <amodra@gmail.com>
5417 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5418 match gnu-user.h startfile.
5419 (ENDFILE_LINUX_SPEC): Similarly.
5421 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5424 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5426 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5428 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5429 * config/aarch64/aarch64.md (mov<mode>): Change.
5430 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5431 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5432 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5434 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5436 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5437 for storage order barriers.
5439 2017-08-11 Martin Liska <mliska@suse.cz>
5442 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5443 variables of void type.
5445 2017-08-11 Martin Liska <mliska@suse.cz>
5447 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5448 TARGET_SUPPORTS_ALIASES.
5449 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5450 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5451 (optimize_weakref): Likewise.
5452 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5453 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5454 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5456 2017-08-11 Martin Liska <mliska@suse.cz>
5459 * config/i386/i386.c (make_resolver_func): Do complete
5460 refactoring of the function.
5462 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5465 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5466 * config/i386/i386.c (ix86_stack_protect_guard): Use
5467 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5468 * doc/invoke.texi (x86 Options): Document
5469 -mstack-protector-guard-symbol= option.
5471 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5473 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5474 * config/i386/i386.c (ix86_split_stack_guard): New function.
5475 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5476 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5477 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5478 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5479 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5480 (split_stack_space_check): Call ix86_split_stack_guard.
5482 2017-08-10 Martin Sebor <msebor@redhat.com>
5484 * print-tree.c (print_node): Print location using the established
5486 Replace spaces with colons.
5487 (debug_raw, debug): Ditto.
5489 2017-08-10 Martin Sebor <msebor@redhat.com>
5492 * pretty-print.c (pp_format): Correct the handling of %s precision.
5494 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5497 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5499 (ix86_finalize_stack_frame_flags): This. Also clear
5500 frame_pointer_needed if -fno-omit-frame-pointer is used without
5502 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5503 with ix86_finalize_stack_frame_flags.
5504 (ix86_expand_epilogue): Likewise.
5505 (ix86_expand_split_stack_prologue): Likewise.
5506 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5508 2017-08-10 Martin Liska <mliska@suse.cz>
5511 * c-attribs.c (handle_target_attribute):
5512 Report warning for an empty string argument of target attribute.
5514 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5517 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5519 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5520 or DECL_NONLOCAL labels.
5521 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5522 or DECL_NONLOCAL labels here.
5524 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5526 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5527 to indicate when early gimple folding has been disabled.
5528 (rs6000_gimple_fold_builtin): Add debug content.
5529 (rs6000_invalid_builtin): Fix whitespace.
5530 (rs6000_expand_builtin): Fix whitespace.
5531 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5533 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5536 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5537 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5539 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5541 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5542 (qdf24xx): Likewise.
5543 * config/aarch64/aarch64.md: Include falkor.md.
5544 * config/aarch64/falkor.md: New.
5546 2017-08-09 Marek Polacek <polacek@redhat.com>
5549 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5550 * diagnostic.c (emit_diagnostic): Add a comment.
5551 (emit_diagnostic_valist): New function.
5553 2017-08-09 Marek Polacek <polacek@redhat.com>
5556 * input.c (make_location): New overload.
5557 * input.h (make_location): Declare.
5559 2017-08-08 Alan Modra <amodra@gmail.com>
5560 H.J. Lu <hongjiu.lu@intel.com>
5563 * gcc.c (NO_PIE_SPEC): Delete.
5564 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5566 (LINK_PIE_SPEC): ..to here.
5567 (LINK_COMMAND_SPEC): Support -no-pie.
5568 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5569 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5571 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5572 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5573 (ENDFILE_CRTEND_SPEC): Similarly.
5575 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5578 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5579 (mstack-protector-guard-offset=): Ditto.
5580 * config/i386/i386.c (ix86_option_override): Handle
5581 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5583 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5584 ix86_stack_protect_guard_offset variables.
5585 (TARGET_STACK_PROTECT_GUARD): Always define.
5586 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5587 and -mstack-protector-guard-offset= options.
5589 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5591 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5592 boundary case for the last candidate.
5594 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5596 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5597 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5599 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5602 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5603 * config/aarch64/aarch64-builtins.c
5604 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5605 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5606 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5608 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5609 Andrew Pinski <pinskia@gmail.com>
5612 * internal-fn.def (XORSIGN): New.
5613 * optabs.def (xorsign_optab): New.
5614 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5615 (convert_expand_mult_copysign): New.
5616 (pass_optimize_widening_mul::execute): Call
5617 convert_expand_mult_copysign.
5619 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5621 PR tree-optimization/81354
5622 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5623 Insert on edges rather than explicitly creating landing pads.
5624 (analyze_candidates_and_replace): Commit edge inserts.
5626 2017-08-08 Richard Biener <rguenther@suse.de>
5629 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5630 (expand_simple_operations): Also look through ADDR_EXPRs with
5631 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5633 2017-08-08 Richard Biener <rguenther@suse.de>
5635 PR tree-optimization/81723
5636 * tree-vect-slp.c (struct bst_traits): New hash traits.
5637 (bst_fail): New global.
5638 (vect_build_slp_tree_2): New worker, split out from ...
5639 (vect_build_slp_tree): ... this now wrapping it with using
5640 bst_fail set to cache SLP tree build fails. Properly handle
5642 (vect_analyze_slp_instance): Allocate and free bst_fail.
5644 2017-08-08 Martin Liska <mliska@suse.cz>
5647 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5648 LABEL_DECLs that can be from a different function.
5650 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5652 PR tree-optimization/81744
5653 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5654 loop's number of iterations.
5656 2017-08-08 Martin Liska <mliska@suse.cz>
5658 * asan.c: Include header files.
5659 * attribs.c (build_decl_attribute_variant): New function moved
5661 (build_type_attribute_qual_variant): Likewise.
5662 (cmp_attrib_identifiers): Likewise.
5663 (simple_cst_list_equal): Likewise.
5664 (omp_declare_simd_clauses_equal): Likewise.
5665 (attribute_value_equal): Likewise.
5666 (comp_type_attributes): Likewise.
5667 (build_type_attribute_variant): Likewise.
5668 (lookup_ident_attribute): Likewise.
5669 (remove_attribute): Likewise.
5670 (merge_attributes): Likewise.
5671 (merge_type_attributes): Likewise.
5672 (merge_decl_attributes): Likewise.
5673 (merge_dllimport_decl_attributes): Likewise.
5674 (handle_dll_attribute): Likewise.
5675 (attribute_list_equal): Likewise.
5676 (attribute_list_contained): Likewise.
5677 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5678 (lookup_attribute_by_prefix): Likewise.
5679 * bb-reorder.c: Include header files.
5680 * builtins.c: Likewise.
5681 * calls.c: Likewise.
5682 * cfgexpand.c: Likewise.
5683 * cgraph.c: Likewise.
5684 * cgraphunit.c: Likewise.
5685 * convert.c: Likewise.
5686 * dwarf2out.c: Likewise.
5687 * final.c: Likewise.
5688 * fold-const.c: Likewise.
5689 * function.c: Likewise.
5690 * gimple-expr.c: Likewise.
5691 * gimple-fold.c: Likewise.
5692 * gimple-pretty-print.c: Likewise.
5693 * gimple.c: Likewise.
5694 * gimplify.c: Likewise.
5695 * hsa-common.c: Likewise.
5696 * hsa-gen.c: Likewise.
5697 * internal-fn.c: Likewise.
5698 * ipa-chkp.c: Likewise.
5699 * ipa-cp.c: Likewise.
5700 * ipa-devirt.c: Likewise.
5701 * ipa-fnsummary.c: Likewise.
5702 * ipa-inline.c: Likewise.
5703 * ipa-visibility.c: Likewise.
5705 * lto-cgraph.c: Likewise.
5706 * omp-expand.c: Likewise.
5707 * omp-general.c: Likewise.
5708 * omp-low.c: Likewise.
5709 * omp-offload.c: Likewise.
5710 * omp-simd-clone.c: Likewise.
5711 * opts-global.c: Likewise.
5712 * passes.c: Likewise.
5713 * predict.c: Likewise.
5714 * sancov.c: Likewise.
5715 * sanopt.c: Likewise.
5716 * symtab.c: Likewise.
5717 * toplev.c: Likewise.
5718 * trans-mem.c: Likewise.
5719 * tree-chkp.c: Likewise.
5720 * tree-eh.c: Likewise.
5721 * tree-into-ssa.c: Likewise.
5722 * tree-object-size.c: Likewise.
5723 * tree-parloops.c: Likewise.
5724 * tree-profile.c: Likewise.
5725 * tree-ssa-ccp.c: Likewise.
5726 * tree-ssa-live.c: Likewise.
5727 * tree-ssa-loop.c: Likewise.
5728 * tree-ssa-sccvn.c: Likewise.
5729 * tree-ssa-structalias.c: Likewise.
5730 * tree-ssa.c: Likewise.
5731 * tree-streamer-in.c: Likewise.
5732 * tree-vectorizer.c: Likewise.
5733 * tree-vrp.c: Likewise.
5735 * ubsan.c: Likewise.
5736 * varasm.c: Likewise.
5737 * varpool.c: Likewise.
5738 * tree.c: Remove functions moved to attribs.[ch].
5740 * config/aarch64/aarch64.c: Add attrs.h header file.
5741 * config/alpha/alpha.c: Likewise.
5742 * config/arc/arc.c: Likewise.
5743 * config/arm/arm.c: Likewise.
5744 * config/avr/avr.c: Likewise.
5745 * config/bfin/bfin.c: Likewise.
5746 * config/c6x/c6x.c: Likewise.
5747 * config/cr16/cr16.c: Likewise.
5748 * config/cris/cris.c: Likewise.
5749 * config/darwin.c: Likewise.
5750 * config/epiphany/epiphany.c: Likewise.
5751 * config/fr30/fr30.c: Likewise.
5752 * config/frv/frv.c: Likewise.
5753 * config/ft32/ft32.c: Likewise.
5754 * config/h8300/h8300.c: Likewise.
5755 * config/i386/winnt.c: Likewise.
5756 * config/ia64/ia64.c: Likewise.
5757 * config/iq2000/iq2000.c: Likewise.
5758 * config/lm32/lm32.c: Likewise.
5759 * config/m32c/m32c.c: Likewise.
5760 * config/m32r/m32r.c: Likewise.
5761 * config/m68k/m68k.c: Likewise.
5762 * config/mcore/mcore.c: Likewise.
5763 * config/microblaze/microblaze.c: Likewise.
5764 * config/mips/mips.c: Likewise.
5765 * config/mmix/mmix.c: Likewise.
5766 * config/mn10300/mn10300.c: Likewise.
5767 * config/moxie/moxie.c: Likewise.
5768 * config/msp430/msp430.c: Likewise.
5769 * config/nds32/nds32-isr.c: Likewise.
5770 * config/nds32/nds32.c: Likewise.
5771 * config/nios2/nios2.c: Likewise.
5772 * config/nvptx/nvptx.c: Likewise.
5773 * config/pa/pa.c: Likewise.
5774 * config/pdp11/pdp11.c: Likewise.
5775 * config/powerpcspe/powerpcspe.c: Likewise.
5776 * config/riscv/riscv.c: Likewise.
5777 * config/rl78/rl78.c: Likewise.
5778 * config/rx/rx.c: Likewise.
5779 * config/s390/s390.c: Likewise.
5780 * config/sh/sh.c: Likewise.
5781 * config/sol2.c: Likewise.
5782 * config/sparc/sparc.c: Likewise.
5783 * config/spu/spu.c: Likewise.
5784 * config/stormy16/stormy16.c: Likewise.
5785 * config/tilegx/tilegx.c: Likewise.
5786 * config/tilepro/tilepro.c: Likewise.
5787 * config/v850/v850.c: Likewise.
5788 * config/vax/vax.c: Likewise.
5789 * config/visium/visium.c: Likewise.
5790 * config/xtensa/xtensa.c: Likewise.
5792 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5795 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5796 constraints since the -mupper-regs-* switches have been
5798 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5799 into a vector from a double word element that was extracted from
5800 another vector, and eliminate extra XXPERMDI instructions.
5801 (vsx_concat_<mode>_2): Likewise.
5802 (vsx_concat_<mode>_3): Likewise.
5803 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5804 concat to allow optimizing inserts from previous extracts.
5806 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5808 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5809 memory reference to a SSP offset in TLS address space.
5810 (ix86_print_operand) <case '@'>: Remove.
5811 (ix86_print_operand_punct_valid_p): Remove '@' code.
5812 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5814 (stack_tls_protect_set_<mode>): Remove.
5815 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5816 (stack_tls_protect_test_<mode>): Remove.
5817 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5819 2017-08-07 Olivier Hainque <hainque@adacore.com>
5822 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5824 2017-08-07 Douglas Rupp <rupp@adacore.com>
5826 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5827 variable was referenced as multidir in command.
5829 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5832 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5835 2017-08-07 Martin Liska <mliska@suse.cz>
5837 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5838 * config/rl78/rl78.c: Add include of attribs.h.
5839 * config/sh/sh.c: Likewise.
5840 * config/v850/v850.c: Likewise.
5842 2017-08-07 Tom de Vries <tom@codesourcery.com>
5845 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5847 2017-08-07 Martin Liska <mliska@suse.cz>
5849 * config/mips/mips.c: Include attribs.h.
5851 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5854 * doc/invoke.texi: Document change in behvaior for -Ofast for
5857 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5859 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5861 (aarch64_pop_regs): Likewise.
5862 (aarch64_gen_load_pair): Likewise.
5863 (aarch64_save_callee_saves): Likewise.
5864 (aarch64_restore_callee_saves): Likewise.
5866 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5868 * config/i386/i386.c: Revert the last change.
5870 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5873 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5875 (ix86_finalize_stack_frame_flags): This. Also clear
5876 frame_pointer_needed if -fno-omit-frame-pointer is used without
5878 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5879 with ix86_finalize_stack_frame_flags.
5880 (ix86_expand_epilogue): Likewise.
5881 (ix86_expand_split_stack_prologue): Likewise.
5883 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5886 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5887 to P_AES for Westmere.
5889 2017-08-07 Jonathan Yong <10walls@gmail.com>
5891 * config/i386/mingw.opt (fset-stack-executable): Removed.
5892 * config/i386/cygming.opt (fset-stack-executable): Moved
5894 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5896 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5898 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5900 2017-08-07 Marek Polacek <polacek@redhat.com>
5903 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5905 2017-08-07 Martin Liska <mliska@suse.cz>
5907 * attribs.h (canonicalize_attr_name): New function.
5908 (cmp_attribs): Move from c-format.c and adjusted.
5909 (is_attribute_p): Moved from tree.h.
5910 * tree-inline.c: Add new includes.
5911 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5912 (private_is_attribute_p): Remove.
5913 (private_lookup_attribute): Likewise.
5914 (private_lookup_attribute_by_prefix): Simplify.
5915 (remove_attribute): Use is_attribute_p.
5916 * tree.h: Remove removed declarations.
5918 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5921 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5922 instead of computing it in the function. Formatting fix.
5923 (expand_case): Don't rely on default_edge being the first edge,
5924 clear it if removing it, pass default_edge to
5925 emit_case_dispatch_table.
5926 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5929 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5931 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5932 insn in the function, emit NOP after the insn.
5934 2017-08-06 Tom de Vries <tom@codesourcery.com>
5936 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5939 2017-08-06 Tom de Vries <tom@codesourcery.com>
5941 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5944 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5946 PR tree-optimization/57371
5947 * match.pd: New pattern.
5949 2017-08-04 Marek Polacek <polacek@redhat.com>
5952 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5953 perform the computation in offset_int.
5955 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5957 PR tree-optimization/81136
5958 * tree-vectorizer.h: Include tree-hash-traits.h.
5959 (vec_base_alignments): New typedef.
5960 (vec_info): Add a base_alignments field.
5961 (vect_record_base_alignments): Declare.
5962 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5964 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5965 (create_data_ref): Add an is_conditional_in_stmt argument.
5966 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5967 the is_conditional_in_stmt field.
5968 (data_ref_loc): Add an is_conditional_in_stmt field.
5969 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5970 (find_data_references_in_stmt): Update call to create_data_ref.
5971 (graphite_find_data_references_in_stmt): Likewise.
5972 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5973 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5974 (vect_record_base_alignment): New function.
5975 (vect_record_base_alignments): Likewise.
5976 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5977 for nested statements even if we fail to compute a misalignment.
5978 Use pooled base alignments for unconditional references.
5979 (vect_find_same_alignment_drs): Compare base addresses instead
5981 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5982 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5984 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5986 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5987 Add an explicit name for the enum. Use auto_vec for slp_instances
5989 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5991 (_bb_vec_info): Add a constructor and destructor.
5992 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5993 (destroy_loop_vec_info): Delete.
5994 (vect_destroy_datarefs): Likewise.
5995 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5996 (vec_info::vec_info): New function.
5997 (vec_info::~vec_info): Likewise.
5998 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5999 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6000 destroy_loop_vec_info.
6001 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6002 (_loop_vec_info::_loop_vec_info): ...this.
6003 (destroy_loop_vec_info): Replace with...
6004 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6005 the stmt_vec_infos. Leave handling of vec_info information to its
6006 destructor. Remove explicit vector releases.
6007 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6008 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6009 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6010 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6011 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6012 (destroy_bb_vec_info): Replace with...
6013 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6014 information to its destructor.
6015 (vect_slp_analyze_bb_1): Use new and delete instead of
6016 new_bb_vec_info and destroy_bb_vec_info.
6017 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6020 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6022 * tree-data-ref.h (subscript): Add access_fn field.
6023 (data_dependence_relation): Add could_be_independent_p.
6024 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6025 (same_access_functions): Move to tree-data-ref.c.
6026 * tree-data-ref.c (ref_contains_union_access_p): New function.
6027 (access_fn_component_p): Likewise.
6028 (access_fn_components_comparable_p): Likewise.
6029 (dr_analyze_indices): Add a reference to access_fn_component_p.
6030 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6032 (constant_access_functions): Likewise.
6033 (add_other_self_distances): Likewise.
6034 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6035 (initialize_data_dependence_relation): Use XCNEW and remove
6036 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6037 of access functions that have the same type. Allow the
6038 subsequence to end with different bases in some circumstances.
6039 Record the chosen access functions in SUB_ACCESS_FN.
6040 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6041 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6042 (subscript_dependence_tester_1): Likewise dra and drb.
6043 (build_classic_dist_vector): Update calls accordingly.
6044 (subscript_dependence_tester): Likewise.
6045 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6046 DDR_COULD_BE_INDEPENDENT_P.
6047 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6048 comp_alias_ddrs instead of may_alias_ddrs.
6049 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6051 (vect_analyze_data_ref_dependence): Use it if
6052 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6053 distance vectors if that fails.
6054 (dependence_distance_ge_vf): New function.
6055 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6056 LOOP_VINFO_MAY_ALIAS_DDRS.
6058 2017-08-04 Richard Biener <rguenther@suse.de>
6061 * fold-const.c (fold_binary_loc): Properly restrict
6062 minus_var0 && minus_var1 case when associating undefined overflow
6065 2017-08-04 Tom de Vries <tom@codesourcery.com>
6067 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6069 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6071 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6072 Don't start diagnostic messages with a capital letter.
6073 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6075 (rs6000_invalid_builtin): Likewise.
6076 (rs6000_trampoline_init): Likewise.
6078 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6081 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6082 after setting changeable df flags.
6084 2017-08-03 Richard Biener <rguenther@suse.de>
6086 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6087 up if the use is in USE - X.
6089 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6091 * toplev.c (dumpfile.h): New include.
6092 (internal_error_reentered): New static function. Use it...
6093 (internal_error_function): ...here to handle reentered internal_error.
6095 2017-08-03 Richard Biener <rguenther@suse.de>
6098 * fold-const.c (split_tree): Add minus_var and minus_con
6099 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6100 here but always use minus_*.
6101 (associate_trees): Assert we never associate with MINUS_EXPR
6102 and NULL first operand. Do not recurse for PLUS_EXPR operands
6103 when associating as MINUS_EXPR either.
6104 (fold_binary_loc): Track minus_var and minus_con.
6106 2017-08-03 Tom de Vries <tom@codesourcery.com>
6109 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6110 ACCEL_COMPILER, apply finish_options on
6111 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6113 2017-08-03 Tom de Vries <tom@codesourcery.com>
6116 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6117 function_entry_patch_area_size > 0.
6119 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6122 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6123 instead of 10??LU, perform unit multiplication in wide_int,
6124 don't change alloc_object_size_limit if the limit is larger
6127 PR tree-optimization/81655
6128 PR tree-optimization/81588
6129 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6130 the case when ranges[i].low and high are 1 for unsigned type with
6134 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6135 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6137 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6139 * tree-vrp.h: Add include guard.
6141 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6144 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6145 (ud2): New insn pattern.
6146 * config/i386/i386.c (ix86_expand_epilogue):
6147 For naked functions, generate ud2 instead of trap insn.
6149 2017-08-02 Marek Polacek <polacek@redhat.com>
6152 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6154 2017-08-02 Tom de Vries <tom@codesourcery.com>
6155 Cesar Philippidis <cesar@codesourcery.com>
6157 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6158 Add missing edge probabilities.
6160 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6162 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6165 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6168 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6169 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6170 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6171 prologue_seq sequences - if any.
6173 2017-08-02 Richard Biener <rguenther@suse.de>
6175 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6176 via vectors if supported, integer extracts via punning if supported
6177 or otherwise vector extracts.
6179 2017-08-02 Richard Biener <rguenther@suse.de>
6181 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6183 (bitmap_insert_into_set): ... this.
6185 2017-08-02 Richard Biener <rguenther@suse.de>
6187 PR tree-optimization/81633
6189 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6191 PR tree-optimization/71752
6192 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6194 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6196 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6197 (machine_function::call_ms2sysv_pad_out): Remove field.
6198 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6199 (ix86_compute_frame_layout): Likewise.
6201 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6204 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6205 attribute with interrupt attribute.
6207 2017-08-01 Andrew Pinski <apinski@cavium.com>
6209 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6210 BIT_INSERT_EXPR's operand 1
6211 to see if the types precision matches.
6213 2017-08-01 Martin Liska <mliska@suse.cz>
6216 * builtins.c (expand_builtin_memcpy_args): Remove.
6217 (expand_builtin_memcpy): Call newly added function
6218 expand_builtin_memory_copy_args.
6219 (expand_builtin_memcpy_with_bounds): Likewise.
6220 (expand_builtin_mempcpy): Remove last argument.
6221 (expand_builtin_mempcpy_with_bounds): Likewise.
6222 (expand_builtin_memory_copy_args): New function created from
6223 expand_builtin_mempcpy_args with small modifications.
6224 (expand_builtin_mempcpy_args): Remove.
6225 (expand_builtin_stpcpy): Remove unused argument.
6226 (expand_builtin): Likewise.
6227 (expand_builtin_with_bounds): Likewise.
6229 2017-08-01 Martin Liska <mliska@suse.cz>
6232 Make mempcpy more optimal (PR middle-end/70140).
6234 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6237 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6238 __builtin_vec_cmpne verify both arguments are compatible vectors
6239 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6240 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6241 move computation of aligned to after checking the argument types.
6245 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6246 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6249 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6250 Doug Rupp <rupp@adacore.com>
6251 Olivier Hainque <hainque@adacore.com>
6253 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6254 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6256 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6257 for TARGET_OS_CPP_BUILTIN.
6258 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6259 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6261 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6262 passing required abi options to the assembler for EABI configurations.
6263 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6264 of .text.hot and .text.unlikely sections for kernel modules when
6265 using ARM style exceptions.
6266 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6267 options. Add EXTRA_CC1_SPEC.
6268 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6270 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6272 (ARM_TARGET2_DWARF_FORMAT): Define.
6273 * config/arm/t-vxworks: Adjust multilib control to removal of the
6274 Diab command line options.
6276 2017-08-01 Martin Liska <mliska@suse.cz>
6278 PR gcov-profile/81561
6279 * gcov.c (unblock): Make unblocking safe as we need to preserve
6280 index correspondence of blocks and block_lists.
6282 2017-08-01 Richard Biener <rguenther@suse.de>
6284 PR tree-optimization/81181
6285 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6286 (compute_antic): ... end of iteration here.
6288 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6290 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6291 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6292 (ftree-slp-vectorize): Likewise.
6293 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6294 can no longer be set independent of flag_tree_loop_vectorize.
6295 * omp-general.c (emp_max_vf): Likewise.
6296 * opts.c (enable_fdo_optimizations): Remove references to
6297 flag_tree_vectorize, these are now implicit.
6298 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6299 and leave it for the options machinery.
6301 2017-08-01 Martin Liska <mliska@suse.cz>
6304 * builtins.c (expand_builtin_memcpy_args): Remove.
6305 (expand_builtin_memcpy): Call newly added function
6306 expand_builtin_memory_copy_args.
6307 (expand_builtin_memcpy_with_bounds): Likewise.
6308 (expand_builtin_mempcpy): Remove last argument.
6309 (expand_builtin_mempcpy_with_bounds): Likewise.
6310 (expand_builtin_memory_copy_args): New function created from
6311 expand_builtin_mempcpy_args with small modifications.
6312 (expand_builtin_mempcpy_args): Remove.
6313 (expand_builtin_stpcpy): Remove unused argument.
6314 (expand_builtin): Likewise.
6315 (expand_builtin_with_bounds): Likewise.
6317 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6320 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6321 print "ds:" only for immediates in generic address space.
6323 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6326 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6327 (ix86_function_ok_for_sibcall): Return false for naked functions.
6329 2017-08-01 Richard Biener <rguenther@suse.de>
6331 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6332 (compute_antic): Seed worklist with exit block predecessors.
6333 * cfganal.c (dfs_find_deadend): For a cycle return the source
6334 of the edge closing it.
6336 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6338 * config/aarch64/aarch64.c
6339 (aarch64_can_const_movi_rtx_p): Move 0 check.
6341 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6343 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6344 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6346 * match.pd: Ditto in address comparison pattern.
6348 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6350 PR tree-optimization/81627
6351 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6352 closed ssa form for store-store chain.
6354 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6356 PR tree-optimization/81620
6357 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6358 for store-store chain.
6360 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6362 PR tree-optimization/81588
6363 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6364 ranges[i].in_p, invert comparison code ccode. For >/>=,
6365 swap rhs1 and rhs2 and comparison code unconditionally,
6366 for </<= don't do that. Don't swap rhs1/rhs2 again if
6367 ranges[i].in_p, instead invert comparison code ccode if
6368 opcode or oe->rank is BIT_IOR_EXPR.
6371 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6372 a direct optab to conversion optab.
6373 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6374 with GET_MODE_INNER as last argument instead of optab_handler.
6375 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6376 vector extraction if possible and optab is available.
6377 * expr.c (store_constructor): Use convert_optab_handler instead
6378 of optab_handler. Use vector initialization from smaller
6379 vectors if possible and optab is available.
6380 * tree-vect-stmts.c (vectorizable_load): Likewise.
6381 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6383 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6384 of vec_init from half-sized vectors with the same element mode.
6385 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6386 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6387 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6388 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6389 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6390 after mode in gen_vec_extract* calls.
6391 (vec_extract<mode>): Renamed to ...
6392 (vec_extract<mode><ssescalarmodelower>): ... this.
6393 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6394 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6395 element mode after mode in gen_vec_init* calls.
6396 (VEC_INIT_HALF_MODE): New mode iterator.
6397 (vec_init<mode>): Renamed to ...
6398 (vec_init<mode><ssescalarmodelower>): ... this.
6399 (vec_init<mode><ssehalfvecmodelower>): New expander.
6400 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6401 (vec_extractv2sfsf): ... this.
6402 (vec_initv2sf): Renamed to ...
6403 (vec_initv2sfsf): ... this.
6404 (vec_extractv2si): Renamed to ...
6405 (vec_extractv2sisi): ... this.
6406 (vec_initv2si): Renamed to ...
6407 (vec_initv2sisi): ... this.
6408 (vec_extractv4hi): Renamed to ...
6409 (vec_extractv4hihi): ... this.
6410 (vec_initv4hi): Renamed to ...
6411 (vec_initv4hihi): ... this.
6412 (vec_extractv8qi): Renamed to ...
6413 (vec_extractv8qiqi): ... this.
6414 (vec_initv8qi): Renamed to ...
6415 (vec_initv8qiqi): ... this.
6416 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6417 (vec_init<mode>): Renamed to ...
6418 (vec_init<mode><VEC_base_l>): ... this.
6419 (vec_extract<mode>): Renamed to ...
6420 (vec_extract<mode><VEC_base_l>): ... this.
6421 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6422 (vec_initv2sfsf): ... this.
6423 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6424 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6425 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6426 element mode after mode in gen_vec_init* calls.
6427 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6428 (vec_init<mode><Vel>): ... this.
6429 (vec_extract<mode>): Renamed to ...
6430 (vec_extract<mode><Vel>): ... this.
6431 * config/aarch64/iterators.md (Vel): New mode attribute.
6432 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6433 Add element mode after mode in gen_vec_extract* calls.
6434 * config/s390/vector.md (non_vec_l): New mode attribute.
6435 (vec_extract<mode>): Renamed to ...
6436 (vec_extract<mode><non_vec_l>): ... this.
6437 (vec_init<mode>): Renamed to ...
6438 (vec_init<mode><non_vec_l>): ... this.
6439 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6440 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6442 * config/arm/iterators.md (V_elem_l): New mode attribute.
6443 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6444 (vec_extract<mode><V_elem_l>): ... this.
6445 (vec_extractv2di): Renamed to ...
6446 (vec_extractv2didi): ... this.
6447 (vec_init<mode>): Renamed to ...
6448 (vec_init<mode><V_elem_l>): ... this.
6449 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6450 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6451 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6452 Add element mode after gen_vec_extract* calls.
6453 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6454 (vec_init<mode><unitmode>): ... this.
6455 (vec_extract<mode>): Renamed to ...
6456 (vec_extract<mode><unitmode>): ... this.
6457 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6458 (vec_init<mode><unitmode>): ... this.
6459 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6460 (vec_initv2sfsf): ... this.
6461 (vec_extractv2sf): Renamed to ...
6462 (vec_extractv2sfsf): ... this.
6463 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6464 Add element mode after gen_vec_extract* calls.
6465 * config/mips/mips.md (unitmode): New mode iterator.
6466 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6467 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6468 * config/spu/spu.md (inner_l): New mode attribute.
6469 (vec_init<mode>): Renamed to ...
6470 (vec_init<mode><inner_l>): ... this.
6471 (vec_extract<mode>): Renamed to ...
6472 (vec_extract<mode><inner_l>): ... this.
6473 * config/sparc/sparc.md (veltmode): New mode iterator.
6474 (vec_init<VMALL:mode>): Renamed to ...
6475 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6476 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6477 (vec_initv2sisi): ... this.
6478 (vec_initv2sf): Renamed to ...
6479 (vec_initv2sfsf): ... this.
6480 (vec_extractv2sf): Renamed to ...
6481 (vec_extractv2sfsf): ... this.
6482 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6483 (vec_init<mode>): Renamed to ...
6484 (vec_init<mode><VEC_base_l>): ... this.
6485 (vec_extract<mode>): Renamed to ...
6486 (vec_extract<mode><VEC_base_l>): ... this.
6487 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6488 (vec_initv2sfsf): ... this.
6489 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6490 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6491 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6492 gen_vec_init* calls.
6494 2017-08-01 Richard Biener <rguenther@suse.de>
6496 PR tree-optimization/81297
6497 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6498 TREE_OVERFLOW from INTEGER_CSTs.
6500 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6502 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6504 2017-07-31 Carl Love <cel@us.ibm.com>
6506 * config/rs6000/rs6000-c: Add support for built-in functions
6507 vector signed char vec_xl_be (signed long long, signed char *);
6508 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6509 vector signed int vec_xl_be (signed long long, signed int *);
6510 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6511 vector signed long long vec_xl_be (signed long long, signed long long *);
6512 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6513 vector signed short vec_xl_be (signed long long, signed short *);
6514 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6515 vector double vec_xl_be (signed long long, double *);
6516 vector float vec_xl_be (signed long long, float *);
6517 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6518 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6519 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6521 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6522 (altivec_expand_builtin): Add switch statement to call
6523 altivec_expand_xl_be for each builtin.
6524 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6525 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6526 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6527 __builtin_vsx_le_be_v16qi.
6528 * doc/extend.texi: Update the built-in documentation file for the
6529 new built-in functions.
6531 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6534 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6536 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6538 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6540 * config.gcc: Add z14.
6541 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6542 CPU model numbers for z13s and z14.
6543 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6545 * config/s390/s390-opts.h (enum processor_type): Rename
6546 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6547 * config/s390/s390.c (processor_table): Add field for CPU name to
6548 be passed to Binutils.
6549 (s390_asm_output_machine_for_arch): Use the new field in
6550 processor_table for Binutils.
6551 (s390_expand_builtin): Replace arch12 with z14.
6552 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6553 (s390_get_sched_attrmask): Likewise.
6554 (s390_get_unit_mask): Likewise.
6555 * config/s390/s390.opt: Add z14 to processor_type enum.
6557 2017-07-31 Martin Jambor <mjambor@suse.cz>
6560 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6561 regardless of optimization level.
6563 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6564 Martin Liska <mliska@suse.cz>
6566 * predict.def: Remove old comment and adjust probability.
6567 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6570 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6573 * config/i386/i386.c (ix86_function_naked): New function.
6574 (ix86_can_use_return_insn_p): Return false for naked functions.
6575 (ix86_expand_prologue): Skip prologue for naked functions.
6576 (ix86_expand_epilogue): Skip epilogue for naked functions
6577 and emit trap instruction.
6578 (ix86_warn_func_return): New function.
6579 (ix86_attribute_table): Add "naked" attribute specification.
6580 (TARGET_WARN_FUNC_RETURN): Define.
6581 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6583 2017-07-31 Martin Liska <mliska@suse.cz>
6585 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6586 (dump_gimple_bb_header): Always dump BB info.
6587 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6589 2017-07-31 Martin Liska <mliska@suse.cz>
6592 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6593 also with current_function_decl non-null equality.
6595 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6598 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6599 change type to the element type, instead add eltype variable and
6600 use it where we are interested in the element type.
6602 PR tree-optimization/81603
6603 * ipa-polymorphic-call.c
6604 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6605 offset arithmetic in offset_int, bail out if the resulting bit offset
6606 doesn't fit into shwi.
6608 2017-07-31 Martin Liska <mliska@suse.cz>
6610 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6611 (gimplify_save_expr): Fix comment.
6613 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6616 * config/i386/i386.c (ix86_function_arg): Update arguments for
6618 (ix86_compute_frame_layout): Set the initial stack offset to
6619 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6620 INCOMING_FRAME_SP_OFFSET.
6621 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6622 stack before exception handler returns.
6623 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6624 the 'ERROR_CODE' for exception handler.
6626 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6628 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6629 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6630 (ASM_OUTPUT_REG_POP): Ditto.
6631 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6632 instead of asm_fprintf to output pure string.
6634 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6636 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6637 to imported_module_or_decl hook.
6638 (debug_nothing_tree_tree_tree_bool): Remove.
6639 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6640 * debug.c (do_nothing_debug_hooks): Use
6641 debug_nothing_tree_tree_tree_bool_bool instead of
6642 debug_nothing_tree_tree_tree_bool.
6643 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6644 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6645 * sdbout.c (sdb_debug_hooks): Likewise.
6646 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6647 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6649 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6650 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6651 attribute, don't add anything.
6653 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6655 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6656 (fold_build2_stat_loc): Likewise.
6657 (fold_build3_stat_loc): Likewise.
6658 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6659 (fold_build1_loc): Remove macro.
6660 (fold_build2_loc): Likewise.
6661 (fold_build3_loc): Likewise.
6663 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6665 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6666 (gimple_build_debug_bind_source_stat): Likewise.
6667 * gimple.h (gimple_build_debug_bind): Remove macro.
6668 (gimple_build_debug_bind_source): Likewise.
6670 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6672 * bitmap.c (bitmap_alloc): Adjust.
6673 (bitmap_gc_alloc): Likewise.
6674 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6676 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6678 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6679 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6680 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6681 (bitmap_gc_alloc_stat): Likewise.
6682 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6686 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6687 * rtl.h (shallow_copy_rtx): Remove macro.
6689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6691 * emit-rtl.c (gen_raw_REG): Adjust.
6692 * gengenrtl.c (gendef): Likewise.
6693 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6694 * rtl.h (rtx_alloc): Remove macro.
6696 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6698 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6699 (build_tree_list_stat): Likewise.
6700 * tree.h (build_tree_list): Remove macro.
6701 (build_tree_list_vec): Likewise.
6703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6705 * tree.c (make_vector_stat): Remove _stat from name.
6706 (build_vector_stat): Likewise.
6707 * tree.h (make_vector_stat): Remove macro.
6708 (build_vector_stat): Likewise.
6710 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6712 * tree.h (build_var_debug_value): Remove prototype.
6714 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6716 * tree.c (tree_cons_stat): Remove _stat from name.
6717 * tree.h (tree_cons): Remove macro.
6719 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6721 * tree.c (build_vl_exp_stat): Remove _stat from name.
6722 * tree.h (build_vl_exp): Remove macro.
6724 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6726 * tree.c (build_decl_stat): Remove _stat from name.
6727 * tree.h (build_decl): Remove macro.
6729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6731 * gimple.c (gimple_build_with_ops_stat): Adjust.
6732 (gimple_alloc_stat): Remove _stat from name.
6733 * gimple.h (gimple_alloc): Remove macro.
6735 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6737 * tree.c (make_tree_vec_stat): Remove _stat from name.
6738 (grow_tree_vec_stat): Likewise.
6739 * tree.h (make_tree_vec_stat): Adjust prototype.
6740 (grow_tree_vec_stat): Likewise.
6741 (make_tree_vec): Remove macro.
6742 (grow_tree_vec): Likewise.
6744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6746 * fold-const.c (fold_build1_stat_loc): Adjust.
6747 (fold_build2_stat_loc): Likewise.
6748 (fold_build3_stat_loc): Likewise.
6749 * tree.c (build0_stat): Remove _stat from name.
6750 (build1_stat): Likewise.
6751 (build2_stat): Likewise.
6752 (build3_stat): Likewise.
6753 (build4_stat): Likewise.
6754 (build5_stat): Likewise.
6755 * tree.h (build1_loc): Remove macro, and rename _stat function
6757 (build2_loc): Likewise.
6758 (build3_loc): Likewise.
6759 (build4_loc): Likewise.
6760 (build5_loc): Likewise.
6762 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6764 * tree.c (make_int_cst_stat): Remove _stat from name.
6765 * tree.h (make_int_cst_stat): Adjust prototype.
6766 (make_int_cst): Remove macro.
6768 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6770 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6771 * tree.h (make_tree_binfo_stat): Adjust prototype.
6772 (make_tree_binfo): Remove.
6774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6776 * tree.c (copy_node_stat): Rename to copy_node.
6777 (build_distinct_type_copy): Adjust.
6778 * tree.h (copy_node_stat): Adjust prototype.
6779 (copy_node): Remove macro.
6781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6783 * tree.c (make_node_stat): rename to make_node.
6784 (build_tree_list_stat): Adjust.
6785 (build0_stat): Likewise.
6786 (build2_stat): Likewise.
6787 (build3_stat): Likewise.
6788 (build4_stat): Likewise.
6789 (build5_stat): Likewise.
6790 (build_decl_stat): Likewise.
6791 * tree.h (make_node_stat): Adjust prototype.
6792 (make_node): remove macro.
6794 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6796 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6797 (PPC_FEATURE2_SCV): Likewise.
6798 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6800 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6802 * config/aarch64/aarch64.c
6803 (aarch64_internal_mov_immediate): Add new special pattern.
6804 * config/aarch64/aarch64.md (*movdi_aarch64):
6805 Add reg/32bit const mov case.
6807 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6808 Richard Sandiford <richard.sandiford@linaro.org>
6810 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6811 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6812 Add integer and movi cases.
6813 (movi-split-hf-df-sf split, fp16): New.
6814 (enabled): Added TARGET_FP_F16INST.
6815 * config/aarch64/iterators.md (GPF_HF): New.
6816 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6818 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6820 * config/aarch64/aarch64.c
6821 (aarch64_simd_container_mode): Add prototype.
6822 (aarch64_expand_mov_immediate): Add HI support.
6823 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6824 (aarch64_can_const_movi_rtx_p): New.
6825 (aarch64_preferred_reload_class):
6826 Remove restrictions of using FP registers for certain SIMD operations.
6827 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6828 (aarch64_valid_floating_const): Add integer move validation.
6829 (aarch64_simd_imm_scalar_p): Remove.
6830 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6831 (aarch64_legitimate_constant_p): Expand list of supported cases.
6832 * config/aarch64/aarch64-protos.h
6833 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6834 (aarch64_reinterpret_float_as_int): New.
6835 (aarch64_simd_imm_scalar_p): Remove.
6836 * config/aarch64/constraints.md (Uvi): New.
6837 (Dd): Split into Ds and new Dd.
6838 * config/aarch64/aarch64.md (*movsi_aarch64):
6840 (*movdi_aarch64): Add SIMD mov case.
6842 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6844 * tree-predcom.c: (struct chain): Handle store-store chain in which
6845 stores for elimination only store loop invariant values.
6846 (execute_pred_commoning_chain): Ditto.
6847 (prepare_initializers_chain_store_elim): Ditto.
6848 (prepare_finalizers): Ditto.
6849 (is_inv_store_elimination_chain): New function.
6850 (initialize_root_vars_store_elim_1): New function.
6852 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6854 * tree-predcom.c: Revise general description of the pass.
6855 (enum chain_type): New enum type for store elimination.
6856 (struct chain): New field supporting store elimination.
6857 (struct component): Ditto.
6858 (dump_chain): Dump store-stores chain.
6859 (release_chain): Release resources.
6860 (split_data_refs_to_components): Compute and create component
6861 contains only stores for elimination.
6862 (get_chain_last_ref_at): New function.
6863 (make_invariant_chain): Initialization.
6864 (make_rooted_chain): Specify chain type in parameter and record it.
6865 (add_looparound_copies): Skip for store-stores chain.
6866 (determine_roots_comp): Compute type of chain and pass it to
6868 (initialize_root_vars_store_elim_2): New function.
6869 (finalize_eliminated_stores): New function.
6870 (remove_stmt): Handle store for elimination.
6871 (execute_pred_commoning_chain): Execute predictive commoning on
6873 (determine_unroll_factor): Skip unroll for store-stores chain.
6874 (prepare_initializers_chain_store_elim): New function.
6875 (prepare_initializers_chain): Hanlde store-store chain.
6876 (prepare_finalizers_chain, prepare_finalizers): New function.
6877 (tree_predictive_commoning_loop): Return integer value indicating
6878 if loop is unrolled or lcssa form is corrupted.
6879 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6881 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6883 * tree-predcom.c (initialize_root): Delete.
6884 (execute_pred_commoning_chain): Initialize root vars and replace
6885 reference of non-combined chain directly, rather than call above
6888 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6890 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6891 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6893 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6895 * tree-predcom.c (struct chain): New field init_seq.
6896 (release_chain): Release init_seq.
6897 (prepare_initializers_chain): Record intialization stmts in above
6899 (insert_init_seqs): New function.
6900 (tree_predictive_commoning_loop): Call insert_init_seqs.
6902 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6904 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6906 2017-07-28 Richard Biener <rguenther@suse.de>
6908 * match.pd: Remove superfluous :c.
6909 * genmatch.c (simplify::id): Add member.
6910 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6912 (current_id): New global.
6913 (dt_node::parent): Move from ...
6914 (dt_operand::parent): ... here. Add for_id member.
6915 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6916 (decision_tree::find_node): Relax order requirement when
6917 merging DT_TRUE nodes to ones inbetween the current simplify
6918 and the one we try to merge with. Add diagnostic whenever
6919 we need to enforce pattern order by not merging.
6920 (decision_tree::insert): Set current_id.
6921 (decision_tree::print_node): Dump parent node and for_id.
6922 (parser::last_id): Add member.
6923 (parser::push_simplify): Assign unique id.
6924 (parser::parser): Initialize last_id.
6926 2017-07-28 Martin Liska <mliska@suse.cz>
6929 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6930 gimple_build_debug_bind.
6932 2017-07-28 Richard Biener <rguenther@suse.de>
6934 PR tree-optimization/81502
6935 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6937 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6939 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6940 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6941 for BIT_FIELD_REF args.
6942 * fold-const.c (make_bit_field_ref): Likewise.
6943 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6945 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6948 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6949 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6950 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6951 Or it into SANITIZER_UNDEFINED.
6952 * ubsan.c: Include gimple-fold.h and varasm.h.
6953 (ubsan_expand_ptr_ifn): New function.
6954 (instrument_pointer_overflow): New function.
6955 (maybe_instrument_pointer_overflow): New function.
6956 (instrument_object_size): Formatting fix.
6957 (pass_ubsan::execute): Call instrument_pointer_overflow
6958 and maybe_instrument_pointer_overflow.
6959 * internal-fn.c (expand_UBSAN_PTR): New function.
6960 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6961 * sanitizer.def (__ubsan_handle_pointer_overflow,
6962 __ubsan_handle_pointer_overflow_abort): New builtins.
6963 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6964 * internal-fn.def (UBSAN_PTR): New internal function.
6965 * opts.c (sanitizer_opts): Add pointer-overflow.
6966 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6967 * fold-const.c (build_range_check): Compute pointer range check in
6968 integral type if pointer arithmetics would be needed. Formatting
6971 2017-07-28 Martin Liska <mliska@suse.cz>
6974 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6975 parameters that are of a variable-length.
6977 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6979 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6981 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6982 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6983 (CRT_CALL_STATIC_FUNCTION): Likewise.
6984 (ASM_DEFAULT_SPEC): New define.
6985 (ASM_SPEC32): Likewise.
6986 (ASM_SPEC64): Likewise.
6987 (ASM_SPEC_COMMON): Likewise.
6988 (ASM_SPEC): Likewise.
6989 (INVALID_64BIT): Likewise.
6990 (LINK_OS_DEFAULT_SPEC): Likewise.
6991 (LINK_OS_SPEC32): Likewise.
6992 (LINK_OS_SPEC64): Likewise.
6993 (POWERPC_LINUX): Likewise.
6994 (PTRDIFF_TYPE): Likewise.
6995 (RESTORE_FP_PREFIX): Likewise.
6996 (RESTORE_FP_SUFFIX): Likewise.
6997 (SAVE_FP_PREFIX): Likewise.
6998 (SAVE_FP_SUFFIX): Likewise.
6999 (SIZE_TYPE): Likewise.
7000 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7001 (TARGET_64BIT): Likewise.
7002 (TARGET_64BIT): Likewise.
7003 (TARGET_AIX): Likewise.
7004 (WCHAR_TYPE_SIZE): Likewise.
7005 (WCHAR_TYPE): Undefine.
7006 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7007 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7008 (CPP_OS_RTEMS_SPEC): Delete.
7009 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7010 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7012 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7014 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7016 PR tree-optimization/81578
7017 * tree-parloops.c (build_new_reduction): Bail out if
7018 reduction_code isn't one of the standard OpenMP reductions.
7019 Move the details printing after that decision.
7021 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7023 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7024 related to reload_in_progress.
7025 (splat_input_operand): Likewise.
7026 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7028 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7030 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7031 (TARGET_INSTANTIATE_DECLS): Likewise.
7032 (legitimate_indexed_address_p): Delete reload_in_progress code.
7033 (rs6000_debug_legitimate_address_p): Likewise.
7034 (rs6000_eliminate_indexed_memrefs): Likewise.
7035 (rs6000_emit_le_vsx_store): Likewise.
7036 (rs6000_emit_move_si_sf_subreg): Likewise.
7037 (rs6000_emit_move): Likewise.
7038 (register_to_reg_type): Likewise.
7039 (rs6000_pre_atomic_barrier): Likewise.
7040 (rs6000_machopic_legitimize_pic_address): Likewise.
7041 (rs6000_allocate_stack_temp): Likewise.
7042 (rs6000_address_for_fpconvert): Likewise.
7043 (rs6000_address_for_altivec): Likewise.
7044 (rs6000_secondary_memory_needed_rtx): Delete function.
7045 (rs6000_check_sdmode): Likewise.
7046 (rs6000_alloc_sdmode_stack_slot): Likewise.
7047 (rs6000_instantiate_decls): Likewise.
7048 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7049 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7050 Delete reload_in_progress.
7051 (*vec_reload_and_plus_<mptrsize>): Likewise.
7052 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7053 (vsx_div_v2di): Likewise.
7054 (vsx_udiv_v2di): Likewise.
7056 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7058 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7059 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7060 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7061 (rs6000_debug_reg_global): Delete print of LRA status.
7062 (rs6000_option_override_internal): Delete dead LRA related code.
7063 (rs6000_lra_p): Delete function.
7064 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7066 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7068 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7069 * config/riscv/rtems.h: New file.
7071 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7072 Sudakshina Das <sudi.das@arm.com>
7074 * config/aarch64/aarch64.md
7075 (define_split for and<mode>3nr_compare): Move
7076 non aarch64_logical_operand to a register.
7077 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7078 register immediate operand to a register.
7079 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7081 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7084 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7086 2017-07-27 Richard Biener <rguenther@suse.de>
7088 PR tree-optimization/81573
7089 PR tree-optimization/81494
7090 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7091 multi defuse cycle case.
7093 2017-07-27 Richard Biener <rguenther@suse.de>
7095 PR tree-optimization/81571
7096 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7099 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7101 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7102 earlier and only if MASK_FPU is set. Adjust formatting.
7104 2017-07-27 Martin Liska <mliska@suse.cz>
7106 * opt-functions.awk: Add validation of value of Init.
7107 * optc-gen.awk: Pass new argument.
7109 2017-07-27 Martin Liska <mliska@suse.cz>
7111 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7112 Fix wrong condition.
7114 2017-07-27 Martin Liska <mliska@suse.cz>
7116 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7117 BBs and edges seen by autoFDO.
7119 2017-07-27 Richard Biener <rguenther@suse.de>
7121 PR tree-optimization/81502
7122 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7123 with incompatible but same sized type.
7124 (execute_update_addresses_taken): Likewise.
7126 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7128 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7129 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7131 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7134 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7135 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7136 Change s_operand to memory_operand.
7138 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7140 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7141 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7142 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7143 Emit instructions rather than returning an expression. Handle TFmode
7144 and KFmode by casting to TImode.
7145 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7146 (rs6000_emit_le_vsx_store): Likewise.
7147 * config/rs6000/vsx.md (VSX_TI): New iterator.
7148 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7149 (*vsx_le_undo_permute_<mode>): Likewise.
7150 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7151 emit the split sequence.
7152 (*vsx_le_perm_store_<mode>): Likewise.
7154 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7156 PR tree-optimization/81555
7157 PR tree-optimization/81556
7158 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7159 if true, force CHANGED for the recursive invocation.
7160 (reassociate_bb): Remember original length of ops array, pass
7161 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7163 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7164 attributes for noipa attribute. For naked attribute use
7165 lookup_attribute first before lookup_attribute_spec.
7166 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7168 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7169 for functions with noipa attribute.
7170 (cgraph_externally_visible_p): Return true for functions with noipa
7172 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7173 for functions with noipa attribute.
7174 * doc/extend.texi: Document noipa function attribute.
7175 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7176 also for functions with noipa attribute.
7177 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7179 2017-07-26 Andrew Pinski <apinski@cavium.com>
7181 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7182 vec_unalign_load_cost and vec_unalign_store_cost.
7184 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7186 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7187 -mvsx-small-integer option.
7188 (ISA_3_0_MASKS_IEEE): Likewise.
7189 (OTHER_VSX_VECTOR_MASKS): Likewise.
7190 (POWERPC_MASKS): Likewise.
7191 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7192 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7193 code, only testing for DImode being allowed in non-VSX floating
7195 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7196 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7198 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7199 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7200 TARGET_P8_VECTOR test.
7201 (rs6000_secondary_reload_simple_move): Likewise.
7202 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7203 since TARGET_P9_VECTOR was already tested.
7204 (rs6000_opt_masks): Remove -mvsx-small-integer.
7205 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7206 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7208 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7209 test for TARGET_VEXTRACTUB was used, and that uses
7211 (p9 extract splitter): Likewise.
7212 (vsx_extract_<mode>_di_p9): Likewise.
7213 (vsx_extract_<mode>_store_p9): Likewise.
7214 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7215 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7216 the elimination of TARGET_VSX_SMALL_INTEGER.
7217 (vsx_extract_<mode>_p8): Likewise.
7218 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7219 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7220 (vsx_set_<mode>_p9): Likewise.
7221 (vsx_set_v4sf_p9): Likewise.
7222 (vsx_set_v4sf_p9_zero): Likewise.
7223 (vsx_insert_extract_v4sf_p9): Likewise.
7224 (vsx_insert_extract_v4sf_p9_2): Likewise.
7225 * config/rs6000/rs6000.md (sign extend splitter): Change
7226 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7227 (floatsi<mode>2_lfiwax_mem): Likewise.
7228 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7229 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7230 since a test for TARGET_P9_VECTOR was used.
7231 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7232 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7233 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7234 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7235 TARGET_P8_VECTOR test.
7236 (fix_trunc<mode>si2_stfiwx): Likewise.
7237 (fix_trunc<mode>si2_internal): Likewise.
7238 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7239 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7241 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7242 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7243 TARGET_P8_VECTOR test.
7244 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7245 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7246 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7248 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7249 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7250 since a test for TARGET_P9_VECTOR was used.
7251 (splitter for loading small constants): Likewise.
7253 2017-07-26 Andrew Pinski <apinski@cavium.com>
7255 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7258 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7261 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7262 (fp_valid_at): Likewise.
7264 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7266 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7267 (qdf24xx_addrcost_table): Likewise.
7268 (cortexa57_tunings): Update to use generic_branch_cost.
7269 (cortexa72_tunings): Likewise.
7270 (cortexa73_tunings): Likewise.
7271 (qdf24xx_tunings): Likewise.
7273 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7275 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7276 (thunderx2t99_branch_cost): Likewise.
7277 (cortexa35_tunings): Update to use generic_branch_cost.
7278 (cortexa53_tunings): Likewise.
7279 (cortexa57_tunings): Likewise.
7280 (cortexa72_tunings): Likewise.
7281 (cortexa73_tunings): Likewise.
7282 (thunderx2t99_tunings): Likewise.
7284 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7286 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7287 (sparc_option_override): Honour MASK_FSMULD.
7288 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7289 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7290 * config/sparc/sparc.opt (mfsmuld): New option.
7291 * doc/invoke.texi (mfsmuld): Document option.
7293 2017-07-26 Marek Polacek <polacek@redhat.com>
7296 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7298 2017-07-26 Richard Biener <rguenther@suse.de>
7300 * gimple-match-head.c (do_valueize): Return OP if valueize
7302 (get_def): New helper to get at the def stmt of a SSA name
7304 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7305 do_valueize to get at the def stmt.
7306 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7308 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7311 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7313 2017-07-26 Martin Liska <mliska@suse.cz>
7316 * function.c (expand_function_start): Make expansion of
7317 nonlocal_goto_save_area after parm_birth_insn.
7319 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7321 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7322 from all CPU target flags enable members.
7324 2017-07-26 Richard Biener <rguenther@suse.de>
7326 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7327 (decision_tree::gen): Make 'type' const.
7328 (write_predicate): Likewise.
7330 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7332 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7333 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7334 (rs6000_option_override_internal): Likewise.
7335 (rs6000_expand_vector_set): Likewise.
7336 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7337 (TARGET_UPPER_REGS_SF): Likewise.
7338 (TARGET_UPPER_REGS_DI): Likewise.
7339 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7340 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7341 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7342 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7343 (Splitters for DI constants in Altivec registers): Likewise.
7344 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7345 (vsx_set_v4sf_p9): Likewise.
7346 (vsx_set_v4sf_p9_zero): Likewise.
7347 (vsx_insert_extract_v4sf_p9): Likewise.
7348 (vsx_insert_extract_v4sf_p9_2): Likewise.
7350 2017-07-25 Carl Love <cel@us.ibm.com>
7352 * doc/extend.texi: Update the built-in documentation file for the
7353 existing built-in functions
7354 vector signed char vec_cnttz (vector signed char);
7355 vector unsigned char vec_cnttz (vector unsigned char);
7356 vector signed short vec_cnttz (vector signed short);
7357 vector unsigned short vec_cnttz (vector unsigned short);
7358 vector signed int vec_cnttz (vector signed int);
7359 vector unsigned int vec_cnttz (vector unsigned int);
7360 vector signed long long vec_cnttz (vector signed long long);
7361 vector unsigned long long vec_cnttz (vector unsigned long long);
7363 2017-07-25 Andrew Pinski <apinski@cavium.com>
7365 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7366 accesses where the use is for the first operand of a BIT_INSERT.
7368 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7371 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7372 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7374 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7376 * config/i386/gstabs.h: Delete.
7377 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7379 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7381 * config/i386/i386.c (ix86_decompose_address): Do not check for
7382 register RTX when looking at index_reg or base_reg.
7383 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7385 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7387 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7388 to update EH info here.
7390 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7392 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7394 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7396 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7398 2017-07-25 Torsten Duwe <duwe@suse.de>
7400 * common.opt: Introduce -fpatchable-function-entry
7401 command line option, and its variables function_entry_patch_area_size
7402 and function_entry_patch_area_start.
7403 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7404 including a two-value parser.
7405 * target.def (print_patchable_function_entry): New target hook.
7406 * targhooks.h (default_print_patchable_function_entry): New function.
7407 * targhooks.c (default_print_patchable_function_entry): Likewise.
7408 * toplev.c (process_options): Switch off IPA-RA if
7409 patchable function entries are being generated.
7410 * varasm.c (assemble_start_function): Look at the
7411 patchable-function-entry command line switch and current
7412 function attributes and maybe generate NOP instructions by
7413 calling the print_patchable_function_entry hook.
7414 * doc/extend.texi: Document patchable_function_entry attribute.
7415 * doc/invoke.texi: Document -fpatchable_function_entry
7416 command line option.
7417 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7419 * doc/tm.texi: Re-generate.
7421 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7424 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7425 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7427 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7429 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7431 2017-07-25 Richard Biener <rguenther@suse.de>
7433 PR tree-optimization/81455
7434 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7435 not walk in cycles when looking for guards.
7437 2017-07-25 Richard Biener <rguenther@suse.de>
7439 PR tree-optimization/81529
7440 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7441 when optimizing backedge uses.
7443 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7445 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7447 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7448 to dl_section_ref. On AIX, append an expression to subtract
7449 the size of the section length to dl_section_ref.
7451 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7453 * configure.ac: If any of the config.* scripts fail, exit 1.
7454 * configure: Regenerate.
7456 2017-07-25 Richard Biener <rguenther@suse.de>
7459 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7460 of immediate uses, be more verbose on errors.
7462 2017-07-25 Richard Biener <rguenther@suse.de>
7464 PR tree-optimization/81510
7465 * tree-vect-loop.c (vect_is_simple_reduction): When the
7466 reduction stmt is not inside the loop bail out.
7468 2017-07-25 Richard Biener <rguenther@suse.de>
7470 PR tree-optimization/81303
7471 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7472 profitability check against LOOP_VINFO_NITERSM1.
7474 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7476 * domwalk.c (cmp_bb_postorder): Simplify.
7477 (sort_bbs_postorder): New function. Use it...
7478 (dom_walker::walk): ...here to optimize common cases.
7480 2017-07-25 Martin Liska <mliska@suse.cz>
7483 * ipa-visibility.c (function_and_variable_visibility): Make the
7484 redirection just on target that supports aliasing.
7485 Fix GNU coding style.
7487 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7490 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7492 * config/arm/rtems.h: Likewise.
7493 * config/bfin/rtems.h: Likewise.
7494 * config/i386/rtemself.h: Likewise.
7495 * config/lm32/rtems.h: Likewise.
7496 * config/m32c/rtems.h: Likewise.
7497 * config/m68k/rtemself.h: Likewise.
7498 * config/microblaze/rtems.h: Likewise.
7499 * config/mips/rtems.h: Likewise.
7500 * config/moxie/rtems.h: Likewise.
7501 * config/nios2/rtems.h: Likewise.
7502 * config/powerpcspe/rtems.h: Likewise.
7503 * config/rs6000/rtems.h: Likewise.
7504 * config/rtems.h: Likewise.
7505 * config/sh/rtems.h: Likewise.
7506 * config/sh/rtemself.h: Likewise.
7507 * config/sparc/rtemself.h: Likewise.
7509 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7512 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7513 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7514 * tree-ssa-structalias.c (alias_get_name): Same.
7516 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7519 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7520 instructions if no du chain is found.
7522 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7524 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7526 2017-07-25 Richard Biener <rguenther@suse.de>
7529 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7532 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7534 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7536 (ISA_2_7_MASKS_SERVER): Likewise.
7537 (ISA_3_0_MASKS_IEEE): Likewise.
7538 (OTHER_P8_VECTOR_MASKS): Likewise.
7539 (OTHER_VSX_VECTOR_MASKS): Likewise.
7540 (POWERPC_MASKS): Likewise.
7541 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7542 duplicate list of options.
7543 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7544 explicit -mupper-regs options.
7545 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7546 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7547 alias for -mupper-regs-df.
7548 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7549 (rs6000_init_hard_regno_mode_ok): Likewise.
7550 (rs6000_option_override_internal): Likewise.
7551 (rs6000_opt_masks): Likewise.
7552 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7553 options in terms of whether -mvsx or -mpower8-vector was used.
7554 (TARGET_UPPER_REGS_DI): Likewise.
7555 (TARGET_UPPER_REGS_SF): Likewise.
7556 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7557 -mupper-regs-* options.
7559 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7561 * passes.c (emergency_dump_function): Print some empty lines and a
7562 header before the RTL dump.
7564 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7566 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7568 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7571 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7572 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7574 2017-07-24 Carl Love <cel@us.ibm.com>
7576 * config/rs6000/rs6000-c.c: Add support for built-in functions
7577 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7578 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7579 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7580 vec_extract_fp_from_shortl): Add defines for the two builtins.
7581 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7582 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7584 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7585 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7586 * doc/extend.texi: Update the built-in documentation file for the
7587 new built-in function.
7589 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7592 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7594 * doc/generic.texi: Likewise.
7595 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7596 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7598 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7600 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7601 (aarch64_mls_elt_merge<mode>): Likewise.
7603 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7605 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7606 having __cxa_atexit.
7608 2017-07-23 Michael Collison <michael.collison@arm.com>
7610 * config/arm/arm.c (arm_option_override): Deprecate
7611 use of -mstructure-size-boundary.
7612 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7613 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7615 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7618 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7619 Reduce cost estimate for direct moves.
7621 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7624 * config/i386/i386.c (ix86_option_override_internal): Disable
7625 BMI, BMI2 and TBM instructions for -m16.
7627 2017-07-21 Carl Love <cel@us.ibm.com>
7629 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7630 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7631 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7632 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7633 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7634 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7635 VMULOSW): New enum "unspec" values.
7636 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7637 altivec_vmulosw): New patterns.
7638 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7639 VMULOSW): Add definitions.
7641 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7643 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7644 (qdf24xx): Likewise.
7645 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7646 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7647 (AARCH64_FL_V8_1): Renumber.
7648 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7649 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7650 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7651 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7652 rdma to feature modifiers list.
7654 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7657 * ipa-visibility (function_and_variable_visibility): Convert
7658 recursive PLT call to direct call if appropriate.
7660 2017-07-21 Andrew Pinski <apinski@cavium.com>
7662 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7663 operand 1 to see if the types precision matches.
7664 * fold-const.c (operand_equal_p): Likewise.
7666 2017-07-21 Richard Biener <rguenther@suse.de>
7668 PR tree-optimization/81303
7669 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7670 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7671 (vect_peeling_hash_get_lowest_cost): Adjust.
7672 (vect_enhance_data_refs_alignment): Likewise. Use
7673 vect_get_peeling_costs_all_drs to compute the penalty for no
7674 peeling to match up costs.
7676 2017-07-21 Richard Biener <rguenther@suse.de>
7678 PR tree-optimization/81500
7679 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7680 we didn't identify a reduction path.
7682 2017-07-21 Tom de Vries <tom@codesourcery.com>
7683 Cesar Philippidis <cesar@codesourcery.com>
7685 PR gcov-profile/81442
7686 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7689 2017-07-21 Tom de Vries <tom@codesourcery.com>
7692 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7694 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7695 nvptx_override_options_after_change.
7697 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7699 * dwarf2out.c (output_file_names): Avoid double testing for
7702 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7704 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7706 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7708 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7710 (try_crossjump_to_edge): Do not punt on partitioned functions.
7712 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7714 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7715 Put all BBs reachable only via paths crossing cold region to cold
7717 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7719 2016-07-21 Richard Biener <rguenther@suse.de>
7721 PR tree-optimization/81303
7722 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7723 into account prologue and epilogue iterations when raising
7724 min_profitable_iters to sth at least covering one vector iteration.
7726 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7728 * config/arm/arm.c (arm_test_cpu_arch_dat):
7731 2017-07-20 Nathan Sidwell <nathan@acm.org>
7733 Remove TYPE_METHODS.
7734 * tree.h (TYPE_METHODS): Delete.
7735 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7736 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7737 (dbxout_type_methods): Scan TYPE_FIELDS.
7738 (dbxout_type): Don't check TYPE_METHODS here.
7739 * function.c (use_register_for_decl): Always ignore register for
7740 class types when not optimizing.
7741 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7742 * tree.c (free_lang_data_in_type): Stitch out member functions and
7743 templates from TYPE_FIELDS.
7744 (build_distinct_type_copy, verify_type_variant,
7745 verify_type): Member fns are on TYPE_FIELDS.
7746 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7747 * tree-pretty-print.c (dump_generic_node): Likewise.
7749 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7752 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7753 V2TImode and V4TImode.
7754 (ix86_expand_vector_extract): Likewise.
7755 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7756 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7757 (ssescalarmode): Handle V4TImode and V2TImode.
7758 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7759 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7760 (VEXTRACTI128_MODE): New mode iterator.
7761 (splitter for *vec_extractv?ti first element): New.
7762 (VEC_INIT_MODE): New mode iterator.
7763 (vec_init<mode>): Consolidate 3 expanders into one using
7764 VEC_INIT_MODE mode iterator.
7766 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7768 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7769 non_spilled_static_chain_regno_p.
7771 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7773 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7775 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7777 * bb-reorder.c (connect_traces): Allow copying of blocks within
7780 2017-07-20 Richard Biener <rguenther@suse.de>
7782 * gimple.h (gimple_phi_result): Add gphi * overload.
7783 (gimple_phi_result_ptr): Likewise.
7784 (gimple_phi_arg): Likewise. Adjust index assert to only
7785 allow actual argument accesses rather than all slots available
7787 (gimple_phi_arg_def): Add gphi * overload.
7788 * tree-phinodes.c (make_phi_node): Initialize only actual
7790 (resize_phi_node): Clear memory not covered by old node,
7791 do not initialize excess argument slots.
7792 (reserve_phi_args_for_new_edge): Initialize new argument slot
7795 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7797 PR tree-optimization/81388
7799 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7801 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7802 by removing computation of may_be_zero.
7804 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7805 Tom de Vries <tom@codesourcery.com>
7808 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7809 when gimple level profile disagrees with what RTL expander did.
7811 2017-07-20 Richard Biener <rguenther@suse.de>
7813 PR tree-optimization/61171
7814 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7815 (vect_analyze_stmt): Add slp instance parameter.
7816 (vectorizable_reduction): Likewise.
7817 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7818 (vect_is_simple_reduction): Deal with chains not detected
7819 as SLP reduction chain, specifically not properly associated
7820 chains containing a mix of plus/minus.
7821 (get_reduction_op): Remove.
7822 (get_initial_defs_for_reduction): Simplify, pass in whether
7823 this is a reduction chain, pass in the SLP node for the PHIs.
7824 (vect_create_epilog_for_reduction): Get the SLP instance as
7826 (vectorizable_reduction): Get the SLP instance as arg.
7827 During analysis remember the SLP node with the PHIs in the
7828 instance. Simplify getting at the vectorized reduction PHIs.
7829 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7830 through SLP instance.
7831 (vect_slp_analyze_operations): Likewise.
7832 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7833 (vect_transform_stmt): Likewise.
7835 2017-07-20 Tom de Vries <tom@codesourcery.com>
7837 PR tree-optimization/81489
7838 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7839 read of phi arg location to before loop that modifies phi.
7841 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7843 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7846 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7849 * except.c (execute): Fix ordering issue.
7851 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7853 PR rtl-optimization/81423
7854 * combine.c (make_compound_operation_int): Don't try to optimize
7855 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7857 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7859 PR rtl-optimization/81423
7860 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7861 with a constant that is -1 in the truncated to mode.
7863 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7865 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7866 (determine_unlikely_bbs): ... here.
7867 * predict.h (propagate_unlikely_bbs_forward): Declare.
7868 * cfgexpand.c (pass_expand::execute): Use it.
7869 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7871 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7872 propagate_unlikely_bbs_forward.
7874 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7877 * except.c (maybe_add_nop_after_section_switch): New function.
7880 2017-07-19 Tom de Vries <tom@codesourcery.com>
7882 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7884 2017-07-19 Tom de Vries <tom@codesourcery.com>
7886 * gimple.h (gimple_phi_arg): Make assert more strict.
7888 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7890 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7891 * config/rs6000/mmintrin.h: New file.
7892 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7894 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7896 PR tree-optimization/81346
7897 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7899 2017-07-19 Tom de Vries <tom@codesourcery.com>
7901 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7903 2017-07-19 Tom de Vries <tom@codesourcery.com>
7905 * config/nvptx/nvptx-modes.def: Add V2DImode.
7906 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7907 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7908 (nvptx_output_mov_insn): Handle lack of mov.b128.
7909 (nvptx_print_operand): Handle 'H' and 'L' codes.
7910 (nvptx_vector_mode_supported): Allow V2DImode.
7911 (nvptx_preferred_simd_mode): New function.
7912 (nvptx_data_alignment): New function.
7913 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7914 nvptx_preferred_simd_mode.
7915 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7917 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7919 2017-07-19 Tom de Vries <tom@codesourcery.com>
7921 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7922 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7923 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7924 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7925 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7926 (mov<VECIM>_insn): New define_insn.
7927 (define_expand "mov<VECIM>): New define_expand.
7929 2017-07-19 Tom de Vries <tom@codesourcery.com>
7931 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7933 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7935 PR tree-optimization/81346
7936 * fold-const.h (fold_div_compare, range_check_type): Declare.
7937 * fold-const.c (range_check_type): New function.
7938 (build_range_check): Use range_check_type.
7939 (fold_div_compare): No longer static, rewritten into
7940 a match.pd helper function.
7941 (fold_comparison): Don't call fold_div_compare here.
7942 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7945 2017-07-19 Nathan Sidwell <nathan@acm.org>
7947 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7948 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7949 * tree.c (find_decls_types_r, verify_type): Use
7950 TYPE_{MIN,MAX}_VALUE_RAW.
7951 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7952 (hash_tree): Likewise.
7953 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7955 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7958 2017-07-18 Tom de Vries <tom@codesourcery.com>
7961 * omp-expand.c (expand_omp_for_static_chunk): Handle
7962 equal-argument loop exit phi.
7964 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7967 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7968 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7969 operand 2 predicate.
7970 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7971 operand 2 predicate.
7972 (ror,rol -> rorx splitters): Use const_int_operand as
7973 operand 2 predicate.
7975 2017-06-18 Richard Biener <rguenther@suse.de>
7977 PR tree-optimization/81410
7978 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7979 the gap in the ! slp_perm SLP case after each group.
7981 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7984 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7987 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7990 * predict.c (set_even_probabilities): Cleanup; do not affect
7991 probabilities that are already known.
7992 (combine_predictions_for_bb): Call even when count is set.
7994 2017-07-18 Nathan Sidwell <nathan@acm.org>
7996 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7999 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8002 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8003 optimization for loop niter analysis.
8005 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8008 * config/avr/avr.c (avr_optimize_casesi): Don't use
8009 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8011 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8013 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8014 body_cost_vec from _vect_peel_extended_info.
8015 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8016 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8019 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8021 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8023 2017-07-18 Richard Biener <rguenther@suse.de>
8025 PR tree-optimization/80620
8026 PR tree-optimization/81403
8027 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8028 info when re-using a VN table entry.
8030 2017-07-18 Richard Biener <rguenther@suse.de>
8032 PR tree-optimization/81418
8033 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8034 vectype_in. Verify that with lane-reducing reduction operations
8035 we have a single def-use cycle.
8037 2017-07-17 Carl Love <cel@us.ibm.com>
8039 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8041 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8042 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8043 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8044 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8045 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8046 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8047 VMULOSW): New enum "unspec" values.
8048 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8049 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8050 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8051 altivec_vmulosw): New patterns.
8052 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8053 VMULOSW): Add definitions.
8054 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8056 * config/alpha/alpha.c: Include predict.h.
8058 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8060 * tree-vrp.c (compare_assert_loc): Fix comparison function
8061 to return predictable results.
8063 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8065 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8068 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8069 * doc/invoke.texi (mexpand-adddi): Update text.
8071 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8073 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8074 that also clobbers the CC register. The old expand code is moved
8076 (*arc_clzsi2): ... here.
8077 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8078 the CC register. The old expand code is moved to ...
8079 (arc_ctzsi2): ... here.
8081 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8083 * config/arc/arc.opt (mindexed-loads): Use initial value
8084 TARGET_INDEXED_LOADS_DEFAULT.
8085 (mauto-modify-reg): Use initial value
8086 TARGET_AUTO_MODIFY_REG_DEFAULT.
8087 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8088 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8089 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8090 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8092 2017-07-17 Martin Liska <mliska@suse.cz>
8095 * opts.c (finish_options): Do not allow -fgnu-tm
8096 w/ -fsanitize={kernel-,}address. Say sorry.
8098 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8101 * tree-loop-distribution.c (classify_partition): Only assert on
8102 numer of iterations.
8103 (merge_dep_scc_partitions): Delete prameter. Update function call.
8104 (distribute_loop): Remove code handling loop with unknown niters.
8105 (pass_loop_distribution::execute): Skip loop with unknown niters.
8107 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8110 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8111 function sort_partitions_by_post_order.
8113 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8115 PR tree-optimization/81374
8116 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8117 the max index of basic blocks, rather than number of basic blocks.
8119 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8121 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8123 (arc_legitimate_pic_operand_p): Likewise.
8124 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8126 (arc_needs_pcl_p): Likewise.
8127 (arc_legitimate_pc_offset_p): Likewise.
8128 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8129 function is also used in constrains.md.
8130 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8131 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8132 PLUS. Only return true/false in known cases, otherwise assert.
8133 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8134 is already called in arc_legitimate_constant_p.
8135 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8137 (LEGITIMATE_PIC_OPERAND_P): Use
8138 arc_raw_symbolic_reference_mentioned_p function.
8139 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8144 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8145 Andrew Burgess <andrew.burgess@embecosm.com>
8147 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8148 (arc_return_address_register): New function.
8149 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8150 (arc_handle_fndecl_attribute): Add naked attribute.
8151 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8152 (TARGET_WARN_FUNC_RETURN): Likewise.
8153 (arc_allocate_stack_slots_for_args): New function.
8154 (arc_warn_func_return): Likewise.
8155 (machine_function): Change type fn_type.
8156 (arc_compute_function_type): Consider new naked function type,
8157 change function return type.
8158 (arc_must_save_register): Adapt to handle new
8159 arc_compute_function_type's return type.
8160 (arc_expand_prologue): Likewise.
8161 (arc_expand_epilogue): Likewise.
8162 (arc_return_address_regs): Delete.
8163 (arc_return_address_register): New function.
8164 (arc_epilogue_uses): Use above function.
8165 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8166 (arc_function_type): Change encoding, add naked type.
8167 (ARC_INTERRUPT_P): Change to handle the new encoding.
8168 (ARC_FAST_INTERRUPT_P): Likewise.
8169 (ARC_NORMAL_P): Define.
8170 (ARC_NAKED_P): Likewise.
8171 (arc_compute_function_type): Delete prototype.
8172 * config/arc/arc.md (in_ret_delay_slot): Use
8173 arc_return_address_register function.
8174 (simple_return): Likewise.
8175 (p_return_i): Likewise.
8177 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8179 PR tree-optimization/81428
8180 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8181 can't be built for those types.
8183 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8185 Remove stuff dead since r239246.
8187 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8188 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8189 (avr_inform_devices): Remove dead stuff.
8191 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8193 * config/arm/arm_neon.h: Fix softp typo.
8195 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8197 PR tree-optimization/81365
8198 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8199 aggregate moves onto bb predecessor edges, make sure there are no
8200 loads that could alias the lhs in between the start of bb and the
8203 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8206 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8207 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8208 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8210 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8212 PR tree-optimization/81396
8213 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8214 (init_symbolic_number): Initialize it to 1.
8215 (perform_symbolic_merge): Add n_ops from both operands into the new
8217 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8218 without base_addr as useless if they need more than one operation.
8219 (bswap_replace): Handle !bswap case for NULL base_addr.
8221 2017-07-17 Tom de Vries <tom@codesourcery.com>
8224 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8227 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8229 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8230 conditional builtin define __FIX_LEON3FT_B2BST.
8232 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8234 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8235 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8238 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8240 PR rtl-optimization/81424
8241 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8242 to remove potential trapping from operands if -fnon-call-exceptions.
8244 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8246 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8247 profile_proability for scalling.
8248 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8250 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8252 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8254 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8256 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8257 fixpoint arithmetics.
8259 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8261 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8262 fixpoint arithmetics.
8264 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8266 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8267 fixpoint arithmetics.
8269 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8271 * profile-count.h (profile_probability::from_reg_br_prob_note,
8272 profile_probability::to_reg_br_prob_note): New functions.
8273 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8274 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8275 * predict.c (probability_reliable_p): Update.
8276 (edge_probability_reliable_p): Update.
8277 (br_prob_note_reliable_p): Update.
8278 (invert_br_probabilities): Update.
8279 (add_reg_br_prob_note): New function.
8280 (combine_predictions_for_insn): Update.
8281 * asan.c (asan_clear_shadow): Update.
8282 * cfgbuild.c (compute_outgoing_frequencies): Update.
8283 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8284 (update_br_prob_note): Update.
8285 (rtl_verify_edges): Update.
8286 (purge_dead_edges): Update.
8287 (fixup_reorder_chain): Update.
8288 * emit-rtl.c (try_split): Update.
8289 * ifcvt.c (cond_exec_process_insns): Update.
8290 (cond_exec_process_if_block): Update.
8291 (dead_or_predicable): Update.
8292 * internal-fn.c (expand_addsub_overflow): Update.
8293 (expand_neg_overflow): Update.
8294 (expand_mul_overflow): Update.
8295 * loop-doloop.c (doloop_modify): Update.
8296 * loop-unroll.c (compare_and_jump_seq): Update.
8297 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8298 * predict.h: Update.
8299 * reorg.c (mostly_true_jump): Update.
8301 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8302 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8303 * config/arc/arc.c: (emit_unlikely_jump): Update.
8304 * config/arm/arm.c: (emit_unlikely_jump): Update.
8305 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8306 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8307 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8308 (ix86_print_operand): Update.
8309 (ix86_split_fp_branch): Update.
8310 (predict_jump): Update.
8311 * config/ia64/ia64.c (ia64_print_operand): Update.
8312 * config/mmix/mmix.c (mmix_print_operand): Update.
8313 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8314 (rs6000_expand_split_stack_prologue): Update.
8315 * config/rs6000/rs6000.c: Update.
8316 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8317 (s390_expand_vec_movstr): Update.
8318 (s390_expand_cs_tdsi): Update.
8319 (s390_expand_split_stack_prologue): Update.
8320 * config/sh/sh.c (sh_print_operand): Update.
8321 (expand_cbranchsi4): Update.
8322 (expand_cbranchdi4): Update.
8323 * config/sparc/sparc.c (output_v9branch): Update.
8324 * config/spu/spu.c (get_branch_target): Update.
8325 (ea_load_store_inline): Update.
8326 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8327 * config/tilepro/tilepro.c: Update.
8329 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8331 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8332 (gimplify_save_expr): Likewise.
8334 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8336 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8338 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8340 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8342 * timevar.def (TV_IPA_FNSUMMARY): Define.
8344 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8346 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8347 to back store errata sensitive sequence from being generated.
8348 (sqrtdf2_fix): Likewise.
8350 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8352 * tree-ssa-threadupdate.c (compute_path_counts,
8353 update_joiner_offpath_counts): Use profile_probability.
8355 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8358 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8360 * config/arm/arm-c.c (arm_cpu_builtins): Define
8361 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8363 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8365 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8366 array entries to represent __ieee128 versions of the
8367 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8368 scalar_extract_sig, and scalar_insert_exp built-in functions.
8369 (altivec_resolve_overloaded_builtin): Add special case handling
8370 for the __builtin_scalar_insert_exp function, as represented by
8371 the P9V_BUILTIN_VEC_VSIEDP constant.
8372 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8373 exponent support for __ieee128 argument.
8374 (VSESQP): Add scalar extract signature support for __ieee128
8376 (VSTDCNQP): Add scalar test negative support for __ieee128
8378 (VSIEQP): Add scalar insert exponent support for __int128 argument
8379 with __ieee128 result.
8380 (VSIEQPF): Add scalar insert exponent support for __ieee128
8381 argument with __ieee128 result.
8382 (VSTDCQP): Add scalar test data class support for __ieee128
8384 (VSTDCNQP): Add overload support for scalar test negative with
8386 (VSTDCQP): Add overload support for scalar test data class
8388 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8390 (UNSPEC_VSX_SIEXPQP): New constant.
8391 (xsxexpqp): New insn for VSX scalar extract exponent quad
8393 (xsxsigqp): New insn for VSX scalar extract significand quad
8395 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8396 precision with floating point argument.
8397 (xststdcqp): New expand for VSX scalar test data class quad
8399 (xststdcnegqp): New expand for VSX scalar test negative quad
8401 (xststdcqp): New insn to match expansions for VSX scalar test data
8402 class quad precision and VSX scalar test negative quad precision.
8403 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8404 special case operand checking to enforce that second operand of
8405 VSX scalar test data class with quad precision argument is a 7-bit
8407 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8408 prototypes and descriptions of __ieee128 versions of
8409 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8410 scalar_test_data_class, and scalar_test_neg built-in functions.
8412 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8414 PR tree-optimization/81162
8415 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8416 replace a negate with an add.
8418 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8420 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8422 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8424 * config/arm/arm-c.c (arm_cpu_builtins): Define
8425 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8427 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8429 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8430 (armv8-r): Set ARM Cortex-R52 as default CPU.
8431 * config/arm/arm-tables.opt: Regenerate.
8432 * config/arm/arm-tune.md: Regenerate.
8433 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8435 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8436 extension for -mcpu=cortex-r52.
8438 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8440 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8441 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8442 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8443 (fp-armv8): Define it as FP_ARMv8 only.
8444 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8445 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8447 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8448 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8449 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8450 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8451 than TARGET_FPU_ARMV8.
8452 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8453 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8454 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8456 * config/arm/neon.md (neon_vrint): Likewise.
8457 (neon_vcvt): Likewise.
8458 (neon_<fmaxmin_op><mode>): Likewise.
8459 (<fmaxmin><mode>3): Likewise.
8460 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8461 * config/arm/predicates.md (arm_cond_move_operator): Check against
8462 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8464 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8466 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8469 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8471 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8472 loop in comment with memset.
8474 2017-07-14 Martin Liska <mliska@suse.cz>
8476 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8477 * dwarf2out.c (is_java): Remove the function.
8478 (output_pubname): Remove usage of the function.
8479 (lower_bound_default): Remove usage of DW_LANG_Java.
8480 (gen_compile_unit_die): Likewise.
8481 * gcc.c: Remove compiler defaults for .java and .zip files.
8482 * gimple-expr.c (remove_suffix): Change as there's no longer
8483 extension than 4-letter one.
8484 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8485 (gimplify_save_expr): Likewise.
8486 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8487 as it's possible even for other languages than Java.
8488 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8489 * lto-opts.c (lto_write_options): Remove reference to Java.
8490 * opts.c (strip_off_ending): Update file extension handling.
8491 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8492 * tree-eh.c (lower_resx): Likewise.
8493 * tree.c (free_lang_data_in_type): Remove dead code.
8494 (find_decls_types_r): Likewise.
8495 (build_common_builtin_nodes): Remove Java from a comment.
8496 (verify_type): Remove dead code.
8497 * varasm.c (assemble_external): Remove Java from a comment.
8499 2017-07-14 Martin Liska <mliska@suse.cz>
8501 * opts.c (finish_options): Add quotes.
8502 (common_handle_option): Likewise.
8504 2017-07-14 Martin Liska <mliska@suse.cz>
8506 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8507 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8509 * dwarf2out.c (lower_bound_default): Do not handle
8511 (gen_compile_unit_die): Likewise.
8512 * gcc.c: Remove default extension binding for GNU Pascal.
8513 * stmt.c: Remove Pascal language from a comment.
8514 * xcoffout.c: Likewise.
8516 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8519 * diagnostic-show-locus.c (fixit_cmp): New function.
8520 (layout::layout): Sort m_fixit_hints.
8521 (column_range::column_range): Assert that the values are valid.
8522 (struct char_span): New struct.
8523 (correction::overwrite): New method.
8524 (struct source_line): New struct.
8525 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8526 calls in terms of classes source_line and char_span, and
8527 correction::overwrite.
8528 (selftest::test_overlapped_fixit_printing_2): New function.
8529 (selftest::diagnostic_show_locus_c_tests): Call it.
8531 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8534 early if there is no lhs.
8536 2017-07-13 Martin Liska <mliska@suse.cz>
8538 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8539 (gen_reference_type_die): Likewise.
8540 * stor-layout.c: Remove Pascal-related comment.
8542 2017-07-13 Martin Liska <mliska@suse.cz>
8544 * opts.c (finish_options): Add quotes to error messages.
8545 (parse_sanitizer_options): Likewise.
8547 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8549 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8551 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8553 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8555 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8557 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8559 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8561 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8564 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8565 provides the hardware capability bits, define the macro
8566 __BUILTIN_CPU_SUPPORTS__.
8567 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8568 if GLIBC does not provide the hardware capability bits. Add a
8569 gcc_unreachable call if the built-in cpu function is neither
8570 __builtin_cpu_is nor __builtin_cpu_supports.
8571 (rs6000_get_function_versions_dispatcher): Change the warning
8572 that an old GLIBC is used which does not export the capability
8573 bits to be an error.
8574 * doc/extend.texi (target_clones attribute): Document the
8575 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8576 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8577 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8578 the macros defined by GCC if the newer GLIBC is available.
8580 2017-07-12 Jeff Law <law@redhat.com>
8582 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8583 remaining includes slightly.
8584 * config/riscv/riscv-builtins.c: Include profile-count.h.
8586 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8589 * config/avr/avr.c (avr_set_current_function): In diagnostic
8590 messages: Quote keywords and (parts of) identifiers.
8591 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8594 2017-07-12 Carl Love <cel@us.ibm.com>
8596 * config/rs6000/rs6000-c.c: Add support for built-in functions
8597 vector bool char vec_revb (vector bool char);
8598 vector bool short vec_revb (vector short char);
8599 vector bool int vec_revb (vector bool int);
8600 vector bool long long vec_revb (vector bool long long);
8601 * doc/extend.texi: Update the built-in documentation file for the
8602 new built-in functions.
8604 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8606 * config/s390/s390.md: Remove movcc splitter.
8608 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8610 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8611 load/store on condition.
8613 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8616 * config/avr/avr.c (avr_encode_section_info)
8617 [progmem && !TREE_READONLY]: Error if progmem object needs
8620 2017-07-11 Michael Collison <michael.collison@arm.com>
8622 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8625 2017-07-11 Carl Love <cel@us.ibm.com>
8627 * config/rs6000/rs6000-c.c: Add support for builtins
8628 vector unsigned int vec_parity_lsbb (vector signed int);
8629 vector unsigned int vec_parity_lsbb (vector unsigned int);
8630 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8631 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8632 vector unsigned long long vec_parity_lsbb (vector signed long long);
8633 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8634 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8635 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8636 * doc/extend.texi: Update the built-in documentation file for the
8637 new built-in functions.
8639 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8641 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8642 (layout::m_primary_loc): New field.
8643 (layout::layout): Initialize new field. Move location filtering
8644 logic from here to...
8645 (layout::maybe_add_location_range): ...this new method. Add
8646 support for filtering to just the lines already specified by other
8648 (layout::will_show_line_p): New method.
8649 (gcc_rich_location::add_location_if_nearby): New method.
8650 (selftest::test_add_location_if_nearby): New test function.
8651 (selftest::diagnostic_show_locus_c_tests): Call it.
8652 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8655 2017-07-11 Tom de Vries <tom@codesourcery.com>
8657 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8658 (bb_first_real_insn): New function.
8659 (nvptx_single): Add extra initialization of broadcasted condition
8662 2017-07-11 Nathan Sidwell <nathan@acm.org>
8664 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8666 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8668 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8669 Binutils doc as TEXI will mess them up.
8670 * doc/invoke.texi (AVR Options): Same here.
8672 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8674 * config/sparc/sparc.opt (mfix-ut700): New option.
8675 (mfix-gr712rc): Likewise.
8676 (sparc_fix_b2bst): New variable.
8677 * doc/invoke.texi (SPARC options): Document them.
8678 (ARM options): Fix warnings.
8679 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8680 instructions to prevent sequences that can trigger the store-store
8681 errata for certain LEON3FT processors.
8682 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8683 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8684 * config/sparc/sparc.md (fix_b2bst): New attribute.
8685 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8687 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8690 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8692 (*rsqrtsf2_sse): Ditto.
8694 (div<mode>3): Macroize insn from divdf3 and divsf3
8695 using MODEF mode iterator.
8697 2017-07-10 Martin Sebor <msebor@redhat.com>
8699 PR tree-optimization/80397
8700 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8701 instead of testing for equality to INTEGER_TYPE.
8703 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8705 * config.gcc: Remove uclibc from arc target spec.
8707 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8709 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8711 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8714 * lto-wrapper.c (remove_option): New function.
8715 (merge_and_complain): Merge PIC/PIE options more realistically.
8717 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8719 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8723 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8724 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8725 * config.in: Regenerate.
8726 * configure: Regenerate.
8727 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8728 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8729 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8730 (TARGET_GASISR_PROLOGUES): ...target mask.
8731 * common/config/avr/avr-common.c
8732 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8733 Set -mgas-isr-prologues.
8734 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8735 INSERT_PASS_BEFORE for it.
8736 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8737 * config/avr/avr.c (avr_option_override)
8738 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8739 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8740 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8741 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8742 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8744 (make_avr_pass_pre_proep): New function.
8745 (emit_push_sfr) <treg>: Add argument to function and use it
8747 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8748 and set machine->gasisr.yes.
8749 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8750 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8751 __gcc_isr.n_pushed to .L__stack_usage.
8752 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8753 (avr_asm_final_postscan_insn): ...this new static function.
8754 * config/avr/avr.h (machine_function)
8755 <is_no_gccisr, use_L__stack_usage>: New fields.
8756 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8757 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8758 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8759 (gasisr, *gasisr): New expander and insn.
8760 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8761 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8762 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8764 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8766 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8769 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8771 Move jump-tables out of .text again.
8774 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8775 (ASM_OUTPUT_ADDR_VEC): New function.
8776 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8777 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8778 INSN_ADDRESSes as asm comment.
8779 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8780 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8781 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8782 * config/avr/avr.md (*tablejump): Adjust comment.
8783 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8784 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8786 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8787 (avr_output_addr_vec): New proto.
8788 (avr_log_t) <insn_addresses>: New field.
8790 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8793 * config/i386/i386.c (ix86_function_arg_advance): Set
8794 outgoing_args_on_stack to true if there are outgoing arguments
8796 (ix86_function_arg): Likewise.
8797 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8798 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8799 * config/i386/i386.h (machine_function): Add
8800 outgoing_args_on_stack.
8802 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8804 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8806 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8808 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8810 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8811 (REAL_H): Remove $(MACHMODE_H).
8812 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8814 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8815 $(MACHMODE_H) and double-int.h.
8816 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8818 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8821 2017-07-07 Andrew Pinski <apinski@cavium.com>
8823 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8824 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8826 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8828 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8829 Add warning if GCC was not configured to link against a GLIBC that
8830 exports the hardware capability bits.
8831 (make_resolver_func): Make resolver function private and not a
8832 COMDAT function. Create the name with clone_function_name instead
8833 of make_unique_name.
8836 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8837 correct operand in doing the split.
8839 2017-07-07 Carl Love <cel@us.ibm.com>
8841 * config/rs6000/rs6000-c: Add support for built-in function
8842 vector unsigned short vec_pack_to_short_fp32 (vector float,
8844 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8845 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8846 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8847 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8848 (convert_4f32_8i16): Add define_expand.
8849 * doc/extend.texi: Update the built-in documentation file for the
8850 new built-in function.
8852 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8854 * config/sparc/m8.md: New file.
8855 * config/sparc/sparc.md: Include m8.md.
8857 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8859 * config/sparc/sparc.opt: New option -mvis4b.
8860 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8861 (sparc_option_override): Handle VIS4B.
8862 (enum sparc_builtins): Define
8863 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8864 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8865 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8866 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8867 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8868 (check_constant_argument): New function.
8869 (sparc_vis_init_builtins): Define builtins
8870 __builtin_vis_dictunpack{8,16,32},
8871 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8872 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8873 __builtin_vis_fpcmpde{8,16,32}shl and
8874 __builtin_vis_fpcmpur{8,16,32}shl.
8875 (sparc_expand_builtin): Check that the constant operands to
8876 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8877 constant and in range.
8878 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8880 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8881 (SPARC_IMM5_P): Likewise.
8882 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8883 (enabled): Handle vis4b.
8884 (UNSPEC_DICTUNPACK): New unspec.
8885 (UNSPEC_FPCMPSHL): Likewise.
8886 (UNSPEC_FPUCMPSHL): Likewise.
8887 (UNSPEC_FPCMPDESHL): Likewise.
8888 (UNSPEC_FPCMPURSHL): Likewise.
8889 (cpu_feature): New CPU feature `vis4b'.
8890 (dictunpack{8,16,32}): New insns.
8891 (FPCSMODE): New mode iterator.
8892 (fpcscond): New code iterator.
8893 (fpcsucond): Likewise.
8894 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8895 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8896 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8897 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8898 * config/sparc/constraints.md: Define constraints `q' for unsigned
8899 2-bit integer constants and `t' for unsigned 5-bit integer
8901 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8903 (imm5_operand_dictunpack16): Likewise.
8904 (imm5_operand_dictunpack32): Likewise.
8905 (imm2_operand): Likewise.
8906 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8907 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8908 ditunpack* and fpcmp*shl builtins.
8910 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8912 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8913 * config.in: Add HAVE_AS_SPARC6 define.
8914 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8916 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8918 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8919 (CPP_CPU_SPEC): Handle m8.
8920 (ASM_CPU_SPEC): Likewise.
8921 * config/sparc/sparc-opts.h (enum processor_type): Add
8923 * config/sparc/sparc.c (m8_costs): New struct.
8924 (sparc_option_override): Handle TARGET_CPU_m8.
8925 (sparc32_initialize_trampoline): Likewise.
8926 (sparc64_initialize_trampoline): Likewise.
8927 (sparc_issue_rate): Likewise.
8928 (sparc_register_move_cost): Likewise.
8929 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8930 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8931 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8932 (CPP_CPU_SPEC): Handle M8.
8933 (ASM_CPU_SPEC): Likewise.
8934 (AS_M8_FLAG): Define.
8935 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8936 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8937 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8939 * configure: Regenerate.
8940 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8943 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8945 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8947 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8948 ("*movdi_insn_sp32"): Do not set v3pipe.
8949 ("*movsi_insn"): Likewise.
8950 ("*movdi_insn_sp64"): Likewise.
8951 ("*movsf_insn"): Likewise.
8952 ("*movdf_insn_sp32"): Likewise.
8953 ("*movdf_insn_sp64"): Likewise.
8954 ("*zero_extendsidi2_insn_sp64"): Likewise.
8955 ("*sign_extendsidi2_insn"): Likewise.
8956 ("*mov<VM32:mode>_insn"): Likewise.
8957 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8958 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8959 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8960 ("<vlop:code><VL:mode>3"): Likewise.
8961 ("*not_<vlop:code><VL:mode>3"): Likewise.
8962 ("*nand<VL:mode>_vis"): Likewise.
8963 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8964 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8965 ("one_cmpl<VL:mode>2"): Likewise.
8966 ("faligndata<VM64:mode>_vis"): Likewise.
8967 ("alignaddrsi_vis"): Likewise.
8968 ("alignaddrdi_vis"): Likweise.
8969 ("alignaddrlsi_vis"): Likewise.
8970 ("alignaddrldi_vis"): Likewise.
8971 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8972 ("bmaskdi_vis"): Likewise.
8973 ("bmasksi_vis"): Likewise.
8974 ("bshuffle<VM64:mode>_vis"): Likewise.
8975 ("cmask8<P:mode>_vis"): Likewise.
8976 ("cmask16<P:mode>_vis"): Likewise.
8977 ("cmask32<P:mode>_vis"): Likewise.
8978 ("pdistn<P:mode>_vis"): Likewise.
8979 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8981 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8983 * config/sparc/sparc.md ("subtype"): New insn attribute.
8984 ("*wrgsr_sp64"): Set insn subtype.
8985 ("*rdgsr_sp64"): Likewise.
8986 ("alignaddrsi_vis"): Likewise.
8987 ("alignaddrdi_vis"): Likewise.
8988 ("alignaddrlsi_vis"): Likewise.
8989 ("alignaddrldi_vis"): Likewise.
8990 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8991 ("fexpand_vis"): Likewise.
8992 ("fpmerge_vis"): Likewise.
8993 ("faligndata<VM64:mode>_vis"): Likewise.
8994 ("bshuffle<VM64:mode>_vis"): Likewise.
8995 ("cmask8<P:mode>_vis"): Likewise.
8996 ("cmask16<P:mode>_vis"): Likewise.
8997 ("cmask32<P:mode>_vis"): Likewise.
8998 ("fchksm16_vis"): Likewise.
8999 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9000 ("fmean16_vis"): Likewise.
9001 ("fp<plusminus_insn>64_vis"): Likewise.
9002 ("<plusminus_insn>v8qi3"): Likewise.
9003 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9004 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9005 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9006 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9007 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9008 ("*movqi_insn"): Likewise.
9009 ("*movhi_insn"): Likewise.
9010 ("*movsi_insn"): Likewise.
9011 ("movsi_pic_gotdata_op"): Likewise.
9012 ("*movdi_insn_sp32"): Likewise.
9013 ("*movdi_insn_sp64"): Likewise.
9014 ("movdi_pic_gotdata_op"): Likewise.
9015 ("*movsf_insn"): Likewise.
9016 ("*movdf_insn_sp32"): Likewise.
9017 ("*movdf_insn_sp64"): Likewise.
9018 ("*zero_extendhisi2_insn"): Likewise.
9019 ("*zero_extendqihi2_insn"): Likewise.
9020 ("*zero_extendqisi2_insn"): Likewise.
9021 ("*zero_extendqidi2_insn"): Likewise.
9022 ("*zero_extendhidi2_insn"): Likewise.
9023 ("*zero_extendsidi2_insn_sp64"): Likewise.
9024 ("ldfsr"): Likewise.
9025 ("prefetch_64"): Likewise.
9026 ("prefetch_32"): Likewise.
9027 ("tie_ld32"): Likewise.
9028 ("tie_ld64"): Likewise.
9029 ("*tldo_ldub_sp32"): Likewise.
9030 ("*tldo_ldub1_sp32"): Likewise.
9031 ("*tldo_ldub2_sp32"): Likewise.
9032 ("*tldo_ldub_sp64"): Likewise.
9033 ("*tldo_ldub1_sp64"): Likewise.
9034 ("*tldo_ldub2_sp64"): Likewise.
9035 ("*tldo_ldub3_sp64"): Likewise.
9036 ("*tldo_lduh_sp32"): Likewise.
9037 ("*tldo_lduh1_sp32"): Likewise.
9038 ("*tldo_lduh_sp64"): Likewise.
9039 ("*tldo_lduh1_sp64"): Likewise.
9040 ("*tldo_lduh2_sp64"): Likewise.
9041 ("*tldo_lduw_sp32"): Likewise.
9042 ("*tldo_lduw_sp64"): Likewise.
9043 ("*tldo_lduw1_sp64"): Likewise.
9044 ("*tldo_ldx_sp64"): Likewise.
9045 ("*mov<VM32:mode>_insn"): Likewise.
9046 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9047 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9049 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9051 * config/sparc/sparc.md ("type"): New insn type viscmp.
9052 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9054 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9055 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9056 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9057 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9059 ("n7_vis_logical_11cycle"): Likewise.
9060 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9061 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9062 * config/sparc/niagara.md ("niag_vis"): Likewise.
9063 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9064 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9066 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9068 * config/sparc/sparc.md: New instruction type `bmask'.
9069 (bmaskdi_vis): Use the `bmask' type.
9070 (bmasksi_vis): Likewise.
9071 * config/sparc/ultra3.md (us3_array): Likewise.
9072 * config/sparc/niagara7.md (n7_array): Likewise.
9073 * config/sparc/niagara4.md (n4_array): Likewise.
9074 * config/sparc/niagara2.md (niag2_vis): Likewise.
9075 (niag3_vis): Likewise.
9076 * config/sparc/niagara.md (niag_vis): Likewise.
9078 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9080 * ipa-comdats.c: Remove optimize check from gate.
9081 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9082 for functions not optimized.
9083 (ipa_fn_summary_read): Skip optimize check.
9084 (ipa_fn_summary_write): Likewise.
9085 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9087 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9089 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9090 (check_callers): Give up on caller which is not optimized.
9091 (inline_small_functions): Likewise.
9092 (ipa_inline): Do not give up when not optimizing.
9093 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9094 away unoptimizes cdtors.
9095 (whole_program_function_and_variable_visibility): Do
9096 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9097 * ipa.c (process_references): Do not check optimize.
9098 (symbol_table::remove_unreachable_nodes): Update optimize check.
9099 (set_writeonly_bit): Update optimize check.
9100 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9101 (pass_ipa_single_use::gate): Remove.
9103 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9105 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9106 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9107 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9108 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9109 permute_load, permute_store, adjust_extract, adjust_splat,
9110 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9111 replace_swap_with_copy, dump_swap_insn_table,
9112 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9113 recombine_lvx_pattern, recombine_stvx_pattern,
9114 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9115 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9116 to file rs6000-p8swap.c.
9117 * config/rs6000/rs6000-p8swap.c: New file.
9118 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9119 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9120 and rs6000*-*-* targets.
9122 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9124 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9126 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9128 * lto-wrapper.c (merge_and_complain): Do not merge
9129 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9130 fsigned_zeros, ftrapping_math, fwrapv.
9131 (append_compiler_options): Do not track these options.
9132 (append_linker_options): Likewie
9134 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9136 * cgraphunit.c (cgraph_node::finalize_function): When
9137 !flag_toplevel_reorde set no_reorder flag.
9138 (varpool_node::finalize_decl): Likewise.
9139 (symbol_table::compile): Drop no toplevel reorder path.
9141 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9143 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9144 edges; zero probability is not better than uninitialized.
9146 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9148 * asan.h (asan_sanitize_allocas_p): Declare.
9149 * asan.c (asan_sanitize_allocas_p): New function.
9150 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9151 (handle_builtin_alloca): Likewise.
9152 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9153 if !asan_sanitize_allocas_p.
9154 * params.def (asan-instrument-allocas): Add new option.
9155 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9156 * opts.c (common_handle_option): Disable allocas sanitization for
9159 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9161 * asan.c: Include gimple-fold.h.
9162 (get_last_alloca_addr): New function.
9163 (handle_builtin_stackrestore): Likewise.
9164 (handle_builtin_alloca): Likewise.
9165 (asan_emit_allocas_unpoison): Likewise.
9166 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9167 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9168 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9169 (instrument_builtin_call): Pass gimple iterator to
9170 get_mem_refs_of_builtin_call.
9171 (last_alloca_addr): New global.
9172 * asan.h (asan_emit_allocas_unpoison): Declare.
9173 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9174 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9175 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9176 if function calls alloca.
9177 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9178 * gimple-fold.h (replace_call_with_value): Declare.
9179 * internal-fn.c: Include asan.h.
9180 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9181 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9183 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9185 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9186 (C_SELFTEST_FLAGS): New.
9187 (CPP_SELFTEST_FLAGS): New.
9188 (SELFTEST_DEPS): New, from deps of s-selftest.
9189 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9190 (CPP_SELFTEST_DEPS): New.
9191 (selftest): Add dependency on s-selftest-c++.
9192 (s-selftest): Rename to...
9193 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9194 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9195 than SELFTEST_FLAGS.
9196 (selftest-gdb): Rename to...
9197 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9199 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9200 (selftest-valgrind): Rename to...
9201 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9203 (selftest-valgrind): Reintroduce as an alias for
9204 selftest-c-valgrind.
9205 (s-selftest-c++): New.
9206 (selftest-c++-gdb): New.
9207 (selftest-c++-valgrind): New.
9209 2017-07-06 Olivier Hainque <hainque@adacore.com>
9211 * gcc.c (process_command): When deciding if undefined variables
9212 should be ignored when processing specs, accept "gcc -v" as well.
9214 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9216 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9217 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9219 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9221 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9222 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9223 * config/arm/arm-tables.opt: Regenerate.
9224 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9226 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9228 2017-07-06 Carl Love <cel@us.ibm.com>
9230 * ChangeLog: Clean up from mid air collision
9232 2017-07-06 Carl Love <cel@us.ibm.com>
9234 * config/rs6000/rs6000-c.c: Add support for built-in functions
9235 vector signed int vec_subc (vector signed int, vector signed int);
9236 vector signed __int128 vec_subc (vector signed __int128,
9237 vector signed __int128);
9238 vector unsigned __int128 vec_subc (vector unsigned __int128,
9239 vector unsigned __int128);
9240 vector signed int vec_sube (vector signed int, vector signed int,
9242 vector unsigned int vec_sube (vector unsigned int,
9243 vector unsigned int,
9244 vector unsigned int);
9245 vector signed __int128 vec_sube (vector signed __int128,
9246 vector signed __int128,
9247 vector signed__int128);
9248 vector unsigned __int128 vec_sube (vector unsigned __int128,
9249 vector unsigned __int128,
9250 vector unsigned __int128);
9251 vector signed int vec_subec (vector signed int, vector signed int,
9253 vector unsigned int vec_subec (vector unsigned int,
9254 vector unsigned int,
9255 vector unsigned int);
9256 vector signed __int128 vec_subec (vector signed __int128,
9257 vector signed __int128,
9258 vector signed__int128);
9259 vector unsigned __int128 vec_subec (vector unsigned __int128,
9260 vector unsigned __int128,
9261 vector unsigned __int128);
9262 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9263 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9264 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9265 BU_ALTIVEC_OVERLOAD_X definitions.
9266 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9267 * doc/extend.texi: Update the built-in documentation file for the new
9270 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9273 * diagnostic-show-locus.c (layout::layout): Use start and finish
9274 spelling location for the start and finish of each range.
9275 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9276 Add unused aspect param.
9277 * input.c (expand_location_1): Add "aspect" param, and use it
9278 to access the correct part of the location.
9279 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9281 (expand_location_to_spelling_point): Likewise.
9282 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9283 param, and pass it to expand_location_1.
9285 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9287 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9288 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9289 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9290 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9291 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9292 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9293 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9294 _mm_maskz_getmant_ss): New intrinsics.
9295 (__builtin_ia32_getexpss128_mask): Changed to ...
9296 __builtin_ia32_getexpss128_round ... this.
9297 (__builtin_ia32_getexpsd128_mask): Changed to ...
9298 __builtin_ia32_getexpsd128_round ... this.
9299 * config/i386/i386-builtin-types.def
9300 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9301 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9302 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9303 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9304 __builtin_ia32_getmantss_mask_round): New builtins.
9305 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9306 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9307 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9308 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9309 * config/i386/sse.md
9310 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9311 avx512f_sgetexp<mode><mask_scalar_name>
9312 <round_saeonly_scalar_name> ... this.
9313 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9314 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9315 vgetexp<ssescalarmodesuffix>
9316 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9317 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9318 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9319 avx512f_vgetmant<mode><mask_scalar_name>
9320 <round_saeonly_scalar_name> ... this.
9321 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9322 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9323 vgetmant<ssescalarmodesuffix>
9324 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9325 %0<mask_scalar_operand4>, %1, %2
9326 <round_saeonly_scalar_mask_op4>, %3} ... this.
9327 * config/i386/subst.md (mask_scalar_operand4,
9328 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9329 round_saeonly_scalar_nimm_predicate): New subst attributes.
9331 2017-07-06 Julia Koval <julia.koval@intel.com>
9333 * config/i386/i386.c (ix86_erase_embedded_rounding):
9334 Remove code for old rounding pattern.
9336 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9338 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9340 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9342 * doc/sourcebuild.texi (Test Directives, Variants of
9343 dg-require-support): Add documentation for dg-require-stack-check.
9345 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9347 * config/i386/subst.md (mask_scalar, round_scalar,
9348 round_saeonly_scalar): New meta-templates.
9349 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9350 round_scalar_mask_operand3, round_scalar_mask_op3,
9351 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9352 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9353 round_saeonly_scalar_constraint,
9354 round_saeonly_scalar_prefix): New subst attribute.
9355 * config/i386/sse.md
9356 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9357 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9358 <round_scalar_name> ... this.
9359 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9360 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9361 <round_scalar_name> ... this.
9362 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9363 <sse>_vm<code><mode>3<mask_scalar_name>
9364 <round_saeonly_scalar_name> ... this.
9365 (v<plusminus_mnemonic><ssescalarmodesuffix>
9366 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9367 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9368 v<plusminus_mnemonic><ssescalarmodesuffix>
9369 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9370 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9371 (v<multdiv_mnemonic><ssescalarmodesuffix>
9372 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9373 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9374 v<multdiv_mnemonic><ssescalarmodesuffix>
9375 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9376 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9377 (v<maxmin_float><ssescalarmodesuffix>
9378 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9379 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9380 v<maxmin_float><ssescalarmodesuffix>
9381 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9382 %0<mask_scalar_operand3>, %1, %<iptr>2
9383 <round_saeonly_scalar_mask_op3>} ... this.
9385 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9387 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9388 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9390 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9391 Alan Hayward <alan.hayward@arm.com>
9392 David Sherwood <david.sherwood@arm.com>
9394 * combine.c (simplify_if_then_else): Remove "enum" before
9396 * compare-elim.c (can_eliminate_compare): Likewise.
9397 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9399 (aarch64_lookup_simd_builtin_type): Likewise.
9400 (aarch64_simd_builtin_type): Likewise.
9401 (aarch64_init_simd_builtin_types): Likewise.
9402 (aarch64_simd_expand_args): Likewise.
9403 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9405 (aarch64_reverse_mask): Likewise.
9406 (aarch64_simd_emit_reg_reg_move): Likewise.
9407 (aarch64_gen_adjusted_ldpstp): Likewise.
9408 (aarch64_ccmp_mode_to_code): Likewise.
9409 (aarch64_operands_ok_for_ldpstp): Likewise.
9410 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9411 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9413 (aarch64_min_divisions_for_recip_mul): Likewise.
9414 (aarch64_reassociation_width): Likewise.
9415 (aarch64_get_condition_code_1): Likewise.
9416 (aarch64_simd_emit_reg_reg_move): Likewise.
9417 (aarch64_simd_attr_length_rglist): Likewise.
9418 (aarch64_reverse_mask): Likewise.
9419 (aarch64_operands_ok_for_ldpstp): Likewise.
9420 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9421 (aarch64_gen_adjusted_ldpstp): Likewise.
9422 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9424 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9425 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9426 (arm_lookup_simd_builtin_type): Likewise.
9427 (arm_simd_builtin_type): Likewise.
9428 (arm_init_simd_builtin_types): Likewise.
9429 (arm_expand_builtin_args): Likewise.
9430 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9431 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9432 (ft32_setup_incoming_varargs): Likewise.
9433 (ft32_function_arg): Likewise.
9434 (ft32_function_arg_advance): Likewise.
9435 (ft32_pass_by_reference): Likewise.
9436 (ft32_arg_partial_bytes): Likewise.
9437 (ft32_valid_pointer_mode): Likewise.
9438 (ft32_addr_space_pointer_mode): Likewise.
9439 (ft32_addr_space_legitimate_address_p): Likewise.
9440 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9442 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9443 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9444 (iamcu_alignment): Likewise.
9445 (canonicalize_vector_int_perm): Likewise.
9446 (ix86_noce_conversion_profitable_p): Likewise.
9447 (ix86_mpx_bound_mode): Likewise.
9448 (ix86_operands_ok_for_move_multiple): Likewise.
9449 * config/microblaze/microblaze-protos.h
9450 (microblaze_expand_conditional_branch_reg): Likewise.
9451 * config/microblaze/microblaze.c
9452 (microblaze_expand_conditional_branch_reg): Likewise.
9453 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9455 (rs6000_reassociation_width): Likewise.
9456 (rs6000_invalid_binary_op): Likewise.
9457 (fusion_p9_p): Likewise.
9458 (emit_fusion_p9_load): Likewise.
9459 (emit_fusion_p9_store): Likewise.
9460 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9462 (riscv_hard_regno_mode_ok_p): Likewise.
9463 (riscv_address_insns): Likewise.
9464 (riscv_split_symbol): Likewise.
9465 (riscv_legitimize_move): Likewise.
9466 (riscv_function_value): Likewise.
9467 (riscv_hard_regno_nregs): Likewise.
9468 (riscv_expand_builtin): Likewise.
9469 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9470 (riscv_build_integer): Likewise.
9471 (riscv_split_integer): Likewise.
9472 (riscv_legitimate_constant_p): Likewise.
9473 (riscv_cannot_force_const_mem): Likewise.
9474 (riscv_regno_mode_ok_for_base_p): Likewise.
9475 (riscv_valid_base_register_p): Likewise.
9476 (riscv_valid_offset_p): Likewise.
9477 (riscv_valid_lo_sum_p): Likewise.
9478 (riscv_classify_address): Likewise.
9479 (riscv_legitimate_address_p): Likewise.
9480 (riscv_address_insns): Likewise.
9481 (riscv_load_store_insns): Likewise.
9482 (riscv_force_binary): Likewise.
9483 (riscv_split_symbol): Likewise.
9484 (riscv_force_address): Likewise.
9485 (riscv_legitimize_address): Likewise.
9486 (riscv_move_integer): Likewise.
9487 (riscv_legitimize_const_move): Likewise.
9488 (riscv_legitimize_move): Likewise.
9489 (riscv_address_cost): Likewise.
9490 (riscv_subword): Likewise.
9491 (riscv_output_move): Likewise.
9492 (riscv_canonicalize_int_order_test): Likewise.
9493 (riscv_emit_int_order_test): Likewise.
9494 (riscv_function_arg_boundary): Likewise.
9495 (riscv_pass_mode_in_fpr_p): Likewise.
9496 (riscv_pass_fpr_single): Likewise.
9497 (riscv_pass_fpr_pair): Likewise.
9498 (riscv_get_arg_info): Likewise.
9499 (riscv_function_arg): Likewise.
9500 (riscv_function_arg_advance): Likewise.
9501 (riscv_arg_partial_bytes): Likewise.
9502 (riscv_function_value): Likewise.
9503 (riscv_pass_by_reference): Likewise.
9504 (riscv_setup_incoming_varargs): Likewise.
9505 (riscv_print_operand): Likewise.
9506 (riscv_elf_select_rtx_section): Likewise.
9507 (riscv_save_restore_reg): Likewise.
9508 (riscv_for_each_saved_reg): Likewise.
9509 (riscv_register_move_cost): Likewise.
9510 (riscv_hard_regno_mode_ok_p): Likewise.
9511 (riscv_hard_regno_nregs): Likewise.
9512 (riscv_class_max_nregs): Likewise.
9513 (riscv_memory_move_cost): Likewise.
9514 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9515 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9516 (rl78_addr_space_address_mode): Likewise.
9517 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9519 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9520 (rs6000_reassociation_width): Likewise.
9521 (rs6000_invalid_binary_op): Likewise.
9522 (fusion_p9_p): Likewise.
9523 (emit_fusion_p9_load): Likewise.
9524 (emit_fusion_p9_store): Likewise.
9525 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9526 (ok_for_simple_move_operands): Likewise.
9527 (ok_for_simple_move_strict_operands): Likewise.
9528 (ok_for_simple_arith_logic_operands): Likewise.
9529 (visium_legitimize_reload_address): Likewise.
9530 (visium_select_cc_mode): Likewise.
9531 (output_cbranch): Likewise.
9532 (visium_split_double_move): Likewise.
9533 (visium_expand_copysign): Likewise.
9534 (visium_expand_int_cstore): Likewise.
9535 (visium_expand_fp_cstore): Likewise.
9536 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9537 (visium_function_arg): Likewise.
9538 (visium_function_arg_advance): Likewise.
9539 (visium_libcall_value): Likewise.
9540 (visium_setup_incoming_varargs): Likewise.
9541 (visium_legitimate_constant_p): Likewise.
9542 (visium_legitimate_address_p): Likewise.
9543 (visium_legitimize_address): Likewise.
9544 (visium_secondary_reload): Likewise.
9545 (visium_register_move_cost): Likewise.
9546 (visium_memory_move_cost): Likewise.
9547 (prepare_move_operands): Likewise.
9548 (ok_for_simple_move_operands): Likewise.
9549 (ok_for_simple_move_strict_operands): Likewise.
9550 (ok_for_simple_arith_logic_operands): Likewise.
9551 (visium_function_value_1): Likewise.
9552 (rtx_ok_for_offset_p): Likewise.
9553 (visium_legitimize_reload_address): Likewise.
9554 (visium_split_double_move): Likewise.
9555 (visium_expand_copysign): Likewise.
9556 (visium_expand_int_cstore): Likewise.
9557 (visium_expand_fp_cstore): Likewise.
9558 (visium_split_cstore): Likewise.
9559 (visium_select_cc_mode): Likewise.
9560 (visium_split_cbranch): Likewise.
9561 (output_cbranch): Likewise.
9562 (visium_print_operand_address): Likewise.
9563 * expmed.c (flip_storage_order): Likewise.
9564 * expmed.h (emit_cstore): Likewise.
9565 (flip_storage_order): Likewise.
9566 * genrecog.c (validate_pattern): Likewise.
9567 * hsa-gen.c (gen_hsa_addr): Likewise.
9568 * internal-fn.c (expand_arith_overflow): Likewise.
9569 * ira-color.c (allocno_copy_cost_saving): Likewise.
9570 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9571 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9572 (process_invariant_for_inheritance): Likewise.
9573 * lra-eliminations.c (move_plus_up): Likewise.
9574 * omp-low.c (lower_oacc_reductions): Likewise.
9575 * simplify-rtx.c (simplify_subreg): Likewise.
9576 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9577 (TARGET_CHKP_BOUND_MODE): Likewise..
9578 * targhooks.c (default_chkp_bound_mode): Likewise.
9579 (default_setup_incoming_vararg_bounds): Likewise.
9580 * targhooks.h (default_chkp_bound_mode): Likewise.
9581 (default_setup_incoming_vararg_bounds): Likewise.
9582 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9583 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9584 (have_whole_vector_shift): Likewise.
9585 * tree-vect-stmts.c (vectorizable_load): Likewise.
9586 * doc/tm.texi: Regenerate.
9588 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9590 Graceful degrade if Binutils PR21472 is not available.
9593 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9594 .rodata in flash test fails.
9595 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9596 * confgure: Regenerate.
9597 * config.in: Regenerate.
9598 * config/avr/avr.c (avr_asm_named_section)
9599 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9600 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9601 (avr_asm_init_sections): Same.
9603 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9605 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9606 (fma<VH:mode>4_intrinsic): Likewise.
9607 (*fmsub<VCVTF:mode>4): Likewise.
9608 (*fmsub<VH:mode>4_intrinsic): Likewise.
9610 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9613 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9614 Don't depend on "optimize > 0".
9615 (out_movhi_r_mr, out_movqi_mr_r): Same.
9616 (out_movhi_mr_r, out_movqi_r_mr): Same.
9617 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9618 io_address_operand on "optimize > 0".
9620 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9622 * tree-loop-distribution.c: Add general explanantion on the pass.
9623 (generate_loops_for_partition): Mark distributed loop.
9624 (pg_add_dependence_edges): New parameter. Handle alias data
9625 dependence specially and record it in the parameter if asked.
9626 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9627 (init_partition_graph_vertices, add_partition_graph_edge): New.
9628 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9629 (free_partition_graph_vdata, build_partition_graph): New.
9630 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9631 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9632 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9633 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9634 (version_for_distribution_p, finalize_partitions): New.
9635 (distribute_loop): Handle alias data dependence specially. Factor
9636 out loop fusion code as functions and call these functions.
9638 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9640 * tree-loop-distribution.c (classify_partition): New parameter and
9641 better handle reduction statement.
9642 (rdg_build_partitions): Revise comment.
9643 (distribute_loop): Compute statements in all partitions and pass it
9644 to classify_partition.
9646 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9648 * tree-loop-distribution.c (enum partition_type): New.
9649 (struct partition): New field type.
9650 (partition_merge_into): Add parameter. Update partition type.
9651 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9652 (build_rdg_partition_for_vertex): Compute partition type.
9653 (rdg_build_partitions): Dump partition type.
9654 (distribute_loop): Update calls to partition_merge_into.
9656 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9658 * tree-loop-distribution.c (struct ddr_hasher): New.
9659 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9661 (classify_partition): Call get_data_dependence.
9662 (pg_add_dependence_edges): Ditto.
9663 (distribute_loop): Release data dependence hash table.
9665 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9667 * tree-loop-distribution.c (ref_base_address): Delete.
9668 (similar_memory_accesses): Rename ...
9669 (share_memory_accesses): ... to this. Check if partitions access
9670 the same memory reference.
9671 (distribute_loop): Call share_memory_accesses.
9673 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9675 * tree-loop-distribution.c (struct partition): New field recording
9677 (partition_alloc, partition_free): Init and release data refs.
9678 (partition_merge_into): Merge data refs.
9679 (build_rdg_partition_for_vertex): Collect data refs for partition.
9680 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9682 (distribute_loop): Remve data refs from vertice data of partition
9685 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9687 * tree-loop-distribution.c (params.h): Include header file.
9688 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9689 (datarefs_vec): New global var.
9690 (create_rdg_vertices): Use datarefs_vec directly.
9691 (free_rdg): Don't free data references.
9692 (build_rdg): Update use. Don't free data references.
9693 (distribute_loop): Compute global variable for data references.
9694 Bail out if there are too many data references.
9696 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9698 * tree-loop-distribution.c (loop_nest): New global var.
9699 (build_rdg): Use loop directly, rather than loop nest.
9700 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9702 (distribute_loop): Compute global variable loop nest. Update use.
9704 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9706 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9707 (partition_merge_into): New parameter. Dump reason for fusion.
9708 (distribute_loop): Update use of partition_merge_into.
9710 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9712 * tree-loop-distribution.c (bb_top_order_index): New.
9713 (bb_top_order_index_size, bb_top_order_cmp): New.
9714 (stmts_from_loop): Use topological order.
9715 (pass_loop_distribution::execute): Compute and release topological
9716 order for basic blocks.
9718 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9720 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9723 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9725 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9726 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9727 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9728 * internal-fn.def (LOOP_DIST_ALIAS): New.
9729 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9730 (fold_loop_internal_call): ... this.
9731 (vect_loop_dist_alias_call): New function.
9732 (set_uid_loop_bbs): Call fold_loop_internal_call.
9733 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9736 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9739 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9740 Require dead FLAGS_REG at the beginning of a peephole.
9742 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9745 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9746 arguments in the call to __builtin_ia32_sbb_u32.
9747 (_subborrow_u64): Swap _X and _Y arguments in the call to
9748 __builtin_ia32_sbb_u64.
9750 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9753 * tree-vrp.c (compare_assert_loc): Turn into a function template
9754 with stable template parameter. Only test if a->e is NULL,
9755 !a->e == !b->e has been verified already. Use e == NULL or
9756 e != NULL instead of e or ! e tests. If stable is true, don't use
9757 iterative_hash_expr, on the other side allow a or b or both NULL
9758 and sort the NULLs last.
9759 (process_assert_insertions): Sort using compare_assert_loc<false>
9760 instead of compare_assert_loc, later sort using
9761 compare_assert_loc<true> before calling process_assert_insertions_for
9762 in a loop. Use break instead of continue once seen NULL pointer.
9764 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9766 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9767 Cortex-R7 and Cortex-R8 processors.
9769 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9771 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9772 uninitialized while src is not.
9774 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9776 * common/config/arm/arm-common.c: Adjust include path for
9778 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9779 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9780 (arm-cpu-data.h): Likewise.
9781 (arm-cpu-cdata.h): Likewise.
9782 * config/arm/arm-cpu.h: Delete.
9783 * config/arm/arm-cpu-cdata.h: Delete.
9784 * config/arm/arm-cpu-data.h: Delete.
9786 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9788 * config/arm/arm-cpus.in (cortex-a55): New.
9789 (cortex-a75): Likewise.
9790 (cortex-a75.cortex-a55): Likewise.
9791 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9793 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9794 * config/arm/arm-cpu-cdata.h: Regenerate.
9795 * config/arm/arm-cpu-data.h: Regenerate.
9796 * config/arm/arm-cpu.h: Regenerate.
9797 * config/arm/arm-tables.opt: Regenerate.
9798 * config/arm/arm-tune.md: Regenerate.
9800 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9802 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9804 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9806 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9809 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9811 PR tree-optimization/81292
9812 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9813 full_string_p, also call adjust_related_strinfos if the adjustment
9814 is simple, otherwise invalidate related strinfos.
9816 2017-07-04 Martin Liska <mliska@suse.cz>
9819 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9820 newly created variable as DECL_IGNORED_P.
9822 2017-07-04 Martin Liska <mliska@suse.cz>
9825 * ipa-inline.c (inline_small_functions):
9826 Use xstrdup_for_dump.
9828 2017-07-04 Tom de Vries <tom@codesourcery.com>
9830 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9832 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9835 * config/darwin.c (darwin_function_switched_text_sections):
9838 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9840 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9842 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9844 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9846 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9848 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9849 min_profitable_iters, and th as inclusive lower bounds.
9850 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9851 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9852 for min_profitable_iters and min_profitable_estimate.
9853 (vect_transform_loop): Treat th as an inclusive lower bound.
9854 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9856 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9859 * config/darwin.c (darwin_function_switched_text_sections):
9860 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9861 in two pieces, and suppress the use of buf.
9863 2017-07-03 Nathan Sidwell <nathan@acm.org>
9865 * hash-table.h (hash_table_mod1): Fix indentation.
9867 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9870 * predict.c (force_edge_cold): Be more careful about propagation
9872 * profile-count.h (profile_probability::guessed,
9873 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9875 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9877 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9879 * doc/invoke.texi (rcpc architecture extension): Document it.
9881 2017-07-03 Richard Biener <rguenther@suse.de>
9883 PR tree-optimization/60510
9884 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9885 the scalar reduction PHI and use it.
9886 (vectorizable_reduction): Properly guard the single_defuse_cycle
9887 path for non-SLP reduction chains where we cannot use it.
9888 Rework reduc_def/index and vector type deduction. Rework
9889 vector operand gathering during reduction op code-gen.
9890 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9891 chains dissolve the chain and leave it to non-SLP reduction
9894 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9896 * tree-data-ref.h (dr_alignment): Declare.
9897 * tree-data-ref.c (dr_alignment): New function.
9898 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9899 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9901 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9903 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9905 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9906 and base_misalignment fields.
9907 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9908 * tree-data-ref.c: Include builtins.h.
9909 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9910 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9911 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9912 * tree-vect-data-refs.c: Include tree-cfg.h.
9913 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9914 fields instead of calculating an alignment here.
9915 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9916 innermost_loop_behavior fields.
9918 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9920 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9922 (DR_STEP_ALIGNMENT): New macro.
9923 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9924 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9925 (create_data_ref): Print it.
9926 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9927 to tell whether the step preserves vector (mis)alignment.
9928 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9929 Move the check for an integer step and generalise to all INTEGER_CST.
9930 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9931 Print the outer step alignment.
9933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9935 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9936 with offset_alignment.
9937 (DR_ALIGNED_TO): Delete.
9938 (DR_OFFSET_ALIGNMENT): New macro.
9939 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9940 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9941 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9942 (create_data_ref): Likewise.
9943 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9944 (vect_analyze_data_refs): Likewise.
9945 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9946 creating dummy innermost behavior.
9948 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9950 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9951 with a "innermost_loop_behavior *" and refeence tree.
9952 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9953 (create_data_ref): Update call accordingly.
9954 * tree-predcom.c (find_looparound_phi): Likewise.
9956 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9958 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9959 fields with dr_wrt_vec_loop.
9960 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9961 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9962 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9963 (vect_dr_behavior): New function.
9964 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9965 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9966 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9967 track whether the step preserves the misalignment.
9968 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9969 Use vect_dr_behavior.
9970 (vect_setup_realignment): Update call accordingly.
9971 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9972 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9973 call to vect_create_addr_base_for_vector_ref.
9974 (vect_create_cond_for_align_checks): Likewise.
9975 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9976 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9977 (vect_recog_mask_conversion_pattern): Likewise.
9978 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9979 (new_stmt_vec_info): Remove redundant zeroing.
9981 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9983 * common/config/arm/arm-common.c (arm_be8_option): New function.
9984 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9985 (ISA_ARMv6): Add isa_bit_be8.
9986 * config/arm/arm.h (arm_be8_option): Add prototype.
9987 (BE8_SPEC_FUNCTION): New define.
9988 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9989 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9990 (mlittle-endian): Similarly.
9991 (mbe8, mbe32): New options.
9992 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9993 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9995 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9997 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9999 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10001 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10002 (cleanup_tree_cfg_bb): Use it.
10003 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10005 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10007 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10010 * loop-doloop.c (add_test): Update profile.
10012 2017-07-03 Martin Liska <mliska@suse.cz>
10015 * sanopt.c (rewrite_usage_of_param): New function.
10016 (sanitize_rewrite_addressable_params): Likewise.
10017 (pass_sanopt::execute): Call rewrite_usage_of_param.
10019 2017-07-03 Richard Biener <rguenther@suse.de>
10021 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10022 back to using VIEW_CONVERT_EXPR.
10024 2017-07-03 Martin Liska <mliska@suse.cz>
10027 * doc/extend.texi: Document when a resolver function is
10028 generated for target_clones.
10030 2017-07-03 Martin Liska <mliska@suse.cz>
10032 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10033 and shadow memory of auto variables which are subject of
10034 use-after-scope sanitization.
10035 (asan_expand_mark_ifn): Add do set only when is_poison.
10037 2016-07-03 Richard Biener <rguenther@suse.de>
10039 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10041 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10042 (vectorizable_reduction): Perform reduction PHI creation when
10043 visiting a reduction PHI and adjust and simplify code generation
10044 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10045 (vect_transform_loop): Visit reduction PHIs.
10046 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10047 defs into the SLP tree.
10048 (vect_build_slp_tree): Reduction defs terminate the recursion.
10049 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10051 (vect_get_vec_defs_for_stmt_copy): Export.
10052 (vect_get_vec_defs): Likewise.
10053 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10055 (vect_get_vec_defs_for_stmt_copy): Declare.
10056 (vect_get_vec_defs): Likewise.
10058 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10060 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10061 parameter with a "loop" parameter and use it instead of the
10062 loop containing DR_STMT. Don't check simple_iv when doing
10063 BB analysis. Describe the two analysis modes in the comment.
10065 2017-07-03 Tom de Vries <tom@codesourcery.com>
10067 PR tree-optimization/69468
10068 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10069 (find_same_succ_bb): Handle ignore_edge_flags.
10071 2017-07-03 Tom de Vries <tom@codesourcery.com>
10073 PR tree-optimization/81192
10074 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10076 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10078 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10080 2017-07-03 Tom de Vries <tom@codesourcery.com>
10082 PR tree-optimization/81192
10083 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10084 BB_SAME_SUCC (bb) == NULL.
10086 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10088 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10091 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10093 * dumpfile.c: Include profile-count.h
10094 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10096 (insert_cond_bb): Update profile.
10097 * tree-cfg.h (insert_cond_bb): Update prototype.
10098 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10099 * tree-dump.c: Do not include tree-cfg.
10101 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10103 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10105 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10107 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10110 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10112 * tree-complex.c (expand_complex_div_wide): update profile.
10114 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10115 Alan Hayward <alan.hayward@arm.com>
10116 David Sherwood <david.sherwood@arm.com>
10118 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10119 (CORETYPES_H): New define.
10120 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10121 (insn-modes-inline.h, s-modes-inline-h): New rules.
10122 (generated_files): Add insn-modes-inline.h.
10123 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10124 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10125 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10126 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10127 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10128 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10129 (build/gencodes.o, build/genconditions.o): Likewise.
10130 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10131 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10132 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10133 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10134 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10135 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10136 * coretypes.h: Include everything up to real.h for generators.
10137 Include insn-modes.h first. Include wide-int-print.h after
10138 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10139 * machmode.h: Don't include insn-modes.h here.
10140 * function-tests.c: Remove includes of signop.h, machmode.h,
10141 double-int.h and wide-int.h.
10143 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10145 * optc-save-gen.awk: Likewise.
10146 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10147 * godump.c: Remove include of wide-int-print.h.
10148 * pretty-print.h: Likewise.
10149 * wide-int-print.cc: Likewise.
10150 * wide-int.cc: Likewise.
10151 * hash-map-tests.c: Remove include of signop.h.
10152 * hash-set-tests.c: Likewise.
10153 * rtl-tests.c: Likewise.
10154 * mkconfig.sh: Remove include of machmode.h.
10155 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10157 (emit_insn_modes_inline_h): ...this new function. Emit the code
10158 into an insn-modes-inline.h header file, adding appropriate
10159 include guards and end comments.
10160 (emit_insn_modes_c_header): Remove include of machmode.h.
10161 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10163 (main): Handle -i flag and call emit_insn_modes_inline_h when
10166 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10168 * tree-ssa-strlen.c (strinfo): Rename the length field to
10169 nonzero_chars. Add a full_string_p field.
10170 (compare_nonzero_chars, zero_length_string_p): New functions.
10171 (get_addr_stridx): Add an offset_out parameter.
10172 Use compare_nonzero_chars.
10173 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10174 (new_strinfo): Update after above changes to strinfo.
10175 (set_endptr_and_length): Set full_string_p.
10176 (get_string_length): Update after above changes to strinfo.
10177 (unshare_strinfo): Update call to new_strinfo.
10178 (maybe_invalidate): Likewise.
10179 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10180 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10181 as a uhwi instead of an shwi. Update after above changes to
10182 strinfo and new_strinfo.
10183 (zero_length_string): Assert that chainsi contains full strings.
10184 Use zero_length_string_p. Update call to new_strinfo.
10185 (adjust_related_strinfos): Update after above changes to strinfo.
10186 Copy full_string_p from origsi.
10187 (adjust_last_stmt): Use zero_length_string_p.
10188 (handle_builtin_strlen): Update after above changes to strinfo and
10189 new_strinfo. Install the lhs as the string length if the previous
10190 entry didn't describe a full string.
10191 (handle_builtin_strchr): Update after above changes to strinfo
10193 (handle_builtin_strcpy): Likewise.
10194 (handle_builtin_strcat): Likewise.
10195 (handle_builtin_malloc): Likewise.
10196 (handle_pointer_plus): Likewise.
10197 (handle_builtin_memcpy): Likewise. Track nonzero characters
10198 that aren't necessarily followed by a nul terminator.
10199 (handle_char_store): Likewise.
10201 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10203 PR tree-optimization/80769
10204 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10205 for malloc and calloc. Document the new invariant that all related
10206 strinfos have delayed lengths or none do.
10207 (verify_related_strinfos): Move earlier in file.
10208 (set_endptr_and_length): New function, split out from...
10209 (get_string_length): ...here. Also set the lengths of related
10211 (zero_length_string): Assert that chainsi has known (rather than
10213 (adjust_related_strinfos): Likewise.
10215 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10217 PR tree-optimization/81136
10218 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10219 assert that two references with the same misalignment have the same
10220 compile-time misalignment if those compile-time misalignments
10223 2017-07-01 Andi Kleen <ak@linux.intel.com>
10225 * print-tree.c (print_node): Print all attributes.
10227 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10229 * cfg.c (scale_bbs_frequencies): New function.
10230 * cfg.h (scale_bbs_frequencies): Declare it.
10231 * cfgloopanal.c (single_likely_exit): Cleanup.
10232 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10234 (scale_loop_profile): Likewise.
10235 (loop_version): Likewise.
10236 (create_empty_loop_on_edge): Update.
10237 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10238 scale_loop_frequencies, scale_loop_profile, loopify,
10239 loop_version): Update prototypes.
10240 * modulo-sched.c (sms_schedule): Update.
10241 * predict.c (unlikely_executed_edge_p): Also check probability.
10242 (probably_never_executed_edge_p): Fix typo.
10243 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10244 * tree-parloops.c (gen_parallel_loop): Update.
10245 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10246 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10247 * tree-ssa-loop-split.c (split_loop): Update.
10248 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10249 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10250 (vect_loop_versioning): Update.
10251 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10253 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10255 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10257 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10259 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10260 to keep profile consistent.
10262 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10264 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10265 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10266 * profile-count.h (max_safe_multiplier): Make unsigned.
10267 (profile_count::guessed_zero): New.
10269 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10271 * bb-reorder.c (fix_up_crossing_landing_pad,
10272 fix_crossing_conditional_branches): Use make_single_succ_edge
10273 to keep profile consistent.
10275 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10277 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10280 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10283 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10284 the right scopes, make sure cond_jump isn't preserved between multiple
10285 iterations. Search for fallthru edge whenever there are 3+ edges and
10286 use find_fallthru_edge for it.
10288 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10290 Patch by Alexander Monakov <amonakov@ispras.ru>
10291 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10292 probabilities consistently.
10294 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10296 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10297 emit_cmp_and_jump_insns.
10299 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10302 * tree-inline.c (expand_call_inline): Combine profile statuses.
10304 2017-06-30 Andrew Pinski <apinski@cavium.com>
10306 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10307 fold_stmt returned true.
10309 2017-06-30 Nathan Sidwell <nathan@acm.org>
10311 * ggc.h (empty_string): Delete.
10312 * cfgexpand.c (expand_asm_stmt): Use plain "".
10313 * optabs.c (expand_asm_memory_barrier): Likewise.
10314 * stringpool.c (empty_string): Delete.
10315 (digit_vector, digit_string): Delete.
10316 (ggc_alloc_string): Use plain "", don't optimize single digit
10317 strings. Use ggc_alloc_atomic.
10319 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10321 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10322 comparison set and one other set, use the cost of the non-comparison
10325 2017-06-30 Nathan Sidwell <nathan@acm.org>
10327 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10330 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10332 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10333 loops. Remove now unneeded calls to gimple_switch_set_label() that
10334 just set removed labels to NULL_TREE.
10336 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10338 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10339 (set_range_info): ...here. Only call set_range_info_raw if domain
10341 (set_nonzero_bits): Call set_range_info_raw.
10342 * tree-ssanames.h (set_range_info_raw): New.
10344 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10347 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10348 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10349 of nonimmediate_operand and <store_mask_constraint> instead of m
10350 for the input operand. For V8FI iterator, always split if input
10351 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10352 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10353 <store_mask_predicate> instead of register_operand and
10354 <store_mask_constraint> instead of v for the input operand. Make
10355 sure both operands aren't MEMs for if not <mask_applied>.
10357 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10359 * lto-wrapper.c (copy_file) Close both file descriptors before
10362 2017-06-30 Martin Liska <mliska@suse.cz>
10365 * multiple_target.c (create_dispatcher_calls): Make ifunc
10366 also for function that don't have calls or are not referenced.
10368 2017-06-30 Richard Biener <rguenther@suse.de>
10370 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10371 analyze the first scalar stmt. Move vector type computation
10372 for the BB case here from ...
10373 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10374 live operation processing in the SLP case properly.
10376 2017-06-30 Richard Biener <rguenther@suse.de>
10378 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10380 2017-06-30 Martin Liska <mliska@suse.cz>
10383 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10384 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10386 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10388 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10391 2017-06-30 Richard Biener <rguenther@suse.de>
10393 PR tree-optimization/81249
10394 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10395 condition reduction result to original scalar type.
10397 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10399 * profile-count.h (enum profile_quality): Fix typos and whitespace
10402 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10404 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10405 type for branch probabilities.
10407 2017-06-29 Julian Brown <julian@codesourcery.com>
10408 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10410 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10411 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10412 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10413 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10415 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10417 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10418 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10419 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10420 CC usage from generic code to here.
10421 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10422 CC usage into the target macros.
10424 2017-06-29 Maya Rashish <coypu@sdf.org>
10426 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10429 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10431 * arm/arm-builtins.c: Include profile-count.h
10432 * except.c (sjlj_emit_function_enter): Use
10433 profile_probability::unlikely.
10435 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10437 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10438 and tocrel_offset be pointer args rather than implicitly using
10440 (legitimate_constant_pool_address_p, rs6000_emit_move,
10441 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10442 tocrel_offset and use in toc_relative_expr_p call.
10443 (print_operand, print_operand_address): Use static tocrel_base_oac
10444 and tocrel_offset_oac.
10445 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10448 2017-06-29 Maya Rashish <coypu@sdf.org>
10450 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10452 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10454 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10455 objects, take into account only the alignment of 'op0' and 'mode1' if
10458 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10460 * ccmp.c (ccmp_tree_comparison_p): New function.
10461 (ccmp_candidate_p): Update to use above function.
10462 (get_compare_parts): New function.
10463 (expand_ccmp_next): Update to use new functions.
10464 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10466 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10467 take mode as argument.
10468 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10469 * expr.c (expand_expr_real_1): Pass mode as argument.
10471 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10473 * combine.c (combine_instructions): Print insns to dump_file, together
10476 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10478 * asan.c (asan_emit_stack_protection): Update.
10479 (create_cond_insert_point): Update.
10480 * auto-profile.c (afdo_propagate_circuit): Update.
10481 * basic-block.h (struct edge_def): Turn probability to
10482 profile_probability.
10483 (EDGE_FREQUENCY): Update.
10484 * bb-reorder.c (find_traces_1_round): Update.
10485 (better_edge_p): Update.
10486 (sanitize_hot_paths): Update.
10487 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10488 (make_single_succ_edge): Update.
10489 (check_bb_profile): Update.
10490 (dump_edge_info): Update.
10491 (update_bb_profile_for_threading): Update.
10492 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10493 probabilitycount to 0.
10494 * cfgbuild.c (compute_outgoing_frequencies): Update.
10495 * cfgcleanup.c (try_forward_edges): Update.
10496 (outgoing_edges_match): Update.
10497 (try_crossjump_to_edge): Update.
10498 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10499 (expand_gimple_tailcall): Update.
10500 (construct_init_block): Use make_single_succ_edge.
10501 (construct_exit_block): Use make_single_succ_edge.
10502 * cfghooks.c (verify_flow_info): Update.
10503 (redirect_edge_succ_nodup): Update.
10504 (split_edge): Update.
10505 (account_profile_record): Update.
10506 * cfgloopanal.c (single_likely_exit): Update.
10507 * cfgloopmanip.c (scale_loop_profile): Update.
10508 (set_zero_probability): Remove.
10509 (duplicate_loop_to_header_edge): Update.
10510 * cfgloopmanip.h (loop_version): Update prototype.
10511 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10512 (force_nonfallthru_and_redirect): Update.
10513 (update_br_prob_note): Update.
10514 (rtl_verify_edges): Update.
10515 (purge_dead_edges): Update.
10516 (rtl_lv_add_condition_to_bb): Update.
10517 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10518 * cgraphunit.c (init_lowered_empty_function): Update.
10519 (cgraph_node::expand_thunk): Update.
10520 * cilk-common.c: Include profile-count.h
10521 * dojump.c (inv): Remove.
10522 (jumpifnot): Update.
10523 (jumpifnot_1): Update.
10524 (do_jump_1): Update.
10526 (do_jump_by_parts_greater_rtx): Update.
10527 (do_compare_rtx_and_jump): Update.
10528 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10529 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10530 * dwarf2cfi.c: Include profile-count.h
10531 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10532 (sjlj_emit_dispatch_table): Likewise.
10533 * explow.c: Include profile-count.h
10534 * expmed.c (emit_store_flag_force): Update.
10535 (do_cmp_and_jump): Update.
10536 * expr.c (compare_by_pieces_d::generate): Update.
10537 (compare_by_pieces_d::finish_mode): Update.
10538 (emit_block_move_via_loop): Update.
10539 (store_expr_with_bounds): Update.
10540 (store_constructor): Update.
10541 (expand_expr_real_2): Update.
10542 (expand_expr_real_1): Update.
10543 * expr.h (try_casesi, try_tablejump): Update prototypes.
10544 * gimple-pretty-print.c (dump_probability): Update.
10545 (dump_profile): New.
10546 (dump_gimple_label): Update.
10547 (dump_gimple_bb_header): Update.
10548 * graph.c (draw_cfg_node_succ_edges): Update.
10549 * hsa-gen.c (convert_switch_statements): Update.
10550 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10551 (find_if_case_1): Update.
10552 (find_if_case_2): Update.
10553 * internal-fn.c (expand_arith_overflow_result_store): Update.
10554 (expand_addsub_overflow): Update.
10555 (expand_neg_overflow): Update.
10556 (expand_mul_overflow): Update.
10557 (expand_vector_ubsan_overflow): Update.
10558 * ipa-cp.c (good_cloning_opportunity_p): Update.
10559 * ipa-split.c (split_function): Use make_single_succ_edge.
10560 * ipa-utils.c (ipa_merge_profiles): Update.
10561 * loop-doloop.c (add_test): Update.
10562 (doloop_modify): Update.
10563 * loop-unroll.c (compare_and_jump_seq): Update.
10564 (unroll_loop_runtime_iterations): Update.
10565 * lra-constraints.c (lra_inheritance): Update.
10566 * lto-streamer-in.c (input_cfg): Update.
10567 * lto-streamer-out.c (output_cfg): Update.
10568 * mcf.c (adjust_cfg_counts): Update.
10569 * modulo-sched.c (sms_schedule): Update.
10570 * omp-expand.c (expand_omp_for_init_counts): Update.
10571 (extract_omp_for_update_vars): Update.
10572 (expand_omp_ordered_sink): Update.
10573 (expand_omp_for_ordered_loops): Update.
10574 (expand_omp_for_generic): Update.
10575 (expand_omp_for_static_nochunk): Update.
10576 (expand_omp_for_static_chunk): Update.
10577 (expand_cilk_for): Update.
10578 (expand_omp_simd): Update.
10579 (expand_omp_taskloop_for_outer): Update.
10580 (expand_omp_taskloop_for_inner): Update.
10581 * omp-simd-clone.c (simd_clone_adjust): Update.
10582 * optabs.c (expand_doubleword_shift): Update.
10583 (expand_abs): Update.
10584 (emit_cmp_and_jump_insn_1): Update.
10585 (expand_compare_and_swap_loop): Update.
10586 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10587 * predict.c (predictable_edge_p): Update.
10588 (edge_probability_reliable_p): Update.
10589 (set_even_probabilities): Update.
10590 (combine_predictions_for_insn): Update.
10591 (combine_predictions_for_bb): Update.
10592 (propagate_freq): Update.
10593 (estimate_bb_frequencies): Update.
10594 (force_edge_cold): Update.
10595 * profile-count.c (profile_count::dump): Add missing space into dump.
10596 (profile_count::debug): Add newline.
10597 (profile_count::differs_from_p): Explicitly convert to unsigned.
10598 (profile_count::stream_in): Update.
10599 (profile_probability::dump): New member function.
10600 (profile_probability::debug): New member function.
10601 (profile_probability::differs_from_p): New member function.
10602 (profile_probability::differs_lot_from_p): New member function.
10603 (profile_probability::stream_in): New member function.
10604 (profile_probability::stream_out): New member function.
10605 * profile-count.h (profile_count_quality): Rename to ...
10606 (profile_quality): ... this one.
10607 (profile_probability): New.
10608 (profile_count): Update.
10609 * profile.c (compute_branch_probabilities): Update.
10610 * recog.c (peep2_attempt): Update.
10611 * sched-ebb.c (schedule_ebbs): Update.
10612 * sched-rgn.c (find_single_block_region): Update.
10613 (compute_dom_prob_ps): Update.
10614 (schedule_region): Update.
10615 * sel-sched-ir.c (compute_succs_info): Update.
10616 * stmt.c (struct case_node): Update.
10617 (do_jump_if_equal): Update.
10618 (get_outgoing_edge_probs): Update.
10619 (conditional_probability): Update.
10620 (emit_case_dispatch_table): Update.
10621 (expand_case): Update.
10622 (expand_sjlj_dispatch_table): Update.
10623 (emit_case_nodes): Update.
10624 * targhooks.c: Update.
10625 * tracer.c (better_p): Update.
10626 (find_best_successor): Update.
10627 * trans-mem.c (expand_transaction): Update.
10628 * tree-call-cdce.c: Update.
10629 * tree-cfg.c (gimple_split_edge): Upate.
10630 (move_sese_region_to_fn): Upate.
10631 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10632 * tree-eh.c (lower_resx): Upate.
10633 (cleanup_empty_eh_move_lp): Upate.
10634 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10635 * tree-inline.c (copy_edges_for_bb): Update.
10636 (copy_cfg_body): Update.
10637 * tree-parloops.c (gen_parallel_loop): Update.
10638 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10639 (gimple_gen_time_profiler): Update.
10640 * tree-ssa-dce.c (remove_dead_stmt): Update.
10641 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10642 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10643 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10644 (unloop_loops): Update.
10645 (try_peel_loop): Update.
10646 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10647 * tree-ssa-loop-split.c (connect_loops): Update.
10648 (split_loop): Update.
10649 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10650 (hoist_guard): Update.
10651 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10652 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10653 (value_replacement): Update.
10654 * tree-ssa-reassoc.c (branch_fixup): Update.
10655 * tree-ssa-tail-merge.c (replace_block_by): Update.
10656 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10657 (create_edge_and_update_destination_phis): Update.
10658 (compute_path_counts): Update.
10659 (recompute_probabilities): Update.
10660 (update_joiner_offpath_counts): Update.
10661 (freqs_to_counts_path): Update.
10662 (duplicate_thread_path): Update.
10663 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10664 (struct switch_conv_info): Update.
10665 (gen_inbound_check): Update.
10666 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10667 (vect_do_peeling): Update.
10668 (vect_loop_versioning): Update.
10669 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10670 (optimize_mask_stores): Update.
10671 * ubsan.c (ubsan_expand_null_ifn): Update.
10672 * value-prof.c (gimple_divmod_fixed_value): Update.
10673 (gimple_divmod_fixed_value_transform): Update.
10674 (gimple_mod_pow2): Update.
10675 (gimple_mod_pow2_value_transform): Update.
10676 (gimple_mod_subtract): Update.
10677 (gimple_mod_subtract_transform): Update.
10678 (gimple_ic): Update.
10679 (gimple_stringop_fixed_value): Update.
10680 (gimple_stringops_transform): Update.
10681 * value-prof.h: Update.
10683 2017-06-29 Carl Love <cel@us.ibm.com>
10685 * config/rs6000/rs6000-c.c: Add support for built-in functions
10686 vector signed int vec_signed (vector float);
10687 vector signed long long vec_signed (vector double);
10688 vector signed int vec_signed2 (vector double, vector double);
10689 vector signed int vec_signede (vector double);
10690 vector signed int vec_signedo (vector double);
10691 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10692 instruction generator.
10693 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10694 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10695 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10697 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10698 vunsignede_v2df): Add define_expands.
10699 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10700 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10701 VEC_UNSIGNEDO): Add definitions.
10702 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10703 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10704 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10705 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10706 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10707 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10708 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10709 vec_unsignede, vec_unsignedo): Add builtin defines.
10710 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10712 * doc/extend.texi: Update the built-in documentation file for the
10713 new built-in functions.
10715 2017-06-29 Richard Biener <rguenther@suse.de>
10717 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10718 reduction chains to LOOP_VINFO_REDUCTIONS.
10719 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10720 SLP reductions after processing reduction chains.
10722 2017-06-29 Nathan Sidwell <nathan@acm.org>
10724 * builtins.c (fold_builtin_FUNCTION): Use
10725 lang_hooks.decl_printable_name.
10727 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10729 PR middle-end/81194
10730 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10731 with only one label.
10732 * stmt.c (expand_case): Assert NCASES is greater than one.
10734 2017-06-29 Richard Biener <rguenther@suse.de>
10736 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10738 (group_case_labels): Likewise.
10739 (find_taken_edge): Push sanity checking on val to workers...
10740 (find_taken_edge_cond_expr): ... here
10741 (find_taken_edge_switch_expr): ... and here, handle cases
10742 with just a default label.
10743 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10744 (group_case_labels): Likewise.
10745 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10746 group_case_labels does anything cleanup the CFG again.
10748 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10750 PR tree-optimization/81196
10751 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10752 exit condition comparing two IVs.
10754 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10756 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10757 profile to the dummy entry at the end of the list of architectures.
10758 * config/arm/arm-cpu-cdata.h: Regenerated.
10760 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10761 Michael Collison <michael.collison@arm.com>
10764 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10766 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10767 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10768 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10769 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10770 of shift when the shift amount is masked with constant equal to
10771 the size of the mode.
10772 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10775 2017-06-29 Martin Liska <mliska@suse.cz>
10777 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10779 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10782 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10785 (edge_growth_cache_entry::edge_growth_cache_entry): New
10787 (reset_edge_growth_cache): Update to use constructor.
10789 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10791 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10792 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10793 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10795 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10797 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10798 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10800 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10802 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10803 (*-linux-uclibc*): Add t-uclibc tmake_file.
10804 * config/t-musl: New.
10805 * config/t-uclibc: New.
10807 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10809 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10811 (gen_comm_data): Emit architectural setting of arch_prof.
10812 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10814 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10815 (armv8-m.base, armv8-m.main): Likewise.
10816 * arm-protos.h (arm_build_target): Add profile field.
10817 (arch_option): Likewise.
10818 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10820 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10821 arm_active_target.profile.
10823 2017-06-28 Richard Biener <rguenther@suse.de>
10825 PR middle-end/81227
10826 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10827 TYPE_OVERFLOW_WRAPS.
10828 * match.pd (negate_expr_p): Likewise.
10829 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10830 fold_build2, not fold_binary.
10832 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10834 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10835 Convert memory address to Pmode.
10836 (aarch64_print_operand): Assert MEM operands are always Pmode.
10838 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10841 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10842 Remove redundant if.
10843 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10844 * config/arm/aarch-common-protos.h
10845 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10846 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10848 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10851 * multiple_target.c (create_dispatcher_calls): Set the default
10852 clone to be static, not public.
10854 2017-06-28 Richard Biener <rguenther@suse.de>
10856 * tree-vect-loop.c (vectorizable_reduction): Move special
10857 cond reduction IV var creation ...
10858 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10859 parameter. Use STMT_VINFO_VECTYPE.
10860 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10863 2017-06-28 Martin Liska <mliska@suse.cz>
10866 * ipa-visibility.c (non_local_p): Handle visibility.
10868 2017-06-28 Martin Liska <mliska@suse.cz>
10871 * common.opt: Add IntegerRange to various options.
10872 * opt-functions.awk (integer_range_info): New function.
10873 * optc-gen.awk: Add integer_range_info to cl_options struct.
10874 * opts-common.c (decode_cmdline_option): Handle
10875 CL_ERR_INT_RANGE_ARG.
10876 (cmdline_handle_error): Likewise.
10877 * opts.c (print_filtered_help): Show valid interval in
10878 when --help is provided.
10879 * opts.h (struct cl_option): Add range_min and range_max fields.
10880 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10882 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10884 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10885 (x * C EQ/NE y * C): New transformation.
10887 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10889 * genmultilib (combination_space): Accept '+' in option names.
10891 2017-06-28 Martin Liska <mliska@suse.cz>
10894 * asan.c (instrument_derefs): Bail out inner references
10895 that are hard register variables.
10897 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10900 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10901 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10903 2017-06-28 Richard Biener <rguenther@suse.de>
10905 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10906 (vect_get_slp_defs): Adjust.
10907 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10909 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10911 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10912 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10913 (vectorizable_reduction): Adjust.
10914 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10916 (vect_get_slp_defs): Likewise.
10917 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10918 (vectorizable_bswap): Adjust.
10919 (vectorizable_call): Likewise.
10920 (vectorizable_conversion): Likewise.
10921 (vectorizable_assignment): Likewise.
10922 (vectorizable_shift): Likewise.
10923 (vectorizable_operation): Likewise.
10924 (vectorizable_store): Likewise.
10925 (vectorizable_condition): Likewise.
10926 (vectorizable_comparison): Likewise.
10928 2017-06-28 Michael Collison <michael.collison@arm.com>
10931 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10933 (arm_gen_movmemqi): Removed unused variable 'i'.
10934 Convert 'for' loop into 'while' loop.
10935 (arm_expand_prologue): Remove last unnecessary set of insn.
10936 (thumb_pop): Remove unused variable 'pushed_words'.
10937 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10939 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10941 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10942 * config/s390/s390-protos.h: Add prototype of
10943 s390_rel_address_ok_p.
10944 * config/s390/s390.c (s390_got_symbol): New function.
10945 (s390_rel_address_ok_p): New function.
10946 (legitimize_pic_address): Use s390_rel_address_ok_p.
10947 (s390_load_got): Use s390_got_symbol.
10948 (s390_option_override): Issue error if
10949 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10950 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10952 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10954 2017-06-27 Andrew Pinski <apinski@cavium.com>
10956 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10957 (X * copysign (1.0, X)): New pattern.
10958 (X * copysign (1.0, -X)): New pattern.
10959 (copysign (-1.0, CST)): New pattern.
10961 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10963 * genmultilib (combination_space): Remove variable.
10964 Validate reuse rules against regular expression for any sequence
10965 of multilib options in any order.
10967 2017-06-27 Michael Collison <michael.collison@arm.com>
10969 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10970 call aarch64_split_simd_combine.
10971 * (aarch64_combine_internal<mode>): Delete pattern.
10972 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10973 Allow register and subreg operands.
10975 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10977 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10978 specific need, just fallback on defaults.
10979 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10981 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10982 Olivier Hainque <hainque@adacore.com>
10984 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10986 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10987 targets. Pick a default if no particular attempt applied.
10988 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10991 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10993 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10994 (x86_64-wrs-vxworks7): Likewise.
10996 2017-06-27 Marek Polacek <polacek@redhat.com>
10999 * ubsan.c (instrument_null): Check get_base_address's result for null.
11001 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11003 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11005 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11007 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11008 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11009 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11010 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11011 New function types.
11012 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11013 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11014 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11015 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11016 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11017 BUILT_IN_FEUPDATEENV): New builtins.
11018 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11019 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11020 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11021 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11023 (builtin_structptr_types): Adjust size.
11024 * tree.c (builtin_structptr_types): Add four entries.
11026 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11027 Olivier Hainque <hainque@adacore.com>
11029 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11030 (TLS_SYM): New local macro, forcing reference to __tls__ on
11031 link command lines for VxWorks 7 RTPs, triggering initialization
11033 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11034 OS features TLS support, true for RTPs on VxWorks 7.
11035 * config/vxworks.c (vxworks_override_options): Setup emutls
11038 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11040 * predict.c (test_prediction_value_range): Use -1U instead of -1
11041 to avoid narrowing conversion warning.
11042 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11043 to avoid narrowing conversion warning.
11044 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11046 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11048 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11050 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11051 64bit configurations.
11052 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11053 (SIZE_TYPE): Likewise.
11054 * config/vxworks.c (vxworks_emutls_var_fields): Use
11055 long_unsigned_type_node instead of unsigned_type_node as the offset
11056 field type, which is "pointer" mode in emutls.c.
11058 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11061 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11063 PR middle-end/81207
11064 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11065 gimple_vuse copying separately from gimple_vdef copying.
11067 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11069 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11072 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11073 Olivier Hainque <hainque@adacore.com>
11075 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11076 for all vxworks7 targets.
11077 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11078 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11079 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11080 variations for VX6/VX7 and 32/64bits later on in ...
11081 (VXWORKS_LIB_SPEC): Leverage new macros.
11082 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11083 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11085 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11087 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11088 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11090 2017-06-26 Carl Love <cel@us.ibm.com>
11092 * config/rs6000/rs6000-c.c: Add support for built-in functions
11093 vector bool char vec_reve (vector bool char);
11094 vector signed char vec_reve (vector signed char);
11095 vector unsigned char vec_reve (vector unsigned char);
11096 vector bool int vec_reve (vector bool int);
11097 vector signed int vec_reve (vector signed int);
11098 vector unsigned int vec_reve (vector unsigned int);
11099 vector bool long long vec_reve (vector bool long long);
11100 vector signed long long vec_reve (vector signed long long);
11101 vector unsigned long long vec_reve (vector unsigned long long);
11102 vector bool short vec_reve (vector bool short);
11103 vector signed short vec_reve (vector signed short);
11104 vector double vec_reve (vector double);
11105 vector float vec_reve (vector float);
11106 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11107 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11108 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11109 (altivec_vreve): New pattern.
11110 * config/rs6000/altivec.h (vec_reve): New define.
11111 * doc/extend.texi (vec_rev): Update the built-in documentation file
11112 for the new built-in functions.
11114 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11116 PR tree-optimization/71815
11117 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11119 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11121 (slsr_process_phi): Likewise.
11122 (replace_uncond_cands_and_profitable_phis): Don't replace a
11123 multiply candidate with a stride of 1 (copy or cast).
11124 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11126 (lowest_cost_path): Likewise.
11127 (total_savings): Likewise.
11129 2017-06-26 Richard Biener <rguenther@suse.de>
11132 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11133 Use def_builtin_pure for all gather builtins.
11135 2017-06-26 Richard Biener <rguenther@suse.de>
11137 PR tree-optimization/81203
11138 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11139 non-dominating BBs.
11141 2017-06-26 Marek Polacek <polacek@redhat.com>
11144 * doc/invoke.texi: Document -Wmultistatement-macros.
11146 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11148 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11149 arm_neon_ok_no_float_abi effective target.
11151 2017-06-26 Richard Biener <rguenther@suse.de>
11153 PR tree-optimization/80928
11154 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11155 (copy_bbs): Set BB_DUPLICATED flag early.
11156 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11158 (execute_on_shrinking_pred): Likewise.
11159 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11160 BB_DUPLICATED blocks.
11161 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11162 iterate over all PHIs considering removal of *gsi.
11164 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11166 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11169 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11171 * config/rs6000/rs6000-string.c: (expand_block_clear,
11172 do_load_for_compare, select_block_compare_mode,
11173 compute_current_alignment, expand_block_compare,
11174 expand_strncmp_align_check, expand_strn_compare,
11175 expand_block_move, rs6000_output_load_multiple)
11176 Move functions related to string/block move/compare
11177 to a separate file.
11178 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11179 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11180 for this function which is now used in two files.
11181 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11182 * config.gcc: Add rs6000-string.o to extra_objs for
11183 targets powerpc*-*-* and rs6000*-*-*.
11185 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11188 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11189 32-bit, since indexed is not valid for DImode.
11190 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11191 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11192 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11193 (define_peephole2 for Altivec d-form store): Likewise.
11196 * multiple_target.c (create_dispatcher_calls): Only create the
11197 dispatcher call if the function is the default clone of a
11198 versioned function.
11200 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11202 PR middle-end/80902
11203 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11204 a call, force the call to not be a tail call.
11206 2017-06-23 Jeff Law <law@redhat.com>
11208 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11211 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11213 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11214 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11215 handling for early expansion of vector shifts (sl,sr,sra,rl).
11216 (builtin_function_type): Add vector shift right instructions
11217 to the unsigned argument list.
11219 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11221 rtl-optimizatoin/79286
11222 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11223 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11224 trap. PIC register plus a const unspec without offset can never trap.
11226 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11228 * tree.h (builtin_structptr_type): New type.
11229 (builtin_structptr_types): Declare new array.
11230 * tree.c (builtin_structptr_types): New array.
11231 (free_lang_data, build_common_tree_nodes): Use it.
11233 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11236 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11239 2017-06-22 Matt Turner <mattst88@gmail.com>
11241 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11242 Lake models to skylake case. Assume skylake for unknown
11243 models with clflushopt.
11245 2017-06-22 Jeff Law <law@redhat.com>
11247 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11248 frame sizes that do not satisfy aarch64_uimm12_shift.
11250 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11252 * profile-count.h (apply_probability,
11253 apply_scale, probability_in): Fix checks for zero.
11255 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11257 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11258 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11260 2016-06-22 Richard Biener <rguenther@suse.de>
11262 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11263 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11264 REDUC_MAX_EXPR support.
11265 (vectorizable_reduction): Likewise.
11266 (vect_create_epilog_for_reduction): Likewise.
11268 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11270 * match.pd (A / (1 << B) -> A >> B): New.
11271 * generic-match-head.c: Include optabs-tree.h.
11272 * gimple-match-head.c: Likewise.
11273 * optabs-tree.h (target_supports_op_p): New.
11274 * optabs-tree.c (target_supports_op_p): New.
11276 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11278 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11279 $gcc_cv_ld --help output.
11280 (gcc_cv_ld_demangle): Likewise.
11281 (gcc_cv_ld_eh_frame_hdr): Likewise.
11282 (gcc_cv_ld_pie): Likewise.
11283 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11284 (gcc_cv_ld_buildid): Likewise.
11285 (gcc_cv_ld_sysroot): Likewise.
11286 (ld_bndplt_support): Likewise.
11287 (ld_pushpopstate_support): Likewise.
11288 * configure: Regenerate.
11289 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11291 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11294 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11295 operands indexes to avoid gap between operands and match_dups.
11297 2017-06-21 Andrew Pinski <apinski@cavium.com>
11299 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11300 Increment Arith_shift and Arith_shift_reg by 1.
11301 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11303 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11304 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11305 (aarch64_strip_extend): Add new argument and test for it.
11306 (aarch64_cheap_mult_shift_p): New function.
11307 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11308 add a cost if it is true.
11309 Update calls to aarch64_strip_extend.
11310 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11312 2017-06-21 Andrew Pinski <apinski@cavium.com>
11314 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11316 (thunderxt88): Likewise.
11317 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11318 (thunderx_prefetch_tune): New variable.
11319 (thunderx2t99_prefetch_tune): Update for the correct values.
11320 (thunderxt88_tunings): New variable.
11321 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11322 generic_prefetch_tune.
11323 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11325 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11327 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11328 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11329 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11330 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11331 (aarch64_atomic_cas<mode>, GPI): Likewise.
11333 2017-06-21 Martin Liska <mliska@suse.cz>
11335 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11336 statements on cold and hot labels.
11337 * predict.c (tree_estimate_probability_bb): Remove the
11338 prediction from this place.
11340 2017-06-21 Martin Liska <mliska@suse.cz>
11342 PR tree-optimization/79489
11343 * gimplify.c (maybe_add_early_return_predict_stmt): New
11345 (gimplify_return_expr): Call the function.
11346 * predict.c (tree_estimate_probability_bb): Remove handling
11348 * predict.def: Update comment about early return predictor.
11349 * gimple-predict.h (is_gimple_predict): New function.
11350 * predict.def: Change default value of early return to 66.
11351 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11353 * passes.def: Put pass_strip_predict_hints to the beginning of
11356 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11358 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11359 FUNCTION_DECL declarations.
11360 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11362 (dwaf2out_decl): Likewise.
11363 * godump.c (go_early_global_decl): Skip call to the real debug hook
11364 for FUNCTION_DECL declarations.
11365 * passes.c (rest_of_decl_compilation): Skip call to the
11366 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11367 -fdump-go-spec is passed.
11369 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11371 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11372 Reorder for compactness.
11373 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11374 (def_builtin_pure, def_builtin_pure2): New functions.
11375 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11377 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11379 * match.pd (nop_convert): New predicate.
11380 ((A +- CST1) +- CST2): Allow some NOP conversions.
11382 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11385 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11386 with ctors/dtors if GOVD_SHARED is set.
11388 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11390 * config/aarch64/aarch64.md (movti_aarch64):
11391 Emit mov rather than orr.
11392 (movtf_aarch64): Likewise.
11393 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11394 Emit mov rather than orr.
11396 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11398 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11399 Swap alternatives, make integer dup more expensive.
11401 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11403 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11404 Return true for non-tls symbols.
11406 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11408 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11409 (cortex-a75): Likewise.
11410 (cortex-a75.cortex-a55): Likewise.
11411 * config/aarch64/aarch64-tune.md: Regenerate.
11412 * doc/invoke.texi (-mtune): Document new values for -mtune.
11414 2017-06-21 Tom de Vries <tom@codesourcery.com>
11416 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11417 stack_size feature.
11418 (Effective-Target Keywords, Other attributes): Suggest using
11419 dg-add-options stack_size feature to get stack limit in stack_size
11420 effective target documentation.
11422 2017-06-21 Julian Brown <julian@codesourcery.com>
11423 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11425 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11426 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11427 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11429 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11430 attribute type list for neon_multiply.
11431 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11432 attribute type list for neon_multiply.
11433 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11434 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11435 attribute type list for neon_multiply.
11436 * config/arm/types.md (crypto_pmull): Add.
11437 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11438 attribute type list.
11440 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11442 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11445 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11447 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11448 to make sure not to dereference a NULL cost_classes_ptr pointer.
11450 2017-06-20 Carl Love <cel@us.ibm.com>
11452 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11453 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11454 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11455 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11456 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11457 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11458 VMULOSW): New enum "unspec" values.
11459 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11460 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11461 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11462 altivec_vmulosw): New patterns.
11463 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11464 VMULOSW): Add definitions.
11466 2017-06-20 Julia Koval <julia.koval@intel.com>
11468 * config/i386/i386.c: Fix rounding expand for new pattern.
11469 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11471 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11473 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11474 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11476 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11478 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11481 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11483 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11484 architecture, then by implementer ID.
11485 * config/aarch64/aarch64-tune.md: Regenerate.
11487 2017-06-20 Richard Biener <rguenther@suse.de>
11489 PR middle-end/81097
11490 * fold-const.c (split_tree): Fold to type before negating.
11492 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11494 * diagnostic-show-locus.c
11495 (selftest::test_fixit_deletion_affecting_newline): New function.
11496 (selftest::diagnostic_show_locus_c_tests): Call it.
11498 2017-06-20 Andreas Schwab <schwab@suse.de>
11501 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11504 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11506 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11507 __ARM_FEATURE_COPROC according to support.
11509 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11511 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11512 Rewritten to avoid overflow for > 32-bit pointers.
11515 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11516 by removing enum keyword.
11517 (ubsan_type_descriptor): Likewise. Formatting fix.
11520 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11521 splitter): Require TARGET_SSE2 in the condition.
11523 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11526 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11527 for doing vector set of SFmode on ISA 3.0.
11528 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11529 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11531 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11532 SFmode value into a V4SF variable that was extracted from another
11533 V4SF variable without converting the element to double precision
11534 and back to single precision vector format.
11535 (vsx_insert_extract_v4sf_p9_2): Likewise.
11537 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11539 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11540 in UWHI to avoid undefined overflow.
11543 * ubsan.h (enum ubsan_encode_value_phase): New.
11544 (ubsan_encode_value): Change second argument to
11545 enum ubsan_encode_value_phase with default value of
11546 UBSAN_ENCODE_VALUE_GENERIC.
11547 * ubsan.c (ubsan_encode_value): Change second argument to
11548 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11549 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11550 create_tmp_var_raw instead of create_tmp_var and use a
11552 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11553 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11554 ubsan_encode_value callers.
11557 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11558 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11559 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11561 2017-06-19 Richard Biener <rguenther@suse.de>
11563 PR middle-end/81118
11564 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11565 estimates if we changed anything.
11567 2017-06-19 Richard Biener <rguenther@suse.de>
11569 PR tree-optimization/80887
11570 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11571 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11572 simplified lookups, then reset mprts_hook.
11573 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11575 (try_to_simplify): Likewise.
11577 2017-06-19 Martin Liska <mliska@suse.cz>
11580 * gimplify.c (gimplify_switch_expr):
11581 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11583 2017-06-19 Martin Liska <mliska@suse.cz>
11585 * doc/install.texi: Document that PGO runs in 4 stages.
11587 2017-06-19 Martin Liska <mliska@suse.cz>
11590 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11591 to dispatcher function name.
11592 * multiple_target.c (replace_function_decl): New function.
11593 (create_dispatcher_calls): Redirect both edges and references.
11595 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11597 * profile-count.c (profile_count::dump): Dump quality.
11598 (profile_count::differs_from_p): Update for unsigned val.
11599 * profile-count.h (profile_count_quality): New enum.
11600 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11602 2017-06-19 Richard Biener <rguenther@suse.de>
11604 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11605 struct function as arg.
11606 (estimate_numbers_of_iterations): Export overload with loop arg.
11607 (free_numbers_of_iterations_estimates_loop): Use an overload of
11608 free_numbers_of_iterations_estimates instead.
11609 * tree-cfg.c (remove_bb): Adjust.
11610 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11611 * tree-parloops.c (gen_parallel_loop): Likewise.
11612 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11614 (tree_unroll_loops_completely): Likewise.
11615 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11616 Use an overload instead and export.
11617 (estimated_loop_iterations): Adjust.
11618 (max_loop_iterations): Likewise.
11619 (likely_max_loop_iterations): Likewise.
11620 (estimate_numbers_of_iterations): Take struct function as arg
11622 (loop_exits_before_overflow): Adjust.
11623 (free_numbers_of_iterations_estimates_loop): Use an overload.
11624 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11625 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11627 2017-06-19 Richard Biener <rguenther@suse.de>
11630 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11631 RANGE_EXPR conservatively.
11633 2017-06-16 Carl Love <cel@us.ibm.com>
11635 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11636 definitions for vec_float, vec_float2, vec_floato,
11637 vec_floate built-ins.
11638 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11639 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11641 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11642 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11643 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11644 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11645 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11646 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11647 vec_floato): Add builtin defines.
11648 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11649 Update the built-in documentation file for the new built-in
11652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11654 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11655 (mthumb): Mark as the negative of -marm.
11657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11659 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11661 (ARM Options, -mtune): Document that this accepts the same
11662 extension options as -mcpu.
11663 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11667 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11668 permitted extensions.
11670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11672 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11673 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11674 (armv8-m.main): Add option +nodsp.
11675 * config/arm/arm-cpu-cdata.h: Regenerated.
11677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11679 * config/arm/t-fuchsia: New file.
11680 * config.gcc (arm*-*-fuchsia*): Use it.
11682 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11684 * config/arm/t-symbian: Rewrite for new option infrastructure.
11686 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11688 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11689 (MULTILIB_REQUIRED): Likewise.
11691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11693 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11694 (MULTILIB_RESUE): Likewise.
11695 (MULTILIB_MATCHES): Likewise.
11696 (MULTLIB_REQUIRED): Likewise.
11698 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11700 * config/arm/t-rtems: Rewrite for new option framework.
11702 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11704 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11705 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11706 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11707 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11708 * config/arm/t-multilib: ... here.
11709 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11710 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11711 armv7-a and armv8*-a when A-profile libraries have not been built.
11712 * config/arm/t-rmprofile: Rewrite.
11714 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11716 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11717 with a backslash. Remove the backslash after substituting unescaped
11719 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11723 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11724 the driver to pass through the default setting of -mfloat-abi.
11725 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11727 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11728 (all_feat_combs): New rule.
11729 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11731 * config/arm/t-aprofile: Rewrite.
11733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11735 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11736 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11737 fpu is not specified by the user/command-line.
11738 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11739 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11740 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11741 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11742 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11743 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11747 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11748 * config/arm/t-arm-elf: Rewritten.
11750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11752 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11753 have some floating-point instructions.
11754 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11755 (TARGET_MAYBE_HARD_FLOAT): New macro.
11756 * config/arm/arm-builtins.c (arm_init_builtins): Use
11757 TARGET_MAYBE_HARD_FLOAT.
11758 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11760 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11762 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11763 (configargs.h): Include it.
11764 (arm_print_hint_for_fpu_option): New function.
11765 (arm_parse_fpu_option): New function.
11766 (candidate_extension): New class.
11767 (arm_canon_for_multilib): New function.
11768 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11769 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11770 (ARCH_CANONICAL_SPECS): New macro.
11771 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11775 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11776 are set after handling multilib fragments. Set target_cpu_default2
11779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11781 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11783 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11784 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11786 (check_arch): Likewise.
11787 * config/arm/arm.c (arm_configure_build_target): Handle
11788 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11789 options in the default.
11791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11793 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11794 when an option is an alias of another.
11795 * config/arm/parsecpu.awk (optalias): New parser token.
11796 (gen_comm_data): Mark non-alias options as such. Emit entries
11797 for extension aliases.
11798 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11799 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11800 (armv6kz, armv6zk, armv6t2): Likewise.
11801 (armv7): Make vfpv3-d16 an alias.
11802 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11804 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11805 Sort in canonical order.
11806 (armv8-a): Sort in canonical order.
11807 (armv8.1-a, armv8.2-a): Likewise.
11808 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11810 (cortex-a9): Sort in canonical order.
11811 * config/arm/arm.c (selftests.h): Include it.
11812 (arm_test_cpu_arch_data): New function.
11813 (arm_run_self_tests): New function.
11814 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11815 (targetm): Move declaration to the end of the file.
11816 * arm-cpu-cdata.h: Regenerated.
11818 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11820 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11821 call to target_mode_check describing the type of option passed.
11822 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11823 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11824 arm_parse_cpu_option_name to match parameters against list of
11826 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11827 arm_arch_core_flags data structure.
11828 * config/arm/arm-cpu_cdata.h: Regenerated.
11830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11832 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11834 (arm_print_hint_for_cpu_option): Likewise.
11835 (arm_print_hint_for_arch_option): Likewise.
11836 (arm_parse_cpu_option_name): Likewise.
11837 (arm_parse_arch_option_name): Likewise.
11838 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11839 of entries in the all_fpus list.
11840 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11841 (arm_parse_cpu_option_name): Declare.
11842 (arm_parse_arch_option_name): Declare.
11843 (arm_parse_option_features): Declare.
11844 (arm_intialize_isa): Declare.
11845 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11847 (gen_comm_data): ... here. Make definitions non-static.
11848 * config/arm/arm-cpu-data.h: Regenerated.
11849 * config/arm/arm-cpu-cdata.h: Regenerated.
11851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11853 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11854 (cpu_arch_extension): New structure.
11855 (cpu_arch_option, arch_option, cpu_option): New structures.
11856 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11857 architecture types.
11858 (gen_data): Generate new format data tables.
11859 * config/arm/arm.c (cpu_tune): New structure.
11860 (cpu_option, processors): Delete.
11861 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11862 (arm_print_hint_for_cpu_option): ... this and ...
11863 (arm_print_hint_for_arch_option): ... this.
11864 (arm_parse_arch_cpu_name): Delete. Replace with ...
11865 (arm_parse_cpu_option_name): ... this and ...
11866 (arm_parse_arch_option_name): ... this.
11867 (arm_unrecognized_feature): Change type of target parameter to
11869 (arm_parse_arch_cpu_features): Delete. Replace with ...
11870 (arm_parse_option_features): ... this.
11871 (arm_configure_build_target): Rework to use new configuration data
11873 (arm_print_tune_info): Rework for new configuration data tables.
11874 * config/arm/arm-cpu-data.h: Regenerated.
11875 * config/arm/arm-cpu.h: Regenerated.
11877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11879 * Makefile.in (OBJS): Move sbitmap.o from here ...
11880 (OBJS-libcommon): ... to here.
11882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11884 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11885 (ISA_ALL_CRYPTO): New macro.
11886 (ISA_ALL_SIMD): New macro
11887 (ISA_ALL_FP): New macro.
11888 * config/arm/arm.c (fpu_bitlist): Update initializer.
11889 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11891 (arm9e): Add fpu. Add option for nofp
11892 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11893 (arm926ej-s, arm1026ej-s): Likewise.
11894 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11895 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11896 neon-fp16, neon-vfpv4, nofp and nosimd.
11897 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11898 (cortex-a8): Add fpu. Add option for nofp.
11899 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11900 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11901 (cortex-r4f): Add fpu.
11902 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11903 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11905 (cortex-r8): Likewise.
11906 (cortex-m4): Add fpu. Add option for nofp.
11907 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11908 (cortex-a17.cortex-a7): Likewise.
11909 (cortex-a32): Add fpu. Add options for crypto and nofp.
11910 (cortex-a35, cortex-a53): Likewise.
11911 (cortex-a57): Add fpu. Add option for crypto.
11912 (cortex-a72, cortex-a73): Likewise.
11913 (exynos-m1): Likewise.
11914 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11915 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11916 (cortex-m33): Add fpu. Add option for nofp.
11917 * config/arm/arm-cpu-cdata.h: Regenerated
11918 * config/arm/arm-cpu-data.h: Regenerated.
11920 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11922 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11923 (armv5te, armv5tej): Likewise.
11924 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11925 (armv7): Add options fp and vfpv3-d16.
11926 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11927 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11929 (armv7ve): Likewise.
11930 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11931 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11932 (armv8-a): Add nocrypto option.
11933 (armv8.1-a, armv8.2-a): Likewise.
11934 (armv8-m.main): add options fp, fp.dp and nofp.
11936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11938 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11940 (armv8-a+crc): Delete.
11941 (armv8.1-a): Add options simd, crypto and nofp.
11942 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11943 (armv8.2-a+fp16): Delete.
11944 (armv8-m.main): Add option dsp.
11945 (armv8-m.main+dsp): Delete.
11946 (cortex-a8): Add fpu. Add nofp option.
11947 (cortex-a9): Add fpu. Add nofp and nosimd options.
11948 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11949 link to main cpu and architecture data structures.
11950 (gen_comm_data): Only put isa attributes from the main architecture
11952 (option): New statement for architecture and CPU entries.
11953 * arm.c (struct cpu_option): New structure.
11954 (struct processors): Add entry for options.
11955 (arm_unrecognized_feature): New function.
11956 (arm_parse_arch_cpu_name): Ignore any characters after the first
11958 (arm_parse_arch_cpu_feature): New function.
11959 (arm_configure_build_target): Separate out any CPU and architecture
11960 features and parse separately. Don't error out if -mfpu=auto is
11961 used with only an architecture string.
11962 (arm_print_asm_arch_directives): New function.
11963 (arm_file_start): Call it.
11964 * config/arm/arm-cpu-cdata.h: Regenerated.
11965 * config/arm/arm-cpu-data.h: Likewise.
11966 * config/arm/arm-tables.opt: Likewise.
11968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11970 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11971 assembler when it is not -mfpu=auto.
11973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11975 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11976 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11977 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11978 (ASM_CPU_SPEC): Rewrite.
11979 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11980 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11981 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11982 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11983 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11984 copied string is NUL-terminated. Also strip any characters prefixed
11986 (arm_rewrite_selected_arch): New function.
11987 (arm_rewrite_march): New function.
11989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11991 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11992 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11993 (march, mcpu, mtune): Convert to string-based options.
11994 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11995 (arm_parse_arch_cpu_name): New function.
11996 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11997 identify selected architecture or CPU.
11998 (arm_option_save): New function.
11999 (TARGET_OPTION_SAVE): Redefine.
12000 (arm_option_restore): Restore string options.
12001 (arm_option_print): Print string options.
12003 2017-06-16 Martin Sebor <msebor@redhat.com>
12005 PR tree-optimization/80933
12006 PR tree-optimization/80934
12007 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12008 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12009 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12010 (gimple_fold_builtin): Call them.
12012 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12014 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12015 as unlikely; update profile.
12017 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12019 * predict.c (force_edge_cold): Handle declaring edges impossible
12022 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12024 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12026 (try_unroll_loop_completely): Fix reporting.
12028 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12030 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12032 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12035 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12036 if given a non-constant argument for an intrinsic which requires a
12039 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12041 * profile.c (compare_freqs): New function.
12042 (branch_prob): Sort edge list.
12043 (find_spanning_tree): Assume that the list is priority sorted.
12045 2017-06-16 Richard Biener <rguenther@suse.de>
12047 PR tree-optimization/81090
12048 * passes.def (pass_record_bounds): Remove.
12049 * tree-pass.h (make_pass_record_bounds): Likewise.
12050 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12051 make_pass_record_bounds): Likewise.
12052 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12053 not free niter estimates at the beginning but at the end.
12054 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12056 2017-06-16 Richard Biener <rguenther@suse.de>
12058 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12059 initializer to workaround ICE in host GCC 4.8.
12061 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12063 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12065 (clone_inlined_nodes): Update.
12067 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12069 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12070 prefetch settings, and enable prefetching by default at -O3.
12072 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12074 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12075 Set flag_prefetch_loop_arrays according to tuning data.
12077 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12079 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12080 New tune structure.
12081 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12082 [Unrelated to main purpose of the patch] Place the pointer field last
12083 to enable type checking errors when tune structure are wrongly merged.
12084 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12085 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12086 (thunderx2t99_prefetch_tune): New tune constants.
12087 (tune_params *_tunings): Update all tunings (no functional change).
12088 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12089 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12090 from tunings structures.
12092 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12095 * ubsan.c (instrument_null): Add T argument, use it instead
12096 of computing it based on IS_LHS.
12097 (instrument_object_size): Likewise.
12098 (pass_ubsan::execute): Adjust instrument_null and
12099 instrument_object_size callers to pass gimple_get_lhs or
12100 gimple_assign_rhs1 result to it. Use instrument_null instead of
12101 calling get_base_address and instrument_mem_ref. Handle
12102 aggregate call arguments for object-size sanitization.
12104 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12106 PR tree-optimization/81089
12107 * tree-vrp.c (is_masked_range_test): Validate operands of
12110 2017-06-15 Martin Sebor <msebor@redhat.com>
12113 * dumpfile.c (dump_register): Avoid calling memset to initialize
12114 a class with a default ctor.
12115 * gcc.c (struct compiler): Remove const qualification.
12116 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12117 * hash-table.h: Ditto.
12118 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12120 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12121 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12123 * params.h (struct param_info): Make struct members non-const.
12124 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12125 with default initialization.
12126 * vec.h (vec_copy_construct, vec_default_construct): New helper
12128 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12129 with vec_copy_construct.
12130 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12131 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12132 * doc/invoke.texi (-Wclass-memaccess): Document.
12134 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12136 * emit-rtl.h (is_leaf): Update comment about local
12137 register allocator.
12139 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12142 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12143 for a variable to have a section before checking if the section has a
12145 Set section to.persistent if persistent attribute is set.
12146 Warn if .persistent attribute is used on an automatic variable.
12148 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12150 PR rtl-optimization/80474
12151 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12153 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12155 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12158 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12160 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12161 (casesi): Emit bounds checking as RTL.
12162 (casesi_internal_mips16_<mode>): Remove bounds checking.
12164 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12166 * config/xtensa/xtensa.c (xtensa_option_override): Append
12167 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12168 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12169 xtensa_doloop_hooks): Define unconditionally.
12170 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12172 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12173 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12174 for it in xtensa_option_override.
12175 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12176 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12178 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12180 * doc/cppopts.texi: Document '-' special value to -MF.
12182 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12184 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12185 (cortex_a53_fconst): Likewise.
12186 (cortex_a53_fpmul): Likewise.
12187 (cortex_a53_f_load_64): Likewise.
12188 (cortex_a53_f_load_many): Likewise.
12189 (cortex_a53_advsimd_alu): Likewise.
12190 (cortex_a53_advsimd_alu_q): Likewise.
12191 (cortex_a53_advsimd_mul): Likewise.
12192 (cortex_a53_advsimd_mul_q): Likewise.
12193 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12194 Add missing fmul, r2f_cvt and fconst cases.
12196 2017-06-14 Richard Biener <rguenther@suse.de>
12198 PR middle-end/81088
12199 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12201 (fold_binary_loc): When associating do not treat pre-existing
12202 TREE_OVERFLOW on literal constants as a reason to allow
12203 TREE_OVERFLOW on associated literal constants.
12205 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12207 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12208 (MASK_FEATURES): New macro.
12209 * config/sparc/sparc.c (sparc_option_override): Remove the special
12210 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12212 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12214 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12215 a division of 0 if non-call exceptions are enabled.
12217 2017-06-14 Andrew Pinski <apinski@cavium.com>
12218 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12221 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12222 Improve vector initialization code gen for only variable case.
12224 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12226 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12228 2017-06-14 Richard Biener <rguenther@suse.de>
12230 PR tree-optimization/81083
12231 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12234 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12236 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12237 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12238 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12239 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12240 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12241 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12243 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12245 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12246 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12248 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12250 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12252 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12254 * config/rs6000/t-rtems: Don't handle SPE.
12256 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12258 * config/rs6000/t-linux: Don't handle SPE.
12260 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12262 * config/rs6000/eabispe.h: Delete file.
12264 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12266 * config/rs6000/t-spe: Delete file.
12268 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12270 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12271 (rs6000_legitimate_offset_address_p): Return false for anything in
12272 V2SImode or V2SFmode.
12274 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12276 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12277 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12278 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12280 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12281 (rs6000_legitimate_offset_address_p): Ditto.
12282 (rs6000_emit_move): Ditto.
12283 (rs6000_init_builtins): Remove V4HI_type_node.
12285 2017-06-13 Martin Liska <mliska@suse.cz>
12288 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12289 (gate_asan): Likewise.
12290 * asan.h (asan_no_sanitize_address_p): Remove the function.
12291 (sanitize_flags_p): New function.
12292 * builtins.def: Fix coding style.
12293 * common.opt: Use renamed enum value.
12294 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12295 * doc/extend.texi: Document no_sanitize attribute.
12296 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12297 to SANITIZE_UNDEFINED_NONDEFAULT.
12298 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12299 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12300 Use sanitize_flags_p.
12301 * gimplify.c (gimplify_function_tree): Likewise.
12302 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12303 * opts.c (parse_no_sanitize_attribute): New function.
12304 (common_handle_option): Use renamed enum value.
12305 * opts.h (parse_no_sanitize_attribute): Declare.
12306 * tree.c (sanitize_flags_p): New function.
12307 * tree.h: Declared here.
12308 * tsan.c: Use sanitize_flags_p.
12309 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12310 (instrument_mem_ref): Likewise.
12311 (instrument_bool_enum_load): Likewise.
12312 (do_ubsan_in_current_function): Remove the function.
12313 (pass_ubsan::execute): Use sanitize_flags_p.
12314 * ubsan.h: Remove do_ubsan_in_current_function
12315 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12316 (dump_function_to_file): Use it here.
12318 2017-06-13 Martin Jambor <mjambor@suse.cz>
12320 PR tree-optimization/80803
12321 PR tree-optimization/81063
12322 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12323 (propagate_subaccesses_across_link): Enqueue subtree whenever
12324 necessary instead of relying on the caller.
12326 2017-06-13 Martin Jambor <mjambor@suse.cz>
12328 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12329 that have a first_link.
12330 (sort_and_splice_var_accesses): Do not check first_link before
12332 (subtree_mark_written_and_enqueue): Likewise.
12333 (propagate_all_subaccesses): Likewise and do not stop at first
12334 parent with a first_link.
12336 2017-06-13 Martin Jambor <mjambor@suse.cz>
12338 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12341 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12343 * match.pd: New pattern.
12345 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12347 * tree-vrp.c (is_masked_range_test): New function.
12348 (register_edge_assert_for): Determine ranges for
12351 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12353 PR tree-optimization/67328
12354 * fold-const.c (maskable_range_p): New function.
12355 (build_range_check): Generate bittests if possible.
12357 2017-06-13 Martin Liska <mliska@suse.cz>
12359 * gimple-pretty-print.c (dump_probability): Add new argument.
12360 (dump_edge_probability): Dump both probability and count.
12361 (dump_gimple_label): Likewise.
12362 (dump_gimple_bb_header): Likewise.
12364 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12367 * config/avr/avr-devices.c: Fix indentation.
12368 * config/avr/gen-avr-mmcu-specs.c: Dito.
12370 2017-06-13 Richard Biener <rguenther@suse.de>
12372 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12373 instead get vector type from stmt_info.
12374 (vectorizable_reduction): Adjust. Remove dead code.
12376 2017-06-13 Richard Biener <rguenther@suse.de>
12378 PR middle-end/81065
12379 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12380 case of C * (x * C2 + C3).
12381 (fold_addr_of_array_ref_difference): Properly fold index difference.
12383 2017-06-12 David S. Miller <davem@davemloft.net>
12386 * config/sparc/sparc.md (return expander): Emit frame blockage if
12387 function uses alloca.
12389 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12391 * combine.c (make_field_assignment): Check len rather than the mode
12392 precision when calling force_to_mode.
12394 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12396 Support multilibs and devices that see flash in RAM address range.
12399 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12400 (avr_mcu_t) <flash_pm_offset>: New field.
12401 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12402 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12403 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12404 (AVR_TINY_PM_OFFSET): Remove macro.
12405 * config/avr/avr.opt (-mshort-calls): New option.
12406 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12407 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12408 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12409 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12410 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12411 instead of avr_arch->have_jmp_call.
12412 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12413 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12414 avr_arch->flash_pm_offset to define.
12415 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12416 new field flash_pm_offset. Add entry for avrxmega3.
12417 (avr_texinfo): Add entry for avrxmega3.
12418 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12419 attiny212, attiny214,
12420 attiny412, attiny414, attiny416, attiny417,
12421 attiny814, attiny816, attiny817,
12422 attiny1614, attiny1616, attiny1617,
12423 attiny3214, attiny3216, attiny3217.
12424 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12425 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12426 (avr_print_operand_address) [AVR_TINY]: Same.
12427 (avr_asm_init_sections) <readonly_data_section>: Only patch
12428 callback if avr_arch->flash_pm_offset = 0.
12429 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12430 for rodata if avr_arch->flash_pm_offset != 0.
12431 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12432 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12433 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12434 (m_options): Append opt_rcall.
12435 (m_dirnames): Append dir_rcall.
12436 * config/avr/t-multilib: Regenerate.
12438 * configure.ac [target=avr]: Check whether avrxmega3 default
12439 linker description file works as needed.
12440 * configure: Regenerate.
12441 * doc/avr-mmcu.texi: Regenerate.
12442 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12443 <__AVR_ARCH__>: Document avrxmega3 and 103.
12444 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12445 <__AVR_SHORT_CALLS__>: Document it.
12446 <__AVR_PM_BASE_ADDRESS__>: Document it.
12447 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12448 (AVR Variable Attributes) <progmem>: Document this is
12449 not needed for avrxmega3.
12450 (AVR Named Address Spaces) <__flash>: Dito.
12452 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12454 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12456 2017-06-12 Doug Rupp <rupp@adacore.com>
12458 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12459 Append vxworks-stdint.h to the tm_file list.
12460 * config/vxworks-stdint.h: New file.
12462 2017-06-12 Martin Liska <mliska@suse.cz>
12464 PR tree-optimization/81041
12465 * tree-profile.c (gimple_gen_ic_func_profiler):
12466 Create an extra BB in profile-generate
12467 (gimple_gen_time_profiler): Likewise.
12469 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12471 PR tree-optimization/81003
12472 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12473 (update_range_test): Use it instead of force_gimple_operand_gsi.
12475 2017-06-12 Richard Biener <rguenther@suse.de>
12477 PR tree-optimization/81053
12478 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12479 with backedge value not defined in loop. Simplify def stmt
12482 2017-06-11 Tom de Vries <tom@codesourcery.com>
12485 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12487 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12488 nvptx_cannot_force_const_mem.
12490 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12492 * opts.c (finish_options): Move test for flag_split_stack after
12493 it has been initialized.
12495 2017-06-11 Jason Merrill <jason@redhat.com>
12497 * tree.h (id_equal): New.
12498 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12499 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12500 instead of strcmp of IDENTIFIER_POINTER.
12502 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12504 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12505 (mark_all_inlined_calls_cdtor): Fix formating.
12506 (inline_transform): Rescale profile before inlining.
12508 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12510 * cgraph.h (cgraph_edge::clone): Update prototype.
12511 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12512 (cgraph_node::create_clone): Update.
12513 (cgraph_node::create_version_clone): Update.
12514 * tree-inline.c (copy_bb): Update.
12515 (expand_call_inline): Update.
12517 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12519 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12520 factored out from ...
12521 (rs6000_emit_prologue): ... here.
12523 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12525 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12526 factored out from ...
12527 (rs6000_emit_prologue): ... here.
12529 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12531 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12533 (handle_missing_profiles): Fix computation of tp_first_run.
12534 (counts_to_freqs): Do not touch freqs when count is 0.
12536 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12538 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12541 2017-06-10 Tom de Vries <tom@codesourcery.com>
12543 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12544 attributes): Document signal effective target.
12546 2017-06-10 Tom de Vries <tom@codesourcery.com>
12548 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12549 Document effective target stack_size.
12551 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12553 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12554 to the edit_context if they can be auto-applied.
12556 2017-06-9 Ian Lance Taylor <iant@golang.org>
12558 * opts.c (finish_options): If -fsplit-stack, disable implicit
12559 -forder-blocks-and-partition.
12560 * doc/invoke.texi (Optimize Options): Document that when using
12561 -fsplit-stack -forder-blocks-and-partition is not implicitly
12564 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12566 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12567 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12568 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12569 * builtins.def (abort, trap, unreachable): Declare cold.
12570 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12571 * tree-core.h (ECF_COLD): New.
12572 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12573 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12575 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12577 * predict.c (unlikely_executed_stmt_p): Cleanup.
12579 2017-06-09 Richard Biener <rguenther@suse.de>
12581 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12582 model if the ref is always written to.
12584 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12586 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12588 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12590 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12593 2017-06-09 Tom de Vries <tom@codesourcery.com>
12596 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12597 "target cannot support label values" when encountering LABEL_REF.
12599 2017-06-09 Martin Liska <mliska@suse.cz>
12601 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12602 (gimple_gen_ic_func_profiler): Emit direct comparison
12603 of __gcov_indirect_call_callee with NULL.
12604 (gimple_gen_time_profiler): Change probability from
12605 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12607 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12609 * profile.c (edge_gcov_counts): Turn to pointer.
12610 (compute_branch_probabilities, compute_branch_probabilities): Update.
12611 (branch_prob): Do not clear edge_gcov_count.
12612 * profile.h (edge_gcov_counts): Turn to pointer.
12613 (edge_gcov_count): Update.
12615 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12617 * gimple.h (gimple_check_failed): Mark cold.
12619 2017-06-09 Richard Biener <rguenther@suse.de>
12621 PR tree-optimization/66623
12622 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12623 refactor check_reduction into two parts, properly computing
12624 whether we have to check reduction validity for outer loop
12627 2017-06-09 Richard Biener <rguenther@suse.de>
12629 PR tree-optimization/79483
12630 * graphite-scop-detection.c (order): New global.
12631 (get_order): Compute bb to order mapping that satisfies code
12632 generation constraints.
12633 (cmp_pbbs): New helper.
12634 (build_scops): Start domwalk at entry block, sort generated
12637 2017-06-09 Richard Biener <rguenther@suse.de>
12639 PR middle-end/81007
12640 * ipa-polymorphic-call.c
12641 (ipa_polymorphic_call_context::restrict_to_inner_class):
12642 Skip FIELD_DECLs with error_mark_node type.
12643 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12646 2017-06-09 Martin Liska <mliska@suse.cz>
12648 * predict.c (struct branch_predictor): New struct.
12649 (test_prediction_value_range): New test.
12650 (predict_c_tests): New function.
12651 * selftest-run-tests.c (selftest::run_tests): Run the function.
12652 * selftest.h: Declare new tests.
12654 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12657 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12658 gen_add3_insn did not fail.
12659 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12660 r0, construct that number in a temporary reg and add that reg to r0.
12661 If asked to put the result in r0 as well, fail.
12663 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12665 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12666 for early expansion of vec_eqv.
12668 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12670 PR middle-end/81005
12671 * ubsan.c (instrument_null): Avoid pointless code temporary.
12672 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12674 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12678 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12681 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12684 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12686 * predict.c (unlikely_executed_edge_p): Move ahead.
12687 (probably_never_executed_edge_p): Use it.
12689 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12691 PR middle-end/79988
12692 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12693 gimple_call_builtin_p call.
12695 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12697 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12698 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12699 rtl_check_failed_type2, rtl_check_failed_code1,
12700 rtl_check_failed_code2, rtl_check_failed_code_mode,
12701 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12702 rtvec_check_failed_bounds, rtl_check_failed_flag,
12703 _fatal_insn_not_found, _fatal_insn): Likewise.
12704 * tree.h (tree_contains_struct_check_failed,
12705 tree_check_failed, tree_not_check_failed,
12706 tree_class_check_failed, tree_range_check_failed,
12707 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12708 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12709 tree_operand_check_failed, omp_clause_check_failed,
12710 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12713 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12715 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12716 flag_branch_probabilities.
12717 * ipa-inline.c (edge_badness): Likewise.
12718 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12719 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12720 * predict.c (maybe_hot_frequency_p): Likewise.
12721 (probably_never_executed): Likewise.
12722 * sched-ebb.c (schedule_ebbs): Likewise.
12723 * sched-rgn.c (find_single_block_region): Likewise.
12724 * tracer.c (tail_duplicate): Likewise.
12726 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12728 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12729 longer requires x_flag_profile_use.
12731 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12733 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12734 instead of flag_reorder_blocks_and_partition.
12735 * dbxout.c (dbxout_function_end): Likewise.
12736 * dwarf2out.c (gen_subprogram_die): Likewise.
12737 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12738 * hw-doloop.c (reorg_loops): Likewise.
12739 * varasm.c (assemble_start_function,
12740 assemble_end_function): Likewise.
12741 (decide_function_section): Do not check for
12742 flag_reorder_blocks_and_partition.
12744 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12746 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12748 (chkp_get_hard_register_fake_addr_expr): Ditto.
12749 (chkp_build_addr_expr): Add check for hard reg case.
12750 (chkp_parse_array_and_component_ref): Ditto.
12751 (chkp_find_bounds_1): Ditto.
12752 (chkp_process_stmt): Don't generate bounds store for
12755 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12757 * predict.c (maybe_hot_bb_p): Do not check profile status.
12758 (maybe_hot_edge_p): Likewise.
12759 (probably_never_executed): Check for zero counts even if profile
12761 (unlikely_executed_edge_p): New function.
12762 (unlikely_executed_stmt_p): New function.
12763 (unlikely_executed_bb_p): New function.
12764 (set_even_probabilities): Use unlikely predicates.
12765 (combine_predictions_for_bb): Likewise.
12766 (predict_paths_for_bb): Likewise.
12767 (predict_paths_leading_to_edge): Likewise.
12768 (determine_unlikely_bbs): New function.
12769 (estimate_bb_frequencies): Use it.
12770 (compute_function_frequency): Use zero counts even if profile is
12772 * profile-count.h: Fix typo.
12774 2017-08-08 Julia Koval <julia.koval@intel.com>
12776 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12777 _mm512_mask_cvtsepi16_storeu_epi8,
12778 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12779 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12780 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12781 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12782 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12783 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12784 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12785 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12786 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12787 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12788 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12789 __builtin_ia32_pmovuswb256mem_mask,
12790 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12791 __builtin_ia32_pmovwb512mem_mask): New builtins.
12793 2017-08-08 Julia Koval <julia.koval@intel.com>
12795 PR target/73350,80862
12796 * config/i386/subst.md (round): Fix round pattern.
12797 * config/i386/i386.c (ix86_erase_embedded_rounding):
12798 Fix erasing rounding for the fixed pattern.
12800 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12802 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12804 2017-06-08 Martin Liska <mliska@suse.cz>
12806 PR gcov-profile/80911
12807 * gcov.c (block_info::block_info): New constructor.
12809 2017-06-07 Carl Love <cel@us.ibm.com>
12811 * config/rs6000/rs6000-c: The return type of the following
12812 built-in functions was implemented as int not long long. Fix sign
12813 of return value for the unsigned version of vec_mulo and vec_mule.
12814 vector unsigned long long vec_bperm (vector unsigned long long,
12815 vector unsigned char)
12816 vector signed long long vec_mule (vector signed int,
12818 vector unsigned long long vec_mule (vector unsigned int,
12819 vector unsigned int)
12820 vector signed long long vec_mulo (vector signed int,
12822 vector unsigned long long vec_mulo (vector unsigned int,
12823 vector unsigned int)
12824 * doc/extend.texi: Fix the documentation for the built-in
12827 2017-06-07 Carl Love <cel@us.ibm.com>
12830 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12833 2017-06-07 Carl Love <cel@us.ibm.com>
12835 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12836 support, Generate doublehv for signed int/float for BE case only.
12838 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12840 * doc/invoke.texi (mcx16): Rewrite.
12842 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12844 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12845 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12846 *mov<mode>_softfloat, and an anonymous splitter): Use
12847 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12849 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12851 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12853 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12854 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12855 (rs6000_debug_reg_global): Adjust.
12856 (rs6000_init_hard_regno_mode_ok): Adjust.
12857 (rs6000_dbx_register_number): Adjust.
12858 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12859 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12860 Remove SPE_ACC and SPEFSCR.
12861 (REG_ALLOC_ORDER): Ditto.
12862 (FRAME_POINTER_REGNUM): Change to 111.
12863 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12864 (REG_CLASS_NAMES): Ditto.
12865 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12866 (REGISTER_NAMES): Ditto.
12867 (ADDITIONAL_REG_NAMES): Ditto.
12868 (rs6000_reg_names): Ditto.
12869 * config/rs6000/rs6000.md: Renumber some register number
12872 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12874 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12876 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12877 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12879 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12880 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12881 Delete the SPE high registers.
12882 (REG_ALLOC_ORDER): Ditto.
12883 (enum reg_class): Remove SPE_HIGH_REGS.
12884 (REG_CLASS_NAMES): Ditto.
12885 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12886 (REGISTER_NAMES): Ditto.
12887 (rs6000_reg_names): Ditto.
12888 * doc/tm.texi.in: Remove SPE as example.
12889 * doc/tm.texi: Regenerate.
12891 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12893 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12894 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12895 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12896 * config/rs6000/rs6000.md (type): Remove "brinc".
12898 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12900 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12901 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12902 * config/rs6000/linuxspe.h: Delete file.
12903 * config/rs6000/rs6000.md: Don't include spe.md.
12904 * config/rs6000/spe.h: Delete file.
12905 * config/rs6000/spe.md: Delete file.
12906 * config/rs6000/t-rs6000: Remove spe.md.
12908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12910 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12911 (reg_or_none500mem_operand): Delete.
12912 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12913 instead of reg_or_none500mem_operand.
12915 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12917 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12918 handling of SPE flags.
12919 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12921 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12923 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12925 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12926 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12927 paired_divv2sf3): Similar.
12928 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12929 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12930 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12932 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12933 Rename the paired_* instruction patterns.
12934 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12936 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12937 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12938 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12939 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12940 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12941 PAIRED_VECTOR_MODE.
12942 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12943 (spe_func_has_64bit_regs_p): Delete.
12944 (spe_expand_predicate_builtin): Delete.
12945 (spe_expand_evsel_builtin): Delete.
12946 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12947 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12948 (invalid_e500_subreg): Delete.
12949 (rs6000_legitimize_address): Always force_reg op2 as well, for
12950 paired single memory accesses.
12951 (rs6000_member_type_forces_blk): Delete.
12952 (rs6000_spe_function_arg): Delete.
12953 (rs6000_expand_unop_builtin): Delete SPE handling.
12954 (rs6000_expand_binop_builtin): Ditto.
12955 (spe_expand_stv_builtin): Delete.
12956 (bdesc_2arg_spe): Delete.
12957 (spe_expand_builtin): Delete.
12958 (spe_expand_predicate_builtin): Delete.
12959 (spe_expand_evsel_builtin): Delete.
12960 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12961 (spe_init_builtins): Delete.
12962 (spe_func_has_64bit_regs_p): Delete.
12963 (savres_routine_name): Delete "info" parameter. Adjust callers.
12964 (rs6000_emit_stack_reset): Ditto.
12965 (rs6000_dwarf_register_span): Delete.
12966 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12967 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12968 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12970 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12972 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12973 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12974 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12975 mulv2sf3, divv2sf3): Delete expanders.
12977 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12979 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12981 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12983 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12984 * config/rs6000/rs6000.c: Ditto.
12986 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12988 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12989 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12990 comparison_operator.
12992 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12994 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12995 * config/rs6000/rs6000.opt: Ditto.
12996 * config/rs6000/t-rtems: Ditto.
12998 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13000 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13001 TARGET_E500_SINGLE by 0, simplify.
13002 * config/rs6000/rs6000.c: Ditto.
13003 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13004 (spe_build_register_parallel): Delete.
13005 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13006 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13007 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13008 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13009 (E500_CONVERT): Delete.
13010 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13012 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13014 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13015 * config/rs6000/dfp.md: Ditto.
13016 (negdd2, *negdd2_fpr): Merge.
13017 (absdd2, *absdd2_fpr): Merge.
13018 (negtd2, *negtd2_fpr): Merge.
13019 (abstd2, *abstd2_fpr): Merge.
13020 * config/rs6000/e500.h: Delete file.
13021 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13022 TARGET_FPRS by 1 and simplify.
13023 * config/rs6000/rs6000-c.c: Ditto.
13024 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13025 TARGET_DF_SPE by 0.
13026 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13028 * config/rs6000/rs6000.md: Ditto.
13029 (floatdidf2, *floatdidf2_fpr): Merge.
13030 (move_from_CR_gt_bit): Delete.
13031 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13032 (E500_CR_IOR_COMPARE): Delete.
13033 (All patterns that require !TARGET_FPRS): Delete.
13034 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13036 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13038 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13040 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13042 * graphds.c (add_edge): Intitialize edge's attached data.
13043 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13044 pointer parameter. Call pointed function on each edge during
13045 graph traversing. Skip traversing the edge when the function
13047 (graphds_dfs, graphds_scc): Ditto.
13048 (for_each_edge): New parameter. Pass the new parameter to callback
13050 * graphds.h (skip_edge_callback): New function pointer type.
13051 (graphds_dfs, graphds_scc): New function pointer parameter.
13052 (graphds_edge_callback, for_each_edge): New parameter.
13054 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13056 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13057 out code checking if runtime alias check is possible to below ...
13058 Call the new function.
13059 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13060 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13062 2017-06-07 Marek Polacek <polacek@redhat.com>
13065 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13066 TYPE_OVERFLOW_WRAPS check.
13068 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13070 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13071 if versioning is required.
13072 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13073 peeling with the check for versioning.
13075 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13077 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13078 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13079 Set true to new parameter if new ssa variable is defined.
13080 (vect_gen_vector_loop_niters): Refactor. Set range information
13081 for the new vector loop bound variable.
13082 (vect_do_peeling): Ditto.
13084 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13086 * tree-affine.c (ssa.h): Include header file.
13087 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13088 has wrapping overflow behavior.
13090 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13092 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13094 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13096 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13097 (tree_to_aff_combination): ... here.
13099 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13101 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13102 reg_pressure model function.
13103 (ivopts_global_cost_for_size): Delete.
13104 (determine_set_costs, iv_ca_recount_cost): Call new model function
13105 ivopts_estimate_reg_pressure.
13107 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13109 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13110 expensive than udiv. Remove floating point cases from mod.
13112 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13114 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13115 Increase idiv cost.
13117 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13119 * config/aarch64/aarch64.md
13120 (copysignsf3): Fix mask generation.
13122 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13124 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13126 (class dump_manager): Add register_dumps method.
13127 * dumpfile.c: Include langhooks.h.
13128 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13129 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13130 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13131 (dump_manager::dump_register): Start with 512 entries instead of 32.
13132 (dump_manager::register_dumps): New method.
13133 * toplev.c (general_init): Instead of invoking register_dumps
13134 langhook, invoke register_dumps method on the dump manager.
13135 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13138 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13140 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13141 Say that # requires an associated define_split to exist, and that
13142 the define_split must be suitable for use after register allocation.
13144 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13146 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13147 (compute_outgoing_frequencies): Also initialize zero counts.
13148 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13149 around loops; preserve more of profile when nothing changes.
13151 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13153 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13155 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13156 * config/arm/arm-cpu-cdata.h: Regenerate.
13157 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13158 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13159 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13160 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13161 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13163 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13165 * config/arm/t-rmprofile: Likewise.
13166 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13168 2017-06-06 David S. Miller <davem@davemloft.net>
13171 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13172 blockage if function uses alloca.
13174 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13176 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13177 New "uid" fields to hold pretty-print IDs of group and ref.
13178 Memory references are now identified as <group_id>:<ref_id>
13179 instead of using [random] addresses.
13180 (dump_mem_details): Simplify, no functional change.
13181 (dump_mem_ref): Simplify and make output more concise.
13182 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13183 (find_or_create_group): Initialize group uid.
13184 (record_ref): Initialize ref uid. Improve debug output.
13185 (prune_group_by_reuse, should_issue_prefetch_p,)
13186 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13187 (mark_nontemporal_store, determine_loop_nest_reuse):
13188 Improve debug output.
13190 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13192 * dbgcnt.def (prefetch): New debug counter.
13193 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13194 (schedule_prefetches): Stop issueing prefetches if debug counter
13197 2017-06-06 Tom de Vries <tom@codesourcery.com>
13199 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13200 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13202 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13204 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13205 Use aarch64_reg_or_zero predicate for operand 4.
13206 (aarch64_compare_and_swap<mode> define_insn_and_split):
13207 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13208 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13210 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13212 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13213 (arm_compute_save_core_reg_mask): This.
13214 (thumb1_compute_save_reg_mask): Rename into ...
13215 (thumb1_compute_save_core_reg_mask): This.
13216 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13217 (arm_compute_frame_layout): Likewise.
13219 2017-06-06 Richard Biener <rguenther@suse.de>
13221 PR tree-optimization/80974
13222 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13223 keep or clear leaders SSA info.
13225 2017-06-06 Tom de Vries <tom@codesourcery.com>
13227 * config/nvptx/nvptx.c (split_mode_p): New function.
13228 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13230 2017-06-06 Tom de Vries <tom@codesourcery.com>
13232 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13234 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13237 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13240 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13242 * shrink-wrap.c (handle_simple_exit): Update profile.
13243 (try_shrink_wrapping): Upate profile.
13245 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13247 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13248 (tree_guess_outgoing_edge_probabilities): New.
13249 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13250 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13252 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13254 * ipa-split.c (split_function): Initialize return bb profile.
13256 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13258 * profile.c (compute_branch_probabilities): Also initialize
13259 EXIT_BLOCK profile.
13261 2017-06-06 Richard Biener <rguenther@suse.de>
13263 PR tree-optimization/80928
13264 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13265 (vect_analyze_loop_operations): Properly guard analysis for
13267 (vect_transform_loop): Likewise.
13268 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13269 (vect_model_induction_cost): Do not cost for pure SLP.
13270 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13271 of induction in inner loop vectorization.
13272 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13273 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13274 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13276 (vect_analyze_slp_cost_1): Cost induction.
13277 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13278 (vect_get_slp_vect_defs): Likewise.
13279 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13280 (vect_transform_stmt): Handle SLP reductions.
13281 * tree-vectorizer.h (vectorizable_induction): Adjust.
13283 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13285 * config/rs6000/rs6000.c (make_resolver_func): Update
13286 init_lowered_empty_function call.
13288 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13290 * doc/invoke.texi: Document the -fprofile-abs-path option.
13291 * common.opt (fprofile-abs-path): New option.
13292 * gcov-io.h (gcov_write_filename): Declare.
13293 * gcov-io.c (gcov_write_filename): New function.
13294 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13295 * profile.c (output_location): Likewise.
13297 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13299 * shring-wrap.c: Revert accidental commit.
13301 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13303 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13305 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13307 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13309 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13310 profile in callgraph edge.
13311 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13312 (apply_scale): Likewise.
13313 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13314 Also scale profile when inlining function with zero profile.
13315 (initialize_cfun): Update exit block profile even when it is zero.
13316 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13317 when profile is read.
13319 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13321 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13322 (CLONE_*): New constants to define the processors we can generate
13323 code for with the target_clone attribute.
13324 (rs6000_clone_map): New array to identify which clone processors
13325 the current program is running on.
13326 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13327 target_clone attribute.
13328 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13329 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13330 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13331 (cpu_expand_builtin): Add support for target_clone attribute.
13332 (rs6000_valid_attribute_p): Allow "default" attribute.
13333 (get_decl_name): New debug function to simplify printing the
13334 current function name in debugging statements.
13335 (rs6000_clone_priority): New functions to support the target_clone
13336 attribute, and be able to generate code to switch between ISA 2.05
13337 through ISA 3.0 (power6 through power9).
13338 (rs6000_compare_version_priority): Likewise.
13339 (rs6000_get_function_versions_dispatcher): Likewise.
13340 (make_resolver_func): Likewise.
13341 (add_condition_to_bb): Likewise.
13342 (dispatch_function_versions): Likewise.
13343 (rs6000_generate_version_dispatcher_body): Likewise.
13344 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13345 (fusion_gpr_load_p): Fix a spacing issue.
13346 * doc/extend.texi (Common Function Attributes): Document that the
13347 PowerPC supports the target_clone attribute.
13349 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13351 * config/arm/arm.h: explain F symbol found in description of ARM
13352 register allocation in its legend.
13354 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13356 * config/mips/frame-header-opt.c: Include profile-count.h.
13357 * config/riscv/riscv.c: Include profile-count.h
13359 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13361 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13363 (sm_set_flag_if_changed): Add bbs field.
13364 (execute_sm_if_changed_flag_set): Pass BBS.
13365 (execute_sm): Update.
13367 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13369 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13372 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13374 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13375 (peephole2): New peephole2 to emit the above.
13376 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13378 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13380 * config/aarch64/aarch64.c (define_peephole2 above
13381 *sub_<shift>_<mode>): New peephole.
13383 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13385 * config/i386/i386.c (make_resolver_func): Update.
13386 * Makefile.in: Add profile-count.h and profile-count.o
13387 * auto-profile.c (afdo_indirect_call): Update to new API.
13388 (afdo_set_bb_count): Update.
13389 (afdo_propagate_edge): Update.
13390 (afdo_propagate_circuit): Update.
13391 (afdo_calculate_branch_prob): Update.
13392 (afdo_annotate_cfg): Update.
13393 * basic-block.h: Include profile-count.h
13394 (struct edge_def): Turn count to profile_count.
13395 (struct basic_block_def): Likewie.
13396 (REG_BR_PROB_BASE): Move to profile-count.h
13397 (RDIV): Move to profile-count.h
13398 * bb-reorder.c (max_entry_count): Turn to profile_count.
13399 (find_traces): Update.
13400 (rotate_loop):Update.
13401 (connect_traces):Update.
13402 (sanitize_hot_paths):Update.
13403 * bt-load.c (migrate_btr_defs): Update.
13404 * cfg.c (RDIV): Remove.
13405 (init_flow): Use alloc_block.
13406 (alloc_block): Uninitialize count.
13407 (unchecked_make_edge): Uninitialize count.
13408 (check_bb_profile): Update.
13409 (dump_edge_info): Update.
13410 (dump_bb_info): Update.
13411 (update_bb_profile_for_threading): Update.
13412 (scale_bbs_frequencies_int): Update.
13413 (scale_bbs_frequencies_gcov_type): Update.
13414 (scale_bbs_frequencies_profile_count): New.
13415 * cfg.h (update_bb_profile_for_threading): Update.
13416 (scale_bbs_frequencies_profile_count): Declare.
13417 * cfgbuild.c (compute_outgoing_frequencies): Update.
13418 (find_many_sub_basic_blocks): Update.
13419 * cfgcleanup.c (try_forward_edges): Update.
13420 (try_crossjump_to_edge): Update.
13421 * cfgexpand.c (expand_gimple_tailcall): Update.
13422 (construct_exit_block): Update.
13423 * cfghooks.c (verify_flow_info): Update.
13424 (dump_bb_for_graph): Update.
13425 (split_edge): Update.
13426 (make_forwarder_block): Update.
13427 (duplicate_block): Update.
13428 (account_profile_record): Update.
13429 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13430 (get_estimated_loop_iterations): Update.
13431 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13432 (single_likely_exit): Update.
13433 * cfgloopmanip.c (scale_loop_profile): Update.
13435 (set_zero_probability): Update.
13436 (lv_adjust_loop_entry_edge): Update.
13437 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13438 (purge_dead_edges): Update.
13439 (rtl_account_profile_record): Update.
13440 * cgraph.c (cgraph_node::create): Uninitialize count.
13441 (symbol_table::create_edge): Uninitialize count.
13442 (cgraph_update_edges_for_call_stmt_node): Update.
13443 (cgraph_edge::dump_edge_flags): Update.
13444 (cgraph_node::dump): Update.
13445 (cgraph_edge::maybe_hot_p): Update.
13446 * cgraph.h: Include profile-count.h
13447 (create_clone), create_edge, create_indirect_edge): Update.
13448 (cgraph_node): Turn count to profile_count.
13449 (cgraph_edge0: Likewise.
13450 (make_speculative, clone): Update.
13451 (create_edge): Update.
13452 (init_lowered_empty_function): Update.
13453 * cgraphclones.c (cgraph_edge::clone): Update.
13454 (duplicate_thunk_for_node): Update.
13455 (cgraph_node::create_clone): Update.
13456 * cgraphunit.c (cgraph_node::analyze): Update.
13457 (cgraph_node::expand_thunk): Update.
13458 * final.c (dump_basic_block_info): Update.
13459 * gimple-streamer-in.c (input_bb): Update.
13460 * gimple-streamer-out.c (output_bb): Update.
13461 * graphite.c (print_global_statistics): Update.
13462 (print_graphite_scop_statistics): Update.
13463 * hsa-brig.c: Include basic-block.h.
13464 * hsa-dump.c: Include basic-block.h.
13465 * hsa-gen.c (T sum_slice): Update.
13466 (convert_switch_statements):Update.
13467 * hsa-regalloc.c: Include basic-block.h.
13468 * ipa-chkp.c (chkp_produce_thunks): Update.
13469 * ipa-cp.c (struct caller_statistics): Update.
13470 (init_caller_stats): Update.
13471 (gather_caller_stats): Update.
13472 (ipcp_cloning_candidate_p): Update.
13473 (good_cloning_opportunity_p): Update.
13474 (get_info_about_necessary_edges): Update.
13475 (dump_profile_updates): Update.
13476 (update_profiling_info): Update.
13477 (update_specialized_profile): Update.
13478 (perhaps_add_new_callers): Update.
13479 (decide_about_value): Update.
13480 (ipa_cp_c_finalize): Update.
13481 * ipa-devirt.c (struct odr_type_warn_count): Update.
13482 (struct decl_warn_count): Update.
13483 (struct final_warning_record): Update.
13484 (possible_polymorphic_call_targets): Update.
13485 (ipa_devirt): Update.
13486 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13487 * ipa-icf.c (sem_function::merge): Update.
13488 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13489 * ipa-inline.c (compute_uninlined_call_time): Update.
13490 (compute_inlined_call_time): Update.
13491 (want_inline_small_function_p): Update.
13492 (want_inline_self_recursive_call_p): Update.
13493 (edge_badness): Update.
13494 (lookup_recursive_calls): Update.
13495 (recursive_inlining): Update.
13496 (inline_small_functions): Update.
13497 (dump_overall_stats): Update.
13498 (dump_inline_stats): Update.
13499 * ipa-profile.c (ipa_profile_generate_summary): Update.
13500 (ipa_propagate_frequency): Update.
13501 (ipa_profile): Update.
13502 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13503 * ipa-utils.c (ipa_merge_profiles): Update.
13504 * loop-doloop.c (doloop_modify): Update.
13505 * loop-unroll.c (report_unroll): Update.
13506 (unroll_loop_runtime_iterations): Update.
13507 * lto-cgraph.c (lto_output_edge): Update.
13508 (lto_output_node): Update.
13509 (input_node): Update.
13510 (input_edge): Update.
13511 (merge_profile_summaries): Update.
13512 * lto-streamer-in.c (input_cfg): Update.
13513 * lto-streamer-out.c (output_cfg): Update.
13514 * mcf.c (create_fixup_graph): Update.
13515 (adjust_cfg_counts): Update.
13516 (sum_edge_counts): Update.
13517 * modulo-sched.c (sms_schedule): Update.
13518 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13519 * predict.c (maybe_hot_count_p): Update.
13520 (probably_never_executed): Update.
13521 (dump_prediction): Update.
13522 (combine_predictions_for_bb): Update.
13523 (propagate_freq): Update.
13524 (handle_missing_profiles): Update.
13525 (counts_to_freqs): Update.
13526 (rebuild_frequencies): Update.
13527 (force_edge_cold): Update.
13528 * predict.h: Include profile-count.h
13529 (maybe_hot_count_p, counts_to_freqs): UPdate.
13530 * print-rtl-function.c: Do not include cfg.h
13531 * print-rtl.c: Include basic-block.h
13532 * profile-count.c: New file.
13533 * profile-count.h: New file.
13534 * profile.c (is_edge_inconsistent): Update.
13535 (correct_negative_edge_counts): Update.
13536 (is_inconsistent): Update.
13537 (set_bb_counts): Update.
13538 (read_profile_edge_counts): Update.
13539 (compute_frequency_overlap): Update.
13540 (compute_branch_probabilities): Update; Initialize and deinitialize
13542 (branch_prob): Update.
13543 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13544 (edge_gcov_count): New.
13545 (bb_gcov_count): New.
13546 * shrink-wrap.c (try_shrink_wrapping): Update.
13547 * tracer.c (better_p): Update.
13548 * trans-mem.c (expand_transaction): Update.
13549 (ipa_tm_insert_irr_call): Update.
13550 (ipa_tm_insert_gettmclone_call): Update.
13551 * tree-call-cdce.c: Update.
13552 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13553 (gimple_duplicate_sese_tail): Update.
13554 (gimple_account_profile_record): Update.
13555 (execute_fixup_cfg): Update.
13556 * tree-inline.c (copy_bb): Update.
13557 (copy_edges_for_bb): Update.
13558 (initialize_cfun): Update.
13559 (freqs_to_counts): Update.
13560 (copy_cfg_body): Update.
13561 (expand_call_inline): Update.
13562 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13563 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13564 (try_unroll_loop_completely): Update.
13565 (try_peel_loop): Update.
13566 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13567 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13568 * tree-ssa-loop-split.c (connect_loops): Update.
13569 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13570 * tree-ssa-reassoc.c (branch_fixup): Update.
13571 * tree-ssa-tail-merge.c (replace_block_by): Update.
13572 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13573 (compute_path_counts): Update.
13574 (update_profile): Update.
13575 (recompute_probabilities): Update.
13576 (update_joiner_offpath_counts): Update.
13577 (estimated_freqs_path): Update.
13578 (freqs_to_counts_path): Update.
13579 (clear_counts_path): Update.
13580 (ssa_fix_duplicate_block_edges): Update.
13581 (duplicate_thread_path): Update.
13582 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13583 (struct switch_conv_info): Update.
13584 * tree-tailcall.c (decrease_profile): Update.
13585 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13586 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13587 * value-prof.c (check_counter): Update.
13588 (gimple_divmod_fixed_value): Update.
13589 (gimple_mod_pow2): Update.
13590 (gimple_mod_subtract): Update.
13591 (gimple_ic_transform): Update.
13592 (gimple_stringop_fixed_value): Update.
13593 * value-prof.h (gimple_ic): Update.
13595 2017-06-02 Carl Love <cel@us.ibm.com>
13597 * config/rs6000/rs6000-c: Add support for built-in functions
13598 vector double vec_doublee (vector signed int);
13599 vector double vec_doublee (vector unsigned int);
13600 vector double vec_doublee (vector float);
13601 vector double vec_doubleh (vector signed int);
13602 vector double vec_doubleh (vector unsigned int);
13603 vector double vec_doubleh (vector float);
13604 vector double vec_doublel (vector signed int);
13605 vector double vec_doublel (vector unsigned int);
13606 vector double vec_doublel (vector float);
13607 vector double vec_doubleo (vector signed int);
13608 vector double vec_doubleo (vector unsigned int);
13609 vector double vec_doubleo (vector float);.
13610 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13611 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13613 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13614 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13615 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13617 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13618 vec_doublel, vec_doubleh.
13619 * doc/extend.texi: Update the built-in documentation file for the
13620 new built-in functions.
13622 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13625 * ipa-inline-analysis.c (free_growth_caches): Set
13626 edge_removal_hook_holder to NULL after removing it.
13628 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13630 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13631 comparision with zero.
13633 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13634 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13635 for early expansion of vec_min and vec_max builtins.
13636 (builtin_function_type): Add min/max unsigned variants to those
13637 identified as having unsigned arguments.
13639 2017-06-02 Olivier Hainque <hainque@adacore.com>
13641 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13643 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13645 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13646 Use VALL_F16 iterator rather than VALL.
13648 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13650 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13651 Emit CBNZ inside loop when doing a strong exchange and comparing
13652 against zero. Generate the CC flags after the loop.
13654 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13656 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13657 (dl_section_ref): New.
13658 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13659 On AIX, append an expression to subtract the size of the
13660 section length to dl_section_ref.
13662 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13664 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13665 for early expansion of vector absolute builtins.
13667 2017-06-02 Richard Biener <rguenther@suse.de>
13669 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13670 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13672 2017-06-02 Richard Biener <rguenther@suse.de>
13674 PR tree-optimization/80948
13675 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13676 stmt order as well.
13678 2017-06-02 Richard Biener <rguenther@suse.de>
13680 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13682 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13683 for inductions as relevant.
13685 2017-06-02 Richard Biener <rguenther@suse.de>
13687 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13688 (vectorizable_induction): ... this. Remove dead code.
13690 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13692 * builtins. (expand_builtin_alloca): Remove second parameter and
13693 infer its value from the first parameter instead.
13694 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13696 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13698 PR rtl-optimization/80903
13699 * loop-doloop.c (add_test): Unshare sequence.
13701 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13703 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13705 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13707 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13709 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13710 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13711 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13712 (xlogue_layout::get_stub_rtx): Make static.
13713 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13714 (xlogue_layout::compute_stub_managed_regs): Rename to...
13715 (xlogue_layout::count_stub_managed_regs): ...this.
13716 (xlogue_layout::is_stub_managed_reg): New function.
13717 (xlogue_layout::m_stub_names): Rename to...
13718 (xlogue_layout::s_stub_names): ...this, make static.
13719 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13720 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13721 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13722 xlogue_layout::s_stub_names): Instantiate statics.
13723 (stub_managed_regs): Remove.
13724 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13725 (disable_call_ms2sysv_xlogues): Rename to...
13726 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13727 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13729 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13730 change after reload_completed.
13731 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13733 (ix86_expand_prologue): Likewise.
13734 (ix86_expand_epilogue): Likewise.
13735 (ix86_expand_split_stack_prologue): Likewise.
13736 (ix86_compute_frame_layout): Remove frame parameter ...
13737 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13738 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13740 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13741 (ix86_frame): Move from here ...
13742 * config/i386/i386.h (ix86_frame): ... to here.
13743 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13744 complete ix86_frame data structure instead. Remove some_ld_name.
13746 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13748 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13749 symbols that hold a DECL_VALUE_EXPR.
13751 2017-06-01 Martin Jambor <mjambor@suse.cz>
13753 PR tree-optimization/80898
13754 * tree-sra.c (process_subtree_disqualification): Removed.
13755 (disqualify_candidate): Do not acll
13756 process_subtree_disqualification.
13757 (subtree_mark_written_and_enqueue): New function.
13758 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13759 RHS has been disqualified and re-queue LHS if necessary. Apart
13760 from that, ignore disqualified RHS.
13762 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13764 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13765 address fetch for z10 or later.
13767 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13769 * config/arc/arc.md (tst_movb): Add guard when splitting.
13771 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13773 * config/arc/arc.c (arc_can_eliminate): Test against
13774 arc_frame_pointer_needed.
13776 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13778 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13779 to prevent store reordering.
13780 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13781 (type): Add block type.
13782 (stack_tie): Define special instruction to be used in
13785 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13787 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13788 constraint. It is not valid for the pattern.
13789 (noncommutative_binary_comparison): Likewise.
13791 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13793 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13796 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13798 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13799 be used by the reg-alloc.
13801 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13803 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13804 reg-alloc when having mul64 or mul32x16 instructions.
13805 (mulsidi3): Likewise.
13806 (umulsidi3): Likewise.
13807 (mulsi32x16): New pattern.
13808 (mulsi64): Likewise.
13809 (mulsidi64): Likewise.
13810 (umulsidi64): Likewise.
13811 (MUL32x16_REG): Define.
13812 (mul64_600): Use MUL32x16_REG.
13813 (mac64_600): Likewise.
13814 (umul64_600): Likewise.
13815 (umac64_600): Likewise.
13817 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13819 * config/arc/arc.md (mulsi3_700): Make it commutative.
13821 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13823 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13825 (*sign_extendsidi2_insn): Likewise for movstosw.
13827 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13829 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13830 the type of the input discriminant value. Convert the
13831 discriminant value of signedness vary.
13833 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13835 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13836 Add to -Wall section.
13838 2017-06-01 Richard Biener <rguenther@suse.de>
13840 PR middle-end/66313
13841 * fold-const.c (fold_plusminus_mult_expr): If the factored
13842 factor may be zero use a wrapping type for the inner operation.
13843 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13844 and handle moved defs.
13845 (process_assignment): Properly guard the unary op case. Return a
13846 tri-state indicating that moving the stmt before the call may allow
13847 to continue. Pass through to_move.
13848 (find_tail_calls): Handle moving unrelated defs before
13851 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13854 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13855 splitter result in the canonical way.
13857 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13859 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13860 also for 32bit target. Update insn attributes.
13861 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13863 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13865 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13866 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13867 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13868 (_mm_maskz_min_ss): New intrinsics.
13870 2017-05-31 Martin Liska <mliska@suse.cz>
13872 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13873 Change comment style to one we normally use.
13874 (vectorizable_reduction): Likewise.
13875 (vectorizable_induction): Likewise.
13876 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13877 (vectorizable_call): Likewise.
13878 (vectorizable_simd_clone_call): Likewise.
13879 (vectorizable_conversion): Likewise.
13880 (vectorizable_assignment): Likewise.
13881 (vectorizable_shift): Likewise.
13882 (vectorizable_operation): Likewise.
13883 (vectorizable_store): Likewise.
13884 (vectorizable_load): Likewise.
13885 * tree-vectorizer.h: Likewise.
13887 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13889 * passes.c (emergency_dump_function): New.
13890 * tree-pass.h (emergency_dump_function): Declare.
13891 * plugin.c (plugins_internal_error_function): Remove.
13892 * plugin.h (plugins_internal_error_function): Remove declaration.
13893 * toplev.c (internal_error_function): New static function. Use it...
13894 (general_init): ...here.
13896 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13898 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13899 (arc_rtx_costs): Add costs for new patterns.
13900 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13901 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13903 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13905 * tree-ssa-strlen.c (get_next_strinfo): New function.
13906 (get_stridx_plus_constant): Use it.
13907 (zero_length_string): Likewise.
13908 (adjust_related_strinfos): Likewise.
13909 (adjust_last_stmt): Likewise.
13911 2017-05-31 Richard Biener <rguenther@suse.de>
13914 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13915 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13917 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13919 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13920 loop_vinfo argument and use of dependence distance vectors.
13921 Check instead whether the two references differ only in their
13922 initial value and assume that they have the same alignment if the
13923 difference is a multiple of the vector alignment.
13924 (vect_analyze_data_refs_alignment): Update call accordingly.
13926 2017-05-31 Martin Liska <mliska@suse.cz>
13929 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13931 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13933 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13934 (create_intersect_range_checks): Move from ...
13935 * tree-data-ref.c (create_intersect_range_checks_index)
13936 (create_intersect_range_checks): ... to here.
13937 (create_runtime_alias_checks): New function factored from ...
13938 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13939 here. Call above function.
13940 * tree-data-ref.h (create_runtime_alias_checks): New function.
13942 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13944 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13945 segment length for dr_b and compute it in wide_int.
13947 2017-05-31 Richard Biener <rguenther@suse.de>
13949 PR tree-optimization/80906
13950 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13951 and pass through iv_map.
13952 (copy_bb_and_scalar_dependences): Adjust.
13953 (translate_pending_phi_nodes): Likewise.
13954 (copy_loop_close_phi_args): Handle code-generating IVs instead
13957 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13959 * diagnostic-color.c (color_dict): Add "type-diff".
13960 (parse_gcc_colors): Update comment.
13961 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13962 -fdiagnostics-show-template-tree and -fno-elide-type.
13963 (GCC_COLORS): Add type-diff to example.
13965 (-fdiagnostics-show-template-tree): New.
13966 (-fno-elide-type): New.
13967 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13968 the pp_format_decoder callback. Call any m_format_postprocessor's
13970 (pretty_printer::pretty_printer): Initialize
13971 m_format_postprocessor.
13972 (pretty_printer::~pretty_printer): Delete any
13973 m_format_postprocessor.
13974 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13975 (class format_postprocessor): New class.
13976 (struct pretty_printer::format_decoder): Document the new parameters.
13977 (struct pretty_printer::m_format_postprocessor): New field.
13978 * tree-diagnostic.c (default_tree_printer): Update for new
13979 bool and const char ** params.
13980 * tree-diagnostic.h (default_tree_printer): Likewise.
13982 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13984 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13985 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13986 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13987 rs6000_gen_cell_microcode code.
13988 (rs6000_final_prescan_insn): Delete.
13989 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13990 "warn-cell-microcode" entries.
13991 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13992 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13993 throughout. Change cc_reg_not_micro_cr0_operand to
13994 cc_reg_not_cr0_operand throughout.
13995 (*extendhi<mode>2_noload): Delete.
13996 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13997 (mwarn-cell-microcode): Delete.
13998 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13999 -mgen-cell-microcode and -mwarn-cell-microcode.
14001 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14004 * config/i386/constraints.md (Yd): New constraint.
14006 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14007 and (?Yd, r) alternatives. Update insn attributes.
14008 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14009 and (?*Yd, r) alternatives. Update insn attributes.
14010 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14012 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14014 * gimplify.c (gimplify_modify_expr): Don't create a
14015 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14018 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14020 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14022 2017-05-30 Richard Biener <rguenther@suse.de>
14024 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14025 and reduc_def fields.
14026 (STMT_VINFO_REDUC_TYPE): New define.
14027 (STMT_VINFO_REDUC_DEF): Likewise.
14028 (vect_force_simple_reduction): Adjust prototype.
14029 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14030 (vect_is_simple_reduction): Remove check_reduction argument.
14031 (vect_force_simple_reduction): Adjust and set
14032 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14033 (vectorizable_reduction): Do not re-do reduction analysis
14034 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14035 * tree-parloops.c (gather_scalar_reductions): Adjust.
14037 2017-05-30 Richard Biener <rguenther@suse.de>
14039 PR middle-end/80901
14040 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14043 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14045 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14046 Introduce unknown_misalignment parameter and remove vf.
14047 (vect_peeling_hash_get_lowest_cost):
14048 Pass unknown_misalignment parameter.
14049 (vect_enhance_data_refs_alignment):
14050 Fix unsupportable data ref treatment.
14052 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14054 * tree-vect-data-refs.c (vect_get_data_access_cost):
14055 Workaround for SLP handling.
14056 (vect_enhance_data_refs_alignment):
14057 Compute costs for doing no peeling at all, compare to the best
14058 peeling costs so far and avoid peeling if cheaper.
14060 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14062 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14063 Return peeling info and set costs to zero for unlimited cost
14065 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14066 with unknown misalignment. Compute and costs for unknown
14067 misalignment, compare them to the costs for known misalignment
14068 and choose the cheapest for peeling.
14070 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14072 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14073 (vect_get_peeling_costs_all_drs): Create function.
14074 (vect_peeling_hash_get_lowest_cost):
14075 Use vect_get_peeling_costs_all_drs.
14076 (vect_peeling_supportable): Create function.
14077 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14079 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14081 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14082 DR_HAS_NEGATIVE_STEP.
14083 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14084 (vect_enhance_data_refs_alignment): Use.
14085 (vect_duplicate_ssa_name_ptr_info): Use.
14086 * tree-vectorizer.h (dr_misalignment): Use.
14087 (known_alignment_for_access_p): Use.
14089 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14092 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14094 (has_section_name): New function.
14096 2017-05-30 Martin Liska <mliska@suse.cz>
14099 * auto-profile.c (get_function_decl_from_block): Fix
14102 2017-05-30 Richard Biener <rguenther@suse.de>
14104 PR middle-end/80876
14105 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14107 2017-05-30 Martin Liska <mliska@suse.cz>
14109 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14110 * dumpfile.h (struct dump_file_info): Remove ctors.
14112 2017-05-30 Martin Liska <mliska@suse.cz>
14114 * predict.def: Fix GNU coding style.
14116 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14118 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14119 Mark 'to' argument with ATTRIBUTE_UNUSED.
14121 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14123 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14124 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14125 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14128 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14130 * doc/install.texi (Options specification): Restore entry of
14131 --enable-sjlj-exceptions.
14133 2017-05-27 Michael Eager <eager@eagercon.com>
14136 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14138 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14140 * config/microblaze/microblaze.h
14141 (FIXED_REGISTERS): Update in macro.
14142 (CALL_USED_REGISTERS): Update in macro.
14144 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14146 * doc/install.texi: Add links to macOS binary distributions.
14148 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14152 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14154 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14156 2017-05-26 Martin Liska <mliska@suse.cz>
14158 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14160 2017-05-26 Martin Liska <mliska@suse.cz>
14162 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14163 always leading ';; '.
14164 (dump_bb_info): Likewise.
14165 (brief_dump_cfg): Likewise.
14166 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14167 * dumpfile.c: Remove usage of TDF_VERBOSE.
14168 * dumpfile.h (enum dump_kind): Likewise.
14169 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14170 * print-tree.c (debug_verbose): Remove.
14171 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14172 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14173 * tree-diagnostic.c (default_tree_printer): Replace
14174 TDF_DIAGNOSTIC with TDF_SLIM.
14176 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14178 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14179 in parameter loop, rather than loop_vinfo.
14180 (create_intersect_range_checks): Ditto.
14181 (vect_create_cond_for_alias_checks): Update call to above functions.
14183 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14185 PR tree-optimization/80815
14186 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14187 for merging runtime alias checks. Handle negative DR_STEPs.
14189 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14191 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14193 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14194 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14195 out code pruning runtime alias checks.
14196 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14197 factored out from above.
14198 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14200 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14202 (prune_runtime_alias_test_list): New decalaration.
14204 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14206 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14207 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14208 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14209 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14210 (operator==, comp_dr_with_seg_len_pair): Ditto.
14211 (vect_prune_runtime_alias_test_list): Ditto.
14213 2017-05-26 Martin Liska <mliska@suse.cz>
14216 * params.def: Bound partial-inlining-entry-probability param.
14218 2017-05-26 Marek Polacek <polacek@redhat.com>
14221 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14224 2017-05-26 Richard Biener <rguenther@suse.de>
14226 PR tree-optimization/80842
14227 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14230 2017-05-26 Richard Biener <rguenther@suse.de>
14232 PR tree-optimization/80844
14233 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14235 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14237 * doc/md.texi (Machine Constraints): Update x86 family
14238 machine constraints section to match 'config/i386/constraints.md'.
14240 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14242 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14244 2017-05-25 Nathan Sidwell <nathan@acm.org>
14246 * doc/invoke.texi (--enable-languages): Update documentation.
14248 2017-05-25 Martin Liska <mliska@suse.cz>
14250 * dumpfile.c: Add TDF_FOLDING.
14251 * dumpfile.h (enum dump_kind): Likewise.
14252 * genmatch.c (dt_simplify::gen_1): Use it.
14254 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14256 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14258 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14260 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14261 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14263 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14265 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14266 * match.pd (X == C): Rewrite it here.
14267 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14268 with_certain_nonzero_bits2): New predicates.
14269 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14271 2017-05-24 Nathan Sidwell <nathan@acm.org>
14273 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14276 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14279 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14281 * config/powerpcspe: New port. Files are copied from the rs6000
14282 port, with "rs6000" in filenames replaced by "powerpcspe".
14284 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14286 PR rtl-optimization/80754
14287 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14289 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14291 * config/sparc/sparc.md (length): Return the correct value for -mflat
14292 sibcalls to match output_sibcall.
14294 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14298 * config/rs6000/rs6000.c (struct machine_function): Add new field
14300 (rs6000_get_separate_components): Init that field, use it.
14301 (rs6000_components_for_bb): Use the field.
14303 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14305 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14307 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14309 PR middle-end/80823
14310 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14312 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14315 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14316 address against address_operand predicate.
14317 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14319 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14321 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14322 parameters passed indirectly.
14324 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14326 * config/i386/i386.md (*movdi_internal): Remove SSE4
14327 alternative 18 (?r, *v). Update insn attributes.
14328 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14329 Update insn attributes.
14330 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14331 Update insn attributes.
14332 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14333 alternative 1 (r, v). Remove isa attribute.
14334 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14335 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14336 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14338 2017-05-23 Tom de Vries <tom@codesourcery.com>
14340 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14343 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14345 * cgraphunit.c (symbol_table::process_new_functions): Update.
14346 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14347 (inline_generate_summary): Rename to ...
14348 (ipa_fn_summary_generate): ... this one.
14349 (inline_read_summary): Rename to ...
14350 (ipa_fn_summary_read): ... this one.
14351 (inline_write_summary): Rename to ...
14352 (ipa_fn_summary_write): ... this one.
14353 (inline_free_summary): Rename to ...
14354 (ipa_free_fn_summary): ... this one.
14355 (pass_data_local_fn_summary, pass_local_fn_summary,
14356 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14357 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14358 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14359 make_pass_ipa_fn_summary): New.
14360 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14361 inline_write_summary, inline_free_summary): Remove.
14362 (ipa_free_fn_summary) : New.
14363 * ipa-inline.c (ipa_inline): Update.
14364 (pass_ipa_inline): Do not generate summaries.
14365 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14367 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14368 and add pass_ipa_fn_summary.
14369 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14371 (make_pass_inline_parameters): Remove.
14373 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14375 * omp-low.c (struct omp_context): Remove "default_kind" member.
14378 * omp-offload.c (execute_oacc_device_lower): Remove the
14379 parallelism dimensions function attributes for unparallelized
14380 OpenACC kernels constructs.
14382 2017-05-23 Martin Liska <mliska@suse.cz>
14384 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14386 (cgraph_edge::make_speculative): Likewise.
14387 (cgraph_edge::resolve_speculation): Likewise.
14388 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14389 (cgraph_node::dump): Likewise.
14390 * cgraph.h: Likewise.
14391 * cgraphunit.c (analyze_functions): Likewise.
14392 (symbol_table::compile): Likewise.
14393 * ipa-cp.c (print_all_lattices): Likewise.
14394 (determine_versionability): Likewise.
14395 (initialize_node_lattices): Likewise.
14396 (ipcp_verify_propagated_values): Likewise.
14397 (estimate_local_effects): Likewise.
14398 (update_profiling_info): Likewise.
14399 (create_specialized_node): Likewise.
14400 (perhaps_add_new_callers): Likewise.
14401 (decide_about_value): Likewise.
14402 (decide_whether_version_node): Likewise.
14403 (identify_dead_nodes): Likewise.
14404 (ipcp_store_bits_results): Likewise.
14405 * ipa-devirt.c (dump_targets): Likewise.
14406 (ipa_devirt): Likewise.
14407 * ipa-icf.c (sem_item::dump): Likewise.
14408 (sem_function::equals): Likewise.
14409 (sem_variable::equals): Likewise.
14410 (sem_item_optimizer::read_section): Likewise.
14411 (sem_item_optimizer::execute): Likewise.
14412 (congruence_class::dump): Likewise.
14413 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14414 (dump_inline_summary): Likewise.
14415 (estimate_node_size_and_time): Likewise.
14416 (inline_analyze_function): Likewise.
14417 * ipa-inline-transform.c (inline_call): Likewise.
14418 * ipa-inline.c (report_inline_failed_reason): Likewise.
14419 (want_early_inline_function_p): Likewise.
14420 (edge_badness): Likewise.
14421 (update_edge_key): Likewise.
14422 (inline_small_functions): Likewise.
14423 * ipa-profile.c (ipa_profile): Likewise.
14424 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14425 (ipa_make_edge_direct_to_target): Likewise.
14426 (remove_described_reference): Likewise.
14427 (ipa_impossible_devirt_target): Likewise.
14428 (propagate_controlled_uses): Likewise.
14429 (ipa_print_node_params): Likewise.
14430 (ipcp_transform_function): Likewise.
14431 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14432 (propagate_pure_const): Likewise.
14433 * ipa-reference.c (generate_summary): Likewise.
14434 (read_write_all_from_decl): Likewise.
14435 (propagate): Likewise.
14436 (ipa_reference_read_optimization_summary): Likewise.
14437 * ipa-utils.c (ipa_merge_profiles): Likewise.
14438 * ipa.c (walk_polymorphic_call_targets): Likewise.
14439 (symbol_table::remove_unreachable_nodes): Likewise.
14440 (ipa_single_use): Likewise.
14441 * passes.c (execute_todo): Likewise.
14442 * predict.c (drop_profile): Likewise.
14443 * symtab.c (symtab_node::get_dump_name): New function.
14444 (symtab_node::dump_name): Likewise.
14445 (symtab_node::dump_asm_name): Likewise.
14446 (symtab_node::dump_references): Likewise.
14447 (symtab_node::dump_referring): Likewise.
14448 (symtab_node::dump_base): Likewise.
14449 (symtab_node::debug_symtab): Likewise.
14450 * tree-sra.c (convert_callers_for_node): Likewise.
14451 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14452 * value-prof.c (init_node_map): Likewise.
14454 2017-05-23 Martin Liska <mliska@suse.cz>
14456 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14457 and symtab_node::debug_symtab to symbol_table::debug.
14458 * cgraphunit.c (analyze_functions): Use the renamed function.
14459 (symbol_table::compile): Likewise.
14460 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14461 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14462 * passes.c (execute_todo): Likewise.
14463 * symtab.c (symbol_table::dump): New function.
14464 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14466 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14468 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14469 that nonconst implies exec.
14471 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14473 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14474 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14475 (inline_edge_summary_vec): Turn into ...
14476 (ipa_call_summaries): ... this one.
14477 (redirect_to_unreachable, edge_set_predicate,
14478 evaluate_properties_for_edge, inline_summary_alloc,
14479 reset_ipa_call_summary, reset_inline_summary,
14480 inline_summary_t::duplicate): Update.
14481 (inline_edge_duplication_hook): Turn to ...
14482 (ipa_call_summary_t::duplicate): ... this one.
14483 (inline_edge_removal_hook): Turn to ...
14484 (ipa_call_summary_t::remove): ... this one.
14485 (dump_inline_edge_summary): Turn to ...
14486 (dump_ipa_call_summary): ... this one.
14487 (estimate_function_body_sizes): Update.
14488 (inline_update_callee_summaries): Update.
14489 (remap_edge_change_prob): Update.
14490 (remap_edge_summaries): Update.
14491 (inline_merge_summary): Update.
14492 (do_estimate_edge_time): Update.
14493 (inline_generate_summary): Update.
14494 (inline_read_section): Update.
14495 (inline_read_summary): Update.
14496 (inline_free_summary): Update.
14497 * ipa-inline.c (can_inline_edge_p): Update.
14498 (compute_inlined_call_time): Update.
14499 (want_inline_small_function_p): Update.
14500 (edge_badness): Update.
14501 (early_inliner): Update.
14502 * ipa-inline.h (inline_edge_summary): Turn to ...
14503 (ipa_call_summary): ... this one.
14504 (ipa_call_summary_t): New class.
14505 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14506 (ipa_call_summaries): New.
14507 (inline_edge_summary): Remove.
14508 (estimate_edge_growth): Update.
14509 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14510 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14511 * ipa-split.c (execute_split_functions): Update.
14512 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14514 2017-05-23 Tom de Vries <tom@codesourcery.com>
14516 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14517 attributes): Document rdrand effective target.
14519 2017-05-23 Tom de Vries <tom@codesourcery.com>
14521 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14522 attributes): Sort alphabetically.
14524 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14526 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14528 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14531 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14532 V2DF/V2DI splat into two separate patterns, one that handles
14533 registers, and the other that only handles memory. Drop support
14534 for splatting from a GPR on ISA 2.07 and then splitting the
14535 splat into direct move and splat.
14536 (vsx_splat_<mode>_reg): Likewise.
14537 (vsx_splat_<mode>_mem): Likewise.
14539 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14541 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14543 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14545 PR middle-end/80809
14546 * omp-low.c (finish_taskreg_remap): New function.
14547 (finish_taskreg_scan): If unit size of ctx->record_type
14548 is non-constant, unshare the size expression and replace
14549 decls in it with possible outer var refs.
14551 PR middle-end/80809
14552 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14553 GOVD_SHARED rather than GOVD_PRIVATE with it.
14554 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14555 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14557 PR middle-end/80853
14558 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14559 as last argument to build_outer_var_ref for pointer bases of array
14560 section reductions.
14562 2017-05-19 Martin Sebor <msebor@redhat.com>
14564 * print-tree.c (print_node): Print DECL_READ_P flag.
14566 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14568 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14569 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14570 * cgraph.c: Likewise.
14571 * cgraphunit.c: Likewise.
14572 * gengtype.c: Likewise.
14573 * ipa-cp.c: Likewise.
14574 * ipa-devirt.c: Likewise.
14575 * ipa-icf.c: Likewise.
14576 * ipa-predicate.c: Likewise.
14577 * ipa-profile.c: Likewise.
14578 * ipa-prop.c: Likewise.
14579 * ipa-split.c: Likewise.
14581 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14582 edge_predicate_pool, dump_inline_hints,
14583 inline_summary::account_size_time, redirect_to_unreachable,
14584 edge_set_predicate, set_hint_predicate,
14585 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14586 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14587 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14588 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14589 ipa_call_summary_t::remove, initialize_growth_caches,
14590 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14591 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14592 mark_modified, unmodified_parm_1, unmodified_parm,
14593 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14594 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14595 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14596 will_be_nonconstant_predicate, record_modified_bb_info,
14597 get_minimal_bb, record_modified, param_change_prob,
14598 phi_result_unknown_predicate, predicate_for_phi_result,
14599 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14600 estimate_function_body_sizes, compute_inline_parameters,
14601 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14602 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14603 inline_update_callee_summaries, remap_edge_change_prob,
14604 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14605 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14606 inline_analyze_function, inline_summary_t::insert,
14607 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14608 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14609 inline_free_summary): Move to ipa-fnsummary.h
14610 (predicate_t): Remove.
14611 * ipa-fnsummary.c: New file.
14612 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14613 (enum inline_hints_vals, inline_hints, agg_position_info,
14614 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14615 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14616 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14617 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14618 inline_read_summary, inline_write_summary, inline_free_summary,
14619 inline_analyze_function, initialize_inline_failed,
14620 inline_merge_summary, inline_update_overall_summary,
14621 compute_inline_parameters): Move to ipa-fnsummary.h
14622 * ipa-fnsummary.h: New file.
14623 * ipa-inline-transform.h: Include ipa-inline.h.
14624 * ipa-inline.c: LIkewise.
14626 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14628 * ipa-inline.c (edge_badness): Use inlined_time instead of
14629 inline_summaries->get.
14631 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14633 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14635 2017-05-22 Nathan Sidwell <nathan@acm.org>
14637 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14638 (fdump-lang): Document 'raw' option.
14639 * dumpfile.h (TDI_tu): Delete.
14640 * dumpfile.c (dump_files): Remove translation-unit.
14641 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14643 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14645 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14646 command option from $(AWK) call.
14647 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14648 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14649 [FORMAT]: Remove handling of variable.
14650 * config/avr/t-multilib: Regenerate.
14652 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14654 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14656 (dump_inline_summary): Do not print self_time.
14657 (estimate_function_body_sizes): Do not set self_time.
14658 (compute_inline_parameters): Likewise.
14659 (inline_read_section, inline_write_summary): Do not stream self_time.
14660 * ipa-inline.h (inline_summary): Drop self_time.
14662 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14664 * ipa-inline-analysis.c (account_size_time): Rename to ...
14665 (inline_summary::account_size_time): ... this one.
14666 (reset_ipa_call_summary): Turn to ...
14667 (ipa_call_summary::reset): ... this one.
14668 (reset_inline_summary): Turn to ...
14669 (inline_summary::reset): ... this one.
14670 (inline_summary_t::remove): Update.
14671 (inline_summary_t::duplicate): Update.
14672 (ipa_call_summary_t::remove): Update.
14673 (dump_inline_summary): Update.
14674 (estimate_function_body_sizes): Update.
14675 (compute_inline_parameters): Update.
14676 (estimate_node_size_and_time): Update.
14677 (inline_merge_summary): Update.
14678 (inline_update_overall_summary): Update.
14679 (inline_read_section): Update.
14680 (inline_write_summary): Update.
14681 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14682 add account_size_time and reset member functions.
14683 (ipa_call_summary): Add reset function.
14684 * ipa-predicate.h (predicate::operator &): Constify.
14686 2017-05-22 Richard Biener <rguenther@suse.de>
14688 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14690 2017-05-19 Jason Merrill <jason@redhat.com>
14692 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14694 2017-05-19 Marek Polacek <polacek@redhat.com>
14697 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14698 TYPE_OVERFLOW_WRAPS checks.
14700 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14702 * tree-core.h (enum omp_clause_default_kind): Add
14703 "OMP_CLAUSE_DEFAULT_PRESENT".
14704 * tree-pretty-print.c (dump_omp_clause): Handle it.
14705 * gimplify.c (enum gimplify_omp_var_data): Add
14706 "GOVD_MAP_FORCE_PRESENT".
14707 (gimplify_adjust_omp_clauses_1): Map it to
14708 "GOMP_MAP_FORCE_PRESENT".
14709 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14711 * gimplify.c (oacc_default_clause): Clarify.
14713 2017-05-19 Nathan Sidwell <nathan@acm.org>
14715 LANG_HOOK_REGISTER_DUMPS
14716 * toplev.c (general_init): Call register dump lang hook.
14717 * doc/invoke.texi: Document -fdump-lang option family.
14718 * dumpfile.c (dump_files): Remove class dump here.
14719 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14720 * dumpfile.h (tree_dump_index): Remove TDI_class.
14721 * langhooks-def.h (lhd_register_dumps): Declare.
14722 (LANG_HOOKS_REGISTER_DUMPS): Define.
14723 (LANG_HOOKS_INITIALIZER): Add it.
14724 * langhooks.c (lhd_register_dumps): Define.
14725 * langhooks.h (struct lang_hooks): Add register_dumps.
14727 2017-05-19 Nathan Sidwell <nathan@acm.org>
14729 * context.h (context::set_passes): New.
14730 * context.c (context::context): Do not create pass manager.
14731 * toplev.c (general_init): Create pass manager here.
14733 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14735 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14736 use this splitter if two add or or instructions would also work for
14737 the constant we want to generate.
14739 2017-05-19 Richard Biener <rguenther@suse.de>
14742 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14743 predicate evaluation.
14745 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14747 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14749 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14752 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14754 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14755 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14756 (inline_edge_summary_vec): Turn into ...
14757 (ipa_call_summaries): ... this one.
14758 (redirect_to_unreachable, edge_set_predicate,
14759 evaluate_properties_for_edge, inline_summary_alloc,
14760 reset_ipa_call_summary, reset_inline_summary,
14761 inline_summary_t::duplicate): Update.
14762 (inline_edge_duplication_hook): Turn to ...
14763 (ipa_call_summary_t::duplicate): ... this one.
14764 (inline_edge_removal_hook): Turn to ...
14765 (ipa_call_summary_t::remove): ... this one.
14766 (dump_inline_edge_summary): Turn to ...
14767 (dump_ipa_call_summary): ... this one.
14768 (estimate_function_body_sizes): Update.
14769 (inline_update_callee_summaries): Update.
14770 (remap_edge_change_prob): Update.
14771 (remap_edge_summaries): Update.
14772 (inline_merge_summary): Update.
14773 (do_estimate_edge_time): Update.
14774 (inline_generate_summary): Update.
14775 (inline_read_section): Update.
14776 (inline_read_summary): Update.
14777 (inline_free_summary): Update.
14778 * ipa-inline.c (can_inline_edge_p): Update.
14779 (compute_inlined_call_time): Update.
14780 (want_inline_small_function_p): Update.
14781 (edge_badness): Update.
14782 (early_inliner): Update.
14783 * ipa-inline.h (inline_edge_summary): Turn to ...
14784 (ipa_call_summary): ... this one.
14785 (ipa_call_summary_t): New class.
14786 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14787 (ipa_call_summaries): New.
14788 (inline_edge_summary): Remove.
14789 (estimate_edge_growth): Update.
14790 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14791 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14792 * ipa-split.c (execute_split_functions): Update.
14793 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14795 2017-05-19 Richard Biener <rguenther@suse.de>
14797 PR middle-end/80764
14798 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14800 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14802 * config/rs6000/rs6000.c (struct machine_function): Add field
14803 fpr_is_wrapped_separately.
14804 (rs6000_get_separate_components): Use 64 components. Handle the
14805 new FPR components.
14806 (rs6000_components_for_bb): Handle the FPR components.
14807 (rs6000_emit_prologue_components): Handle the FPR components.
14808 (rs6000_emit_epilogue_components): Handle the FPR components.
14809 (rs6000_set_handled_components): Handle the FPR components.
14810 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14811 that are already separately shrink-wrapped.
14812 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14813 that are already separately shrink-wrapped.
14815 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14818 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14821 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14822 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14823 cases where the register allocator uses a move and an offsettable
14824 memory operation to/from a FPR register on ISA 2.06/2.07.
14825 (define_peephole2 for Altivec d-form store): Likewise.
14827 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14830 * config/i386/mmx.md (*mov<mode>_internal): Enable
14831 alternatives 11, 12, 13 and 14 also for 32bit targets.
14832 Remove alternatives 15, 16, 17 and 18.
14833 * config/i386/sse.md (vec_concatv2di): Change
14834 alternative (!x, *y) to (x, ?!*Yn).
14836 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14838 * dumpfile.h (enum dump_kind): Remove stray comma.
14840 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14842 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14843 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14844 predicate::num_conditions
14845 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14846 (CHANGED): turn into predicate::changed.
14847 (agg_position_info): Move to ipa-predicate.h
14848 (add_condition, predicate::add_clause, predicate::operator &=,
14849 predicate::or_with, predicate::evaluate, predicate::probability,
14850 dump_condition, dump_clause, predicate::dump,
14851 predicate::remap_after_duplication, predicate::remap_after_inlining,
14852 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14853 (evaluate_conditions_for_known_args): Update.
14854 (set_cond_stmt_execution_predicate): Update.
14855 * ipa-inline.h: Include ipa-predicate.h
14856 (condition, inline_param_summary, conditions, agg_position_info,
14857 predicate): Move to ipa-predicate.h
14858 * ipa-predicate.c: New file.
14859 * ipa-predicate.h: New file.
14861 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14863 * final.c (leaf_function_p): Check we are not in a sequence.
14865 2017-05-18 Martin Liska <mliska@suse.cz>
14867 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14868 * dumpfile.c (dump_register): Use new enum dump_kind.
14869 (get_dump_file_name): Likewise.
14870 (dump_enable_all): Likewise.
14871 (dump_switch_p_1): Likewise.
14872 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14873 * dumpfile.h (enum dump_kind): New enum type.
14874 (struct dump_file_info): Create constructor and
14875 format fields and comments.
14876 * passes.c (pass_manager::register_one_dump_file):
14878 * statistics.c (statistics_early_init): Likewise.
14879 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14880 TDF_TREE with TDF_SLIM.
14881 (gather_memory_references_ref): Likewise.
14883 2017-05-18 Martin Liska <mliska@suse.cz>
14885 * vec.h (struct vnull): Use it.
14887 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14889 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14890 (true_predicate, false_predicate, true_predicate_p,
14891 false_predicate_p): Remove.
14892 (single_cond_predicate, not_inlined_predicate): Turn to member function
14894 (add_condition): Update.
14895 (add_clause): Turn to...
14896 (predicate::add_clause): ... this one; update; allow passing NULL
14898 (and_predicates): Turn to ...
14899 (predicate::operator &=): ... this one.
14900 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14901 (or_predicates): Turn to ...
14902 (predicate::or_with): ... this one.
14903 (evaluate_predicate): Turn to ...
14904 (predicate::evaluate): ... this one.
14905 (predicate_probability): Turn to ...
14906 (predicate::probability): ... this one.
14907 (dump_condition): Update.
14908 (dump_predicate): Turn to ...
14909 (predicate::dump): ... this one.
14910 (account_size_time): Update.
14911 (edge_set_predicate): Update.
14912 (set_hint_predicate): UPdate.
14913 (evaluate_conditions_for_known_args): Update.
14914 (evaluate_properties_for_edge): Update.
14915 (remap_predicate_after_duplication): Turn to...
14916 (predicate::remap_after_duplication): ... this one.
14917 (remap_hint_predicate_after_duplication): Update.
14918 (inline_summary_t::duplicate): UPdate.
14919 (dump_inline_edge_summary): Update.
14920 (dump_inline_summary): Update.
14921 (set_cond_stmt_execution_predicate): Update.
14922 (set_switch_stmt_execution_predicate): Update.
14923 (compute_bb_predicates): Update.
14924 (will_be_nonconstant_expr_predicate): Update.
14925 (will_be_nonconstant_predicate): Update.
14926 (phi_result_unknown_predicate): Update.
14927 (predicate_for_phi_result): Update.
14928 (array_index_predicate): Update.
14929 (estimate_function_body_sizes): Update.
14930 (estimate_node_size_and_time): Update.
14931 (estimate_ipcp_clone_size_and_time): Update.
14932 (remap_predicate): Rename to ...
14933 (predicate::remap_after_inlining): ... this one.
14934 (remap_hint_predicate): Update.
14935 (inline_merge_summary): Update.
14936 (inline_update_overall_summary): Update.
14937 (estimate_size_after_inlining): Update.
14938 (read_predicate): Rename to ...
14939 (predicate::stream_in): ... this one.
14940 (read_inline_edge_summary): Update.
14941 (write_predicate): Rename to ...
14942 (predicate::stream_out): ... this one.
14943 (write_inline_edge_summary): Update.
14944 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14945 (clause_t): Turn to uint32_t
14946 (predicate): Turn to class; implement constructor and operators
14948 (size_time_entry): Update.
14949 (inline_summary): Update.
14950 (inline_edge_summary): Update.
14952 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14954 * fold-const.c (fold_binary_loc): Move transformation...
14955 * match.pd (C - X CMP X): ... here.
14957 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14959 * config/sparc/sparc.c (sparc_option_override): Set function
14960 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14961 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14963 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14965 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14967 2017-05-18 Marek Polacek <polacek@redhat.com>
14970 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14971 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14972 gimple_assign_load_p.
14974 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14976 PR middle-end/80692
14977 * real.c (do_compare): Give decimal_do_compare preference over
14978 comparing just the signs.
14980 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14982 * doc/md.texi (Canonicalization of Instructions): Describe the
14983 canonical form of instructions that inherently set a condition
14986 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14988 PR middle-end/80775
14989 * tree-cfg.c: Move deletion of unreachable case statements to after
14990 the merging of consecutive case labels.
14992 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14994 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14995 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14996 restoring of callee-saved registers.
14998 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15000 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15001 * config/visium/visium.c (single_set_and_flags): Likewise.
15002 * config/visium/visium.md (Substitutions): Likewise.
15004 2017-05-17 Martin Liska <mliska@suse.cz>
15006 * cfg.c: Introduce dump_flags_t type and
15007 use it instead of int type.
15009 * cfghooks.c: Likewise.
15010 * cfghooks.h (struct cfg_hooks): Likewise.
15011 * cfgrtl.c: Likewise.
15012 * cfgrtl.h: Likewise.
15013 * cgraph.c (cgraph_node::get_body): Likewise.
15014 * coretypes.h: Likewise.
15015 * domwalk.c: Likewise.
15016 * domwalk.h: Likewise.
15017 * dumpfile.c (struct dump_option_value_info): Likewise.
15018 (dump_enable_all): Likewise.
15019 (dump_switch_p_1): Likewise.
15020 (opt_info_switch_p): Likewise.
15021 * dumpfile.h (enum tree_dump_index): Likewise.
15022 (struct dump_file_info): Likewise.
15023 * genemit.c: Likewise.
15024 * generic-match-head.c: Likewise.
15025 * gengtype.c (open_base_files): Likewise.
15026 * gimple-pretty-print.c: Likewise.
15027 * gimple-pretty-print.h: Likewise.
15028 * graph.c (print_graph_cfg): Likewise.
15029 * graphite-scop-detection.c (dot_all_sese): Likewise.
15030 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15031 * loop-unroll.c (report_unroll): Likewise.
15032 * passes.c (pass_manager::register_one_dump_file): Likewise.
15033 * print-tree.c: Likewise.
15034 * statistics.c: Likewise.
15035 * tree-cfg.c: Likewise.
15036 * tree-cfg.h: Likewise.
15037 * tree-dfa.c: Likewise.
15038 * tree-dfa.h: Likewise.
15039 * tree-dump.c (dump_function): Likewise.
15040 * tree-dump.h (struct dump_info): Likewise.
15041 * tree-pretty-print.c: Likewise.
15042 * tree-pretty-print.h: Likewise.
15043 * tree-ssa-live.c: Likewise.
15044 * tree-ssa-live.h: Likewise.
15045 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15046 * tree-vect-loop.c: Likewise.
15047 * tree-vect-slp.c: Likewise.
15049 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15050 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15052 PR tree-optimization/80457
15053 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15054 of all arguments to a statement as scalar_to_vec operations.
15055 (vectorizable_call): Adjust call to vect_model_simple_cost for
15057 (vectorizable_conversion): Likewise.
15058 (vectorizable_assignment): Likewise.
15059 (vectorizable_shift): Likewise.
15060 (vectorizable_operation): Likewise.
15061 (vectorizable_comparison): Likewise.
15062 (vect_is_simple_cond): Record the def types for operands.
15063 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15064 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15065 for statement argument count.
15067 2017-05-16 Carl Love <cel@us.ibm.com>
15069 * config/rs6000/rs6000-c: Add support for built-in functions
15070 vector unsigned long long vec_bperm (vector unsigned long long,
15071 vector unsigned char)
15072 vector signed long long vec_mule (vector signed int,
15074 vector unsigned long long vec_mule (vector unsigned int,
15075 vector unsigned int)
15076 vector signed long long vec_mulo (vector signed int,
15078 vector unsigned long long vec_mulo (vector unsigned int,
15079 vector unsigned int)
15080 vector signed char vec_sldw (vector signed char,
15081 vector signed char,
15083 vector unsigned char vec_sldw (vector unsigned char,
15084 vector unsigned char,
15086 vector signed short vec_sldw (vector signed short,
15087 vector signed short,
15089 vector unsigned short vec_sldw (vector unsigned short,
15090 vector unsigned short,
15092 vector signed int vec_sldw (vector signed int,
15095 vector unsigned int vec_sldw (vector unsigned int,
15096 vector unsigned int,
15098 vector signed long long vec_sldw (vector signed long long,
15099 vector signed long long,
15101 vector unsigned long long vec_sldw (vector unsigned long long,
15102 vector unsigned long long,
15104 * config/rs6000/rs6000-c: Add support for built-in functions
15105 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15106 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15107 * doc/extend.texi: Update the built-in documentation for the
15108 new built-in functions.
15110 2017-05-16 Marek Polacek <polacek@redhat.com>
15114 * tree.c (save_expr): Don't fold the expression.
15116 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15118 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15119 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15120 and (?*y,m). Update insn attributes.
15122 2017-05-16 Martin Liska <mliska@suse.cz>
15124 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15125 flags argument of print_gimple_stmt, print_gimple_expr,
15126 print_generic_stmt and print_generic_expr.
15127 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15128 * coretypes.h: Likewise.
15129 * except.c (dump_eh_tree): Likewise.
15130 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15131 * gimple-pretty-print.h: Likewise.
15132 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15133 (backprop::push_to_worklist): Likewise.
15134 (backprop::pop_from_worklist): Likewise.
15135 (backprop::process_use): Likewise.
15136 (backprop::intersect_uses): Likewise.
15137 (note_replacement): Likewise.
15138 * gimple-ssa-store-merging.c
15139 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15140 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15141 (pass_store_merging::execute): Likewise.
15142 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15143 (ssa_base_cand_dump_callback): Likewise.
15144 (dump_incr_vec): Likewise.
15145 (replace_refs): Likewise.
15146 (replace_mult_candidate): Likewise.
15147 (create_add_on_incoming_edge): Likewise.
15148 (create_phi_basis): Likewise.
15149 (insert_initializers): Likewise.
15150 (all_phi_incrs_profitable): Likewise.
15151 (introduce_cast_before_cand): Likewise.
15152 (replace_one_candidate): Likewise.
15153 * gimplify.c (gimplify_expr): Likewise.
15154 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15155 (set_rename): Likewise.
15156 (rename_uses): Likewise.
15157 (copy_loop_phi_nodes): Likewise.
15158 (add_close_phis_to_merge_points): Likewise.
15159 (copy_loop_close_phi_args): Likewise.
15160 (copy_cond_phi_args): Likewise.
15161 (graphite_copy_stmts_from_block): Likewise.
15162 (translate_pending_phi_nodes): Likewise.
15163 * graphite-poly.c (print_pdr): Likewise.
15164 (dump_gbb_cases): Likewise.
15165 (dump_gbb_conditions): Likewise.
15166 (print_scop_params): Likewise.
15167 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15168 (build_cross_bb_scalars_use): Likewise.
15169 (gather_bbs::before_dom_children): Likewise.
15170 * hsa-dump.c (dump_hsa_immed): Likewise.
15171 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15172 (get_replacement_map): Likewise.
15173 * ipa-inline-analysis.c (dump_condition): Likewise.
15174 (estimate_function_body_sizes): Likewise.
15175 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15176 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15177 * ipa-prop.c (ipa_dump_param): Likewise.
15178 (ipa_print_node_jump_functions_for_edge): Likewise.
15179 (ipa_modify_call_arguments): Likewise.
15180 (ipa_modify_expr): Likewise.
15181 (ipa_dump_param_adjustments): Likewise.
15182 (ipa_dump_agg_replacement_values): Likewise.
15183 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15184 * ipa-pure-const.c (check_stmt): Likewise.
15185 (pass_nothrow::execute): Likewise.
15186 * ipa-split.c (execute_split_functions): Likewise.
15187 * omp-offload.c (dump_oacc_loop_part): Likewise.
15188 (dump_oacc_loop): Likewise.
15189 * trans-mem.c (tm_log_emit): Likewise.
15190 (tm_memopt_accumulate_memops): Likewise.
15191 (dump_tm_memopt_set): Likewise.
15192 (dump_tm_memopt_transform): Likewise.
15193 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15194 (print_loop): Likewise.
15195 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15196 (chkp_gather_checks_info): Likewise.
15197 (chkp_get_check_result): Likewise.
15198 (chkp_remove_check_if_pass): Likewise.
15199 (chkp_use_outer_bounds_if_possible): Likewise.
15200 (chkp_reduce_bounds_lifetime): Likewise.
15201 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15202 (chkp_mark_completed_bounds): Likewise.
15203 (chkp_register_incomplete_bounds): Likewise.
15204 (chkp_mark_invalid_bounds): Likewise.
15205 (chkp_maybe_copy_and_register_bounds): Likewise.
15206 (chkp_build_returned_bound): Likewise.
15207 (chkp_get_bound_for_parm): Likewise.
15208 (chkp_build_bndldx): Likewise.
15209 (chkp_get_bounds_by_definition): Likewise.
15210 (chkp_generate_extern_var_bounds): Likewise.
15211 (chkp_get_bounds_for_decl_addr): Likewise.
15212 * tree-chrec.c (chrec_apply): Likewise.
15213 * tree-data-ref.c (dump_data_reference): Likewise.
15214 (dump_subscript): Likewise.
15215 (dump_data_dependence_relation): Likewise.
15216 (analyze_overlapping_iterations): Likewise.
15217 * tree-inline.c (expand_call_inline): Likewise.
15218 (tree_function_versioning): Likewise.
15219 * tree-into-ssa.c (dump_defs_stack): Likewise.
15220 (dump_currdefs): Likewise.
15221 (dump_names_replaced_by): Likewise.
15222 (dump_update_ssa): Likewise.
15223 (update_ssa): Likewise.
15224 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15225 * tree-parloops.c (build_new_reduction): Likewise.
15226 (try_create_reduction_list): Likewise.
15227 (ref_conflicts_with_region): Likewise.
15228 (oacc_entry_exit_ok_1): Likewise.
15229 (oacc_entry_exit_single_gang): Likewise.
15230 * tree-pretty-print.h: Likewise.
15231 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15232 (get_scalar_evolution): Likewise.
15233 (add_to_evolution): Likewise.
15234 (get_loop_exit_condition): Likewise.
15235 (analyze_evolution_in_loop): Likewise.
15236 (analyze_initial_condition): Likewise.
15237 (analyze_scalar_evolution): Likewise.
15238 (instantiate_scev): Likewise.
15239 (number_of_latch_executions): Likewise.
15240 (gather_chrec_stats): Likewise.
15241 (final_value_replacement_loop): Likewise.
15242 (scev_const_prop): Likewise.
15243 * tree-sra.c (dump_access): Likewise.
15244 (disqualify_candidate): Likewise.
15245 (create_access): Likewise.
15246 (reject): Likewise.
15247 (maybe_add_sra_candidate): Likewise.
15248 (create_access_replacement): Likewise.
15249 (analyze_access_subtree): Likewise.
15250 (analyze_all_variable_accesses): Likewise.
15251 (sra_modify_assign): Likewise.
15252 (initialize_constant_pool_replacements): Likewise.
15253 (find_param_candidates): Likewise.
15254 (decide_one_param_reduction): Likewise.
15255 (replace_removed_params_ssa_names): Likewise.
15256 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15257 * tree-ssa-copy.c (dump_copy_of): Likewise.
15258 (copy_prop_visit_cond_stmt): Likewise.
15259 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15260 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15261 (record_equivalences_from_stmt): Likewise.
15262 * tree-ssa-dse.c (compute_trims): Likewise.
15263 (delete_dead_call): Likewise.
15264 (delete_dead_assignment): Likewise.
15265 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15266 (forward_propagate_into_cond): Likewise.
15267 (pass_forwprop::execute): Likewise.
15268 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15269 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15271 (move_computations_worker): Likewise.
15272 (execute_sm): Likewise.
15273 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15274 (remove_exits_and_undefined_stmts): Likewise.
15275 (remove_redundant_iv_tests): Likewise.
15276 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15277 (adjust_iv_update_pos): Likewise.
15278 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15279 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15280 (value_replacement): Likewise.
15281 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15282 * tree-ssa-pre.c (print_pre_expr): Likewise.
15283 (get_representative_for): Likewise.
15284 (create_expression_by_pieces): Likewise.
15285 (insert_into_preds_of_block): Likewise.
15286 (eliminate_insert): Likewise.
15287 (eliminate_dom_walker::before_dom_children): Likewise.
15288 (eliminate): Likewise.
15289 (remove_dead_inserted_code): Likewise.
15290 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15291 * tree-ssa-reassoc.c (get_rank): Likewise.
15292 (eliminate_duplicate_pair): Likewise.
15293 (eliminate_plus_minus_pair): Likewise.
15294 (eliminate_not_pairs): Likewise.
15295 (undistribute_ops_list): Likewise.
15296 (eliminate_redundant_comparison): Likewise.
15297 (update_range_test): Likewise.
15298 (optimize_range_tests_var_bound): Likewise.
15299 (optimize_vec_cond_expr): Likewise.
15300 (rewrite_expr_tree): Likewise.
15301 (rewrite_expr_tree_parallel): Likewise.
15302 (linearize_expr): Likewise.
15303 (break_up_subtract): Likewise.
15304 (linearize_expr_tree): Likewise.
15305 (attempt_builtin_powi): Likewise.
15306 (attempt_builtin_copysign): Likewise.
15307 (transform_stmt_to_copy): Likewise.
15308 (transform_stmt_to_multiply): Likewise.
15309 (dump_ops_vector): Likewise.
15310 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15311 (print_scc): Likewise.
15312 (set_ssa_val_to): Likewise.
15313 (visit_reference_op_store): Likewise.
15314 (visit_use): Likewise.
15315 (sccvn_dom_walker::before_dom_children): Likewise.
15316 (run_scc_vn): Likewise.
15317 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15319 (expr_hash_elt::print): Likewise.
15320 (const_and_copies::pop_to_marker): Likewise.
15321 (const_and_copies::record_const_or_copy_raw): Likewise.
15322 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15323 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15324 (dump_predicates): Likewise.
15325 (find_uninit_use): Likewise.
15326 (warn_uninitialized_phi): Likewise.
15327 (pass_late_warn_uninitialized::execute): Likewise.
15328 * tree-ssa.c (verify_vssa): Likewise.
15329 (verify_ssa): Likewise.
15330 (maybe_optimize_var): Likewise.
15331 * tree-vrp.c (dump_value_range): Likewise.
15332 (dump_all_value_ranges): Likewise.
15333 (dump_asserts_for): Likewise.
15334 (register_edge_assert_for_2): Likewise.
15335 (vrp_visit_cond_stmt): Likewise.
15336 (vrp_visit_switch_stmt): Likewise.
15337 (vrp_visit_stmt): Likewise.
15338 (vrp_visit_phi_node): Likewise.
15339 (simplify_cond_using_ranges_1): Likewise.
15340 (fold_predicate_in): Likewise.
15341 (evrp_dom_walker::before_dom_children): Likewise.
15342 (evrp_dom_walker::push_value_range): Likewise.
15343 (evrp_dom_walker::pop_value_range): Likewise.
15344 (execute_early_vrp): Likewise.
15346 2017-05-16 Richard Biener <rguenther@suse.de>
15348 * dwarf2out.c (loc_list_from_tree_1): Do not create
15349 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15351 2017-05-16 Richard Biener <rguenther@suse.de>
15353 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15355 (note_variable_value_in_expr): If we resolved the decl ref
15356 do not push to the stack.
15358 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15360 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15361 operations in fast-math mode.
15362 (vaddq_f16): Likewise.
15363 (vmul_f16): Likewise.
15364 (vmulq_f16): Likewise.
15365 (vsub_f16): Likewise.
15366 (vsubq_f16): Likewise.
15367 * config/arm/neon.md (add<mode>3): New.
15369 (fma:<VH:mode>3): New. Also remove outdated comment.
15372 2017-05-16 Martin Liska <mliska@suse.cz>
15376 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15377 (odr_types_equivalent_p): Likewise.
15379 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15381 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15383 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15386 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15387 non-interunit SSE move alternatives with '?'.
15388 (zero-extendsidi peephole2): New peephole to skip intermediate
15389 general register in SSE zero-extend sequence.
15391 2017-05-15 Jeff Law <law@redhat.com>
15393 * reorg.c (relax_delay_slots): Create a new variable to hold
15394 the temporary target rather than clobbering TARGET_LABEL.
15396 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15397 missing argument to extract_bit_field call.
15398 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15400 2017-05-15 Martin Liska <mliska@suse.cz>
15403 * gcc.c (process_command): Do not allow empty argument of -o option.
15405 2017-05-15 Renlin Li <renlin.li@arm.com>
15407 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15408 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15409 * config/aarch64/constraints.md (Usf): Add long call check.
15410 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15411 (call_value): Likewise.
15412 (sibcall): Likewise.
15413 (sibcall_value): Likewise.
15415 (call_value_insn): New.
15416 (sibcall_insn): Update rtx pattern.
15417 (sibcall_value_insn): Likewise.
15418 (call_internal): Remove.
15419 (call_value_internal): Likewise.
15420 (sibcall_internal): Likewise.
15421 (sibcall_value_internal): Likewise.
15422 (call_reg): Likewise.
15423 (call_symbol): Likewise.
15424 (call_value_reg): Likewise.
15425 (call_value_symbol): Likewise.
15427 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15430 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15432 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15434 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15435 compatible with CCGOCmode and with CCZmode.
15437 2017-05-14 Martin Sebor <msebor@redhat.com>
15439 PR middle-end/77671
15440 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15441 (gimple_fold_builtin_snprintf): Same.
15442 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15443 (gimple_fold_builtin_snprintf): Same.
15444 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15445 of character types.
15446 (is_call_safe): New function.
15447 (try_substitute_return_value): Call it.
15448 (try_simplify_call): New function.
15449 (pass_sprintf_length::handle_gimple_call): Call it.
15451 2017-05-14 Martin Sebor <msebor@redhat.com>
15453 PR middle-end/80669
15454 * builtins.c (expand_builtin_stpncpy): Simplify.
15456 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15458 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15459 * config/i386/i386.h
15460 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15461 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15462 (struct machine_function): Add new members call_ms2sysv,
15463 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15464 (struct machine_frame_state): New fields sp_realigned and
15465 sp_realigned_offset.
15466 * config/i386/i386.c
15467 (enum xlogue_stub): New enum.
15468 (enum xlogue_stub_sets): New enum.
15469 (class xlogue_layout): New class.
15470 (struct ix86_frame): New fields stack_realign_allocate_offset,
15471 stack_realign_offset and outlined_save_offset. Modify comments to
15472 detail stack layout when using out-of-line stubs.
15473 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15474 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15475 -mcall-ms2sysv-xlogues.
15476 (stub_managed_regs): New static variable.
15477 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15478 registers managed by out-of-line stub.
15479 (disable_call_ms2sysv_xlogues): New function.
15480 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15481 m->call_ms2sysv when appropriate and compute frame layout for
15483 (sp_valid_at, fp_valid_at): New inline functions.
15484 (choose_basereg): New function.
15485 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15487 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15488 Use align parameter of choose_baseaddr to generated aligned SSE movs
15490 (pro_epilogue_adjust_stack): Modify to track
15491 machine_frame_state::sp_realigned.
15492 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15493 (ix86_nsaved_sseregs): Likewise.
15494 (ix86_emit_save_regs): Likewise.
15495 (ix86_emit_save_regs_using_mov): Likewise.
15496 (ix86_emit_save_sse_regs_using_mov): Likewise.
15497 (get_scratch_register_on_entry): Likewise.
15498 (gen_frame_set): New function.
15499 (gen_frame_load): Likewise.
15500 (gen_frame_store): Likewise.
15501 (emit_outlined_ms2sysv_save): Likewise.
15502 (emit_outlined_ms2sysv_restore): Likewise.
15503 (ix86_expand_prologue): Modify stack re-alignment code and call
15504 emit_outlined_ms2sysv_save when appropriate.
15505 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15506 parameter rtx_insn *insn, which allows the function to be used to only
15507 generate the notes.
15508 (ix86_expand_epilogue): Modify validity checks of frame and stack
15509 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15510 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15511 * config/i386/predicates.md
15512 (save_multiple): New predicate.
15513 (restore_multiple): Likewise.
15514 * config/i386/sse.md
15515 (save_multiple<mode>): New pattern.
15516 (save_multiple_realign<mode>): Likewise.
15517 (restore_multiple<mode>): Likewise.
15518 (restore_multiple_and_return<mode>): Likewise.
15519 (restore_multiple_leave_return<mode>): Likewise.
15520 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15522 2017-05-14 Julia Koval <julia.koval@intel.com>
15524 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15525 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15526 (__builtin_ia32_xsetbv): New builtins.
15527 * config/i386/i386.c (ix86_expand_special_args_builtin):
15529 (ix86_expand_builtin): Special expand for new intrinsics.
15530 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15531 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15532 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15534 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15536 * cfganal.c (inverted_post_order_compute): Change argument type
15538 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15539 * df-core.c (rest_of_handle_df_initialize): Adjust.
15540 (rest_of_handle_df_finish): Likewise.
15541 (df_analyze_1): Likewise.
15542 (df_analyze): Likewise.
15543 (loop_inverted_post_order_compute): Change argument to be a vec *.
15544 (df_analyze_loop): Adjust.
15545 (df_get_n_blocks): Likewise.
15546 (df_get_postorder): Likewise.
15547 * df.h (struct df_d): Change field to be a vec.
15548 * lcm.c (compute_laterin): Adjust.
15549 (compute_available): Likewise.
15550 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15551 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15552 * tree-ssa-pre.c (compute_antic): Likewise.
15554 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15556 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15557 (depth_first_search::depth_first_search): Change structure init
15558 function to this constructor.
15559 (depth_first_search::add_bb): Rename function to this member.
15560 (depth_first_search::execute): Likewise.
15561 (flow_dfs_compute_reverse_finish): Adjust.
15563 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15565 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15566 (longest_simple_path): Likewise.
15567 * shrink-wrap.c (spread_components): Likewise.
15568 (disqualify_problematic_components): Likewise.
15569 (emit_common_heads_for_components): Likewise.
15570 (emit_common_tails_for_components): Likewise.
15571 (insert_prologue_epilogue_for_components): Likewise.
15573 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15575 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15578 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15580 * df-core.c (df_set_blocks): Start using auto_bitmap.
15581 (df_compact_blocks): Likewise.
15582 * df-problems.c (df_rd_confluence_n): Likewise.
15583 * df-scan.c (df_insn_rescan_all): Likewise.
15584 (df_process_deferred_rescans): Likewise.
15585 (df_update_entry_block_defs): Likewise.
15586 (df_update_exit_block_uses): Likewise.
15587 (df_entry_block_bitmap_verify): Likewise.
15588 (df_exit_block_bitmap_verify): Likewise.
15589 (df_scan_verify): Likewise.
15590 * lra-constraints.c (lra_constraints): Likewise.
15591 (undo_optional_reloads): Likewise.
15592 (lra_undo_inheritance): Likewise.
15593 * lra-remat.c (calculate_gen_cands): Likewise.
15594 (do_remat): Likewise.
15595 * lra-spills.c (assign_spill_hard_regs): Likewise.
15596 (spill_pseudos): Likewise.
15597 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15598 (bitmap_set_subtract_values): Likewise.
15600 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15602 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15603 management with auto_bitmap.
15604 (fix_inter_tick): Likewise.
15605 (fix_recovery_deps): Likewise.
15606 * ira.c (add_store_equivs): Likewise.
15607 (find_moveable_pseudos): Likewise.
15608 (split_live_ranges_for_shrink_wrap): Likewise.
15609 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15610 (rtx_reuse_manager::seen_def_p): Likewise.
15611 (rtx_reuse_manager::set_seen_def): Likewise.
15612 * print-rtl.h (class rtx_reuse_manager): Likewise.
15614 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15616 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15618 (migrate_btr_def): Likewise.
15619 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15620 * df-core.c (loop_post_order_compute): Likewise.
15621 (loop_inverted_post_order_compute): Likewise.
15622 * hsa-common.h: Likewise.
15623 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15624 * init-regs.c (initialize_uninitialized_regs): Likewise.
15625 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15626 (inline_small_functions): Likewise.
15627 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15628 * ira.c (combine_and_move_insns): Likewise.
15629 (build_insn_chain): Likewise.
15630 * loop-invariant.c (find_invariants): Likewise.
15631 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15632 * predict.c (tree_predict_by_opcode): Likewise.
15633 (predict_paths_leading_to): Likewise.
15634 (predict_paths_leading_to_edge): Likewise.
15635 (estimate_loops_at_level): Likewise.
15636 (estimate_loops): Likewise.
15637 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15638 (spread_components): Likewise.
15639 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15640 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15641 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15642 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15643 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15644 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15645 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15646 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15647 (mark_threaded_blocks): Likewise.
15648 (thread_through_all_blocks): Likewise.
15649 * tree-ssa.c (verify_ssa): Likewise.
15650 (execute_update_addresses_taken): Likewise.
15651 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15653 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15655 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15657 (post_order_compute): Likewise.
15658 (inverted_post_order_compute): Likewise.
15659 (pre_and_rev_post_order_compute_fn): Likewise.
15661 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15663 * genrecog.c (int_set::int_set): Explicitly construct our
15664 auto_vec base class.
15665 * vec.h (auto_vec::auto_vec): New constructor.
15667 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15669 * bitmap.h (class auto_bitmap): New constructor taking
15670 bitmap_obstack * argument.
15672 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15674 * bitmap.h (class auto_bitmap): Change type of m_bits to
15675 bitmap_head, and adjust ctor / dtor and member operators.
15677 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15679 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15680 when returned register mode doesn't match original mode.
15682 2017-05-12 Jeff Law <law@redhat.com>
15683 Jakub Jelinek <jakub@redhat.com>
15685 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15686 we look for cc setter after the compare-elim changes.
15687 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15688 within the vector to match what compare-elim now expects.
15689 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15690 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15692 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15693 after the compare-elim changes.
15694 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15695 the vector to match what compare-elim now expects.
15696 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15697 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15698 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15699 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15700 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15702 * config/visium/visium.c (single_set_and_flags): Fix where
15703 we look for cc setter after the compare-elim changes.
15704 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15705 with the vector to match what compare-elim now expects.
15706 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15707 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15708 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15709 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15710 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15711 (neg<mode>2_insn_set_overflow): Likewise.
15713 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15715 PR middle-end/79794
15716 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15717 maybe_expand_insn call, set ops[0].target. If still set after call,
15718 set alt_rtl. Add extra arg to recursive calls.
15719 (extract_bit_field): Add alt_rtl argument. Pass to
15721 * expmed.h (extract_bit_field): Fix prototype.
15722 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15723 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15724 to extract_bit_field_calls.
15725 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15726 Pass alt_rtl to extract_bit_field calls.
15727 * calls.c (store_unaligned_arguments_into_psuedos)
15728 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15729 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15731 * optabs.h (struct expand_operand): Add target bitfield.
15733 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15735 * compare-elim.c (try_eliminate_compare): Canonicalize
15736 operation with embedded compare to
15737 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15738 (set (reg) (operation)].
15740 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15742 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15745 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15746 cost of adding a carry flag for ADC instruction.
15747 [case MINUS]: Ignore the cost of subtracting a carry flag
15748 for SBB instruction.
15750 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15752 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15754 * config/rs6000/bmiintrin.h: New file.
15755 * config/rs6000/bmi2intrin.h: New file.
15756 * config/rs6000/x86intrin.h: New file.
15758 2017-05-12 Jeff Law <law@redhat.com>
15760 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15763 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15765 PR middle-end/80707
15766 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15768 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15770 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15771 early expansion of vector divide builtins.
15772 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15773 builtins identified as having unsigned arguments.
15775 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15777 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15778 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15779 expansion of vector logical operations (and, andc, or, xor,
15782 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15784 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15785 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15787 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15789 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15790 early GIMPLE expansion of vector multiplies.
15792 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15794 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15795 TARGET_HAVE_MOVT conditional.
15796 (movt splitter): Likewise.
15798 2017-05-12 Richard Biener <rguenther@suse.de>
15800 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15801 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15802 Fold all stmts not inplace.
15804 2017-05-12 Richard Biener <rguenther@suse.de>
15806 PR tree-optimization/80713
15807 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15808 inserted_exprs bit for not removed stmts.
15810 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15812 PR middle-end/69921
15813 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15814 parallelized" attribute for parallelized OpenACC kernels.
15815 * omp-offload.c (execute_oacc_device_lower): Use it.
15817 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15818 Set "oacc kernels" attribute.
15819 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15820 parameter. Adjust all users.
15821 (oacc_fn_attrib_kernels_p): Remove function.
15822 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15823 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15824 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15825 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15826 assert "oacc kernels" attribute is set.
15828 2017-05-11 Carl Love <cel@us.ibm.com>
15830 * config/rs6000/rs6000-c: Add support for built-in functions
15831 vector unsigned char vec_popcnt (vector signed char)
15832 vector unsigned char vec_popcnt (vector unsigned char)
15833 vector unsigned short vec_popcnt (vector signed short)
15834 vector unsigned short vec_popcnt (vector unsigned short)
15835 vector unsigned int vec_popcnt (vector signed int)
15836 vector unsigned int vec_popcnt (vector unsigned int)
15837 vector unsigned long long vec_popcnt (vector signed long long)
15838 vector unsigned long long vec_popcnt (vector unsigned long long)
15839 vector signed long long vec_slo (vector signed long long,
15840 vector signed char)
15841 vector signed long long vec_slo (vector signed long long,
15842 vector unsigned char)
15843 vector unsigned long long vec_slo (vector unsigned long long,
15844 vector signed char)
15845 vector unsigned long long vec_slo (vector unsigned long long,
15846 vector unsigned char)
15847 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15848 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15849 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15850 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15851 * doc/extend.texi: Update the built-in documentation file for the
15852 new built-in functions.
15854 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15856 * attribs.h (sorted_attr_string): Move machine independent
15857 functions for target clone support from the i386 port to common
15858 code. Rename ix86_function_versions to common_function_versions.
15859 Rename make_name to make_unique_name.
15860 (common_function_versions): Likewise.
15861 (make_unique_name): Likewise.
15862 (make_dispatcher_decl): Likewise.
15863 (is_function_default_version): Likewise.
15864 * attribs.c (attr_strcmp): Likewise.
15865 (sorted_attr_string): Likewise.
15866 (common_function_versions): Likewise.
15867 (make_unique_name): Likewise.
15868 (make_dispatcher_decl): Likewise.
15869 (is_function_default_version): Likewise.
15870 * config/i386/i386.c (attr_strcmp): Likewise.
15871 (sorted_attr_string): Likewise.
15872 (ix86_function_versions): Likewise.
15873 (make_name): Likewise.
15874 (make_dispatcher_decl): Likewise.
15875 (is_function_default_version): Likewise.
15876 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15878 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15881 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15882 Account for direct move costs for vec_construct of integer
15885 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15888 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15889 (UNSPEC_STX_ATOMIC): Ditto.
15890 (loaddi_via_sse): New insn.
15891 (storedi_via_sse): Ditto.
15892 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15893 Update corresponding peephole2 patterns.
15894 (atomic_storedi_fpu): Ditto.
15896 2017-05-11 Julia Koval <julia.koval@intel.com>
15898 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15899 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15901 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15902 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15903 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15905 2017-05-11 Nathan Sidwell <nathan@acm.org>
15907 * graphite-poly.c: Include dumpfile.h.
15909 * dumpfle.h (dump_function): Declare here ...
15910 * tree-dump.h (dump_function): ... not here.
15911 * dumpfile.c: #include tree-cfg.h.
15912 (dump_function): Move here from ...
15913 * tree-dump.c (dump_function): ... here.
15914 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15915 * graphite-poly.c: Don't include tree-dump.h.
15916 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15917 * print-tree.c: Likewise.
15918 * stor-layout.c: Likewise.
15919 * tree-nested.c: Likewise.
15921 * dumpfile.c (dump_start): Use TDF_FLAGS.
15922 (dump_enable_all): Fix TDF_KIND check thinko.
15924 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15927 array entries to represent two legal parameterizations of the
15928 overloaded __builtin_cmpb function, as represented by the
15929 P6_OV_BUILTIN_CMPB constant.
15930 (altivec_resolve_overloaded_builtin): Add special case handling
15931 for the __builtin_cmpb function, as represented by the
15932 P6_OV_BUILTIN_CMPB constant.
15933 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15934 (BU_P6_64BIT_2): New macro.
15935 (BU_P6_OVERLOAD_2): New macro
15936 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15937 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15938 (CMPB): Add overload support to represent both 32-bit and 64-bit
15939 compare-bytes function.
15940 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15941 support for TARGET_CMPB.
15942 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15943 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15944 documentation of the __builtin_cmpb overloaded built-in function.
15946 2017-05-11 Richard Biener <rguenther@suse.de>
15948 PR tree-optimization/80705
15949 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15950 bases are not vectorizable.
15952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15954 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15955 when counting register pressure.
15957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15959 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15960 (iv_ca_more_deps): Renamed to ...
15961 (iv_ca_compare_deps): ... this.
15962 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15964 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15966 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15968 (determine_group_iv_costs): ... here.
15969 (find_inv_vars_cb): Record inv var if it's not recorded before.
15971 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15973 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15974 (get_shiftadd_cost): Ditto.
15976 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15978 * tree-ssa-address.c: Include header file.
15979 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15981 (add_to_parts): Refactor.
15982 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15983 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15986 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15988 PR tree-optimization/53090
15989 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15991 (extract_cond_operands): Detect condition with IV on both sides
15992 and return COMP_IV_EXPR_2.
15993 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15994 (rewrite_use_compare): Simplify by removing call to function
15995 extract_cond_operands.
15997 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15999 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16000 (extract_cond_operands): Detect condition comparing against non-
16001 invariant bound and return appropriate enum value.
16002 (find_interesting_uses_cond): Update use of extract_cond_operands.
16003 Handle its return value accordingly.
16004 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16006 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16008 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16009 nonlinear iv_use computation in loop invariant sensitive way.
16011 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16013 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16014 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16016 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16018 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16019 (dump_cand): Support iv_cand.inv_exprs.
16020 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16022 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16027 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16029 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16030 as local function. Include necessary header files.
16031 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16033 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16035 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16037 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16039 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16040 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16041 RSHIFT_EXPR and BIT_NOT_EXPR.
16043 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16045 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16046 (adjust_setup_cost): New parameter supporting round up adjustment.
16047 (struct address_cost_data): Delete.
16048 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16049 (split_address_cost, ptr_difference_cost): Delete.
16050 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16051 (struct ainc_cost_data): New struct.
16052 (get_address_cost_ainc): New function.
16053 (get_address_cost, get_computation_cost): Reimplement.
16054 (determine_group_iv_cost_address): Record inv_expr for all uses of
16056 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16057 (iv_ca_has_deps): Reimplemented to ...
16058 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16060 (iv_ca_extend): Call iv_ca_more_deps.
16062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16064 * tree-ssa-address.c (struct mem_address): Move to header file.
16065 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16066 * tree-ssa-address.h (struct mem_address): Move from C file.
16067 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16069 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16071 * tree-affine.h (aff_combination_type): New interface.
16072 (aff_combination_zero_p): Remove static.
16073 (aff_combination_const_p): New interface.
16074 (aff_combination_singleton_var_p): New interfaces.
16076 2017-05-11 Richard Biener <rguenther@suse.de>
16078 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16079 Skip unreachable blocks and destinations.
16080 (eliminate): Move stmt removal and fixup ...
16081 (fini_eliminate): ... here. Skip inserted exprs.
16082 (pass_pre::execute): Move fini_pre after fini_eliminate.
16083 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16084 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16085 PRE to get rid of dead code that has invalid SSA form and
16086 split critical edges again.
16088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16090 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16092 2017-05-11 Richard Biener <rguenther@suse.de>
16094 * passes.c (execute_function_todo): Verify loops if they are
16095 said to be up-to-date.
16096 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16097 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16099 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16102 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16103 handle calling assemble_external ourself.
16106 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16107 modes with zero size. Enhance comment.
16109 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16111 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16112 built-ins for vec_xl and vec_xst with short and char pointer
16115 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16117 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16118 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16119 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16120 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16121 (_mm_maskz_min_round_ss): New intrinsics.
16122 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16123 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16124 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16125 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16126 (__builtin_ia32_minss_mask_round): New builtins.
16127 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16128 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16129 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16131 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16132 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16134 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16137 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16139 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16140 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16141 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16142 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16143 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16144 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16145 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16146 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16147 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16148 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16149 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16150 (__builtin_ia32_mulss_mask_round): New builtins.
16151 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16152 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16153 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16155 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16156 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16158 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16161 2017-05-10 Julia Koval <julia.koval@intel.com>
16163 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16164 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16165 (_mm256_setr_m128i): New intrinsics.
16167 2017-05-10 Julia Koval <julia.koval@intel.com>
16169 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16170 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16171 (_mm_maskz_rcp14_ss): New intrinsics.
16172 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16173 (__builtin_ia32_rcp14ss_mask): New builtins.
16174 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16176 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16178 PR tree-optimization/51513
16179 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16180 (assert_unreachable_fallthru_edge_p): Use it.
16181 (group_case_labels_stmt): Likewise.
16182 * tree-cfg.h: Prototype it.
16183 * stmt.c: Include cfghooks.h and tree-cfg.h.
16184 (emit_case_dispatch_table) <gap_label>: New local variable.
16185 Use it to fill dispatch table gaps.
16186 Test for default_label before updating probabilities.
16187 (expand_case) <default_label>: Remove unneeded initialization.
16188 Test for unreachable default case statement and remove its edge.
16189 Set default_label accordingly.
16190 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16192 2017-05-10 Carl Love <cel@us.ibm.com>
16194 * config/rs6000/rs6000-c: Add support for built-in functions
16195 vector signed char vec_neg (vector signed char)
16196 vector signed short int vec_neg (vector short int)
16197 vector signed int vec_neg (vector signed int)
16198 vector signed long long vec_neg (vector signed long long)
16199 vector float vec_neg (vector float)
16200 vector double vec_neg (vector double)
16201 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16203 * config/rs6000/altivec.h: Add define for vec_neg
16204 * doc/extend.texi: Update the built-in documentation for the
16205 new built-in functions.
16207 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16209 PR tree-optimization/77644
16210 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16212 2017-05-10 Nathan Sidwell <nathan@acm.org>
16214 * dumpfile.h (TDI_lang_all): New.
16215 (TDF_KIND): New. Renumber others
16216 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16218 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16220 (get_dump_file_name): Adjust suffix generation.
16221 (dump_enable_all): Use TDF_KIND.
16222 * doc/invoke.texi (-fdump-lang-all): Document.
16224 * dumpfile.h: Tabify.
16226 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16229 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16230 Move member access before delete.
16232 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16234 * tree-inline.c (expand_call_inline): Split block at stmt
16237 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16240 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16241 are now unused after splitting mov{sf,sd}_hardfloat.
16242 (f32_lr2): Likewise.
16243 (f32_lm): Likewise.
16244 (f32_lm2): Likewise.
16245 (f32_li): Likewise.
16246 (f32_li2): Likewise.
16247 (f32_lv): Likewise.
16248 (f32_sr): Likewise.
16249 (f32_sr2): Likewise.
16250 (f32_sm): Likewise.
16251 (f32_sm2): Likewise.
16252 (f32_si): Likewise.
16253 (f32_si2): Likewise.
16254 (f32_sv): Likewise.
16255 (f32_dm): Likewise.
16256 (f32_vsx): Likewise.
16257 (f32_av): Likewise.
16258 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16259 For movsf, order stores so the VSX stores occur before the GPR
16260 store which encourages the register allocator to use a traditional
16261 FPR instead of a GPR. For movsd, order the stores so that the GPR
16262 store comes before the VSX stores to allow the power6 to work.
16263 This is due to the power6 not having a 32-bit integer store
16264 instruction from a FPR.
16265 (movsf_hardfloat): Likewise.
16266 (movsd_hardfloat): Likewise.
16268 2017-05-09 Martin Sebor <msebor@redhat.com>
16270 PR translation/80280
16271 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16274 PR translation/80280
16275 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16276 data member added in r247778.
16277 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16279 2017-05-09 Nathan Sidwell <nathan@acm.org>
16281 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16283 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16286 2017-05-09 Marek Polacek <polacek@redhat.com>
16288 * doc/invoke.texi: Fix typo.
16290 2017-05-09 Richard Biener <rguenther@suse.de>
16292 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16293 (vrp_val_is_min): Likewise.
16294 (set_value_range_to_value): Likewise.
16295 (set_value_range_to_nonnegative): Likewise.
16296 (gimple_assign_nonzero_p): Likewise.
16297 (gimple_stmt_nonzero_p): Likewise.
16298 (vrp_int_const_binop): Likewise. Remove unreachable case.
16299 (adjust_range_with_scev): Adjust comments.
16300 (compare_range_with_value): Likewise.
16301 (extract_range_from_phi_node): Likewise.
16302 (test_for_singularity): Likewise.
16304 2017-05-09 Richard Biener <rguenther@suse.de>
16306 * tree-vrp.c (get_single_symbol): Add assert that we don't
16307 get overflowed constants as invariant part.
16308 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16309 checks. Use wi::cmp instead of recursing for integer constants.
16310 (compare_values): Just ignore whether we assumed undefined
16311 overflow instead of failing the compare.
16312 (extract_range_for_var_from_comparison_expr): Add comment before the
16313 TREE_NO_WARNING sets.
16314 (test_for_singularity): Likewise.
16315 (extract_range_from_comparison): Do not disable optimization
16316 when we assumed undefined overflow.
16317 (extract_range_basic): Remove init of unused var.
16319 2017-05-09 Richard Biener <rguenther@suse.de>
16321 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16322 (extract_range_from_multiplicative_op_1): Adjust.
16323 (extract_range_from_binary_expr_1): Use int_const_binop.
16325 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16328 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16329 rs6000_store_data_bypass_p in seven define_bypass directives and
16330 in several comments.
16331 * config/rs6000/rs6000-protos.h: Add prototype for
16332 rs6000_store_data_bypass_p function.
16333 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16334 function implements slightly different (rs6000-specific) semantics
16335 than store_data_bypass_p, returning false rather than aborting
16336 with assertion error when arguments do not satisfy the
16337 requirements of store data bypass.
16338 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16339 rs6000_store_data_bypass_p.
16341 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16343 * config/xtensa/xtensa-protos.h
16344 (xtensa_initial_elimination_offset): New declaration.
16345 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16346 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16347 macro definition, add case for FRAME_POINTER_REGNUM when
16348 FRAME_GROWS_DOWNWARD.
16349 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16350 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16351 xtensa_initial_elimination_offset.
16353 2017-05-08 Nathan Sidwell <nathan@acm.org>
16355 * doc/invoke.texi: Alphabetize -fdump options.
16357 2017-05-08 Martin Sebor <msebor@redhat.com>
16359 PR translation/80280
16360 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16362 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16364 * target.def (compute_frame_layout): New optional target hook.
16365 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16366 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16367 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16369 * reload1.c (verify_initial_elim_offsets): Likewise.
16370 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16371 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16372 (arm_get_frame_offsets): Split up into this ...
16373 (arm_compute_frame_layout): ... and this function.
16375 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16377 * config/aarch64/constraints.md (Usa): New constraint.
16378 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16380 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16382 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16383 with_multilib_list after it has been checked.
16385 2017-05-08 Richard Biener <rguenther@suse.de>
16387 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16388 (bitmap_set_subtract_values): Likewise.
16390 2017-05-08 Richard Biener <rguenther@suse.de>
16392 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16393 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16395 (gimple_stmt_nonzero_warnv_p): Rename to ...
16396 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16398 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16399 (extract_range_basic): Adjust, do not disable propagation on
16400 strict overflow sensitive simplification.
16401 (vrp_visit_cond_stmt): Likewise.
16403 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16405 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16406 body size unconditionally.
16408 2017-05-07 Jeff Law <law@redhat.com>
16411 2017-05-06 Jeff Law <law@redhat.com>
16412 PR tree-optimization/78496
16413 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16416 PR tree-optimization/78496
16417 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16418 (simplify_stmt_using_ranges): Call it.
16419 (vrp_dom_walker::before_dom_children): Extract equivalences
16420 from an ASSERT_EXPR with an equality comparison against a
16423 2017-05-06 Jeff Law <law@redhat.com>
16425 PR tree-optimization/78496
16426 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16429 PR tree-optimization/78496
16430 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16431 (simplify_stmt_using_ranges): Call it.
16432 (vrp_dom_walker::before_dom_children): Extract equivalences
16433 from an ASSERT_EXPR with an equality comparison against a
16436 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16438 * lra-constraints.c (lra_copy_reg_equiv): New function.
16439 (split_reg): Use it to copy equivalence information from the
16440 original register to the spill register.
16442 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16444 PR rtl-optimization/75964
16445 * simplify-rtx.c (simplify_const_relational_operation): Remove
16446 invalid handling of comparisons of integer ABS.
16448 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16450 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16451 initialize to zero.
16452 (init_regs): Remove declaration.
16453 (function_arg_advance_32): Initialize error_p as boolean variable.
16455 2017-05-05 Nathan Sidwell <nathan@acm.org>
16457 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16458 lines. Use for (;;).
16460 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16462 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16463 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16464 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16465 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16466 VF=2 that require versioning.
16468 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16470 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16473 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16475 * diagnostic.h (diagnostic_override_option_index): Convert from
16476 macro to inline function.
16478 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16480 * diagnostic.c (last_module_changed_p): New function.
16481 (set_last_module): New function.
16482 (diagnostic_report_current_module): Convert macro usage to
16483 the above functions.
16484 * diagnostic.h (diagnostic_context::last_module): Strengthen
16485 from const line_map * to const line_map_ordinary *.
16486 (diagnostic_last_module_changed): Delete macro.
16487 (diagnostic_set_last_module): Delete macro.
16489 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16491 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16492 with diagnostic_report_diagnostic.
16493 (diagnostic_n_impl_richloc): Likewise.
16494 * diagnostic.h (report_diagnostic): Delete macro.
16495 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16496 with diagnostic_report_diagnostic.
16497 * substring-locations.c (format_warning_va): Likewise.
16499 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16501 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16502 save/restor of format_spec. Move option-printing code to...
16503 (print_option_information): ...this new function, and
16504 reimplement by simply printing to the pretty_printer,
16505 rather than appending to the format string.
16507 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16509 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16510 handling logic into...
16511 (update_effective_level_from_pragmas): ...this new function.
16513 2017-05-04 Andrew Waterman <andrew@sifive.com>
16515 * config/riscv/riscv.opt (mstrict-align): New option.
16516 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16517 (SLOW_UNALIGNED_ACCESS): Define.
16518 (riscv_slow_unaligned_access): Declare.
16519 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16521 (riscv_slow_unaligned_access): New variable.
16522 (rocket_tune_info): Set slow_unaligned_access to true.
16523 (optimize_size_tune_info): Set slow_unaligned_access to false.
16524 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16525 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16526 (riscv_option_override): Set riscv_slow_unaligned_access.
16527 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16529 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16531 * config/riscv/riscv.md: Unify indentation.
16533 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16538 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16540 (extendsi<mode>2): Add support for doing sign extension via
16541 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16542 don't have ISA 3.0 instructions.
16543 (extendsi<mode>2 splitter): Likewise.
16544 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16545 generate the normal insns since SImode can now go in vector
16546 registers. Disallow the special UNSPECs needed for previous
16547 machines to hide SImode being used. Add new insns
16548 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16549 (fix_trunc<mode>si2_stfiwx): Likewise.
16550 (fix_trunc<mode>si2_internal): Likewise.
16551 (fixuns_trunc<mode>si2): Likewise.
16552 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16553 (fctiw<u>z_<mode>_smallint): Likewise.
16554 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16555 of floating point to 32-bit integer from doing a direct move to
16556 the GPR registers to do a store.
16557 (fctiwz_<mode>): Break long line.
16559 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16561 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16562 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16563 (addr_list, addr_offset_valid_p): New.
16564 (split_address_groups): Check offset validity with above function.
16565 (gt-tree-ssa-loop-ivopts.h): Include header file.
16567 2017-05-05 Nathan Sidwell <nathan@acm.org>
16569 * config.gcc (arm*-*-*): Add missing 'fi'.
16571 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16573 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16574 included in -fopt-info does not matter.
16575 * doc/optinfo.texi (-fopt-info): Fix description of default
16576 behavour. Explicitly say order of options included in -fopt-info
16579 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16581 * config.gcc: Allow combinations of aprofile and rmprofile values for
16582 --with-multilib-list.
16583 * config/arm/t-multilib: New file.
16584 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16585 variables. Remove setting of ISA and floating-point ABI in
16586 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16587 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16588 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16589 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16591 * config/arm/t-rmprofile: Likewise except for the matches changes.
16592 * doc/install.texi (--with-multilib-list): Document the combination of
16593 aprofile and rmprofile values and warn about pitfalls in doing that.
16595 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16597 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16598 (movdi_aarch64): Likewise.
16600 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16602 PR tree-optimization/80632
16603 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16605 (build_arrays): Initialize it for virtual phis.
16606 (fix_phi_nodes): Use it for virtual phis.
16608 PR tree-optimization/80558
16609 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16610 [x, y] op z into [x op, y op z] for op & or | if conditions
16613 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16614 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16617 * config/arm/arm.md (use_literal_pool): Remove.
16618 (64-bit immediate split): No longer takes cost into consideration
16619 if arm_disable_literal_pool is enabled.
16620 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16621 used when arm_disable_literal_pool is enabled.
16622 (arm_max_const_double_inline_cost): Remove use of
16623 arm_disable_literal_pool.
16624 (push_minipool_fix): Add assert.
16625 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16626 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16627 (no_literal_pool_sf_immediate): New.
16629 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16631 PR tree-optimization/80613
16632 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16633 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16635 2017-05-05 Richard Biener <rguenther@suse.de>
16637 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16639 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16641 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16642 of this flag from insn conditions due to removal from r247495.
16644 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16646 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16648 (arm_early_store_addr_dep_ptr): Likewise.
16649 * config/arm/aarch-common-protos.h
16650 (arm_early_load_addr_dep_ptr): Add prototype.
16651 (arm_early_store_addr_dep_ptr): Likewise.
16652 * config/arm/cortex-a53.md: Add new bypasses.
16654 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16656 * tree.c (next_type_uid): Change type to unsigned.
16657 (type_hash_canon): Decrement back next_type_uid if
16658 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16659 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16662 2017-05-04 Martin Sebor <msebor@redhat.com>
16664 * builtins.c: Fix a trivial typo in a comment.
16666 PR middle-end/79234
16667 * builtins.c (check_sizes): Adjust to handle reading past the end.
16668 Avoid printing excessive upper bound of ranges. Use %E to print
16669 tree nodes instead of converting them to %wu.
16670 (expand_builtin_memchr): New function.
16671 (compute_dest_size): Rename...
16672 (compute_objsize): ...to this.
16673 (expand_builtin_memcpy): Adjust.
16674 (expand_builtin_mempcpy): Adjust.
16675 (expand_builtin_strcat): Adjust.
16676 (expand_builtin_strcpy): Adjust.
16677 (check_strncat_sizes): Adjust.
16678 (expand_builtin_strncat): Adjust.
16679 (expand_builtin_strncpy): Adjust and simplify.
16680 (expand_builtin_memset): Adjust.
16681 (expand_builtin_bzero): Adjust.
16682 (expand_builtin_memcmp): Adjust.
16683 (expand_builtin): Handle memcmp.
16684 (maybe_emit_chk_warning): Check strncat just once.
16686 2017-05-04 Martin Sebor <msebor@redhat.com>
16688 PR preprocessor/79214
16689 PR middle-end/79222
16690 PR middle-end/79223
16691 * builtins.c (check_sizes): Add inlining context and issue
16692 warnings even when -Wno-system-headers is set.
16693 (check_strncat_sizes): Same.
16694 (expand_builtin_strncat): Same.
16695 (expand_builtin_memmove): New function.
16696 (expand_builtin_stpncpy): Same.
16697 (expand_builtin): Handle memmove and stpncpy.
16699 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16701 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16702 which is not used any more.
16704 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16706 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16708 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16710 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16711 (cortexa53_tunings): Likewise.
16712 (cortexa57_tunings): Likewise.
16713 (cortexa72_tunings): Likewise.
16714 (cortexa73_tunings): Likewise.
16716 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16718 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16719 Set loop alignment to 8.
16721 2017-05-04 Martin Sebor <msebor@redhat.com>
16723 PR translation/80280
16724 * builtins.c (expand_builtin_object_size): Add missing quoting to
16725 %D and like directives.
16726 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16727 (hsa_type_for_tree_type): Same.
16728 (verify_function_arguments): Same.
16729 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16730 * varasm.c (get_section): Same.
16733 2017-05-04 Martin Sebor <msebor@redhat.com>
16735 PR translation/80280
16736 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16738 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16740 * config/aarch64/aarch64.c (generic_addrcost_table):
16741 Change HI/TI mode setting.
16743 2017-05-04 Martin Jambor <mjambor@suse.cz>
16745 PR tree-optimization/80622
16746 * tree-sra.c (comes_initialized_p): New function.
16747 (build_accesses_from_assign): Only set write lazily when
16748 comes_initialized_p is false.
16749 (analyze_access_subtree): Use comes_initialized_p.
16750 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16751 instead of testing for PARM_DECL.
16753 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16755 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16756 constraint on operand 0 to allow more general addressing modes.
16757 Adjust output template.
16758 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16760 * config/aarch64/aarch64-protos.h
16761 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16762 * config/aarch64/constraints.md (Dp): New address constraint.
16763 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16766 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16768 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16769 update use of estimate_ipcp_clone_size_and_time.
16770 (estimate_local_effects): Update use of
16771 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16772 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16773 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16774 Return nonspecialized time.
16776 2017-05-04 Richard Biener <rguenther@suse.de>
16778 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16779 for the last VUSE which def dominates the PHI. Directly call
16781 (get_continuation_for_phi_1): Remove.
16783 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16785 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16786 to explain the use of truncating division. Cap the number of
16787 iterations to the maximum given by nb_iterations_upper_bound,
16790 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16792 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16793 * configure: Regenerate.
16794 * config.in: Regenerate.
16795 * config/i386/driver-mingw32.c: new file.
16796 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16797 * config.host: Link driver-mingw32.o on MinGW host.
16798 * doc/install.texi: Document new --enable-mingw-wildcard configure
16801 2017-05-04 Marek Polacek <polacek@redhat.com>
16803 PR tree-optimization/80612
16804 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16806 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16807 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16809 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16810 (movt splitter): Likewise.
16811 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16812 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16813 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16814 block for Thumb-1 with MOVT.
16815 (thumb2_legitimate_address_p): Move code block ...
16816 (can_avoid_literal_pool_for_label_p): ... into this new function.
16817 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16819 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16820 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16821 "M-profile targets with the MOVT instruction".
16823 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16825 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16826 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16827 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16829 2017-05-04 Martin Liska <mliska@suse.cz>
16831 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16832 variable cond_code.
16834 2017-05-04 Richard Biener <rguenther@suse.de>
16836 * tree.c (array_at_struct_end_p): Handle arrays at struct
16837 end with flexarrays more conservatively. Refactor and treat
16838 arrays of arrays or aggregates more strict. Fix
16839 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16840 * tree.c (array_at_struct_end_p): Adjust prototype.
16841 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16842 * gimple-fold.c (get_range_strlen): Likewise.
16843 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16845 2017-05-04 Richard Biener <rguenther@suse.de>
16847 PR tree-optimization/31130
16848 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16850 (supports_overflow_infinity): Likewise.
16851 (is_negative_overflow_infinity): Likewise.
16852 (is_positive_overflow_infinity): Likewise.
16853 (is_overflow_infinity): Likewise.
16854 (stmt_overflow_infinity): Likewise.
16855 (overflow_infinity_range_p): Likewise.
16856 (usable_range_p): Remove as always returning true.
16857 (make_overflow_infinity): Remove.
16858 (negative_overflow_infinity): Likewise.
16859 (positive_overflow_infinity): Likewise.
16860 (avoid_overflow_infinity): Likewise.
16861 (set_value_range): Adjust accordingly.
16862 (set_value_range_to_nonnegative): Likewise, remove now unused
16863 overflow_infinity arg.
16864 (vrp_operand_equal_p): Adjust.
16865 (update_value_range): Likewise.
16866 (range_int_cst_singleton_p): Likewise.
16867 (operand_less_p): Likewise.
16868 (compare_values_warnv): Likewise.
16869 (extract_range_for_var_from_comparison_expr): Likewise.
16870 (vrp_int_const_binop): Likewise.
16871 (zero_nonzero_bits_from_vr): Likewise.
16872 (extract_range_from_multiplicative_op_1): Likewise.
16873 (extract_range_from_binary_expr_1): Likewise.
16874 (extract_range_from_unary_expr): Likewise.
16875 (extract_range_from_comparison): Likewise.
16876 (extract_range_basic): Likewise.
16877 (adjust_range_with_scev): Likewise.
16878 (compare_ranges): Likewise.
16879 (compare_range_with_value): Likewise.
16880 (dump_value_range): Likewise.
16881 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16883 (simplify_cond_using_ranges): Adjust.
16885 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16887 * brig-builtins.def: Added a builtin for class_f64.
16888 * builtin-types.def: Added a builtin type needed by class_f64.
16890 2017-05-03 Jason Merrill <jason@redhat.com>
16892 * timevar.def: Add TV_CONSTEXPR.
16894 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16896 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16898 2017-05-03 Martin Jambor <mjambor@suse.cz>
16900 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16901 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16902 * ipa-cp.c (ipcp_propagate_stage): Do not call
16903 ipa_update_after_lto_read.
16904 * ipa-inline.c (ipa_inline): Likewise.
16906 2017-05-03 Martin Jambor <mjambor@suse.cz>
16908 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16909 tag. Added a default constructor and a destructor.
16910 (ipa_edge_args_sum_t): New class;
16911 (ipa_edge_args_sum): Declare.
16912 (ipa_edge_args_vector): Remove declaration.
16913 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16914 (ipa_free_edge_args_substructures): Remove declaration.
16915 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16916 (ipa_edge_args_info_available_for_edge_p): Likewise.
16917 * ipa-prop.c (ipa_edge_args_vector): Removed.
16918 (edge_removal_hook_holder): Likewise.
16919 (edge_duplication_hook_holder): Likewise.
16920 (ipa_edge_args_sum): New variable.
16921 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16922 ipa_edge_args_vector.
16923 (ipa_free_edge_args_substructures): Likewise.
16924 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16925 ipa_edge_args_vector.
16926 (ipa_edge_removal_hook): Turned into method
16927 ipa_edge_args_sum_t::remove.
16928 (ipa_edge_duplication_hook): Turned into method
16929 ipa_edge_args_sum_t::duplicate.
16930 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16931 registering edge hooks.
16932 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16933 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16934 ipa_edge_args_sum instead of ipa_edge_args_vector.
16935 * ipa-profile.c (ipa_profile): Likewise.
16937 2017-05-03 Martin Jambor <mjambor@suse.cz>
16939 * symbol-summary.h (function_summary): New method exists.
16940 (function_summary::symtab_removal): Deallocate through release.
16941 (call_summary): New class.
16942 (gt_ggc_mx): New overload.
16943 (gt_pch_nx): Likewise.
16944 (gt_pch_nx): Likewise.
16946 2017-05-03 Jeff Law <law@redhat.com>
16948 PR tree-optimization/78496
16949 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16950 from simplify_cond_using_ranges. Split off code to walk
16951 backwards through casts into ...
16952 (simplify_cond_using_ranges_2): New function.
16953 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16954 (execute_vrp): After identifying jump threads, call
16955 simplify_cond_using_ranges_2.
16957 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16960 * ipa-inline.h (inline_summary): Add ctor.
16961 (create_ggc): Do not use ggc_cleared_alloc.
16963 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16964 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16966 * gcc.c (handle_braces): Support escaping in switch matching
16968 * doc/invoke.texi (Spec Files): Document it.
16969 Remove superfluous @code markup in items.
16971 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16973 * diagnostic-show-locus.c (struct column_range): New struct.
16974 (get_affected_columns): New function.
16975 (get_printed_columns): New function.
16976 (struct correction): New struct.
16977 (correction::ensure_capacity): New function.
16978 (correction::ensure_terminated): New function.
16979 (struct line_corrections): New struct.
16980 (line_corrections::~line_corrections): New dtor.
16981 (line_corrections::add_hint): New function.
16982 (layout::print_trailing_fixits): Reimplement in terms of the new
16984 (selftest::test_overlapped_fixit_printing): New function.
16985 (selftest::diagnostic_show_locus_c_tests): Call it.
16987 2017-05-03 Nathan Sidwell <nathan@acm.org>
16989 Canonicalize canonical type hashing
16990 * tree.h (type_hash_canon_hash): Declare.
16991 * tree.c (type_hash_list, attribute_hash_list): Move into
16992 type_hash_canon_hash.
16993 (build_type_attribute_qual_variant): Break out hash code calc into
16994 type_hash_canon_hash.
16995 (type_hash_canon_hash): New. Generic type hash computation.
16996 (build_range_type_1, build_array_type_1, build_function_type,
16997 build_method_type_directly, build_offset_type, build_complex_type,
16998 make_vector_type): Call it.
17000 2017-05-03 Richard Biener <rguenther@suse.de>
17002 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17003 When all DRs have unknown misaligned do not always peel
17004 when there is a store but apply the same costing model as if
17005 there were only loads.
17007 2017-05-03 Richard Biener <rguenther@suse.de>
17010 PR tree-optimization/80492
17011 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17012 compare_base_decls returning dont-know properly.
17014 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17016 * config/arm/iterators.md (CCSI): New mode iterator.
17017 (arch): New mode attribute.
17018 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17019 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17020 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17021 code iterator for success result mode.
17022 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17023 the corresponding new insn generators.
17025 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17028 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17029 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17031 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17033 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17034 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17035 (DDR_A): Wrap DDR argument in brackets.
17036 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17037 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17038 (DDR_REVERSED_P): Likewise.
17040 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17042 PR tree-optimization/79472
17043 * tree-switch-conversion.c (struct switch_conv_info): Add
17044 contiguous_range and default_case_nonstandard fields.
17045 (collect_switch_conv_info): Compute contiguous_range and
17046 default_case_nonstandard fields, don't clear final_bb if
17047 contiguous_range and only the default case doesn't have the required
17049 (check_all_empty_except_final): Set default_case_nonstandard instead
17050 of failing if contiguous_range and the default case doesn't have empty
17052 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17053 and only the default case doesn't have the required constants. Skip
17055 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17056 if default_case_nonstandard.
17057 (build_constructors): Build constant 1 just once. Assert that default
17058 values aren't inserted in between cases if contiguous_range. Skip
17060 (build_arrays): Skip virtual phis.
17061 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17062 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17063 Handle virtual phis.
17064 (gen_inbound_check): Handle default_case_nonstandard case.
17065 (process_switch): Adjust check_final_bb caller. Call
17066 gather_default_values with the first non-default case instead of
17067 default case if default_case_nonstandard.
17069 2017-05-02 Nathan Sidwell <nathan@acm.org>
17071 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17072 check. Fix formatting.
17074 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17076 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17077 errors when comparing specialized and unspecialized times.
17079 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17081 * diagnostic-show-locus.c
17082 (layout::should_print_annotation_line_p): Make private.
17083 (layout::print_annotation_line): Make private.
17084 (layout::annotation_line_showed_range_p): Make private.
17085 (layout::show_ruler): Make private.
17086 (layout::print_source_line): Make private. Pass in line and
17087 line_width, rather than calling location_get_source_line. Drop
17089 (layout::print_leading_fixits): New method.
17090 (layout::print_any_fixits): Rename to...
17091 (layout::print_trailing_fixits): ...this, and make private.
17092 Don't print newline fixits.
17093 (diagnostic_show_locus): Move logic for printing one row into...
17094 (layout::print_line): ...this new function. Move the
17095 location_get_source_line call and error-handling from
17096 print_source_line to here. Call print_leading_fixits, and rename
17097 print_any_fixits to print_trailing_fixits.
17098 (selftest::test_fixit_insert_containing_newline): Update now that
17099 newlines are partially supported.
17100 (selftest::test_fixit_insert_containing_newline_2): New test.
17101 (selftest::test_fixit_replace_containing_newline): Update comments.
17102 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17103 * edit-context.c (class added_line): New class.
17104 (class edited_line): Describe newline handling in comment.
17105 (edited_line::actually_edited_p): New method.
17106 (edited_line::print_content): Delete redundant decl.
17107 (edited_line::m_predecessors): New field.
17108 (edited_file::print_content): Call edited_line::print_content.
17109 (edited_file::print_diff): Update to support newlines.
17110 (edited_file::print_diff_hunk): Likewise.
17111 (edited_file::print_run_of_changed_lines): New function.
17112 (edited_file::print_diff_line): Convert to...
17113 (print_diff_line): ...this.
17114 (edited_file::get_effective_line_count): New function.
17115 (edited_line::edited_line): Initialize new field m_predecessors.
17116 (edited_line::~edited_line): Clean up m_predecessors.
17117 (edited_line::apply_fixit): Handle newlines.
17118 (edited_line::get_effective_line_count): New function.
17119 (edited_line::print_content): New function.
17120 (edited_line::print_diff_lines): New function.
17121 (selftest::test_applying_fixits_insert_containing_newline): New
17123 (selftest::test_applying_fixits_replace_containing_newline): New
17125 (selftest::insert_line): New function.
17126 (selftest::test_applying_fixits_multiple_lines): Add example of
17128 (selftest::edit_context_c_tests): Call the new tests.
17130 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17132 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17133 parameter cand. Update dump information.
17134 (get_computation_cost): Update uses.
17136 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17138 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17139 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17140 (get_computation_at, rewrite_use_address): Update use of
17141 get_computation_aff.
17143 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17145 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17146 (get_computation): Delete.
17147 (get_computation_cost): Implement like get_computation_cost_at.
17148 Use get_computation_at.
17149 (get_computation_cost_at): Delete.
17150 (rewrite_use_nonlinear_expr): Use get_computation_at.
17151 (rewrite_use_compare, remove_unused_ivs): Ditto.
17153 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17155 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17157 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17159 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17160 (ivopts_global_cost_for_size): Rename parameter and update uses.
17161 (iv_ca_recount_cost): Update uses.
17162 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17163 candidates seperately in n_invs and n_cands.
17164 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17166 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17168 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17169 (find_inv_vars_cb): New.
17170 (find_depends): Renamed to ...
17171 (find_inv_vars): ... this.
17172 (add_candidate_1, force_var_cost): Call find_inv_vars.
17173 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17175 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17177 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17178 inv_vars. Add inv_exprs.
17179 (struct iv_cand): Rename depends_on to inv_vars.
17180 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17181 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17182 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17183 (dump_cand): Dump inv_vars.
17184 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17185 (record_invariant, find_depends, add_candidate_1): Ditto.
17186 (set_group_iv_cost, force_var_cost): Ditto.
17187 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17188 (get_computation_cost_at, get_computation_cost): Ditto.
17189 (determine_group_iv_cost_generic): Ditto.
17190 (determine_group_iv_cost_address): Ditto.
17191 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17192 (determine_group_iv_costs): Ditto.
17193 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17194 (iv_ca_set_remove_invariants): Renamed to ...
17195 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17196 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17197 (iv_ca_set_add_invariants): Renamed to ...
17198 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17199 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17200 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17201 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17202 (create_new_ivs): Remove useless dump.
17204 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17206 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17208 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17209 (iv_ca_set_no_cp, create_new_iv): Ditto.
17211 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17213 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17215 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17217 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17218 function tree_check2.
17220 2017-05-02 Martin Liska <mliska@suse.cz>
17222 * doc/gcov.texi: Add missing preposition.
17223 * gcov.c (function_info::function_info): Properly fill up
17224 all member variables.
17226 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17228 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17230 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17232 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17234 2017-05-02 Martin Liska <mliska@suse.cz>
17237 * lto-streamer-in.c (lto_read_tree_1): Remove
17238 LTO_STREAMER_DEBUG.
17239 * lto-streamer.c (struct tree_hash_entry): Likewise.
17240 (struct tree_entry_hasher): Likewise.
17241 (tree_entry_hasher::hash): Likewise.
17242 (tree_entry_hasher::equal): Likewise.
17243 (lto_streamer_init): Likewise.
17244 (lto_orig_address_map): Likewise.
17245 (lto_orig_address_get): Likewise.
17246 (lto_orig_address_remove): Likewise.
17247 * lto-streamer.h: Likewise.
17248 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17249 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17251 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17253 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17254 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17255 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17256 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17257 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17258 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17259 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17260 (mm_maskz_sub_ss): New intrinsics.
17261 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17262 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17263 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17264 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17265 (__builtin_ia32_subss_mask_round): New builtins.
17266 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17267 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17268 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17270 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17271 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17273 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17276 2017-05-02 Martin Jambor <mjambor@suse.cz>
17278 PR tree-optimization/78687
17279 * tree-sra.c (access): New field parent.
17280 (process_subtree_disqualification): New function.
17281 (disqualify_candidate): Call it.
17282 (build_accesses_from_assign): Reset write flag if creating an
17284 (build_access_subtree): Fill in parent field and also prpagate
17285 down grp_write flag.
17286 (create_artificial_child_access): New parameter set_grp_write, set
17287 grp_write to its value.
17288 (propagate_subaccesses_across_link): Also propagate grp_write flag
17290 (propagate_all_subaccesses): Push the closest parent back to work
17291 queue if add_access_to_work_queue returned true.
17293 2017-05-02 Richard Biener <rguenther@suse.de>
17295 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17296 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17297 -fstrict-overflow documentation.
17298 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17299 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17300 flag_strict_overflow.
17301 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17302 * lto-opts.c (lto_write_options): Do not stream it.
17303 * lto-wrapper.c (merge_and_complain): Do not handle it.
17304 * opts.c (default_options_table): Do not set -fstrict-overflow.
17305 (finish_options): Likewise do not clear it when sanitizing.
17306 * simplify-rtx.c (simplify_const_relational_operation): Do not
17307 test flag_strict_overflow.
17309 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17311 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17312 using enabled attribute.
17313 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17314 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17315 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17316 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17317 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17318 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17319 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17320 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17321 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17322 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17324 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17326 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17328 2017-05-02 Richard Biener <rguenther@suse.de>
17330 PR tree-optimization/80591
17332 2017-04-10 Richard Biener <rguenther@suse.de>
17334 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17337 2017-05-02 Richard Biener <rguenther@suse.de>
17339 PR tree-optimization/80549
17340 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17341 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17342 headers if they do not have a preheader.
17344 2017-05-02 Martin Liska <mliska@suse.cz>
17347 * common.opt: Fix typo.
17348 * doc/invoke.texi: Likewise.
17350 2017-05-01 Jan Beulich <jbeulich@suse.com>
17352 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17353 swapping, add (x,x,m,x,n) alternative.
17355 2017-05-01 Nathan Sidwell <nathan@acm.org>
17357 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17358 unnecessary unadjusted_alignment check.
17360 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17363 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17365 * gimplify.c (gimplify_cilk_detach): New function.
17366 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17367 * tree-core.h: Document EXPR_CILK_SPAWN.
17368 * tree.h (EXPR_CILK_SPAWN): Define.
17370 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17372 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17373 to use new fixit_hint representation, using the "replace" logic.
17374 (get_line_span_for_fixit_hint): Likewise.
17375 (layout::print_any_fixits): Likewise.
17376 (selftest::test_one_liner_many_fixits): Rename to...
17377 (selftest::test_one_liner_many_fixits_1): ...this, and update
17378 comment and expected output to reflect that the multiple fix-it
17379 hints are now consolidated into one insertion.
17380 (selftest::test_one_liner_many_fixits_2): New test.
17381 (selftest::test_diagnostic_show_locus_one_liner): Update for
17383 (selftest::test_fixit_consolidation): Update for fix-it API
17385 * diagnostic.c (print_parseable_fixits): Likewise.
17386 * edit-context.c (edited_line::m_line_events): Convert from
17387 auto_vec <line_event *> to auto_vec <line_event>.
17388 (class line_event): Convert from abstract base class to a concrete
17389 class, taking over the role of replace_event.
17390 (class insert_event): Delete.
17391 (class replace_event): Rename to class line_event. Convert to
17393 (edit_context::add_fixits): Reimplement.
17394 (edit_context::apply_insert): Delete.
17395 (edit_context::apply_replace): Rename to...
17396 (edit_context::apply_fixit): ...this. Convert to half-open range.
17397 (edited_file::apply_insert): Delete.
17398 (edited_file::apply_replace): Rename to...
17399 (edited_file::apply_fixit): ...this.
17400 (edited_line::~edited_line): Drop deletion of events.
17401 (edited_line::apply_insert): Delete.
17402 (edited_line::apply_replace): Rename to...
17403 (edited_line::apply_fixit): ...this. Convert to half-open range.
17404 Update for change to type of m_line_events.
17405 * edit-context.h (edit_context::apply_insert): Delete.
17406 (edit_context::apply_replace): Rename to...
17407 (edit_context::apply_fixit): ...this.
17409 2017-05-01 Martin Sebor <msebor@redhat.com>
17411 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17414 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17417 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17418 __get_cpuid_max returns 0.
17419 (__get_cpuid_count): Ditto.
17421 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17423 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17424 replacement expression is another instance of one of its arguments.
17426 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17429 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17430 check for stack push/pop autoinc.
17431 * config/i386/i386.c (ix86_agi_dependent): Return false
17432 if the only reason why modified_in_p returned true is that
17433 addr is SP based and set_insn is a push or pop.
17435 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17437 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17440 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17443 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17444 (account_size_time): Use two predicates - exec_pred and
17446 (evaluate_conditions_for_known_args): Compute both clause and
17448 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17449 (inline_summary_t::duplicate): Update.
17450 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17452 (compute_inline_parameters): Likewise.
17453 (estimate_edge_size_and_time): Update caluclation of time.
17454 (estimate_node_size_and_time): Compute both time and nonspecialized
17456 (estimate_ipcp_clone_size_and_time): Update.
17457 (inline_merge_summary): Update.
17458 (do_estimate_edge_time): Update.
17459 (do_estimate_edge_size): Update.
17460 (do_estimate_edge_hints): Update.
17461 (inline_read_section, inline_write_summary): Stream both new predicates.
17462 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17464 (compute_inlined_call_time): Cleanup.
17465 (big_speedup_p): Update.
17466 (edge_badness): Update.
17467 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17468 (size_time_entry): Replace predicate by exec_predicate and
17469 nonconst_predicate.
17470 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17471 (estimate_edge_time): Return also nonspec_time.
17472 (reset_edge_growth_cache): Update.
17474 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17476 PR rtl-optimization/80491
17477 * ifcvt.c (noce_process_if_block): When looking for x setter
17478 with missing else_bb, don't check only the insn right before
17479 cond_earliest, but look for the last insn that x is modified in
17480 within the same bb.
17482 PR rtl-optimization/80491
17483 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17485 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17487 PR tree-optimization/80487
17488 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17490 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17492 PR tree-optimization/79697
17493 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17494 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17495 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17497 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17498 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17500 2017-04-28 Martin Sebor <msebor@redhat.com>
17502 PR tree-optimization/80523
17503 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17504 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17506 (maybe_warn, format_directive, parse_directive): Use new functions.
17507 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17509 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17511 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17513 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17515 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17516 target_header_dir): Set correctly.
17517 * configure: Regenerated.
17518 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17519 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17520 instead of SYSTEM_HEADER_DIR.
17522 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17524 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17525 (estimate_local_effects): Likewise.
17526 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17527 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17528 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17529 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17530 do_estimate_edge_time, estimate_edge_time): Likewise.
17531 * ipa-inline-analysis.c (estimate_node_size_and_time,
17532 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17533 (estimate_time_after_inlining): Remove.
17535 2017-04-28 Martin Liska <mliska@suse.cz>
17537 * doc/gcov.texi: Enhance documentation of gcov.
17539 2017-04-28 Martin Liska <mliska@suse.cz>
17541 * doc/gcov.texi: Sort options in alphabetic order.
17542 * doc/gcov-dump.texi: Likewise.
17543 * doc/gcov-tool.texi: Likewise.
17544 * gcov.c (print_usage): Likewise.
17545 * gcov-dump.c (print_usage): Likewise.
17546 * gcov-tool.c (print_merge_usage_message): Likewise.
17547 (print_rewrite_usage_message): Likewise.
17548 (print_overlap_usage_message): Likewise.
17550 2017-04-28 Martin Liska <mliska@suse.cz>
17552 PR gcov-profile/53915
17553 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17555 2017-04-28 Martin Liska <mliska@suse.cz>
17557 PR gcov-profile/79891
17558 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17559 is marked by compiler as living on a line.
17560 (get_cycles_count): Remove usage of the union.
17561 (output_intermediate_file): Likewise.
17562 (find_source): Fix GNU coding style.
17563 (accumulate_line_counts): Remove old non-all block mode.
17564 (output_lines): Remove usage of the union.
17565 * profile.c (output_location): Include all BBs, even if
17566 belonging to a same line (and file) as a previous BB.
17568 2017-04-28 Martin Liska <mliska@suse.cz>
17570 * gcov.c (process_args): Handle new argument 'w'.
17571 (read_graph_file): Assign ID to BBs.
17572 (output_branch_count): Display BB # if verbose flag is set.
17573 (output_lines): Likewise for arcs.
17574 (print_usage): Add '--verbose' option help.
17575 * doc/gcov.texi: Document --verbose (-w) option.
17577 2017-04-28 Martin Liska <mliska@suse.cz>
17579 * gcov.c (struct block_location_info): New struct.
17580 (process_file): Fill up the new structure.
17581 (read_graph_file): Replace usage of encoding by the newly added
17583 (add_line_counts): Likewise.
17584 (accumulate_line_counts): Remove usage of the union.
17585 (function_info::function_info): New function.
17586 (function_info::~function_info): Likewise.
17587 (process_file): Call delete instead of release_function.
17588 (release_function): Release the function.
17589 (release_structures): Call delete instead of release_function.
17590 (solve_flow_graph): Replace usage of num_blocks.
17591 (find_exception_blocks): Likewise.
17592 (output_lines): Fix GNU coding style.
17594 2017-04-28 Martin Liska <mliska@suse.cz>
17597 * coverage.c (coverage_remove_note_file): New function.
17598 * coverage.h: Declare the function.
17599 * toplev.c (finalize): Clean if an error has been seen.
17601 2017-04-28 Martin Liska <mliska@suse.cz>
17603 PR gcov-profile/80031
17604 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17605 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17606 * gcov.c (read_graph_file): Read just number of blocks.
17607 * profile.c (branch_prob): Do not stream 0 flags per a basic
17610 2017-04-28 Martin Liska <mliska@suse.cz>
17612 * gcov-dump.c (tag_*): Add new argument to declarations.
17613 (dump_gcov_file): Likewise.
17614 (tag_blocks): Add and use new argument depth.
17615 (tag_arcs): Likewise.
17616 (tag_lines): Likewise.
17617 (tag_counters): Likewise.
17618 (tag_summary): Likewise.
17619 (dump_working_sets): Use depth to do a proper indentation.
17621 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17624 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17625 * symtab.c (symtab_node::debug_symtab): Move definition here.
17627 2017-04-28 Richard Biener <rguenther@suse.de>
17629 * lto-streamer.h (LTO_major_version): Bump to 7.
17631 2017-04-28 Richard Biener <rguenther@suse.de>
17633 * tree-vrp.c (assert_info): New struct.
17634 (add_assert_info): New helper.
17635 (register_edge_assert_for_2): Refactor to add asserts to a vector
17637 (register_edge_assert_for_1): Likewise.
17638 (register_edge_assert_for): Likewise.
17639 (finish_register_edge_assert_for): New helper actually registering
17640 asserts where live on edge.
17641 (find_conditional_asserts): Adjust.
17642 (find_switch_asserts): Likewise.
17643 (evrp_dom_walker::try_find_new_range): Generalize.
17644 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17646 2017-04-27 Marek Polacek <polacek@redhat.com>
17649 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17650 arg10 and arg11 to itype.
17652 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17654 * doc/extend.texi (Object Size Checking): Improve grammar.
17656 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17659 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17660 that the logic for permitting reciprocal estimates matches that
17663 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17666 * tree.c (type_cache_hasher::equal): Only compare
17667 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17668 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17669 non-aggregate element types.
17670 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17671 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17673 2017-04-27 Richard Biener <rguenther@suse.de>
17675 PR middle-end/80533
17676 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17677 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17678 keeping a reference to a trailing array.
17680 2017-04-27 Richard Biener <rguenther@suse.de>
17682 PR middle-end/80539
17683 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17684 being in loop-closed SSA form conservatively.
17685 (chrec_fold_multiply_poly_poly): Likewise.
17687 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17689 PR middle-end/79665
17690 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17691 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17693 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17696 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17697 (aarch64_function_arg_alignment): Return unsigned int again, but still
17698 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17699 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17700 Don't emit -Wpsabi note.
17701 (aarch64_function_arg_boundary): Likewise.
17702 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17705 2017-04-26 Nathan Sidwell <nathan@acm.org>
17707 * tree.h (crc32_unsigned_n): Declare.
17708 (crc32_unsigned, crc32_unsigned): Make inline.
17709 * tree.c (crc32_unsigned_bits): Replace with ...
17710 (crc32_unsigned_n): ... this.
17711 (crc32_unsigned, crc32_byte): Remove.
17712 (crc32_string): Remove unnecessary braces.
17714 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17716 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17717 * ipa-inline-analysis.c (MAX_TIME): Remove.
17718 (account_size_time): Use sreal for time.
17719 (dump_inline_summary): Update.
17720 (estimate_function_body_sizes): Update.
17721 (estimate_edge_size_and_time): Update.
17722 (estimate_calls_size_and_time): Update.
17723 (estimate_node_size_and_time): Update.
17724 (inline_merge_summary): Update.
17725 (inline_update_overall_summary): Update.
17726 (estimate_time_after_inlining): Update.
17727 (inline_read_section): Update.
17728 (inline_write_summary): Update.
17729 * ipa-inline.c (compute_uninlined_call_time): Update.
17730 (compute_inlined_call_time): Update.
17731 (recursive_inlining): Update.
17732 (inline_small_functions): Update.
17733 (dump_overall_stats): Update.
17734 * ipa-inline.h: Include sreal.h.
17735 (size_time_entry): Turn time to sreal.
17736 (inline_summary): Turn self_time nad time to sreal.
17738 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17740 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17742 (sreal::stream_out, sreal::stream_in): New.
17743 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17745 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17747 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17750 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17753 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17755 (dimode_scalar_chain::compute_convert_gain): Ditto.
17756 (dimode_scalar_chain::make_vector_copies): Ditto.
17757 (dimode_scalar_chain::convert_reg): Ditto.
17758 (dimode_scalar_chain::convert_insn): Ditto.
17759 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17760 (VI248_AVX512BW_1): New mode iterator.
17761 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17762 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17765 2017-04-25 Martin Sebor <msebor@redhat.com>
17767 PR tree-optimization/80497
17768 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17769 constants are representable in HOST_WIDE_INT.
17770 (parse_directive): Ditto.
17772 2017-04-25 Martin Sebor <msebor@redhat.com>
17775 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17776 (new_zero_array): Adjust signature.
17777 (dom_info::dom_init): Used unsigned rather that size_t.
17778 (dom_info::dom_info): Same.
17780 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17781 Jakub Jelinek <jakub@redhat.com>
17784 * config/arm/arm.c: Include gimple.h.
17785 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17786 returns negative, increment ncrn only if it returned positive.
17787 (arm_needs_doubleword_align): Return int instead of bool,
17788 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17789 members, but if there is any such non-FIELD_DECL
17790 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17791 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17792 returns negative, increment nregs only if it returned positive.
17793 (arm_setup_incoming_varargs): Likewise.
17794 (arm_function_arg_boundary): Emit -Wpsabi note if
17795 arm_needs_doubleword_align returns negative, return
17796 DOUBLEWORD_ALIGNMENT only if it returned positive.
17798 2017-04-25 Marek Polacek <polacek@redhat.com>
17801 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17802 first argument to type.
17804 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17807 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17808 type checks to test for compatibility instead of equality.
17810 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17811 Jakub Jelinek <jakub@redhat.com>
17814 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17816 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17817 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17818 the alignment computation, but return their maximum in warn_alignment.
17819 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17820 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17822 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17823 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17826 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17828 * config/arc/simdext.md (dmpyh): Fix typo.
17830 2017-04-25 Richard Biener <rguenther@suse.de>
17832 PR tree-optimization/80492
17833 * alias.c (compare_base_decls): Handle registers with asm
17834 specification conservatively.
17835 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17836 compare_base_decls returning dont-know properly.
17838 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17840 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17841 (legitimate_offset_address_p): New function.
17842 (arc_legitimate_address_p): Use above function.
17844 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17846 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17848 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17850 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17851 ACCH registers whenever they are available.
17853 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17855 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17856 double regs fix when not used.
17858 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17860 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17862 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17863 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17865 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17867 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17868 when emitting short ADD instructions.
17870 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17872 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17874 (cmpsi_cc_c_insn): Likewise.
17875 (cbranchsi4_scratch): Compute proper instruction length using
17876 compact_hreg_operand.
17877 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17879 2017-04-25 Richard Biener <rguenther@suse.de>
17881 PR middle-end/80509
17882 * passes.c (pass_manager::pass_manager): Initialize
17883 m_name_to_pass_map.
17885 2017-04-25 Richard Biener <rguenther@suse.de>
17887 PR tree-optimization/79201
17888 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17890 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17893 * config/s390/vector.md: Split MEM->GPR vector moves for
17894 non-s_operand addresses.
17896 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17899 * config/s390/predicates.md (reload_const_wide_int_operand): New
17901 * config/s390/s390.md ("movti"): Remove d/P alternative.
17902 ("movti_bigconst"): New pattern definition.
17904 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17907 * s390-protos.h (s390_expand_cs_hqi): Removed.
17908 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17909 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17910 modes as well as CCZ1mode and CCZmode.
17911 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17912 signature of s390_emit_compare_and_swap.
17913 (s390_expand_cs_hqi): Likewise, make static.
17914 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17915 compare-and-swap, in some cases.
17916 (s390_expand_cs): Wrapper function.
17917 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17919 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17920 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17921 patterns for small and large integers. Forbid symref memory operands.
17922 Move expander to s390.c. Require cc register.
17923 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17924 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17925 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17926 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17927 symref memory operands. Remove CC mode and call s390_match_ccmode
17929 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17931 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17933 * config/s390/s390.md (define_peephole2): New peephole to help
17934 combining the load-and-test pattern with volatile memory.
17936 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17938 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17939 with CCZmode for TARGET_Z196.
17941 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17943 PR rtl-optimization/80501
17944 * combine.c (make_compound_operation_int): Set subreg_code to SET
17945 even for AND with mask of the sign bit of mode.
17947 PR rtl-optimization/80500
17948 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17949 sum's initial value.
17951 2017-04-25 Julian Brown <julian@codesourcery.com>
17952 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17954 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17956 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17958 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17960 2017-04-25 Julian Brown <julian@codesourcery.com>
17961 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17963 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17964 (thunderx2t99_sha): New Reservation.
17966 2017-04-25 Julian Brown <julian@codesourcery.com>
17967 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17969 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17970 type for 1-element load.
17972 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17974 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17976 2017-04-24 Martin Jambor <mjambor@suse.cz>
17978 PR tree-optimization/80293
17979 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17980 char arrays not totally scalarizable if it is false.
17981 (analyze_all_variable_accesses): Pass correct value in the new
17982 parameter. Add a statistics counter.
17984 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17986 PR middle-end/79931
17987 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17989 2017-04-24 Richard Biener <rguenther@suse.de>
17991 PR tree-optimization/80494
17992 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17993 out for complex types.
17995 2017-04-24 Richard Biener <rguenther@suse.de>
17997 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17998 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17999 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18000 (DFS): Adjust and never fail.
18001 (sccvn_dom_walker::fail): Remove.
18002 (sccvn_dom_walker::before_dom_children): Adjust.
18003 (run_scc_vn): Likewise and never fail.
18004 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18005 (pass_fre::execute): Likewise.
18007 2017-04-24 Richard Biener <rguenther@suse.de>
18009 PR tree-optimization/79725
18010 * tree-ssa-sink.c (statement_sink_location): Return whether
18011 failure reason was zero uses. Move that check later.
18012 (sink_code_in_bb): Deal with zero uses by removing the stmt
18015 2017-04-24 Richard Biener <rguenther@suse.de>
18018 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18019 pointer-based references.
18021 2017-04-24 Richard Biener <rguenther@suse.de>
18024 * pass_manager.h (pass_manager::operator new): Remove.
18025 (pass_manager::operator delete): Likewise.
18026 * passes.c (pass_manager::operator new): Remove.
18027 (pass_manager::operator delete): Likewise.
18028 (pass_manager::pass_manager): Zero individual pass members.
18030 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18033 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18034 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18035 Check "XEXP (src, 1)" operand here.
18036 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18037 Check "XEXP (src, 1)" operand here.
18038 (dimode_scalar_chain::make_vector_copies): Detect count register
18039 of a shift instruction. Zero extend count register from QImode
18040 to DImode to satisfy vector shift pattern count operand predicate.
18041 Substitute vector shift count operand with a DImode copy.
18042 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18045 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18047 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18048 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18049 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18050 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18051 (UNSPEC_NOREX_MEM): Remove definition.
18053 2017-04-21 Richard Biener <rguenther@suse.de>
18055 PR tree-optimization/79547
18056 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18057 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18058 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18059 without any constraints.
18061 2017-04-21 Richard Biener <rguenther@suse.de>
18063 PR tree-optimization/78847
18064 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18066 2017-04-21 Richard Biener <rguenther@suse.de>
18068 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18069 (build_distinct_type_copy): Likewise.
18070 (build_variant_type_copy): Likewise.
18071 * tree.c (build_qualified_type): Pass down mem-stat info.
18072 (build_distinct_type_copy): Likewise.
18073 (build_variant_type_copy): Likewise.
18075 2017-04-21 Richard Biener <rguenther@suse.de>
18077 PR tree-optimization/80237
18078 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18080 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18081 for a simplified result.
18083 2016-04-21 Richard Biener <rguenther@suse.de>
18085 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18086 sth as strict as a simple_iv but a chrec without symbols and an
18087 operand defined in the loop we are peeling (and not some subloop).
18088 (propagate_constants_for_unrolling): Propagate all constants.
18090 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18093 * config/i386/i386.c (print_reg): Remove assert for disalowed
18094 regno values, call output_operand_lossage instead.
18096 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18099 * config/i386/constraints.md (Yc): New register constraint.
18100 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18101 Use Yc constraint for alternative 2 of operand 0. Remove
18102 preferred_for_speed attribute.
18104 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18106 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18107 lastprivate clauses in SIMT case.
18109 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18111 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18113 2017-04-20 Richard Biener <rguenther@suse.de>
18115 PR tree-optimization/57796
18116 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18117 as N scalar stores.
18118 (vect_model_load_cost): Cost gathers as N scalar loads.
18120 2017-04-20 Richard Biener <rguenther@suse.de>
18122 * ggc-page.c (ggc_allocated_p): Rename to ...
18123 (safe_lookup_page_table_entry): ... this and return the lookup
18125 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18127 2017-04-20 Richard Biener <rguenther@suse.de>
18129 PR tree-optimization/80453
18130 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18131 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18132 from the conditions.
18133 (vn_phi_eq): Pass them down.
18134 (vn_phi_lookup): Record them.
18135 (vn_phi_insert): Likewise.
18137 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18139 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18140 uninitialized variable warning to avoid buffer overrun.
18142 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18145 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18146 is suppressed for '{ 0 }' in C.
18148 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18150 * BASE-VER: Set to 8.0.0.
18152 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18154 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18155 priority .init_array and .fini_array section with SECTION_NOTYPE
18158 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18160 PR middle-end/80423
18161 * tree.h (build_array_type): Add typeless_storage default argument.
18162 * tree.c (type_cache_hasher::equal): Also compare
18163 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18164 (build_array_type): Add typeless_storage argument, set
18165 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18167 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18168 (build_array_type): Likewise. Add typeless_storage argument.
18170 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18171 Jakub Jelinek <jakub@redhat.com>
18173 PR tree-optimization/80426
18174 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18175 operation on symbolic operands, also compute the overflow for the
18176 invariant part when the operation degenerates into a negation.
18178 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18181 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18182 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18185 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18187 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18190 * config/avr/avr.c (tree.h): Include it.
18191 (cgraph.h): Include it.
18192 (avr_encode_section_info): Don't warn for uninitialized progmem
18193 variable if it's just an alias.
18195 2017-04-19 Richard Biener <rguenther@suse.de>
18198 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18199 when needed by AutoPGO.
18201 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18204 * doc/lto.texi: Remove an extra 'that'.
18206 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18208 PR rtl-optimization/80429
18209 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18210 are only used in debug insns.
18212 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18213 Vladimir Makarov <vmakarov@redhat.com>
18215 * config/sparc/predicates.md (input_operand): Add comment. Return
18216 true for any memory operand when LRA is in progress.
18217 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18219 2017-04-18 Jeff Law <law@redhat.com>
18222 * mips.md ({return,simple_return}_internal): Do not overwrite
18225 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18227 PR tree-optimization/80443
18228 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18229 instead of adding 1, subtract -1 and similarly instead of subtracting
18232 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18234 PR rtl-optimization/80357
18235 * haifa-sched.c (tmp_bitmap): New variable.
18236 (model_recompute): Handle duplicate use records.
18237 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18238 (free_global_sched_pressure_data): Free it.
18240 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18243 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18244 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18245 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18246 instead of SYSTEM_HEADER_DIR.
18248 2017-04-18 Jeff Law <law@redhat.com>
18250 PR middle-end/80422
18251 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18252 predecessors after walking up the insn chain.
18254 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18257 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18258 sizetype type into debug info.
18260 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18263 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18264 unneeded test for TARGET_UPPER_REGS_SF.
18265 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18267 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18270 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18271 instead of gsi_after_labels.
18273 2017-04-18 Jeff Law <law@redhat.com>
18275 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18279 2017-04-13 Jeff Law <law@redhat.com>
18280 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18281 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18283 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18286 * config/avr/avr.c (intl.h): Include it.
18287 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18289 2017-04-18 Martin Liska <mliska@suse.cz>
18291 PR gcov-profile/78783
18292 * gcov-tool.c (gcov_output_files): Validate that destination
18293 file is either removed by the tool or by a user.
18295 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18296 Guy Benyei <guybe@mellanox.com>
18298 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18299 block, and do not negate it, the stored id is already negative.
18301 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18303 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18305 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18308 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18309 masks of options that should be turned off if the VSX vector
18310 options are turned off.
18311 (OTHER_P8_VECTOR_MASKS): Likewise.
18312 (OTHER_VSX_VECTOR_MASKS): Likewise.
18313 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18314 rs6000_disable_incompatible_switches to validate no type switches
18316 (rs6000_incompatible_switch): New function to disallow turning on
18317 other vector options if -mno-vsx, -mno-power8-vector, or
18318 -mno-power9-vector are specified.
18320 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18322 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18324 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18326 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18327 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18328 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18329 (ARG_POINTER_CFA_OFFSET): Likewise.
18331 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18333 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18334 conditions to take advantage of various optimizations.
18336 2017-04-13 Jeff Law <law@redhat.com>
18338 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18339 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18340 (zero_extendsidi2_dext): Likewise.
18342 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18345 * fold-const.c (fold_ternary_loc): Revert
18346 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18349 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18351 PR rtl-optimization/80343
18352 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18353 new scratch pseudo.
18355 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18358 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18359 to ubsan_encode_value.
18361 2017-04-13 Jeff Law <law@redhat.com>
18363 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18364 appearing in DEBUG_INSNs.
18366 2017-04-13 Martin Liska <mliska@suse.cz>
18368 PR gcov-profile/80413
18369 * gcov-io.c (gcov_write_string): Copy to buffer just when
18370 allocated size is greater than zero.
18372 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18375 * dwarf2out.c (decls_for_scope): Ignore declarations of
18376 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18378 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18381 * ipa-visibility.c (non_local_p): Fix typos.
18382 (localize_node): When localizing symbol in same comdat group,
18383 dissolve the group only when we know external symbols are going
18385 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18387 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18389 PR tree-optimization/79390
18390 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18391 order does not result in usable sequence, retry with reversed operand
18397 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18398 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18399 op0 instead of fold_convert_loc (loc, type, arg0).
18401 2017-04-12 Jeff Law <law@redhat.com>
18403 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18404 has a delay slot in the generated code.
18406 * config/cris/cris.md (cris_preferred_reload_class): Return
18407 GENNONACR_REGS rather than GENERAL_REGS.
18409 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18412 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18413 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18414 signedness of the result type.
18416 2017-04-12 Richard Biener <rguenther@suse.de>
18417 Jeff Law <law@redhat.com>
18419 PR tree-optimization/80359
18420 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18421 trim stores to TARGET_MEM_REFs.
18423 2017-04-12 Richard Biener <rguenther@suse.de>
18425 PR tree-optimization/79390
18426 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18427 threading case even more.
18429 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18432 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18433 for quad_address_p for TImode, instead of just not indexed_address.
18435 2017-04-12 Richard Biener <rguenther@suse.de>
18436 Bernd Edlinger <bernd.edlinger@hotmail.de>
18438 PR middle-end/79671
18439 * alias.c (component_uses_parent_alias_set_from): Handle
18440 TYPE_TYPELESS_STORAGE.
18441 (get_alias_set): Likewise.
18442 * tree-core.h (tree_type_common): Add typeless_storage flag.
18443 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18444 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18445 for types containing members with TYPE_TYPELESS_STORAGE.
18446 (place_field): Likewise.
18447 (layout_type): Likewise for ARRAY_TYPE.
18448 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18449 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18450 TYPE_TYPELESS_STORAGE.
18451 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18453 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18456 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18457 first argument to type.
18459 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18463 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18464 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18465 (rs6000_expand_binop_builtin): Likewise.
18466 (rs6000_expand_ternop_builtin): Likewise; also add missing
18467 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18468 vshasigma built-ins.
18469 * doc/extend.texi: Document that vec_xxpermdi's third argument
18470 must be a constant.
18472 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18474 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18475 Use shift_const cost parameter when calculating gain of STV shifts.
18477 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18479 PR rtl-optimization/70478
18480 * lra-constraints.c (process_alt_operands): Check memory for
18481 disfavoring memory insn operand.
18483 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18485 PR middle-end/80100
18486 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18487 left shift in unsigned HOST_WIDE_INT type.
18489 PR rtl-optimization/80385
18490 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18491 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18494 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18495 if they have any depend clauses.
18497 2017-04-11 Martin Liska <mliska@suse.cz>
18500 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18501 * ipa-split.c (split_function): Create a local comdat symbol
18502 if caller is in a comdat group.
18504 2017-04-11 Martin Liska <mliska@suse.cz>
18507 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18510 2017-04-11 Martin Sebor <msebor@redhat.com>
18512 PR middle-end/80364
18513 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18514 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18516 (directive::set_width, directive::set_precision, format_character):
18518 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18521 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18524 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18525 conflict, set target->arch_name instead of target->cpu_name.
18527 2017-04-11 Richard Biener <rguenther@suse.de>
18529 PR tree-optimization/80374
18530 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18531 build_zero_cst, remove fold_convertible_p check again.
18533 2017-04-11 Martin Liska <mliska@suse.cz>
18536 * ubsan.c (instrument_object_size): Do not instrument register
18539 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18542 * config/i386/i386-builtin-types.def
18543 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18544 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18545 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18546 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18547 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18548 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18549 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18550 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18551 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18552 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18553 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18554 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18555 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18556 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18557 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18558 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18559 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18560 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18561 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18562 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18563 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18564 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18565 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18566 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18567 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18568 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18569 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18570 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18571 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18572 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18573 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18574 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18575 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18576 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18577 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18578 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18579 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18580 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18581 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18582 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18583 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18584 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18585 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18586 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18587 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18588 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18590 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18591 flag to second_arg_count, handle 4 argument function type _COUNT
18592 aliases, handle second_arg_count on second argument rather than last.
18594 2017-04-10 Jeff Law <law@redhat.com>
18596 PR tree-optimization/80374
18597 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18598 record anything if we can not convert integer_zero_node to the
18601 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18604 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18605 Enhance special handling given to the TARGET_P9_MINMAX option in
18606 relation to certain other options.
18608 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18610 PR tree-optimization/80153
18611 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18612 remove POINTER_PLUS_EXPR's base part directly, rather than through
18615 2017-04-10 Richard Biener <rguenther@suse.de>
18616 Bin Cheng <bin.cheng@arm.com>
18618 PR tree-optimization/80153
18619 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18620 the first element of pointer type aff_tree. Build result expr in
18622 (add_elt_to_tree): Convert to type unconditionally. Remove other
18623 fold_convert calls.
18624 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18625 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18627 2017-04-10 Richard Biener <rguenther@suse.de>
18629 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18632 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18634 PR rtl-optimization/70478
18635 * lra-constraints.c (curr_small_class_check): New.
18636 (update_and_check_small_class_inputs): New.
18637 (process_alt_operands): Update curr_small_class_check. Disfavor
18638 alternative insn memory operands. Check available regs for small
18641 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18644 * config/mips/mips.opt (-mvirt): Update description.
18645 * doc/invoke.texi (-mvirt): Likewise.
18647 2017-04-10 Richard Biener <rguenther@suse.de>
18649 PR middle-end/80362
18650 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18651 looking for NEGATE_EXPR in -A / -B to A / B folding.
18653 2017-04-10 Martin Liska <mliska@suse.cz>
18655 PR gcov-profile/80224
18656 * gcov.c (print_usage): Fix usage string.
18657 (get_gcov_intermediate_filename): Remove.
18658 (output_gcov_file): Use both for normal and intermediate format.
18659 (generate_results): Do not initialize special file for
18660 intermediate format.
18662 2017-04-10 Richard Biener <rguenther@suse.de>
18664 PR tree-optimization/80304
18665 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18668 2017-04-10 Nathan Sidwell <nathan@acm.org>
18671 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18672 (rs6000_init_builtins): Use it.
18674 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18676 * config/arm/arm.md (<mrc>): Add mode to SET source.
18677 (<mrrc>): Likewise.
18679 2017-04-10 Richard Biener <rguenther@suse.de>
18681 PR middle-end/80344
18682 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18684 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18687 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18688 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18689 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18690 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18691 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18692 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18693 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18694 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18695 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18696 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18697 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18698 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18699 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18700 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18701 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18702 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18703 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18704 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18705 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18706 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18707 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18708 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18709 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18711 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18713 PR rtl-optimization/70478
18714 * lra-constraints.c: Reverse the last patch.
18716 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18718 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18719 Add comment for WCHAR_T.
18721 2017-04-08 Martin Liska <mliska@suse.cz>
18724 2017-04-07 Martin Liska <mliska@suse.cz>
18727 * ipa-split.c (split_function): Add function part to a same comdat
18730 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18733 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18735 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18737 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18738 (vec_pairing): ...to this.
18739 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18740 (rs6000_sched_init): Adjust for name change.
18741 (struct rs6000_sched_context): Likewise.
18742 (rs6000_init_sched_context): Likewise.
18743 (rs6000_set_sched_context): Likewise.
18745 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18751 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18753 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18754 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18755 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18757 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18759 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18761 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18763 PR rtl-optimization/70703
18764 * ira-color.c (update_conflict_hard_regno_costs): Use
18765 int64_t instead of HOST_WIDE_INT.
18767 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18769 PR rtl-optimization/70478
18770 * lra-constraints.c (process_alt_operands): Disfavor alternative
18771 insn memory operands.
18773 2017-04-07 Jeff Law <law@redhat.com>
18775 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18776 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18778 2017-04-07 Martin Liska <mliska@suse.cz>
18781 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18782 Show error message instead of an ICE.
18784 2017-04-07 Martin Liska <mliska@suse.cz>
18787 * ipa-split.c (split_function): Add function part to a same comdat
18790 2017-04-07 Richard Biener <rguenther@suse.de>
18792 PR middle-end/80341
18793 * tree.c (get_unwidened): Also handle ! for_type case for
18795 * convert.c (do_narrow): Split out from ...
18796 (convert_to_integer_1): ... here. Do not pass final truncation
18797 type to get_unwidened for TRUNC_DIV_EXPR.
18799 2017-04-07 Richard Biener <rguenther@suse.de>
18801 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18803 (aff_combination_const): Adjust.
18804 (aff_combination_scale): Likewise.
18805 (aff_combination_add_elt): Likewise.
18806 (aff_combination_add_cst): Likewise.
18807 (aff_combination_convert): Likewise.
18808 (add_elt_to_tree): Likewise. Remove unused argument.
18809 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18811 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18813 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18815 * config/arm/arm.c (arm_default_short_enums): Use
18816 ARM_DEFAULT_SHORT_ENUMS.
18817 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18819 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18822 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18823 members with redundant out-of-class redeclaration.
18825 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18828 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18829 * config/i386/i386.md (*zero_extendsidi2):
18830 Add (?*x,*x) and (?*v,*v) alternatives.
18832 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18835 * config/i386/i386.c (ix86_expand_builtin)
18836 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18837 mode from insn data. Convert operands to insn operand mode.
18838 Copy operands that don't satisfy insn predicate to a register.
18840 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18842 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18845 2017-04-06 Richard Biener <rguenther@suse.de>
18847 PR tree-optimization/80334
18848 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18849 preserve alignment of accesses.
18851 2017-04-06 Richard Biener <rguenther@suse.de>
18853 PR tree-optimization/80262
18854 * tree-sra.c (build_ref_for_offset): Preserve address-space
18856 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18857 Drop useless address-space information on MEM_REF offsets.
18859 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18861 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18863 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18865 PR rtl-optimization/70703
18866 * ira-color.c (update_conflict_hard_regno_costs): Use
18867 HOST_WIDE_INT instead of long.
18869 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18872 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18873 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18875 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18876 for x86_64 target. Handle -m3dnowa option.
18878 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18880 PR rtl-optimization/70703
18881 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18882 (update_conflict_hard_regno_costs): Use long instead of unsigned
18883 arithmetic for cost calculation.
18885 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18886 Bernd Edlinger <bernd.edlinger@hotmail.de>
18889 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18892 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18895 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18896 ptr_mode with Pmode throughout.
18897 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18898 into probe_stack_range and use DImode.
18900 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18903 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18904 call_eh_return is true.
18906 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18908 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18909 Initialize last_match_fntype_index.
18911 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18914 * tree-nvr.c: Include internal-fn.h.
18915 (pass_return_slot::execute): Ignore internal calls without
18918 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18919 Richard Biener <rguenther@suse.de>
18922 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18923 captures used multiple times, except for the last use.
18924 * generic-match-head.c: Include gimplify.h.
18926 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18928 PR tree-optimization/79390
18929 * target.h (struct noce_if_info): Declare.
18930 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18931 * target.def (noce_conversion_profitable_p): New target hook.
18932 * ifcvt.h (struct noce_if_info): New type, moved from ...
18933 * ifcvt.c (struct noce_if_info): ... here.
18934 (noce_conversion_profitable_p): Renamed to ...
18935 (default_noce_conversion_profitable_p): ... this. No longer
18937 (noce_try_store_flag_constants, noce_try_addcc,
18938 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18939 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18940 instead of noce_conversion_profitable_p.
18941 * config/i386/i386.c: Include ifcvt.h.
18942 (ix86_option_override_internal): Don't override
18943 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18944 (ix86_noce_conversion_profitable_p): New function.
18945 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18946 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18947 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18948 * doc/tm.texi: Regenerated.
18950 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18952 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18955 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18958 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18959 instructions for small multiply cores.
18961 2017-04-04 Jeff Law <law@redhat.com>
18963 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18965 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18966 that are not set by the loop over the elements.
18968 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18971 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18972 int mode, convert_modes it to mode as unsigned, otherwise use
18973 lowpart_subreg to mode rather than SImode.
18974 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18975 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18976 Use DImode instead of SImode for the shift count operand.
18977 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18980 2017-04-04 Richard Biener <rguenther@suse.de>
18982 PR middle-end/80281
18983 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18984 arithmetic done for the negate or the plus. Simplify.
18985 (A - (-B) -> A + B): Likewise.
18986 * fold-const.c (split_tree): Make sure to not negate pointers.
18988 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18990 PR rtl-optimization/60818
18991 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18992 a compare of comparisons with the thing compared if this results
18993 in a different machine mode.
18995 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18997 * alias.c (base_alias_check): Fix typo in comment.
18998 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18999 * cgraphunit.c (symbol_table::compile): Likewise.
19000 * collect2.c (maybe_run_lto_and_relink): Likewise.
19001 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19002 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19003 * config/avr/avr.c (avr_map_op_t): Likewise.
19004 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19005 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19006 * config/epiphany/epiphany.md (movcc): Likewise.
19007 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19008 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19010 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19011 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19012 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19013 * config/sh/sh.c (sh_rtx_costs): Likewise.
19014 * fold-const.c (fold_truth_andor): Likewise.
19015 * genautomata.c (collapse_flag): Likewise.
19016 * gengtype.h (struct type::u::s): Likewise.
19017 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19018 * input.c (FORMAT_AMOUNT): Likewise.
19019 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19020 (known_aggs_to_agg_replacement_list): Likewise.
19021 * ipa-inline-analysis.c: Likewise.
19022 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19023 * ipa-polymorphic-call.c
19024 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19025 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19026 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19028 * modulo-sched.c (apply_reg_moves): Likewise.
19029 * omp-expand.c (build_omp_regions_1): Likewise.
19030 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19031 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19032 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19033 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19034 * value-prof.c: Likewise.
19035 * var-tracking.c (val_reset): Likewise.
19037 2017-04-03 Richard Biener <rguenther@suse.de>
19039 PR tree-optimization/80275
19040 * fold-const.c (split_address_to_core_and_offset): Handle
19043 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19045 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19046 descriptors is at least equal to that of functions.
19048 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19050 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19052 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19055 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19056 (mov<IMOD4:mode>): New expander.
19057 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19059 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19061 PR rtl-optimization/79405
19062 * fwprop.c (propagations_left): New variable.
19063 (forward_propagate_into): Decrement it.
19064 (fwprop_init): Initialize it.
19065 (fw_prop): If the variable has reached zero, stop propagating.
19066 (fwprop_addr): Ditto.
19068 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19071 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19072 a FUNCTION_DECL, pass it as decl instead of origin to
19075 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19077 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19080 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19083 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19084 TARGET_VSX_SMALL_INTEGER.
19086 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19088 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19089 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19091 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19093 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19094 extraction from odd-numbered MSA register.
19096 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19098 PR middle-end/80173
19099 * expmed.c (store_bit_field_1): Don't attempt to create
19100 a word subreg out of hard registers wider than word if they
19101 have HARD_REGNO_NREGS of 1 for their mode.
19103 PR middle-end/80163
19104 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19105 conversions to integer types wider than word and pointer.
19108 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19109 (rtx_equal_for_cselib_p): Pass 0 to it.
19110 * cselib.c (cselib_hasher::equal): Likewise.
19111 (rtx_equal_for_cselib_1): Add depth argument. If depth
19112 is 128, don't look up VALUE locs and punt. Increment
19113 depth in recursive calls when walking VALUE locs.
19115 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19117 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19118 (make_gcov_file_name): Use the canonical path name for generating
19120 (read_line): Fix handling of files with ascii null bytes.
19122 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19124 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19125 to initialise a vector register instead
19126 of using a const_int.
19128 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19130 PR translation/80189
19131 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19132 diagnostic messages.
19134 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19137 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19138 (dfp_diex_<mode>): Update mode of operand 1.
19139 * doc/extend.texi (dxex, dxexq): Document change to return type.
19140 (diex, diexq): Document change to argument type.
19142 2017-03-30 Martin Jambor <mjambor@suse.cz>
19145 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19146 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19147 it reflects the signature changes performed at the callee side.
19148 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19149 to cgraph_build_function_type_skip_args.
19150 (build_function_decl_skip_args): Adjust call to the above function.
19152 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19155 * config/i386/sse.md
19156 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19157 register as dest whenever it is a MEM not rtx_equal_p to the
19158 corresponding dup operand, and when forcing into reg move the
19159 reg into the memory afterwards.
19160 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19161 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19162 for the force_reg mode.
19163 (avx512vl_vextractf128<mode>): Use register as dest either
19164 always when a MEM, or when it is a MEM not rtx_equal_p to the
19165 corresponding dup operand, or even not when it is a CONST_VECTOR
19166 depending on the mode and lo vs. hi.
19167 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19169 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19170 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19171 Likewise. Require that operands[2] is even.
19172 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19173 Remove extraneous parens. Require that operands[2] is a multiple
19175 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19176 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19177 disallow memory then.
19179 2017-03-30 Richard Biener <rguenther@suse.de>
19181 PR tree-optimization/77498
19182 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19183 to non-constants over backedges.
19185 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19187 PR rtl-optimization/80233
19188 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19189 as last_combined_insn. Do not test for BARRIER_P separately.
19191 2017-03-29 Andreas Schwab <schwab@suse.de>
19194 * calls.c (prepare_call_address): Convert funexp to Pmode before
19195 copying to temp reg.
19197 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19199 PR tree-optimization/80158
19200 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19201 Handle possible future case of more than one alternate
19203 (replace_rhs_if_not_dup): Likewise.
19204 (replace_one_candidate): Likewise.
19206 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19208 PR rtl-optimization/80193
19209 * ira.c (ira): Do not check allocation for LRA.
19211 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19213 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19214 (nvptx_output_simt_exit): Declare.
19215 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19216 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19217 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19218 (nvptx_declare_function_name): ...here. Emit declaration of local
19219 memory space buffer for omp_simt_enter insn.
19220 (nvptx_output_unisimt_switch): New.
19221 (nvptx_output_softstack_switch): New.
19222 (nvptx_output_simt_enter): New.
19223 (nvptx_output_simt_exit): New.
19224 * config/nvptx/nvptx.h (struct machine_function): New fields
19225 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19226 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19227 (UNSPECV_SIMT_EXIT): Ditto.
19228 (omp_simt_enter_insn): New insn.
19229 (omp_simt_enter): New expansion.
19230 (omp_simt_exit): New insn.
19231 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19233 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19234 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19235 (expand_GOMP_SIMT_EXIT): New.
19236 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19237 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19238 (GOMP_SIMT_EXIT): Ditto.
19239 * target-insns.def (omp_simt_enter): New insn.
19240 (omp_simt_exit): Ditto.
19241 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19243 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19244 (lower_rec_input_clauses): Likewise.
19245 (lower_lastprivate_clauses): Handle SIMT privatization.
19247 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19248 (ompdevlow_adjust_simt_enter): New.
19249 (find_simtpriv_var_op): New.
19250 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19251 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19253 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19254 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19255 (copy_decl_for_dup_finish): Ditto.
19257 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19259 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19262 * config/i386/i386.c (ix86_option_override_internal): Always
19263 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19265 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19267 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19269 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19271 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19272 mark new edge's irreducible flag accordign to it.
19273 (vect_do_peeling): Check loop preheader edge's irreducible flag
19274 and pass it to function slpeel_add_loop_guard.
19276 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19278 PR tree-optimization/80218
19279 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19280 Update block frequencies and counts.
19282 2017-03-28 Richard Biener <rguenther@suse.de>
19284 PR tree-optimization/78644
19285 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19286 of a simplification result we may not use it at all.
19288 2017-03-28 Richard Biener <rguenther@suse.de>
19291 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19292 without arguments, generate default definition of a SSA name.
19294 2017-03-28 Richard Biener <rguenther@suse.de>
19296 PR middle-end/80222
19297 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19298 TYPE_REF_CAN_ALIAS_ALL references.
19299 * fold-const.c (fold_indirect_ref_1): Likewise.
19301 2017-03-28 Martin Liska <mliska@suse.cz>
19304 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19305 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19307 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19308 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19310 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19311 (EXTRA_SPECS): Define.
19312 (SUBTARGET_EXTRA_SPECS): Likewise.
19313 (SUBTARGET_CPP_SPEC): Likewise.
19314 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19315 SUBTARGET_EXTRA_SPECS.
19316 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19318 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19320 * config/arc/simdext.md (vst64_insn): Update pattern.
19321 (vld32wh_insn): Likewise.
19322 (vld32wl_insn): Likewise.
19323 (vld64_insn): Likewise.
19324 (vld32_insn): Likewise.
19326 2017-03-28 Marek Polacek <polacek@redhat.com>
19329 * fold-const.c (fold_comparison): Use protected_set_expr_location
19330 instead of SET_EXPR_LOCATION.
19332 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19334 * tree.c (add_expr): Avoid name lookup warning.
19336 2017-03-27 Jeff Law <law@redhat.com>
19338 PR tree-optimization/80216
19339 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19340 function name. Limit recursion depth.
19341 (record_temporary_equivalences): Corresponding changes.
19343 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19345 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19348 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19351 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19353 * cfgcleanup.c (reg_note_cfa_p): New array.
19354 (insns_have_identical_cfa_notes): New function.
19355 (old_insns_match_p): Don't cross-jump in between /f
19356 and non-/f instructions. If both i1 and i2 are frame related,
19357 verify all CFA notes, their order and content.
19359 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19362 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19363 HImode and SImode with zero extend to DImode to one insn.
19364 (bswap<mode>2_extenddi): Likewise.
19365 (bswapsi2_extenddi): Likewise.
19366 (bswaphi2_extendsi): Likewise.
19367 (bswaphi2): Combine bswap HImode and SImode into one insn.
19368 Separate memory insns from swapping register.
19369 (bswapsi2): Likewise.
19370 (bswap<mode>2): Likewise.
19371 (bswaphi2_internal): Delete, no longer used.
19372 (bswapsi2_internal): Likewise.
19373 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19374 store, and gpr<-gpr swap insns.
19375 (bswap<mode>2_store): Likewise.
19376 (bswaphi2_reg): Register only splitter, combine with the splitter.
19377 (bswaphi2 splitter): Likewise.
19378 (bswapsi2_reg): Likewise.
19379 (bswapsi2 splitter): Likewise.
19380 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19381 the insns into load, store, and register/register insns.
19382 (bswapdi2_ldbrx): Likewise.
19383 (bswapdi2_load): Likewise.
19384 (bswapdi2_store): Likewise.
19385 (bswapdi2_reg): Likewise.
19387 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19389 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19390 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19392 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19395 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19398 special handling for target option conflicts between dform
19399 options (-mpower9-dform, -mpower9-dform-vector,
19400 -mpower9-dform-scalar) and -mno-direct-move.
19402 2017-03-27 Richard Biener <rguenther@suse.de>
19404 PR tree-optimization/80181
19405 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19407 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19409 * config/arc/predicates.md (move_double_src_operand): Replace the
19410 call to move_double_src_operand with a call to address_operand.
19412 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19414 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19415 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19416 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19418 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19420 * config/arc/predicates.md (long_immediate_loadstore_operand):
19421 Consider scaled addresses cases.
19423 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19425 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19426 restored when in interrupt.
19427 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19428 doesn't have delay slot.
19430 2017-03-27 Richard Biener <rguenther@suse.de>
19433 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19434 inlined thunk clones.
19436 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19439 * asan.c (instrument_derefs): Copy over last operand from
19440 original COMPONENT_REF to the new COMPONENT_REF with
19441 DECL_BIT_FIELD_REPRESENTATIVE.
19442 * ubsan.c (instrument_object_size): Likewise.
19444 2017-03-27 Richard Biener <rguenther@suse.de>
19446 PR tree-optimization/80170
19447 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19448 sure DR/SCEV didnt fold in constants we do not see when looking
19449 at the reference base alignment.
19451 2017-03-27 Richard Biener <rguenther@suse.de>
19453 PR middle-end/80171
19454 * gimple-fold.c (fold_ctor_reference): Properly guard against
19455 NULL return value from canonicalize_constructor_val.
19457 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19460 * config/i386/i386.c (ix86_expand_builtin)
19461 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19462 flags reg setting and flags reg using instructions.
19463 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19464 clobbering instructions to zero extend op2.
19466 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19468 * doc/install.texi (Configuration) <--with-aix-soname>:
19469 Update link to AIX ld.
19471 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19473 PR rtl-optimization/80160
19474 PR rtl-optimization/80159
19475 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19476 reg_alternate_class into account.
19478 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19481 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19482 to consider in curr_insn_transform.
19484 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19486 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19487 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19488 and emit_mode_inner.
19490 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19492 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19493 argument to the overloaded builtin variants. Use the new flag to
19494 deprecate certain builtin variants.
19495 * config/s390/s390-builtin-types.def: Add new builtin types.
19496 * config/s390/s390-builtins.h: Support new flags field for
19497 overloaded builtins.
19498 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19499 (s390_macro_to_expand): Enable vector float data type.
19500 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19501 builtins by incrementing the __VEC__ version number.
19502 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19504 (s390_resolve_overloaded_builtin): Emit error messages depending
19505 on the builtin flags.
19506 * config/s390/s390.c (s390_expand_builtin): Support additional
19507 flags argument. Change error message to match the messages
19508 emitted in s390-c.c.
19509 * config/s390/s390.md: New UNSPEC_* constants.
19510 (op_type): Add new instruction types.
19511 * config/s390/vecintrin.h: Add new builtins and test data class
19513 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19514 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19515 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19516 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19517 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19518 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19520 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19521 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19522 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19523 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19525 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19526 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19527 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19528 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19530 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19531 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19532 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19533 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19534 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19535 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19536 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19538 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19539 ("vec_scatter_element<V_HW_4:mode>_DI")
19540 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19541 ("vec_fpint<mode>", "vflls")
19542 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19543 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19544 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19545 ("*vec_cmphe<mode>_cc"): ... these.
19547 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19548 mode constant instead of magic value.
19550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19552 * config/s390/s390.c (s390_expand_vec_compare): Support other
19553 vector floating point modes than just V2DF.
19554 (s390_expand_vcond): Likewise.
19555 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19556 (s390_cannot_change_mode_class): Prevent mode changes between TF
19557 and V1TF in vector registers.
19558 * config/s390/s390.md (DF, SF): New mode attributes.
19559 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19560 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19561 SFmode support for VRs.
19562 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19564 (VFT, VF_HW): New mode iterators.
19565 (vw, sdx): New mode attributes.
19566 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19567 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19568 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19569 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19570 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19571 also the new vector floating point modes. Renaming to ...
19573 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19574 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19575 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19576 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19577 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19578 ("vec_unordered<mode>"): ... these.
19580 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19581 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19582 ("*vec_extendv2df"): New insn definitions.
19584 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19586 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19587 ("mulditi3_2", "*muldi3_sign"): New patterns.
19588 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19589 rename the pattern definition.
19591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19593 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19595 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19597 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19599 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19600 instruction if possible.
19601 * config/s390/vector.md (vec_halfnumelts): New mode
19603 ("*vec_vllezlf<mode>"): New pattern.
19605 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19607 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19608 ("popcountv4si2", "popcountv2di2"): Rename to ...
19609 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19610 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19612 ("popcount<mode>2_vxe"): New pattern.
19614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19616 * common/config/s390/s390-common.c (processor_flags_table): Add
19618 * config.gcc: Add arch12.
19619 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19620 Default to arch12 for unknown CPU model numbers.
19621 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19622 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19623 PROCESSOR_max sanity check.
19624 * config/s390/s390-opts.h (enum processor_type): Add
19626 * config/s390/s390.c (processor_table): Add arch12.
19627 (s390_expand_builtin): Add check for B_VXE flag.
19628 (s390_issue_rate): Add PROCESSOR_ARCH12.
19629 (s390_get_sched_attrmask): Likewise.
19630 (s390_get_unit_mask): Likewise.
19631 (s390_sched_score): Enable z13 scheduling for arch12.
19632 (s390_sched_reorder): Likewise.
19633 (s390_sched_variable_issue): Likewise.
19634 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19636 (s390_tune_attr): Use z13 scheduling also for arch12.
19637 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19638 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19639 (TARGET_VXE_P): New macros.
19640 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19641 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19642 * config/s390/s390.opt: Add arch12 as processor_type.
19644 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19646 * config/s390/s390.md
19647 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19648 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19649 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19651 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19652 Rename expanders to ...
19654 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19655 ("fixuns_truncdddi2_emu"): ... these.
19657 ("fixuns_trunc<mode>si2_emu"): New expander.
19659 ("*fixuns_truncdfdi2_z13"): Rename to ...
19660 ("*fixuns_truncdfdi2_vx"): ... this.
19662 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19664 * config/s390/2964.md: Remove the single element vector compare
19665 instructions which are no longer used.
19666 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19668 (s390_canonicalize_comparison): Remove handling of DFmode
19670 (s390_expand_vec_compare_scalar): Remove function.
19671 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19672 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19674 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19676 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19678 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19680 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19681 will anyway by matched by mov<mode>_64dfp.
19683 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19685 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19686 vlef/vstef. Add missing operand to vleif.
19688 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19690 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19691 pair for all vector types with 64 bit elements.
19692 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19693 * config/s390/vector.md (V_HW_64): ... here.
19694 (V_128_NOSINGLE): New mode iterator.
19695 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19696 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19697 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19698 ("*vec_load_pairv2di"): Change to ...
19699 ("*vec_load_pair<mode>"): ... this one.
19701 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19703 * config/s390/constraints.md: Add comments.
19704 (jKK): Reject element sizes > 8 bytes.
19705 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19707 * config/s390/s390.md: Add the s_operand checks formerly in
19708 s390_split_ok_p to various splitters where they are still
19710 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19711 for 128 bit vectors. Plus two splitters.
19713 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19715 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19718 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19721 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19722 error if the boundary argument is not constant.
19724 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19726 PR rtl-optimization/80112
19727 * loop-doloop.c (doloop_condition_get): Don't check condition
19728 if cmp isn't SET with IF_THEN_ELSE src.
19730 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19732 PR tree-optimization/80158
19733 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19734 replacing a candidate statement, also replace it for the
19735 candidate's alternate interpretation.
19736 (replace_rhs_if_not_dup): Likewise.
19737 (replace_one_candidate): Likewise.
19739 2017-03-24 Richard Biener <rguenther@suse.de>
19741 PR tree-optimization/80167
19742 * graphite-isl-ast-to-gimple.c
19743 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19745 (translate_isl_ast_to_gimple::get_rename): Likewise.
19747 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19749 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19750 handling of certain combinations of target options, including the
19751 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19752 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19754 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19757 * config/arm/arm.md (*load_multiple): Add reload_completed to
19758 matching condition.
19760 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19761 Richard Biener <rguenth@suse.de>
19763 PR tree-optimization/79908
19764 PR tree-optimization/80136
19765 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19766 been cast away, gimplify_and_add suffices.
19768 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19770 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19772 2017-03-23 Richard Biener <rguenther@suse.de>
19774 PR tree-optimization/80032
19775 * gimplify.c (gimple_push_cleanup): Forced unconditional
19776 cleanups still have to go to the conditional_cleanups
19779 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19781 PR tree-optimization/80072
19782 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19784 (next_operand_entry_id): Change type to unsigned int.
19785 (sort_by_operand_rank): Make sure to return the right return value
19786 even if unsigned fields are bigger than INT_MAX.
19787 (struct oecount): Change cnt and id type to unsigned int.
19788 (oecount_hasher::equal): Formatting fix.
19789 (oecount_cmp): Make sure to return the right return value
19790 even if unsigned fields are bigger than INT_MAX.
19791 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19794 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19795 TREE_READONLY on result if writing it more than once.
19798 * doc/invoke.texi (-fsanitize=thread): Document that with
19799 -fnon-call-exceptions atomics are not able to throw
19803 * tsan.c: Include tree-eh.h.
19804 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19805 maybe_clean_or_replace_eh_stmt where needed.
19806 (instrument_memory_accesses): Add cfg_changed argument.
19807 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19808 if it returned true.
19809 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19811 PR rtl-optimization/63191
19812 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19813 wrapper function, moved the whole old content into ...
19814 (ix86_delegitimize_address_1): ... this. New inline function.
19815 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19816 true as last argument instead of ix86_delegitimize_address.
19818 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19820 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19821 cortexa57_branch_cost.
19823 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19825 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19827 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19830 * doc/md.texi (Constraints): Document wA constraint.
19831 * config/rs6000/constraints.md (wA): New.
19832 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19833 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19834 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19835 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19837 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19840 * gimplify.c (is_oacc_declared): New function.
19841 (oacc_default_clause): Use it to set default flags for acc declared
19842 variables inside parallel regions.
19843 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19844 declared variables.
19845 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19846 declare attribute to any decl as necessary.
19848 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19851 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19852 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19853 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19854 (arm_arch_lpae): This.
19855 * config/arm/arm.c (arm_arch7ve): Rename into ...
19856 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19857 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19860 2017-03-22 Martin Liska <mliska@suse.cz>
19863 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19864 error message instead of an ICE.
19866 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19868 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19870 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19872 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19874 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19877 2017-03-21 Martin Sebor <msebor@redhat.com>
19879 * doc/extend.texi: Use "cannot" instead of "can't."
19880 * doc/hostconfig.texi: Same.
19881 * doc/install.texi: Same.
19882 * doc/invoke.texi: Same.
19883 * doc/loop.texi: Same.
19884 * doc/md.texi: Same.
19885 * doc/objc.texi: Same.
19886 * doc/rtl.texi: Same.
19887 * doc/tm.texi: Same.
19888 * doc/tm.texi.in: Same.
19889 * doc/trouble.texi: Same.
19891 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19894 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19895 (collect_checksum_attributes): Set it.
19896 (die_checksum_ordered): Use it.
19898 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19900 PR tree-optimization/79908
19901 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19902 change: For a VA_ARG whose LHS has been cast away, use
19903 force_gimple_operand to construct the side effects.
19905 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19907 PR translation/80001
19908 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19909 more amenable to translation.
19910 (oacc_loop_auto_partitions): Likewise.
19912 2017-03-21 Marek Polacek <polacek@redhat.com>
19913 Martin Sebor <msebor@redhat.com>
19915 PR tree-optimization/80109
19916 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19917 on INTEGRAL_TYPE_P.
19919 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19920 Segher Boessenkool <segher@kernel.crashing.org>
19923 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19924 check reg_used_between_p between insn and one of succ or succ2
19925 depending on if succ is artificial insn not inserted into insn
19928 2017-03-21 Martin Liska <mliska@suse.cz>
19930 PR gcov-profile/80081
19931 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19932 * doc/gcc.texi: Include gcov-dump stuff.
19933 * doc/gcov-dump.texi: New file.
19935 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19937 PR rtl-optimization/79150
19938 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19939 conditional jump, if the jump is the last insn of the loop.
19941 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19942 Richard Biener <rguenth@suse.de>
19944 PR tree-optimization/79908
19945 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19946 been cast away, use force_gimple_operand to construct the side
19949 2017-03-21 Martin Liska <mliska@suse.cz>
19951 PR libfortran/79956
19952 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19955 2017-03-21 Brad Spengler <spender@grsecurity.net>
19958 * plugin.c (htab_hash_plugin): New function.
19959 (add_new_plugin): Use it and adjust.
19960 (parse_plugin_arg_opt): Adjust.
19961 (init_one_plugin): Likewise.
19963 2017-03-21 Richard Biener <rguenther@suse.de>
19965 PR tree-optimization/80032
19966 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19967 if set force the cleanup to happen unconditionally.
19968 (gimplify_target_expr): Push inserted clobbers with force_uncond
19969 to avoid them being removed by control-dependent DCE.
19971 2017-03-21 Richard Biener <rguenther@suse.de>
19973 PR tree-optimization/80122
19974 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19975 va_arg_pack_len when the inlined call stmt requires pack
19977 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19979 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19982 * tsan.c (instrument_builtin_call): If the memory model argument
19983 is not a constant, assume it is valid.
19986 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19989 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19991 PR rtl-optimization/79910
19992 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19993 if its dest is used by an insn before I2 (other than the combined
19994 insns themselves, which are properly handled already).
19996 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19999 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20001 * combine.c (record_used_regs): New static function.
20002 (try_combine): Handle situations where there is an additional
20003 instruction between I2 and I3 which needs to have a LOG_LINK
20007 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20009 * combine.c (try_combine): Delete redundant i1 test. Call
20010 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20012 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20015 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20016 alternatives 13/14.
20018 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20020 PR tree-optimization/80054
20021 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20022 the optimization if a PHI or any of its arguments is not dominated
20023 by the candidate's basis. Use gphi* rather than gimple* as
20025 (replace_profitable_candidates): Clean up a gimple* variable that
20026 should be a gphi* variable.
20028 2017-03-20 Martin Sebor <msebor@redhat.com>
20031 * doc/extend.texi (attribute constructor): Document present limitation.
20033 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20036 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20037 __POWER9_VECTOR__ #ifdef control, change template definition to
20038 use Power9-specific built-in function.
20039 (vec_any_eq): Likewise.
20040 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20041 to control outcomes from this test.
20042 (vector_ae_<mode>p): For VEC_F modes, likewise.
20044 2017-03-20 Ian Lance Taylor <iant@google.com>
20046 * config/i386/i386.c (ix86_function_regparm): Save an extra
20047 register for -fsplit-stack with DECL_STATIC_CHAIN.
20049 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20052 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20053 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20055 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20057 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20059 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20062 2017-03-20 Marek Polacek <polacek@redhat.com>
20065 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20067 2017-03-20 Richard Biener <rguenther@suse.de>
20069 PR tree-optimization/80113
20070 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20071 allocate extra SSA name for PHI def.
20072 (add_close_phis_to_outer_loops): Likewise.
20073 (add_close_phis_to_merge_points): Likewise.
20074 (copy_loop_close_phi_args): Likewise.
20075 (copy_cond_phi_nodes): Likewise.
20077 2017-03-20 Martin Liska <mliska@suse.cz>
20079 PR middle-end/79753
20080 * tree-chkp.c (chkp_build_returned_bound): Do not build
20081 returned bounds for a LHS that's not a BOUNDED_P type.
20083 2017-03-20 Martin Liska <mliska@suse.cz>
20087 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20088 COMPLEX_CST and VECTOR_CST.
20090 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20093 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20094 target operand. A new splitter adds the clobber statement in case
20095 the target operand is dead anyway.
20097 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20099 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20100 to age-old versions of binutils and glibc.
20102 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20104 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20106 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20108 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20110 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20112 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20113 requirement for binutils 2.13.
20115 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20117 * combine.c (try_combine): Delete redundant i1 test. Call
20118 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20120 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20122 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20123 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20125 <riscv64-*-elf>: Re-arrange section
20126 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20127 <riscv32-*-linux>: Likewise.
20128 <riscv64-*-elf>: Likewise
20129 <riscv64-*-linux>: Likewise.
20131 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20134 * aarch64.opt(verbose-cost-dump): Fix typo.
20136 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20139 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20140 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20142 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20144 * reload.c (find_reloads): When reloading a nonoffsettable address,
20145 use RELOAD_OTHER for it and its address reloads.
20147 PR rtl-optimization/79910
20148 * combine.c (record_used_regs): New static function.
20149 (try_combine): Handle situations where there is an additional
20150 instruction between I2 and I3 which needs to have a LOG_LINK
20153 2017-03-17 Jeff Law <law@redhat.com>
20155 PR tree-optimization/71437
20156 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20157 conditional in the hash table first.
20158 (vrp_dom_walker::before_dom_children): Extract condition from
20159 ASSERT_EXPR. Record condition, its inverion and any implied
20160 conditions as well.
20162 2017-03-17 Marek Polacek <polacek@redhat.com>
20163 Markus Trippelsdorf <markus@trippelsdorf.de>
20165 PR tree-optimization/80079
20166 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20169 2017-03-17 Richard Biener <rguenther@suse.de>
20171 PR middle-end/80075
20172 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20173 Properly verify the LHS before the RHS possibly claims to be
20175 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20178 2017-03-17 Martin Jambor <mjambor@suse.cz>
20180 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20181 (List of -O2 options): Likewise.
20182 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20185 2017-03-17 Tom de Vries <tom@codesourcery.com>
20187 * gcov-dump.c (print_usage): Print bug_report_url.
20189 2017-03-17 Richard Biener <rguenther@suse.de>
20191 PR middle-end/80050
20192 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20193 (parser::peek): Likewise.
20195 2017-03-17 Richard Biener <rguenther@suse.de>
20197 PR tree-optimization/80048
20198 * sese.c (free_sese_info): Properly release rename_map and
20199 copied_bb_map elements.
20201 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20203 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20204 Add linked-list forward and backlinks. Insert on
20205 construction, remove on destruction.
20206 (class pass_store_merging): Add m_stores_head field.
20207 (pass_store_merging::terminate_and_process_all_chains):
20208 Iterate over m_stores_head list.
20209 (pass_store_merging::terminate_all_aliasing_chains):
20211 (pass_store_merging::execute): Check for debug stmts first.
20212 Push new chains onto the m_stores_head stack.
20214 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20217 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20218 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20219 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20221 2017-03-16 Jeff Law <law@redhat.com>
20223 PR tree-optimization/71437
20224 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20225 member function. Implementation moved into after_dom_children
20226 member function and into the threader's thread_outgoing_edges
20228 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20229 some code into new thread_outgoing_edges.
20230 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20231 definition. Simplify marker handling (do it here). Assume we always
20232 have the available expression and the const/copies tables.
20233 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20235 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20236 * tree-vrp.c (equiv_stack): No longer file scoped.
20237 (vrp_dom_walker): New class.
20238 (vrp_dom_walker::before_dom_children): New member function.
20239 (vrp_dom_walker::after_dom_children): Likewise.
20240 (identify_jump_threads): Setup domwalker. Use it rather than
20241 walking edges in a random order by hand. Simplify setup/finalization.
20242 (finalize_jump_threads): Remove.
20243 (vrp_finalize): Do not call identify_jump_threads here.
20244 (execute_vrp): Do it here instead and call thread_through_all_blocks
20247 PR tree-optimization/71437
20248 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20250 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20252 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20253 (dom_opt_dom_walker::thread_across_edge): Remove
20254 handle_dominating_asserts argument. All callers changed.
20255 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20256 changes. Remove calls to lhs_of_dominating_assert. Other
20257 uses of handle_dominating_asserts turn into unconditional code
20258 (simplify_control_stmt_condition_1): Likewise.
20259 (simplify_control_stmt_condition): Likewise.
20260 (thread_through_normal_block, thread_across_edge): Likewise.
20261 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20262 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20263 object if it is not an SSA_NAME.
20264 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20265 before calling into the VRP specific simplifiers.
20266 (identify_jump_threads): Remove handle_dominating_asserts
20269 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20272 * tree-diagnostic.c (default_tree_printer): No longer static.
20273 * tree-diagnostic.h (default_tree_printer): New prototype.
20275 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20277 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20278 Change ins into fmov.
20280 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20282 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20283 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20284 Use h_con constraint for operand 1.
20285 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20286 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20288 2017-03-15 Jeff Law <law@redhat.com>
20290 PR tree-optimization/71437
20291 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20292 (record_temporary_equivalences): Use it.
20294 PR tree-optimization/71437
20295 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20296 tree-ssa-scopedtables.
20297 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20298 (record_conditions, record_cond, vuse_eq): Likewise.
20299 (record_edge_info): Adjust to API tweak of record_conditions.
20300 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20301 (record_temporary_equivalences, optimize_stmt): Likewise.
20302 (eliminate_redundant_computations): Likewise.
20303 (record_equivalences_from_stmt): Likewise.
20304 * tree-ssa-scopedtables.c: Include options.h and params.h.
20305 (vuse_eq): New function, moved from tree-ssa-dom.c
20306 (build_and_record_new_cond): Likewise.
20307 (record_conditions): Likewise. Accept vector of conditions rather
20308 than edge_equivalence structure for first argument.
20309 for the first argument.
20310 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20311 from tree-ssa-dom.c.
20312 (avail_exprs_stack::record_cond): Likewise.
20313 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20314 from tree-ssa-dom.c.
20315 (avail_exprs_stack): Add new member functions lookup_avail_expr
20317 (record_conditions): Declare.
20319 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20322 * lra-constraints.c (process_alt_operands): Increase reject for
20323 reloading an input/output operand.
20325 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20328 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20329 insns to convert from signed/unsigned char/short to IEEE 128-bit
20331 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20333 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20336 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20337 subreg of inner mode for values already in registers.
20339 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20341 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20342 iteration reg is used after the loop.
20344 2017-03-14 Martin Sebor <msebor@redhat.com>
20346 PR tree-optimization/79800
20347 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20348 precision in negative-positive range.
20349 (format_floating): Call non-const overload with adjusted precision.
20351 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20354 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20357 2017-03-14 Martin Sebor <msebor@redhat.com>
20359 PR middle-end/80020
20360 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20361 * builtins.def (aligned_alloc): Use it.
20364 * Makefile.in (GTFILES): Add calls.c.
20365 * calls.c: Include "gt-calls.h".
20367 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20369 PR rtl-optimization/79728
20370 * regs.h (struct target_regs): New field
20371 x_contains_allocatable_regs_of_mode.
20372 (contains_allocatable_regs_of_mode): New macro.
20373 * reginfo.c (init_reg_sets_1): Initialize it, and change
20374 contains_reg_of_mode so it includes global regs as well.
20375 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20376 rather than contains_regs_of_mode.
20378 2017-03-14 Martin Liska <mliska@suse.cz>
20380 * doc/invoke.texi: Document options that can't be combined with
20381 -fcheck-pointer-bounds.
20383 2017-03-14 Martin Liska <mliska@suse.cz>
20385 PR middle-end/79831
20386 * doc/invoke.texi (-Wchkp): Document the option.
20388 2017-03-14 Martin Liska <mliska@suse.cz>
20390 * Makefile.in: Install gcov-dump.
20392 2017-03-14 Martin Liska <mliska@suse.cz>
20394 * multiple_target.c (expand_target_clones): Bail out for
20395 an invalid attribute.
20397 2017-03-14 Richard Biener <rguenther@suse.de>
20399 * alias.c (struct alias_set_entry): Pack properly.
20400 * cfgloop.h (struct loop): Likewise.
20401 * cse.c (struct set): Likewise.
20402 * ipa-utils.c (struct searchc_env): Likewise.
20403 * loop-invariant.c (struct invariant): Likewise.
20404 * lra-remat.c (struct cand): Likewise.
20405 * recog.c (struct change_t): Likewise.
20406 * rtl.h (struct address_info): Likewise.
20407 * symbol-summary.h (function_summary): Likewise.
20408 * tree-loop-distribution.c (struct partition): Likewise.
20409 * tree-object-size.c (struct object_size_info): Likewise.
20410 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20411 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20412 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20413 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20414 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20415 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20416 (struct _stmt_vec_info): Likewise.
20418 2017-03-14 Martin Liska <mliska@suse.cz>
20421 * multiple_target.c (create_dispatcher_calls): Check that
20422 a target can create a function dispatcher.
20424 2017-03-14 Martin Liska <mliska@suse.cz>
20427 * multiple_target.c (expand_target_clones): Drop local.local
20428 flag for default implementation.
20430 2017-03-14 Richard Biener <rguenther@suse.de>
20432 PR tree-optimization/80030
20433 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20435 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20437 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20438 gcc_fallthrough() instead of __attribute__((fallthrough));
20440 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20442 * doc/gcc.texi: Remove "up" link to (DIR).
20443 * doc/gccint.texi: Ditto.
20445 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20447 * doc/install.texi (Specific) <avr>: Remove reference to
20450 2017-03-13 Jeff Law <law@redhat.com>
20452 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20453 attribute rather than comments.
20455 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20456 match_scratch operand is highest.
20458 2017-03-13 Martin Liska <mliska@suse.cz>
20460 PR middle-end/78339
20461 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20462 is a CHKP clone, use original declaration.
20464 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20466 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20467 (arc_conditional_register_usage): Use a different allocation order
20468 when optimizing for size.
20469 * common/config/arc/arc-common.c (arc_option_optimization_table):
20470 Section anchors default on when optimizing for size.
20472 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20474 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20476 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20478 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20479 * config/arc/arc.md (cpu_facility): Add cd variant.
20480 (*movqi_insn): Add code density variant.
20481 (*movhi_insn): Likewise.
20482 (*movqi_insn): Likewise.
20483 (*addsi3_mixed): Likewise.
20484 (subsi3_insn): Likewise.
20486 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20488 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20490 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20492 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20493 expressions with MINUS and UNARY ops.
20495 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20498 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20500 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20501 between vec_select and vector argument.
20502 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20503 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20504 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20505 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20506 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20507 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20509 2017-03-13 Richard Biener <rguenther@suse.de>
20512 * params.def (vect-max-peeling-for-alignment): Fix typo.
20514 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20516 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20517 issue that only occurred with binutils below 2.18.
20519 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20521 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20522 refer to binutils 2.11/2.12 minimum.
20524 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20526 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20527 ftp.kernel.org and simplify binutils requirement.
20529 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20531 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20533 (Optimize Options): Ditto. Also remove redundancy.
20535 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20537 PR translation/79848
20538 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20540 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20541 to G_ to avoid double translation.
20543 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20545 PR translation/79923
20546 * auto-profile.c (get_combined_location): Convert leading
20547 character of diagnostics to lower case and remove trailing period.
20548 (read_profile): Likewise for various diagnostics.
20549 * config/arm/arm.c (arm_option_override): Remove trailing period
20550 from various diagnostics.
20551 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20552 (msp430_expand_delay_cycles): Likewise.
20554 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20557 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20558 full command-line argument, rather than just "str".
20559 (aarch64_validate_march): Likewise.
20560 (aarch64_validate_mtune): Likewise.
20562 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20564 PR rtl-optimization/78911
20565 * lra-assigns.c (must_not_spill_p): New function.
20566 (spill_for): Use it.
20568 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20570 PR tree-optimization/79981
20571 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20572 ATOMIC_COMPARE_EXCHANGE ifn result.
20573 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20574 IFN_ATOMIC_COMPARE_EXCHANGE.
20576 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20579 * opts.c (parse_sanitizer_options): Add missing question mark to
20580 "did you mean" message.
20582 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20584 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20586 (VMULEUH_UNS): Likewise.
20587 (VMULOUB_UNS): Likewise.
20588 (VMULOUH_UNS): Likewise.
20589 * config/rs6000/rs6000.c (builtin_function_type): Remove
20590 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20592 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20595 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20596 x with result of extra_parsing_for_operand_code_0.
20597 (function_reader::extra_parsing_for_operand_code_0): Convert
20598 return type from void to rtx, returning x. When reading
20599 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20600 larger size containing struct block_symbol.
20602 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20605 -mfloat128-hardware without -m64.
20607 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20610 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20611 entries to the case statement that marks unsigned arguments to
20612 overloaded functions.
20614 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20616 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20617 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20619 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20622 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20623 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20625 2017-03-10 Martin Liska <mliska@suse.cz>
20629 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20630 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20633 2017-03-10 Olivier Hainque <hainque@adacore.com>
20635 * tree-switch-conversion (array_value_type): Start by resetting
20636 candidate type to it's main variant.
20638 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20640 PR rtl-optimization/79909
20641 * combine.c (try_combine): Use simplify_replace_rtx on individual
20642 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20643 of the whole CALL_INSN_FUNCTION_USAGE.
20645 PR tree-optimization/79972
20646 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20647 get_range_info on SSA_NAMEs. Formatting fixes.
20649 2017-03-10 Richard Biener <rguenther@suse.de>
20650 Jakub Jelinek <jakub@redhat.com>
20652 PR tree-optimization/77975
20653 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20654 edge to be constant.
20655 (get_val_for): For constant x return it. Formatting fix.
20656 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20657 would use the same bases as the current one.
20659 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20661 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20662 instead of vec_select for V1TImode.
20663 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20665 (VSX_LE_128): Add V1TI to this mode iterator.
20666 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20667 (*vsx_le_perm_store_<mode>): Likewise.
20668 (pre-reload splitter for VSX stores): Likewise.
20669 (post-reload splitter for VSX stores): Likewise.
20670 (*vsx_xxpermdi2_le_<mode>): Likewise.
20671 (*vsx_lxvd2x2_le_<mode>): Likewise.
20672 (*vsx_stxvd2x2_le_<mode>): Likewise.
20674 2017-03-09 Michael Eager <eager@eagercon.com>
20676 Correct failures with --enable-checking=yes,rtl.
20678 * config/microblaze/microblaze.c (microblaze_expand_shift):
20679 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20680 test for const0_rtx.
20681 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20682 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20684 2017-03-09 Richard Biener <rguenther@suse.de>
20686 PR tree-optimization/79977
20687 * graphite-scop-detection.c (scop_detection::merge_sese):
20688 Handle the case of extra exits to blocks dominating the entry.
20690 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20692 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20695 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20697 PR rtl-optimization/79949
20698 * lra-constraints.c (process_alt_operands): Check memory when
20699 trying to predict a cycle. Print about the overall increase.
20701 2017-03-09 Richard Biener <rguenther@suse.de>
20703 PR middle-end/79971
20704 * gimple-expr.c (useless_type_conversion_p): Preserve
20705 TYPE_SATURATING for fixed-point types.
20707 2017-03-09 Richard Biener <rguenther@suse.de>
20710 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20711 alignment of BLKmode params.
20713 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20716 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20717 (VALL_NO_V2Q): Likewise.
20719 * config/aarch64/aarch64-simd.md
20720 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20722 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20723 VALL_NO_V2Q mode iterator.
20724 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20726 2017-03-09 Martin Liska <mliska@suse.cz>
20728 PR tree-optimization/79631
20729 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20730 tree_int_cst_sign_bit just for INTEGER constants.
20732 2017-03-09 Martin Liska <mliska@suse.cz>
20736 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20739 2017-03-09 Marek Polacek <polacek@redhat.com>
20742 * tree.c (inchash::add_expr): Handle TREE_VEC.
20744 2017-03-09 Martin Liska <mliska@suse.cz>
20747 (chkp_narrow_size_and_offset): New function.
20748 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20749 (void chkp_parse_bit_field_ref): New function.
20750 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20751 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20753 2017-03-09 Martin Liska <mliska@suse.cz>
20756 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20757 (chkp_find_bounds_1): Remove gcc_unreachable.
20759 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20762 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20763 BUILT_IN_SYNC*, determine the access type from the size suffix and
20764 always build a MEM_REF with that type. Handle forgotten
20765 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20768 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20769 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20770 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20771 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20772 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20773 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20774 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20775 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20776 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20777 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20778 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20779 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20780 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20781 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20782 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20783 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20784 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20785 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20786 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20787 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20788 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20789 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20790 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20791 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20792 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20793 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20794 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20795 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20796 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20797 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20798 definitions outside of __OPTIMIZE__ guarded section.
20801 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20802 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20803 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20804 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20807 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20809 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20810 ("vfenez<mode>"): Add missing constraints.
20812 2017-03-08 Martin Sebor <msebor@redhat.com>
20815 * config/nds32/nds32.c (nds32_option_override):
20816 Fix misspelled diagnostic.
20818 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20821 * gimplify.c (gimplify_omp_for): Replace index var in outer
20822 taskloop statement with an artificial variable and add
20823 OMP_CLAUSE_PRIVATE clause for it.
20825 2017-03-08 Richard Biener <rguenther@suse.de>
20827 PR tree-optimization/79955
20828 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20829 for accesses that are completely outside of the variable.
20831 2017-03-08 Andrew Haley <aph@redhat.com>
20833 PR tree-optimization/79943
20834 * tree-ssa-loop-split.c (compute_new_first_bound): When
20835 calculating the new upper bound, (END-BEG) should be added, not
20838 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20840 * config/avr/avr.md (setmemhi): Make sure match_dup
20841 operand number comes before match_scratch.
20843 2017-03-08 Richard Biener <rguenther@suse.de>
20845 PR tree-optimization/79920
20846 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20847 with ncopies == 1 to ...
20848 (vect_transform_slp_perm_load): ... here. Properly compute
20849 all element loads by iterating VF times over the group. Do
20850 not handle ncopies (computed in a broken way) in
20851 vect_create_mask_and_perm.
20853 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20856 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20857 is a uniform vector, use uniform_vector_p return value instead of
20858 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20860 2017-03-07 Marek Polacek <polacek@redhat.com>
20862 PR middle-end/79809
20863 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20864 (alloca_call_type): Likewise.
20866 2017-03-07 Martin Liska <mliska@suse.cz>
20868 * gcov.c (process_args): Put comment to correct location.
20870 2017-03-07 Martin Liska <mliska@suse.cz>
20872 PR middle-end/68270
20873 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20874 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20875 (chkp_narrow_bounds_for_field): Likewise.
20876 (chkp_parse_array_and_component_ref): Pass one more argument to
20879 2017-03-07 Richard Biener <rguenther@suse.de>
20881 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20884 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20886 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20887 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20889 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20892 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20893 to end of description.
20894 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20896 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20898 PR rtl-optimization/79901
20899 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20901 (*avx512f_<code><mode>3<mask_name>): ... this.
20902 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20903 iterator instead of VI8_AVX2_AVX512BW.
20905 PR rtl-optimization/79901
20906 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20907 min/max expander, expand it using expand_vec_cond_expr.
20910 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20913 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20916 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20917 to void * for PCH reasons.
20918 * dwarf2out.c (output_loc_operands, output_die): Cast
20919 v.val_vec.array to unsigned char *.
20921 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20924 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20927 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20929 PR rtl-optimization/79571
20930 * lra-constraints.c (process_alt_operands): Calculate static
20931 reject and subtract it from overall when only addresses will be
20934 2017-03-06 Julia Koval <julia.koval@intel.com>
20937 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20938 incoming stack boundary to 128 for 64-bit targets.
20940 2017-03-06 Richard Biener <rguenther@suse.de>
20942 PR tree-optimization/79894
20943 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20944 to NULL after folding it.
20946 2017-03-06 Richard Biener <rguenther@suse.de>
20948 PR tree-optimization/79824
20949 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20950 check disabling peeling for gaps.
20952 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20954 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20955 attributes): Document gettimeofday.
20957 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20959 * config/s390/s390.c (s390_option_override_internal): Set
20960 PARAM_MIN_VECT_LOOP_BOUND
20962 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20964 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20965 * config/s390/s390.md: Likewise.
20967 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20970 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20971 (<avx2_avx512>_perm<mode>): Rename to ...
20972 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20974 (<avx512>_perm<mode>_mask): Rename to ...
20975 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20977 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20978 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20979 instead of VI8F_256_512.
20980 (avx512f_perm<mode>): New define_expand.
20981 (avx512f_perm<mode>_mask): Likewise.
20982 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20983 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20985 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20987 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20988 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20990 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20992 2017-03-06 Martin Liska <mliska@suse.cz>
20995 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20996 when having a SSA NAME w/o VAR_DECL assigned to it.
20998 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21000 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21001 msa_dpsub_<su>_d): Fix MODE for vec_select.
21003 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21005 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21007 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21009 2017-03-06 Richard Biener <rguenther@suse.de>
21011 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21012 * plugin.c (register_plugin_info): Likewise.
21013 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21015 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21017 * config/i386/sse.md (sse_storehps, sse_storelps,
21018 avx_<castmode><avxsizesuffix>_<castmode>,
21019 avx512f_<castmode><avxsizesuffix>_<castmode>,
21020 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21021 in condition that at least one operand is not a MEM.
21023 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21025 PR middle-end/79805
21026 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21027 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21029 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21030 gimple_call_nothrow_p flag based on whether original builtin can throw.
21031 If it can, emit following stmts on the fallthrough edge.
21032 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21033 don't create new bb if inserting just debug stmts on the edge, try to
21034 insert them on the fallthru bb or just reset debug stmts.
21036 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21039 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21040 restore recog_data (including the operand rtxes inside it) around
21041 the call to get_insn_template.
21043 2017-03-03 Martin Sebor <msebor@redhat.com>
21045 PR tree-optimization/79699
21046 * context.c (context::~context): Free MPFR caches to avoid
21047 a memory leak on program exit.
21049 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21051 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21052 Use wide_int::ulow () instead of .elt (0).
21054 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21056 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21057 (*pushxf): Limit oF constraint to 32bit targets and add oC
21058 constraint for 64bit targets.
21059 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21060 (*pushdf): Change rmF constraint to rmC.
21062 2017-03-03 Martin Liska <mliska@suse.cz>
21064 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21065 Remove unused variable.
21067 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21070 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21071 is a memory operand, increase num_memory.
21072 (ix86_expand_args_builtin): Likewise.
21074 2017-03-03 Jan Hubicka <jh@suse.cz>
21077 * ipa-devirt.c (maybe_record_node): Properly handle
21078 __cxa_pure_virtual visibility.
21080 2017-03-03 Martin Liska <mliska@suse.cz>
21082 PR tree-optimization/79803
21083 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21085 (pass_loop_prefetch::execute): Disabled optimization if an
21086 assumption about L1 cache size is not met.
21088 2017-03-03 Martin Liska <mliska@suse.cz>
21090 PR rtl-optimization/79574
21091 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21092 (hash_scan_set): Likewise.
21093 (dump_hash_table): Likewise.
21094 (hoist_code): Likewise.
21096 2017-03-03 Richard Biener <rguenther@suse.de>
21098 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21099 in place of uhigh ().
21100 (fixed_convert_from_real): Likewise.
21102 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21105 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21107 2017-03-03 Richard Biener <rguenther@suse.de>
21109 PR middle-end/79818
21110 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21111 TYPE_OVERFLOW_UNDEFINED check.
21113 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21115 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21117 (vector_ae_<mode>_p): Likewise.
21118 (vector_nez_<mode>_p): Likewise.
21119 (vector_ne_v2di_p): Likewise.
21120 (vector_ae_v2di_p): Likewise.
21121 (vector_ne_<mode>_p): Likewise.
21122 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21124 (vsx_tsqrt<mode>2_fe): Likewise.
21126 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21129 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21131 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21133 PR rtl-optimization/79780
21134 * cprop.c (one_cprop_pass): When second and further conditional trap
21135 in a single basic block is turned into an unconditional trap, turn it
21136 into a deleted note to avoid RTL verification failures.
21138 2017-03-02 Richard Biener <rguenther@suse.de>
21140 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21142 2017-03-02 Richard Biener <rguenther@suse.de>
21144 PR tree-optimization/79345
21146 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21147 param and abort the walk, returning -1 if it is hit.
21148 (walk_aliased_vdefs): Take a limit param and pass it on.
21149 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21150 defaulting to 0 and return a signed int.
21151 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21152 (check_defs): New helper.
21153 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21154 about uninitialized memory.
21155 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21156 bogus uninitialized warning.
21157 (fixed_convert_from_real): Likewise.
21159 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21161 PR tree-optimization/66768
21162 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21163 iv_use if base object can't be determined.
21165 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21167 PR tree-optimization/79345
21168 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21169 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21170 (get_pattern_stats): Initialize it.
21171 * genemit.c (gen_expand): Verify match_scratch numbers come after
21172 match_operand/match_dup numbers.
21173 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21174 match_scratch numbers.
21175 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21177 * config/s390/s390.md (trunctdsd2): Likewise.
21179 2017-03-02 Richard Biener <rguenther@suse.de>
21181 * wide-int.h (wide_int_storage::operator=): Implement in terms
21184 2017-03-02 Richard Biener <rguenther@suse.de>
21186 PR tree-optimization/79777
21187 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21188 the to insert expression to sth existing.
21190 2017-03-01 Martin Sebor <msebor@redhat.com>
21192 PR middle-end/79692
21193 * gimple-ssa-sprintf.c
21194 (directive::known_width_and_precision): New function.
21195 (format_integer): Use it.
21196 (get_mpfr_format_length): Consider the full range of precision
21197 when computing %g output with the # flag. Set the likely byte
21198 count to 3 rather than 1 when precision is indeterminate.
21199 (format_floating): Correct the lower bound of precision.
21201 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21203 * doc/invoke.texi: Document default code model for 64-bit Linux.
21205 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21208 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21209 udiv rather than div since input pattern is unsigned.
21211 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21213 * config/i386/i386.c (print_reg): Warn for values of
21214 unsupported size in integer register.
21216 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21219 * config/rs6000/predicates.md (current_file_function_operand): Do
21220 not allow self calls to be local if the function is replaceable.
21222 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21225 * config/rs6000/altivec.h (vec_ctz and others): Change the
21226 preprocessor macro that controls conditional compilation from
21227 _ARCH_PWR9 to __POWER9_VECTOR__.
21228 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21229 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21230 control (instead of _ARCH_PWR9 control) so that template
21231 definition uses power9-specific function.
21232 (vec_any_eq): Likewise.
21233 (vec_all_ne): Change macro definition to use a power9-specific
21234 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21235 _ARCH_PWR9 control).
21236 (vec_any_eq) Likewise.
21237 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21238 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21239 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21240 support for xvcmpnedp instruction.
21241 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21242 macro expansion so that Power9 implementation of vec_all_ne does
21243 not use the AltiVec predicate framework.
21244 (VCMPNEH_P): Likewise.
21245 (VCMPNEW_P): Likewise.
21246 (VCMPNED_P): Likewise.
21247 (VCMPNEFP_P): Likewise.
21248 (VCMPNEDP_P): Likewise.
21249 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21250 implementation of vec_any_eq to not use AltiVec predicate
21252 (VCMPAEH_P): Likewise.
21253 (VCMPAEW_P): Likewise.
21254 (VCMPAED_P): Likewise.
21255 (VCMPAEFP_P): Likewise.
21256 (VCMPAEDP_P): Likewise.
21257 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21258 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21259 not use the AltiVec predicate framework.
21260 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21261 of vec_any_eq to not use AltiVec predicate framework.
21262 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21263 support for predefined __POWER9_VECTOR__ macro to indicate that
21264 Power9 instruction selection is enabled.
21265 (altivec_overloaded_builtins): Remove extraneous
21266 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21267 function argument types RS6000_BTI_bool_V16QI and
21268 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21269 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21270 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21271 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21272 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21273 Power9 for implementations of vec_cmpne. Change the signature for
21274 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21275 (representing vec_all_ne) to remove the previously described first
21276 argument of type RS6000_BTI_INTSI, as this was an artifact of
21277 reliance on the AltiVec predicate framework, which is no longer
21278 used in the implementation of these functions. Add
21279 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21280 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21281 since, unlike the AltiVec predicate framework implementation, we
21282 do not share function descriptors between vec_alle and vec_anyeq.
21283 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21284 set of modes that receive special treatment even when
21285 TARGET_P9_VECTOR is true. The special treatment emits code that
21286 does not depend on Power9 instructions.
21287 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21288 define_expand to not rely on AltiVec predicate framework.
21289 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21291 (vector_ne_v2di_p): Change this define_expand to not rely on
21292 AltiVec predicate framework.
21293 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21295 (vector_ne_<mode>_p): Change this define_expand to not rely on
21296 AltiVec predicate framework.
21297 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21299 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21300 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21301 define_insn pattern.
21302 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21303 define_insn pattern because the xvcmpne<VSs>. instruction is not
21305 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21306 instruction is not supported.
21308 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21310 * config/nvptx/nvptx.c: Include intl.h.
21312 2017-03-01 Martin Jambor <mjambor@suse.cz>
21315 * ipa-prop.h (ipa_bits): Removed field known.
21316 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21317 to pointers. Adjusted their comments to warn about their sharing.
21318 (ipcp_transformation_summary): Change bits to a vector of pointers.
21319 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21320 (ipa_get_ipa_bits_for_value): Declare.
21321 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21322 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21323 (ipa_bits_hash_table): Likewise.
21324 (ipa_vr_ggc_hash_traits): Likewise.
21325 (ipa_vr_hash_table): Likewise.
21326 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21327 being pointers and vr_known being removed.
21328 (ipa_set_jf_unknown): Likewise.
21329 (ipa_get_ipa_bits_for_value): New function.
21330 (ipa_set_jfunc_bits): Likewise.
21331 (ipa_get_value_range): New overloaded functions.
21332 (ipa_set_jfunc_vr): Likewise.
21333 (ipa_compute_jump_functions_for_edge): Use the above functions to
21334 construct bits and vr parts of jump functions.
21335 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21336 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21338 (ipcp_grow_transformations_if_necessary): Also allocate
21339 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21341 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21342 them. Fix too long lines.
21343 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21344 vr_known being removed.
21345 (ipa_read_jump_function): Use new setter functions to construct bits
21346 and vr parts of jump functions or set them to NULL.
21347 (write_ipcp_transformation_info): Adjust for bits being pointers.
21348 (read_ipcp_transformation_info): Likewise.
21349 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21351 Include gt-ipa-prop.h.
21352 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21354 (ipcp_store_bits_results): Likewise.
21355 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21356 Do not write to existing jump functions but use a temporary instead.
21358 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21361 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21362 attempt to use its first operand as BIT_FIELD_REF base.
21364 2017-03-01 Richard Biener <rguenther@suse.de>
21366 PR middle-end/79721
21367 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21368 interpolating formula in wrapping arithmetic.
21369 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21371 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21373 PR tree-optimization/79734
21374 * tree-vect-generic.c (expand_vector_condition): Optimize
21375 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21376 Handle VEC_COND_EXPR where comparison has different inner width from
21377 type's inner width.
21379 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21381 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21382 markup, and similar issues. Remove @opindex entries for things
21383 that aren't options. Add missing -mmpy-option entries.
21385 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21387 PR tree-optimization/79737
21388 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21389 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21390 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21391 instead of byte_size. Formatting fix.
21392 (shift_bytes_in_array_right): Formatting fix.
21394 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21397 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21398 condition on optimize for the leaf function test.
21400 2017-02-28 Martin Liska <mliska@suse.cz>
21403 * read-rtl-function.c (function_reader::handle_unknown_directive):
21404 Bail out when one uses -flto.
21406 2017-02-28 Martin Liska <mliska@suse.cz>
21408 * common.opt: Replace space with tabular for options of <number>
21410 * config/i386/i386.opt: Show <number> value for
21411 -mlarge-data-threshold.
21412 * opts.c (print_filtered_help): Do not display number in hexadecimal
21415 2017-02-28 Martin Liska <mliska@suse.cz>
21417 * common.opt: Fix --help=option -Q for options which are of
21420 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21422 * config/i386/i386.c (print_reg): Error out for values
21423 of 8-bit size in invalid integer register.
21425 2017-02-28 Martin Sebor <msebor@redhat.com>
21427 PR tree-optimization/79691
21428 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21430 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21433 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21434 gcc_unreachable with output_operand_lossage.
21436 2017-02-28 Richard Biener <rguenther@suse.de>
21438 PR tree-optimization/79740
21439 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21441 (visit_nary_op): Insert the nary into the hashtable if we
21442 pattern-matched sth.
21443 * tree-ssa-pre.c (eliminate_insert): Robustify.
21445 2017-02-28 Richard Biener <rguenther@suse.de>
21447 PR middle-end/79731
21448 * fold-const.c (decode_field_reference): Reject out-of-bound
21451 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21453 * config/i386/i386.c: Include intl.h.
21454 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21455 instead of just cond ? "..." : "...".
21456 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21457 * coverage.c (read_counts_file): Likewise.
21458 * omp-offload.c: Include intl.h.
21459 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21460 of just cond ? "..." : "...".
21461 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21462 of just cond ? "..." : "...".
21464 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21467 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21469 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21470 'tune for' CPU name.
21471 * config/arm/arm-cpu-data.h: Regenerated.
21473 2017-02-28 Richard Biener <rguenther@suse.de>
21475 PR tree-optimization/79732
21476 * tree-inline.c (expand_call_inline): Do not shadow var.
21478 2017-02-28 Richard Biener <rguenther@suse.de>
21480 PR tree-optimization/79723
21481 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21482 address-space properly.
21484 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21486 * doc/optinfo.texi (Optimization groups): Fix option used for
21488 * doc/invoke.texi (-fopt-info): Document "omp".
21489 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21490 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21491 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21492 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21493 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21495 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21497 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21498 OPTGROUP_OMP with "omp".
21500 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21503 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21504 for arithmetic shift of unsigned V2DI.
21506 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21508 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21509 arc/linux.h headers.
21510 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21511 (LINK_SPEC): Likewise.
21512 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21513 (EXTRA_SPECS): Likewise.
21514 (STARTFILE_SPEC): Likewise.
21515 (ENDFILE_SPEC): Likewise.
21516 (LIB_SPEC): Likewise.
21517 (TARGET_SDATA_DEFAULT): Likewise.
21518 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21519 (MULTILIB_DEFAULTS): Likewise.
21520 (DWARF2_UNWIND_INFO): Likewise.
21521 * config/arc/big.h: New file.
21522 * config/arc/elf.h: Likewise.
21523 * config/arc/linux.h: Likewise.
21524 * config/arc/t-uClibc: Remove.
21526 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21528 PR tree-optimization/77536
21529 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21530 (tree_transform_and_unroll_loop): Use above function to compute the
21531 estimated niter of unrolled loop and use it when scaling profile.
21532 Also use count info rather than frequency if it's non-zero.
21533 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21534 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21535 (vect_transform_loop): Call above function.
21537 2017-02-27 Richard Biener <rguenther@suse.de>
21539 PR tree-optimization/45397
21540 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21541 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21542 (visit_nary_op): Add pattern matching for CSEing sign-changed
21543 or truncated operations with wider ones.
21545 2017-02-27 Richard Biener <rguenther@suse.de>
21547 PR tree-optimization/79690
21548 * tree-vect-stmts.c (vectorizable_store): Use vector type
21549 built from the DR with address-space.
21551 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21553 * doc/invoke.texi (Optimize Options): Refine the description
21554 of asan-use-after-return.
21556 2017-02-25 Alan Modra <amodra@gmail.com>
21558 PR rtl-optimization/79584
21559 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21560 base, not ad->base_term, the reg within base. Remove assertion
21561 that ad->base == ad->base_term. Replace gen_int_mode using
21562 bogus mode with const0_rtx.
21564 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21566 PR middle-end/79396
21567 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21568 FMA_EXPR like tcc_binary or tcc_unary.
21570 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21573 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21575 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21576 (output_loc_operands): Handle DW_OP_call_ref and
21577 DW_OP_GNU_variable_value.
21578 (struct variable_value_struct): New type.
21579 (struct variable_value_hasher): Likewise.
21580 (variable_value_hash): New variable.
21581 (string_types): Remove.
21582 (copy_loc_descr): New function.
21583 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21584 (prepend_loc_descr_to_each): New function.
21585 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21586 instead of add_loc_descr_to_each if the first argument is single
21587 location list and the second has multiple.
21588 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21589 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21590 when looking for variable value which doesn't have other location info.
21591 (loc_list_from_tree): Formatting fix.
21592 (gen_array_type_die): Simplify DW_AT_string_length handling.
21593 (adjust_string_types): Remove.
21594 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21595 string_types. Call resolve_variable_values.
21596 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21597 (resolve_addr_in_expr): Likewise. Add A argument.
21598 (copy_deref_exprloc): Remove deref argument. Adjust for the
21599 original expression being DW_OP_GNU_variable_value with optionally
21600 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21601 optionally after it.
21602 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21603 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21604 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21605 (variable_value_hasher::hash, variable_value_hasher::equal): New
21607 (resolve_variable_value_in_expr, resolve_variable_value,
21608 resolve_variable_values, note_variable_value_in_expr,
21609 note_variable_value): New functions.
21610 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21613 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21616 * opts.h (handle_generated_option): Add GENERATED_P argument.
21617 * opts-common.c (handle_option): Adjust function comment.
21618 (handle_generated_option): Add GENERATED_P argument, pass it to
21620 (control_warning_option): Pass false to handle_generated_option
21622 * opts.c (maybe_default_option): Pass true to handle_generated_option
21624 * optc-gen.awk: Likewise.
21626 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21628 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21629 a REG, look at the REG it is a SUBREG of.
21630 (splitter for cmpeqsi_t): Ditto.
21632 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21634 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21635 the special USEs with the pattern of the insn, not the insn itself.
21637 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21640 * doc/invoke.texi: Document -mload-store-pairs.
21642 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21643 Sandra Loosemore <sandra@codesourcery.com>
21645 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21646 argument isn't a CONST_INT.
21647 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21648 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21649 (nios2_validate_compare): Bypass alternate compare logic if *op2
21650 is not a CONST_INT.
21651 (ldstwm_operation_p): Return false if first_base is not a REG or
21652 if first_offset is not a CONST_INT.
21654 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21656 * config/cris/cris.md: Use correct operand in a define_peephole2.
21658 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21660 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21662 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21664 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21665 this_insn if it is an INSN or JUMP_INSN.
21666 (force_offsettable): Look at base, not at addr.
21667 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21668 on things that aren't necessarily CONST_INTs.
21670 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21672 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21673 -mfpmath=sse is the default also for x86-32 targets with SSE2
21674 instruction set when @option{-ffast-math} is enabled
21676 2017-02-24 Jeff Law <law@redhat.com>
21678 PR rtl-optimizatoin/79286
21679 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21682 2017-02-24 Richard Biener <rguenther@suse.de>
21684 PR tree-optimization/79389
21685 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21688 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21690 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21691 function comment to reflect reality.
21692 (loop_exits_before_overflow): Fix typo in function description.
21694 2017-02-24 Richard Biener <rguenther@suse.de>
21696 PR tree-optimization/79389
21697 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21698 properly that a threading opportunity exists. Detect conditional
21699 copy/constant propagation opportunities.
21701 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21703 * config/visium/visium.md (type): Add trap.
21704 (b): New mode attribute.
21705 (*btst): Rename into...
21706 (*btst<mode>): ...this and adjust.
21707 (*cbranchsi4_btst_insn): Rename into...
21708 (*cbranch<mode>4_btst_insn): ...this and adjust.
21709 (trap): New define_insn.
21711 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21713 PR tree-optimization/79389
21714 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21715 (noce_reversed_cond_code): New function.
21716 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21717 reversed_comparison_code. Formatting fix.
21718 (noce_try_store_flag): Test rev_cond != NULL in addition to
21719 reversed_comparison_code.
21720 (noce_try_store_flag_constants): Likewise.
21721 (noce_try_store_flag_mask): Likewise.
21722 (noce_try_addcc): Use rev_cond if non-NULL instead of
21723 reversed_comparison_code.
21724 (noce_try_cmove_arith): Likewise. Formatting fixes.
21725 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21726 (noce_find_if_block): Initialize rev_cond.
21727 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21728 instead of false as last argument never attempt to reverse it
21731 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21733 PR tree-optimization/79663
21734 * tree-predcom.c (combine_chains): Process refs in reverse order
21735 only for ZERO length chains, and add explaining comment.
21737 2017-02-23 Jeff Law <law@redhat.com>
21739 PR tree-optimization/79578
21740 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21741 in call to operand_equal_p.
21743 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21746 * config/i386/cpuid.h: Fix another undefined behavior.
21748 2017-02-23 Richard Biener <rguenther@suse.de>
21750 PR tree-optimization/79683
21751 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21752 vector types for data-refs.
21754 2017-02-23 Martin Liska <mliska@suse.cz>
21756 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21758 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21760 PR middle-end/79665
21761 * internal-fn.c (get_range_pos_neg): Moved to ...
21762 * tree.c (get_range_pos_neg): ... here. No longer static.
21763 * tree.h (get_range_pos_neg): New prototype.
21764 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21765 are known to be in between 0 and signed maximum inclusive, try to
21766 expand both unsigned and signed divmod and use the cheaper one from
21769 2017-02-22 Jeff Law <law@redhat.com>
21771 PR tree-optimization/79578
21772 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21773 to compare base operands.
21775 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21778 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21779 gpc_reg_operand instead of fpr_reg_operand.
21781 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21783 * config/mips/mips.c (mips_return_in_memory): Force FP
21784 vector types to be returned in memory for o32 ABI.
21786 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21788 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21789 instead of DW_TAG_member for static data member declarations and don't
21790 set no_linkage_name for static inline data members.
21791 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21794 2017-02-22 Martin Liska <mliska@suse.cz>
21796 * doc/invoke.texi: Replace inequality signs with square brackets
21799 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21801 PR tree-optimization/68644
21802 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21804 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21807 * lra-constraints.c (simplify_operand_subreg): Handle
21808 WORD_REGISTER_OPERATIONS targets.
21810 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21813 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21814 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21815 elimination by swapping fld*.
21817 2017-02-22 Richard Biener <rguenther@suse.de>
21819 PR tree-optimization/79673
21820 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21821 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21822 irrelevant address-space qualifiers and avoiding a
21823 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21825 2017-02-22 Richard Biener <rguenther@suse.de>
21827 PR tree-optimization/79666
21828 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21829 to not symbolically negate if that may introduce undefined
21832 2017-02-22 Martin Liska <mliska@suse.cz>
21835 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21836 * data-streamer-out.c (streamer_write_gcov_count_stream):
21838 * value-prof.c (stream_out_histogram_value): Make assert more
21839 precise based on type of counter.
21841 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21844 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21845 Use nonimmediate_operand instead of memory_operand for operand 1.
21846 (float-extend standard_x87sse_constant_load splitter): Ditto.
21848 2017-02-21 Jeff Law <law@redhat.com>
21850 PR tree-optimization/79621
21851 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21852 blocks with edges to themselves.
21854 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21857 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21858 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21859 Use gimple_call_builtin_p.
21862 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21863 on temporarily removed DEBUG_INSNs.
21865 PR tree-optimization/79649
21866 * tree-loop-distribution.c (classify_partition): Give up on
21867 non-generic address space loads/stores.
21869 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21871 * doc/loop.texi (Loop manipulation): Remove nonexistent
21872 tree_ssa_loop_version from the documentation.
21873 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21875 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21878 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21879 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21880 * config/rs6000/rs6000.c: Include except.h.
21881 (rs6000_expand_split_stack_prologue): Call
21882 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21884 2017-02-21 Martin Jambor <mjambor@suse.cz>
21887 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21888 have been analyzed.
21890 2017-02-21 Martin Jambor <mjambor@suse.cz>
21892 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21893 for backward compatibility only.
21894 * doc/invoke.texi (Option Summary): Remove all references to
21895 -fipa-cp-alignment.
21897 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21901 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21903 * lra-constraints.c (curr_insn_transform): Handle
21904 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21906 2017-02-21 Martin Liska <mliska@suse.cz>
21908 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21910 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21912 PR translation/79638
21913 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21915 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21918 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21919 (arm_function_ok_for_sibcall): Return false for an indirect call by
21920 descriptor if all the argument registers are used.
21921 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21922 alignment of the function.
21924 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21926 PR tree-optimization/61441
21927 * simplify-rtx.c (simplify_const_unary_operation): For
21928 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21929 the sNaN unmodified.
21931 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21933 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21934 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21935 instead of SYSTEM_HEADER_DIR.
21937 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21938 Martin Liška <mliska@suse.cz>
21940 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21941 Fix typos and grammar, use active voice, and clarify.
21943 2017-02-20 Marek Polacek <polacek@redhat.com>
21945 PR middle-end/79537
21946 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21949 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21951 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21954 * config/i386/i386.c (ix86_expand_builtin): Handle
21955 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21956 ix86_builtins_isa[fcode].isa as a requirement of those
21957 flags and any other flag in the bitmask.
21958 (ix86_init_mmx_sse_builtins): Use 0 instead of
21959 ~OPTION_MASK_ISA_64BIT as mask.
21960 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21961 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21962 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21963 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21965 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21968 * lra-constraints.c (split_reg): Check requested split mode
21969 is supported by the register.
21971 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21973 * lra-constraints.c (simplify_operand_subreg): Remove early
21976 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21979 * lra-constraints.c (curr_insn_transform): Tighten condition
21980 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21982 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21985 * lra-constraints.c (curr_insn_transform): Handle
21986 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21988 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21991 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21993 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21995 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21998 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22001 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22003 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22006 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22008 * final.c (last_columnnum, override_columnnum): New variables.
22009 (final_start_function): Set last_columnnum, pass it to begin_prologue
22010 hook and pass 0 to dwarf2out_begin_prologue.
22011 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22012 to source_line debug hook.
22013 (notice_source_line): Compute last_columnnum and for debug_column_info
22014 return true on column changes.
22015 * debug.h (struct gcc_debug_hooks): Add column argument to
22016 source_line and begin_prologue hooks.
22017 (debug_nothing_int_charstar_int_bool): Remove prototype.
22018 (debug_nothing_int_int_charstar,
22019 debug_nothing_int_int_charstar_int_bool): New prototypes.
22020 (dwarf2out_begin_prologue): Add column argument.
22021 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22022 begin_prologue hooks.
22023 (debug_nothing_int_charstar_int_bool): Remove.
22024 (debug_nothing_int_int_charstar,
22025 debug_nothing_int_int_charstar_int_bool): New functions.
22026 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22027 through to dwarf2out_source_line.
22028 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22029 (dwarf2out_source_line): Add column argument, emit it if requested.
22030 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22032 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22033 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22034 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22035 through to dwarf2out_begin_prologue.
22036 (vmsdbgout_source_line): Add column argument, pass it through to
22037 dwarf2out_source_line.
22038 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22039 dbxout_source_line caller.
22040 (dbxout_source_line): Add column argument.
22042 * common.opt (gno-column-info, gcolumn-info): New options.
22043 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22044 (check_die): Also test for multiple DW_AT_decl_column attributes.
22045 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22046 DW_AT_decl_column if requested.
22047 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22049 (gen_variable_die): Likewise.
22050 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22051 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22054 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22055 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22056 (ix86_handle_option): Handle OPT_m3dnowa.
22057 * doc/invoke.texi (-m3dnowa): Document.
22058 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22059 -m3dnowa instead of -m3dnow -march=athlon.
22062 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22063 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22065 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22068 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22069 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22070 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22071 generator for vsx_xxpermdi_<mode>_be.
22072 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22073 force big-endian semantics.
22074 (vsx_xxpermdi_<mode>_be): New define_expand with same
22075 implementation as previous version of vsx_xxpermdi_<mode>.
22077 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22079 PR tree-optimization/79327
22080 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22081 variable, its initialization and use.
22083 2017-02-17 Julia Koval <julia.koval@intel.com>
22085 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22086 (OPTION_MASK_ISA_PKU_UNSET): New.
22087 (ix86_handle_option): Handle -mrdpid.
22088 * config/i386/cpuid.h (bit_RDPID): New.
22089 * config/i386/driver-i386.c (host_detect_local_cpu):
22090 Detect RDPID feature.
22091 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22092 * config/i386/i386-c.c (ix86_target_macros_internal):
22094 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22095 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22096 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22097 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22098 * config/i386/i386.md (define_insn "rdpid"): New.
22099 * config/i386/i386.opt Add -mrdpid.
22100 * config/i386/immintrin.h (_rdpid_u32): New.
22102 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22104 PR rtl-optimization/79541
22105 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22106 instead of transforming it into USE.
22108 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22110 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22111 If HONOR_SNANS (SFmode) force the input to a register.
22112 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22113 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22114 an frsp or similar insn.
22116 2017-02-17 Martin Liska <mliska@suse.cz>
22118 PR rtl-optimization/79577
22119 * params.def (selsched-max-sched-times): Increase minimum to 1.
22121 2017-02-17 Martin Liska <mliska@suse.cz>
22123 PR rtl-optimization/79574
22124 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22126 2017-02-17 Martin Liska <mliska@suse.cz>
22128 PR tree-optimization/79529
22129 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22130 ssa_defined_default_def_p to handle cases which are implicitly
22132 * tree-ssa.c (ssa_defined_default_def_p): New function.
22133 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22134 which are implicitly defined.
22135 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22137 2017-02-17 Richard Biener <rguenther@suse.de>
22139 PR middle-end/79576
22140 * params.def (max-ssa-name-query-depth): Limit to 10.
22142 2017-02-17 Richard Biener <rguenther@suse.de>
22144 PR tree-optimization/79552
22145 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22148 2017-02-17 Richard Biener <rguenther@suse.de>
22151 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22153 2017-02-17 Marek Polacek <polacek@redhat.com>
22155 PR middle-end/79536
22156 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22157 (fold_negate_expr): New wrapper.
22159 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22161 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22162 Correct terminology and de-emphasize pre-standard behavior.
22164 2017-02-16 Alan Modra <amodra@gmail.com>
22166 PR rtl-optimization/79286
22167 * ira.c (def_dominates_uses): New function.
22168 (update_equiv_regs): Don't create an equivalence for insns that
22169 may trap where the register def does not dominate the use.
22171 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22173 PR rtl-optimization/78127
22174 * lra.c (lra): Call lra_eliminate before finish the loop after
22177 2017-02-16 Richard Biener <rguenther@suse.de>
22179 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22181 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22182 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22183 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22184 (isl_val_int_from_wi): New function.
22185 (extract_affine_gmp): Rename to ...
22186 (extract_affine_wi): ... this, take a widest_int.
22187 (extract_affine_int): Just wrap extract_affine_wi.
22188 (add_param_constraints): Use isl_val_int_from_wi.
22189 (add_loop_constraints): Likewise, and extract_affine_wi.
22191 2017-02-15 Jeff Law <law@redhat.com>
22193 PR middle-end/79521
22194 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22195 ira_init_register_move_cost_if_necessary.
22197 2017-02-15 Martin Sebor <msebor@redhat.com>
22199 PR middle-end/32003
22200 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22201 removed in a prior commit.
22203 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22205 PR tree-optimization/79347
22206 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22207 counters during peeling.
22209 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22211 * Makefile.in (site.exp): Remove "set ISLVER".
22213 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22216 * real.c (real_from_integer): Call real_convert even for decimal.
22218 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22221 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22223 2017-02-14 Andrew Pinski <apinski@cavium.com>
22225 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22226 cores and change the partno/implementer to be correct.
22227 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22228 the 'B" as the implementer.
22229 * config/aarch64/aarch64-tune.md: Regenerate.
22231 2017-02-14 Carl Love <cel@us.ibm.com>
22233 * config/rs6000/rs6000.c: Add case statement entry to make the
22234 xvcvuxdsp built-in argument unsigned.
22235 * config/rs6000/vsx.md: Fix the source and return operand types so they
22236 match the instruction definitions from the ISA document. Fix typo
22237 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22240 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22243 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22244 member early_clobber_alts.
22245 * lra-lives.c (reg_early_clobber_p): New.
22246 (process_bb_lives): Use it.
22247 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22248 (debug_operand_data): Initialize early_clobber_alts.
22249 (setup_operand_alternative): Set up early_clobber_alts.
22250 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22251 alternatives to new_insn_reg.
22252 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22254 (lra_update_insn_regno_info): Pass the new arg.
22256 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22258 PR middle-end/79505
22259 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22260 (new_oacc_loop_raw): Don't clear already cleared fields.
22263 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22264 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22265 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22267 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22270 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22272 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22275 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22276 the extra instruction to the right place to store 128-bit constant
22279 2017-02-14 Martin Sebor <msebor@redhat.com>
22281 PR middle-end/79448
22282 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22283 warning for strings of unknown length.
22285 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22287 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22289 2017-02-14 Jeff Law <law@redhat.com>
22292 * ira-costs.c (scan_one_insn): Initialize register move costs
22293 for pseudos seen in USE/CLOBBER insns.
22295 PR tree-optimization/79095
22296 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22297 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22298 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22299 if the operands are known to be not equal, then the resulting range
22301 (intersect_ranges): If the new range is ~[0,0] and the old range is
22302 wide, then prefer ~[0,0].
22303 * tree-vrp.c (overflow_comparison_p_1): New function.
22304 (overflow_comparison_p): New function.
22305 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22306 if NAME is used in an overflow test.
22307 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22308 overflow check that can be expressed as an equality test, then adjust
22309 ops to be that equality test.
22311 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22313 * config/s390/s390-builtin-types.def: Remove flags argument.
22314 * config/s390/s390.c (s390_init_builtins): Likewise.
22316 2017-02-14 Martin Liska <mliska@suse.cz>
22318 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22319 vector. Fix trailing white spaces.
22321 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22323 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22326 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22328 PR rtl-optimization/68664
22329 * config/arm/arm.c (arm_sched_can_speculate_insn):
22330 New function. Declare prototype.
22331 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22333 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22335 PR rtl-optimization/68664
22336 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22338 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22340 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22342 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22343 max skip bytes for function, loop and jump.
22345 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22347 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22348 ABS_EXPR for gimple dump.
22350 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22353 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22355 PR tree-optimization/79408
22356 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22357 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22358 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22359 also if rhs1 is INTEGER_CST.
22361 2017-02-14 Richard Biener <rguenther@suse.de>
22363 PR middle-end/79432
22364 * tree-into-ssa.c (insert_phi_nodes): When the function can
22365 have abnormal edges rewrite SSA names with broken use-def
22366 dominance out of SSA and register them for PHI insertion.
22368 2017-02-13 Martin Sebor <msebor@redhat.com>
22370 PR middle-end/79496
22371 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22372 clearing info.nowrite flag when snprintf size argument is a range.
22374 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22376 * cprop.c (cprop_jump): Add missing space in string literal.
22377 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22378 (get_constraint_for_component_ref): Likewise.
22379 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22380 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22381 * lra-constraints.c (process_alt_operands): Likewise.
22382 * ipa-inline.c (inline_small_functions): Likewise.
22383 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22384 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22385 * trans-mem.c (diagnose_tm_1_op): Likewise.
22386 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22387 (grid_parallel_clauses_gridifiable): Likewise.
22389 * config/nvptx/mkoffload.c (process): Add space in between
22392 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22393 "MOD4_SSE_REGS" and "ALL_REGS".
22395 * spellcheck.c (test_data): Add , in between "foo" and "food".
22397 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22400 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22401 boundary crossing check and subsequent code generation agree.
22403 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22405 * config/aarch64/aarch64.c (has_memory_op): Delete.
22406 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22409 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22411 PR rtl-optimization/79388
22412 PR rtl-optimization/79450
22413 * combine.c (distribute_notes): When removing TEM_INSN for which
22414 corresponding dest has last value recorded, invalidate that last
22417 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22419 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22420 of explicit '@'. Add missing assembly comment marker on branch costs
22423 2017-02-13 Nathan Sidwell <nathan@acm.org>
22425 * gengtype-lex.l (<in_struct>): Add '/'.
22427 2017-02-13 Martin Liska <mliska@suse.cz>
22430 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22432 2017-02-13 Richard Biener <rguenther@suse.de>
22434 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22436 * configure: Re-generate.
22437 * config.in: Likewise.
22438 * graphite-dependences.c: Simplify as if
22439 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22440 * graphite-isl-ast-to-gimple.c: Likewise.
22441 * graphite-optimize-isl.c: Likewise.
22442 * graphite-poly.c: Likewise.
22443 * graphite-sese-to-poly.c: Likewise.
22444 * graphite.h: Likewise.
22445 * toplev.c: Include isl/version.h and use isl_version () for
22446 printing the ISL version.
22447 * doc/install.texi: Update ISL requirement.
22449 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22451 * doc/standards.texi (Standards): Update reference to
22454 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22456 * doc/extend.texi (Named Address Spaces): sourceware.org now
22458 * doc/install.texi (Binaries): Ditto.
22461 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22463 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22464 standard terminology "stringize"/"stringizing" throughout.
22465 * doc/cppinternals.texi: Likewise.
22467 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22469 * doc/extend.texi: Fix some spelling mistakes and typos.
22470 * doc/invoke.texi: Likewise.
22472 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22475 * params.def (inline-min-speedup) Change from 10 to 8.
22477 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22479 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22482 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22485 * ipa-inline-analysis.c (get_minimal_bb): New function.
22486 (record_modified): Use it.
22487 (remap_edge_change_prob): Handle also ancestor functions.
22489 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22491 * doc/contrib.texi (Contributors): Remove broken link into
22492 the Mauve CVS repository.
22494 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22496 PR middle-end/79454
22497 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22498 result computation whenever lhs doesn't have vector mode, not
22499 just when it has BLKmode.
22501 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22503 * doc/makefile.texi (profiledbootstrap): Refer to the
22504 installation instructions only in textual form.
22506 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22509 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22511 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22513 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22514 (Specific): Update mingw-w64 reference.
22516 (Specific): Remove broken link to Renesas RX processor.
22518 2017-02-10 Richard Biener <rguenther@suse.de>
22520 * toplev.c (process_options): Do not mention obsolete graphite
22521 options when printing sorry message about missing graphite support.
22522 Mention -floop-nest-optimize.
22524 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22526 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22527 (vtst_p16): Likewise.
22528 (vtstq_p8): Likewise.
22529 (vtstq_p16): Likewise.
22531 (vtstq_p64): Likewise.
22532 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22533 (vset_lane_p64): New.
22534 (vsetq_lane_p64): New.
22536 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22538 PR tree-optimization/79411
22539 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22540 stmt operands are SSA_NAMEs used in abnormal phis.
22541 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22544 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22547 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22550 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22552 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22554 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22556 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22560 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22561 not arbitrary TREE_CONSTANT.
22564 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22565 "omp declare target link" attribute unless is_global_var.
22566 * omp-offload.c (find_link_var_op): Likewise.
22568 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22569 Chung-Lin Tang <cltang@codesourcery.com>
22571 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22573 (gimplify_adjust_omp_clauses): Don't delete TILE.
22574 (gimplify_omp_for): Deal with TILE.
22575 * internal-fn.c (expand_GOACC_TILE): New function.
22576 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22578 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22579 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22581 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22582 avoid DIV for outermost collapse var.
22583 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22584 Remove out of date comments, fix whitespace.
22585 * omp-general.c (omp_extract_for_data): Deal with tiling.
22586 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22587 adjust OLF_DIM_BASE value.
22588 (struct omp_for_data): Add tiling field.
22589 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22590 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22591 for auto loops. Remove default auto determining, moved to
22592 oacc_loop_fixed_partitions.
22593 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22594 stmts, add e_mask field.
22595 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22596 (oacc_thread_numbers): Use oacc_dim_call.
22597 (oacc_xform_tile): New.
22598 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22599 (finish_oacc_loop): Adjust for ifns vector.
22600 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22601 add case for GOACC_TILE fns.
22602 (oacc_loop_xform_loop): Delete.
22603 (oacc_loop_process): Iterate over call list directly, and add
22604 handling for GOACC_TILE fns.
22605 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22607 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22608 vector partitioning to outer loops. Assign 2 partitions to loops
22609 when available. Add TILE handling.
22610 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22611 (execite_oacc_device_lower): Process GOACC_TILE fns,
22612 ignore unknown specs.
22613 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22614 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22615 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22617 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22619 * configure.ac (ACX_BUGURL): Update.
22620 * configure: Regenerate.
22622 2017-02-09 Richard Biener <rguenther@suse.de>
22624 PR tree-optimization/69823
22625 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22626 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22628 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22630 * config/arc/arc-c.def: Add __NPS400__ definition.
22631 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22632 (TARGET_NPS400): Define.
22634 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22636 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22638 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22639 pointer, arch_info.
22640 (arc_cpu_types): Fill the arch_info field with a pointer into the
22641 arc_arch_types table.
22642 (arc_selected_cpu): Declare.
22643 * config/arc/arc.c (arc_selected_cpu): Make global.
22644 (arc_selected_arch): Delete.
22645 (arc_base_cpu): Delete.
22646 (arc_override_options): Remove references to deleted variables,
22647 update access to arch information.
22648 (ARC_OPT): Update access to arch information.
22649 (ARC_OPTX): Likewise.
22650 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22651 (TARGET_ARC600): Update access to arch information.
22652 (TARGET_ARC601): Likewise.
22653 (TARGET_ARC700): Likewise.
22654 (TARGET_EM): Likewise.
22655 (TARGET_HS): Likewise.
22656 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22659 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22662 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22663 condition/operands for integer GE/LE/GEU/LEU operations.
22665 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22667 PR translation/79397
22668 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22671 2017-02-08 Martin Jambor <mjambor@suse.cz>
22674 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22675 whether allocation happened.
22676 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22677 nothing was allocated.
22679 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22681 PR tree-optimization/79408
22682 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22683 constant, but SSA_NAME with a known integer range, use the minimum
22684 of that range instead of op1 to determine if modulo can be replaced
22685 with its first operand.
22687 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22689 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22691 2017-02-08 Richard Biener <rguenther@suse.de>
22693 PR tree-optimization/71824
22694 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22695 Check all loops contained in the merged region.
22697 2017-02-07 Andrew Pinski <apinski@cavium.com>
22699 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22701 2017-02-07 Andrew Pinski <apinski@cavium.com>
22703 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22704 (thunderxt88): Likewise.
22705 (thunderxt81): Disable LSE and change v8.1 to v8.
22706 (thunderxt83): Likewise.
22708 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22709 Richard Biener <rguenther@suse.de>
22711 PR middle-end/79399
22712 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22713 type from int to size_t.
22714 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22716 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22718 PR rtl-optimization/79386
22719 * cprop.c (bypass_conditional_jumps): Initialize
22720 bypass_last_basic_block already before splitting bbs after
22721 unconditional traps...
22722 (bypass_conditional_jumps): ... rather than here.
22725 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22726 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22727 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22728 fix -masm=intel patterns.
22730 2017-02-07 Richard Biener <rguenther@suse.de>
22732 PR tree-optimization/79256
22733 PR middle-end/79278
22734 * builtins.c (get_object_alignment_2): Use min_align_of_type
22735 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22736 and ADJUST_FIELD_ALIGN.
22738 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22740 * doc/tm.texi: Regenerate.
22741 * stor-layout.c (layout_decl): Adjust.
22742 (update_alignment_for_field): Likewise.
22743 (place_field): Likewise.
22744 (min_align_of_type): Likewise.
22745 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22746 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22747 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22748 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22749 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22750 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22751 * config/i386/i386.c (x86_field_alignment): Likewise.
22752 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22753 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22754 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22755 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22756 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22757 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22761 2017-01-30 Richard Biener <rguenther@suse.de>
22763 PR tree-optimization/79256
22764 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22765 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22768 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22770 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22771 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22772 builtins to SImode and emit a zero-extend, if necessary.
22774 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22776 * docs/invoke.texi (RISC-V Options): Alphabetize.
22778 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22780 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22783 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22785 * config/riscv/riscv.c: New file.
22786 * common/config/riscv/riscv-common.c: Likewise.
22787 * config.gcc: Likewise.
22788 * config/riscv/constraints.md: Likewise.
22789 * config/riscv/elf.h: Likewise.
22790 * config/riscv/generic.md: Likewise.
22791 * config/riscv/linux.h: Likewise.
22792 * config/riscv/multilib-generator: Likewise.
22793 * config/riscv/peephole.md: Likewise.
22794 * config/riscv/pic.md: Likewise.
22795 * config/riscv/predicates.md: Likewise.
22796 * config/riscv/riscv-builtins.c: Likewise.
22797 * config/riscv/riscv-c.c: Likewise.
22798 * config/riscv/riscv-ftypes.def: Likewise.
22799 * config/riscv/riscv-modes.def: Likewise.
22800 * config/riscv/riscv-opts.h: Likewise.
22801 * config/riscv/riscv-protos.h: Likewise.
22802 * config/riscv/riscv.h: Likewise.
22803 * config/riscv/riscv.md: Likewise.
22804 * config/riscv/riscv.opt: Likewise.
22805 * config/riscv/sync.md: Likewise.
22806 * config/riscv/t-elf-multilib: Likewise.
22807 * config/riscv/t-linux: Likewise.
22808 * config/riscv/t-linux-multilib: Likewise.
22809 * config/riscv/t-riscv: Likewise.
22810 * configure.ac: Likewise.
22811 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22812 Waterman as RISC-V maintainers.
22813 * doc/install.texi: Add RISC-V entries.
22814 * doc/invoke.texi: Add RISC-V options section.
22815 * doc/md.texi: Add RISC-V constraints section.
22816 * configure: Regenerated.
22818 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22821 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22822 false values to be constant vectors with all 0 or all 1 bits set.
22823 (vcondu<mode><mode>): Likewise.
22824 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22826 (fpmask_comparison_operator): Update comment.
22827 (vecint_comparison_operator): New predicate.
22828 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22829 vector conditionals when the true and false values are constant
22830 vectors with all 0 bits or all 1 bits set.
22832 2017-02-06 Martin Sebor <msebor@redhat.com>
22834 PR tree-optimization/79376
22835 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22837 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22839 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22840 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22841 to simplify split condition.
22843 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22845 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22846 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22849 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22851 PR rtl-optimization/68664
22852 * target.def (can_speculate_insn): New hook.
22853 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22854 * doc/tm.texi: Regenerate.
22855 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22856 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22857 (rs6000_sched_can_speculate_insn): New function.
22859 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22861 PR tree-optimization/79284
22862 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22863 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22864 vectorizable_mask_load_store, vectorizable_operation,
22865 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22866 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22867 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22868 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22869 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22870 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22871 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22872 is_gimple_assign (stmt). Replace another such test with
22873 is_gimple_assign (stmt).
22875 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22878 * config/avr/avr.c (rtl-iter.h): Include it.
22879 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22880 (avr_legitimate_combined_insn): ...and implementation.
22882 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22884 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22885 * config/s390/s390.c (s390_const_operand_ok)
22886 (s390_canonicalize_comparison, s390_extract_part)
22887 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22888 (s390_contiguous_bitmask_p, s390_rtx_costs)
22889 (legitimize_pic_address): Likewise.
22890 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22891 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22892 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22893 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22894 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22896 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22898 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22899 REGNO($0) == REGNO($1).
22901 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22903 * config/s390/linux.h(SIZE_TYPE): Add comment.
22905 2017-02-06 Julian Brown <julian@codesourcery.com>
22906 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22907 Virendra Pathak <virendra.pathak@broadcom.com>
22909 * config/aarch64/aarch64-cores.def: Change the scheduler
22911 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22912 * config/aarch64/thunderx2t99.md: New file.
22914 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22916 * doc/standards.texi (Go Language): Update link to language
22919 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22921 * tree-eh.c (lower_resx): Sanitize profile.
22922 (cleanup_empty_eh_move_lp): Likewise.
22924 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22927 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22929 * cfgloopmanip.h (loop_version): Update prototype.
22930 * modulo-sched.c (sms_schedule): Update call of loop_version.
22931 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22932 * tree-parloops.c (gen_parallel_loop): Likewise.
22933 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22934 * tree-ssa-loop-split.c (split_loop): Likewise.
22935 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22936 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22938 2017-02-05 Martin Liska <mliska@suse.cz>
22941 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22943 (print_operand_address): Initialize a struct to zero.
22945 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22947 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22948 garbage collector only in textual form.
22950 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22952 * doc/extend.texi (x86 specific memory model extensions for
22953 transactional memory): Simplify a phrase.
22955 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22958 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22959 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22960 (atomic_storedi_1): Likewise.
22962 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22964 PR tree-optimization/79338
22965 * tree-parloops.c (gather_scalar_reductions): Don't call
22966 vect_analyze_loop_form for loop->inner before destroying loop's
22969 2017-02-03 Martin Sebor <msebor@redhat.com>
22971 PR tree-optimization/79327
22972 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22973 when precision has resulted in leading zeros.
22974 (format_integer): Adjust the likely counter to assume an unknown
22975 argument that may be zero is non-zero.
22977 2017-02-03 Jason Merrill <jason@redhat.com>
22980 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22981 avoid copying non-taken branch.
22983 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22985 PR tree-optimization/79340
22986 * tree-vect-loop.c (vectorizable_reduction): Release
22987 vec_defs elements after safe_splicing them into other vectors.
22990 PR tree-optimization/79327
22991 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22992 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22994 (format_integer): Use wide_int_to_tree instead of build_int_cst
22995 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22996 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22997 of shortest and longest sequence.
22999 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23001 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23002 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23004 2017-02-03 Walter Lee <walt@tilera.com>
23007 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23008 after initial stackframe link reg save.
23009 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23011 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23014 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23015 wu for stxssp alternative.
23017 2017-02-03 Martin Sebor <msebor@redhat.com>
23019 PR tree-optimization/79352
23020 * gimple-fold.c (get_range_strlen): Add argument.
23021 (get_range_strlen): Change return type to bool.
23022 (get_maxval_strlen): Pass in a dummy argument.
23023 * gimple-fold.h (get_range_strlen): Change return type to bool.
23024 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23025 * tree.h (array_at_struct_end_p): Add argument.
23026 * tree.c (array_at_struct_end_p): Handle it.
23028 2017-02-03 Martin Liska <mliska@suse.cz>
23031 * multiple_target.c (create_dispatcher_calls): Redirect edge
23032 from a caller of a dispatcher.
23033 (expand_target_clones): Make the clones local.
23034 (ipa_target_clone): Do both target clones and resolvers.
23035 (ipa_dispatcher_calls): Remove the pass.
23036 (pass_dispatcher_calls::gate): Likewise.
23037 (make_pass_dispatcher_calls): Likewise.
23038 * passes.def (pass_target_clone): Put as very first IPA early
23041 2017-02-03 Martin Liska <mliska@suse.cz>
23043 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23044 in case of a function with ifunc attribute.
23046 2017-02-03 Martin Liska <mliska@suse.cz>
23048 * cgraph.c (cgraph_node::dump): Dump function version info.
23049 * symtab.c (symtab_node::dump_base): Add missing new line.
23051 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23053 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23054 (ifcombine_ifandif): Use it.
23056 2017-02-03 Martin Liska <mliska@suse.cz>
23058 * doc/invoke.texi: Document default value for
23059 use-after-scope-direct-emission-threshold.
23061 2017-02-03 Martin Liska <mliska@suse.cz>
23063 PR tree-optimization/79339
23064 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23065 (format_floating): Likewise.
23067 2017-02-03 Martin Liska <mliska@suse.cz>
23070 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23072 (ipa_node_params_t::remove): Likewise.
23073 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23074 initialization from removed ipa_node_params_t::insert.
23075 (ipa_node_params::~ipa_node_params): Move from removed
23076 ipa_node_params_t::release.
23077 * symbol-summary.h (symbol_summary::m_released): New member.
23078 Do not release a summary twice. Do not allow to call finalizer
23079 for types of a summary that live in GGC memory.
23081 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23083 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23086 2017-02-02 Martin Sebor <msebor@redhat.com>
23088 PR middle-end/79275
23089 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23090 (format_string): Tighten up the range of output for non-constant
23091 strings and correct the expected range for wide non-constant strings.
23093 2017-02-02 Martin Sebor <msebor@redhat.com>
23095 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23097 PR middle-end/32003
23098 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23100 (-fdump-tree-@var): Add to index and document how to come up
23101 with pass-specific option and dump file names.
23102 (-fdump-passes): Clarify where to look for output.
23104 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23106 PR middle-end/77445
23107 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23108 statistics of the analyzed path; allow threading for speed when
23109 any of BBs along the path are optimized for speed.
23111 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23113 PR middle-end/78468
23114 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23115 settings of the virtual registers.
23118 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23120 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23121 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23124 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23126 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23127 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23129 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23131 * config/s390/s390.md: Add missing comments with the expanded
23133 * config/s390/vector.md: Likewise.
23134 * config/s390/vx-builtins.md: Likewise.
23136 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23139 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23140 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23141 conditions on a single line.
23143 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23145 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23146 __S390_VX__ to __VX__.
23148 2017-02-01 Andrew Pinski <apinski@cavium.com>
23150 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23151 stmt_info to record_stmt_cost.
23152 (vect_get_known_peeling_cost): Pass stmt_info if known to
23154 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23155 cpu_vector_cost field into
23156 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23157 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23158 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23159 splitting of scalar_stmt_cost and vec_stmt_cost.
23160 (thunderx_vector_cost): Likewise.
23161 (cortexa57_vector_cost): LIkewise.
23162 (exynosm1_vector_cost): Likewise.
23163 (xgene1_vector_cost): Likewise.
23164 (thunderx2t99_vector_cost): Improve after the splitting of the two
23166 (aarch64_builtin_vectorization_cost): Update for the splitting of
23167 scalar_stmt_cost and vec_stmt_cost.
23169 2017-02-01 Torvald Riegel <triegel@redhat.com>
23170 Richard Henderson <rth@redhat.com>
23172 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23173 conditional on existance of a fast atomic load.
23174 * optabs-query.c (can_atomic_load_p): New function.
23175 * optabs-query.h (can_atomic_load_p): Declare it.
23176 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23177 no fast atomic load is available for the particular size of access.
23178 (expand_atomic_compare_and_swap): Likewise.
23179 (expand_atomic_load): Likewise.
23180 (expand_atomic_store): Likewise.
23181 (expand_atomic_fetch_op): Likewise.
23182 * testsuite/lib/target-supports.exp
23183 (check_effective_target_sync_int_128): Remove x86 because it provides
23184 no fast atomic load.
23185 (check_effective_target_sync_int_128_runtime): Likewise.
23187 2017-02-01 Richard Biener <rguenther@suse.de>
23189 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23190 (graphite_transform_loops): Provide opt-info for optimized nests.
23191 * tree-parloop.c (parallelize_loops): Provide opt-info for
23192 parallelized loops.
23194 2017-02-01 Richard Biener <rguenther@suse.de>
23196 PR middle-end/79315
23197 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23198 was not set before.
23200 2017-02-01 Richard Biener <rguenther@suse.de>
23202 PR tree-optimization/71824
23203 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23204 Verify the loops are valid in the merged SESE region.
23205 (scop_detection::can_represent_loop_1): Check analyzing the
23206 evolution of the number of iterations in the region succeeds.
23208 2017-01-31 Ian Lance Taylor <iant@golang.org>
23210 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23211 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23213 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23215 PR preprocessor/79210
23216 * input.c (get_substring_ranges_for_loc): Replace line_width
23217 assertion with error-handling.
23219 2017-01-31 Richard Biener <rguenther@suse.de>
23221 PR tree-optimization/77318
23222 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23223 (create_pw_aff_from_tree): Take loop parameter.
23224 (add_condition_to_pbb): Pass loop of the condition to
23225 create_pw_aff_from_tree.
23227 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23229 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23230 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23232 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23236 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23238 (convert_int_to_float128): Likewise.
23239 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23240 (convert_int_to_float128): Likewise.
23241 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23242 (UNSPEC_IEEE128_CONVERT): Likewise.
23243 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23244 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23245 Use local variables for IBM extended format.
23246 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23247 (fix_trunc<mode>si2_fprs): Likewise.
23248 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23249 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23250 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23251 to know that we can now have integers of all sizes in vector
23253 (fix<uns>_<mode>di2_hw): Likewise.
23254 (float<uns>_<mode>si2_hw): Likewise.
23255 (fix_<mode>si2_hw): Likewise.
23256 (fixuns_<mode>si2_hw): Likewise.
23257 (float<uns>_<mode>di2_hw): Likewise.
23258 (float_<mode>di2_hw): Likewise.
23259 (float_<mode>si2_hw): Likewise.
23260 (floatuns_<mode>di2_hw): Likewise.
23261 (floatuns_<mode>si2_hw): Likewise.
23262 (xscvqp<su>wz_<mode>): Delete, no longer used.
23263 (xscvqp<su>dz_<mode>): Likewise.
23264 (xscv<su>dqp_<mode>): Likewise.
23265 (ieee128_mfvsrd_64bit): Likewise.
23266 (ieee128_mfvsrd_32bit): Likewise.
23267 (ieee128_mfvsrwz): Likewise.
23268 (ieee128_mtvsrw): Likewise.
23269 (ieee128_mtvsrd_64bit): Likewise.
23270 (ieee128_mtvsrd_32bit): Likewise.
23272 2017-01-31 Martin Liska <mliska@suse.cz>
23275 * ipa-prop.c (ipa_free_all_node_params): Call release method
23276 instead of ~sumbol_summary to not to trigger double times
23279 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23281 PR tree-optimization/71691
23282 * bitmap.h (class auto_bitmap): New.
23283 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23284 is_maybe_undefined instead of ssa_undefined_value_p.
23286 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23288 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23289 __S390_ARCH_LEVEL__ to __ARCH__.
23291 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23293 PR tree-optimization/79267
23294 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23295 if should_remove_lhs_p is true.
23297 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23300 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23301 (add_alignment_attribute): New.
23302 (base_type_die): Add alignment attribute.
23303 (subrange_type_die): Likewise.
23304 (modified_type_die): Likewise.
23305 (gen_array_type_die): Likewise.
23306 (gen_descr_array_type_die: Likewise.
23307 (gen_enumeration_type_die): Likewise.
23308 (gen_subprogram_die): Likewise.
23309 (gen_variable_die): Likewise.
23310 (gen_field_die): Likewise.
23311 (gen_ptr_to_mbr_type_die): Likewise.
23312 (gen_struct_or_union_type_die): Likewise.
23313 (gen_subroutine_type_die): Likewise.
23314 (gen_typedef_die): Likewise.
23315 (base_type_cmp): Compare alignment attribute.
23317 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23320 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23321 (setb_unsigned) New pattern for setb with CCUNS.
23322 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23323 subfc./subfe sequence to avoid overflow problems. Generate a
23324 shorter sequence with cmpld/setb for power9.
23325 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23326 for generating subfc. instruction.
23327 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23328 now uses this instruction.
23330 2017-01-30 Ian Lance Taylor <iant@google.com>
23333 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23334 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23336 2017-01-30 Martin Sebor <msebor@redhat.com>
23338 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23339 Move constant to the right of a relational operator.
23340 (get_mpfr_format_length, format_character, format_string): Ditto.
23341 (should_warn_p, maybe_warn): Same.
23343 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23345 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23348 * asan.c (get_translation_unit_decl): Remove function.
23349 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23351 2017-01-30 Martin Liska <mliska@suse.cz>
23353 PR gcov-profile/79259
23354 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23355 -fprofile-generate.
23357 2017-01-30 Martin Liska <mliska@suse.cz>
23360 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23361 Initialize variables with NULL value.
23363 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23366 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23368 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23370 2017-01-30 Richard Biener <rguenther@suse.de>
23372 PR tree-optimization/79276
23373 * tree-vrp.c (process_assert_insertions): Properly adjust common
23374 when removing a duplicate.
23376 * gcc.dg/torture/pr79276.c: New testcase.
23378 2017-01-30 Richard Biener <rguenther@suse.de>
23380 PR tree-optimization/79256
23381 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23382 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23384 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23386 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23389 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23390 ("*r<noxa>sbg_<mode>_sll_bitmask")
23391 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23392 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23393 Use contiguous_bitmask_nowrap_operand.
23395 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23398 * config/rs6000/altivec.h (vec_xl): Revise #define.
23399 (vec_xst): Likewise.
23401 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23403 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23405 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23407 PR rtl-optimization/79194
23408 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23409 traps before call to bypass_conditional_jumps.
23411 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23413 PR tree-optimization/71374
23414 * lra-constraints.c (check_conflict_input_operands): New.
23415 (match_reload): Use it.
23417 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23420 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23421 account to calculate conflict_set.
23423 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23425 PR rtl-optimization/78559
23426 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23427 other_insn in combine.
23429 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23431 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23432 uint16_type_node for BT_UINT16.
23434 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23436 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23437 "RTL Tests" to menu.
23438 (GIMPLE Tests): New node.
23439 (RTL Tests): New node.
23441 2017-01-27 Richard Biener <rguenther@suse.de>
23443 PR tree-optimization/79245
23444 * tree-loop-distribution.c (distribute_loop): Apply cost
23445 modeling also to detected patterns.
23447 2017-01-27 Richard Biener <rguenther@suse.de>
23449 PR tree-optimization/71433
23450 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23451 (compare_assert_loc): New function.
23452 (process_assert_insertions): Sort and optimize assert locations
23453 to remove duplicates and push down identical assertions on
23454 edges to their destination block.
23456 2017-01-27 Richard Biener <rguenther@suse.de>
23458 PR tree-optimization/79244
23459 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23460 out SSA names even if abnormal.
23462 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23464 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23465 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23466 instead of MPFR_RNDN.
23468 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23471 * arm.c (arm_option_override): Don't call build_target_option_node
23472 until after doing all option overrides.
23473 (arm_valid_target_attribute_tree): Likewise.
23475 2017-01-27 Martin Liska <mliska@suse.cz>
23477 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23478 for {cd}tors and C++ {cd}tors.
23480 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23482 * config/s390/s390.md ("*setmem_long_and")
23483 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23485 2017-01-26 Martin Sebor <msebor@redhat.com>
23487 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23490 2017-01-26 Martin Sebor <msebor@redhat.com>
23492 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23493 HAVE_DFmode before using XFmode or DFmode.
23494 (parse_directive): Avoid using the z length modifier to avoid
23495 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23497 PR middle-end/78703
23498 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23499 to accept adjustment as an array.
23500 (get_int_range): New function.
23501 (struct directive): Make width and prec arrays.
23502 (directive::set_width, directive::set_precision): Call get_int_range.
23503 (format_integer, format_floating): Handle width and precision ranges.
23504 (format_string, parse_directive): Same.
23506 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23509 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23510 comdat_type_p set, just clone them, but keep the children in the
23514 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23515 which have direct callers with -fvar-tracking-assignments enabled
23517 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23518 inside of type units.
23520 2017-01-26 Martin Sebor <msebor@redhat.com>
23522 PR middle-end/78703
23523 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23525 (struct format_result): Replace number_chars, number_chars_min,
23526 and number_chars_max with a single member of struct result_range.
23528 (format_result::operator+=): Adjust.
23529 (struct fmtresult): Remove bounded. Handle likely and unlikely
23531 (fmtresult::adjust_for_width_or_precision): New function.
23532 (fmtresult:type_max_digits): New function.
23533 (bytes_remaining): Handle likely and unlikely counters.
23534 (min_bytes_remaining): Remove.
23535 (format_percent): Simplify.
23536 (format_integer, format_floating): Set likely and unlikely counters.
23537 (get_string_length, format_character, format_string): Same.
23538 (format_plain, should_warn_p): New function.
23539 (maybe_warn): Call should_warn_p. Update diagnostic messages
23540 and handle those for all directives, including plain strings.
23541 (format_directive): Handle likely and unlikely counters.
23542 Remove unnecessary quoting from diagnostics. Add an informational
23544 (add_bytes): Remove.
23545 (pass_sprintf_length::compute_format_length): Simplify.
23546 (try_substitute_return_value): Handle likely and unlikely counters.
23548 2017-01-26 Carl Love <cel@us.ibm.com>
23550 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23551 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23553 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23556 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23557 endianess for subregs into account.
23558 * lra-constraints.c (lra_constraints): Do risky transformations
23559 always on the first iteration.
23560 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23561 last_call_used_reg_set.
23562 (process_bb_lives): Define and use last_call_used_reg_set.
23563 * lra.c (lra): Always continue after lra_constraints on the first
23566 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23568 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23570 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23572 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23574 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23575 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23576 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23577 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23578 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23579 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23580 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23581 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23582 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23584 2017-01-26 Marek Polacek <polacek@redhat.com>
23587 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23588 for the third operand.
23590 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23592 PR middle-end/79236
23593 * omp-low.c (struct omp_context): Add simt_stmt field.
23594 (scan_omp_for): Return omp_context *.
23595 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23596 context to the _simt_ SIMD stmt.
23597 (lower_omp_for): For combined SIMD with sibling _simt_
23598 SIMD, make sure to use the same decls in _looptemp_
23599 clauses as in the sibling.
23601 2017-01-26 David Sherwood <david.sherwood@arm.com>
23603 PR middle-end/79212
23604 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23607 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23610 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23611 emit fld b; fld a; if possible.
23613 * brig-builtins.def: Update copyright years.
23614 * config/arm/arm_acle_builtins.def: Update copyright years.
23616 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23619 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23620 constraint instead of o for the stxsd instruction.
23622 2017-01-25 Carl Love <cel@us.ibm.com>
23624 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23625 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23627 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23629 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23631 2017-01-25 Richard Biener <rguenther@suse.de>
23633 PR tree-optimization/69264
23634 * target.def (vector_alignment_reachable): Improve documentation.
23635 * doc/tm.texi: Regenerate.
23636 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23638 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23639 earlier changes with respect to TYPE_USER_ALIGN.
23640 (vector_alignment_reachable_p): Likewise. Improve dumping.
23642 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23645 * config/arm/arm.md (xordi3): Force constant operand into a register
23648 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23650 * doc/invoke.texi (-fstore-merging): Correct default optimization
23651 levels at which it is enabled.
23652 (-O): Move -fstore-merging from list to...
23655 2017-01-25 Richard Biener <rguenther@suse.de>
23658 * omp-expand.c: Include debug.h.
23659 (expand_omp_taskreg): Make sure to generate early debug before
23660 outlining anything from a function.
23661 (expand_omp_target): Likewise.
23662 (grid_expand_target_grid_body): Likewise.
23664 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23667 * asan.c (get_translation_unit_decl): New function.
23668 (asan_add_global): Extract modules file name from globals
23669 TRANSLATION_UNIT_DECL name.
23671 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23674 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23675 for long calls with APCS frame and VFP.
23677 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23679 * cfg.c (original_copy_tables_initialized_p): New function.
23680 * cfg.h (original_copy_tables_initialized_p): New decl.
23681 * cfgrtl.c (relink_block_chain): Guard the call to
23682 free_original_copy_tables with a call to
23683 original_copy_tables_initialized_p.
23684 * cgraph.h (symtab_node::native_rtl_p): New decl.
23685 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23686 (symtab_node::needed_p): Don't assert for early assembly output
23687 for __RTL functions.
23688 (cgraph_node::finalize_function): Set "force_output" for __RTL
23690 (cgraph_node::analyze): Bail out early for __RTL functions.
23691 (analyze_functions): Update assertion to support __RTL functions.
23692 (cgraph_node::expand): Bail out early for __RTL functions.
23693 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23695 * function.h (struct function): Update comment for field
23697 * gimple-expr.c: Include "tree-pass.h".
23698 (gimple_has_body_p): Return false for __RTL functions.
23699 * Makefile.in (OBJS): Add run-rtl-passes.o.
23700 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23702 (gcc::pass_manager::get_clean_slate): New accessor.
23703 * passes.c: Include "insn-addr.h".
23704 (should_skip_pass_p): Add logging. Update logic for running
23705 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23706 property-provider override so it is only done for gimple passes.
23708 (skip_pass): New function.
23709 (execute_one_pass): Call skip_pass when skipping passes.
23710 * read-md.c (md_reader::read_char): Support filtering
23711 the input to a subset of line numbers.
23712 (md_reader::md_reader): Initialize fields
23713 m_first_line and m_last_line.
23714 (md_reader::read_file_fragment): New function.
23715 * read-md.h (md_reader::read_file_fragment): New decl.
23716 (md_reader::m_first_line): New field.
23717 (md_reader::m_last_line): New field.
23718 * read-rtl-function.c (function_reader::create_function): Only
23719 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23720 curr_properties. Set DECL_INITIAL to a dummy block.
23721 (read_rtl_function_body_from_file_range): New function.
23722 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23724 * run-rtl-passes.c: New file.
23725 * run-rtl-passes.h: New file.
23727 2017-01-24 Jeff Law <law@redhat.com>
23729 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23732 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23734 PR tree-optimization/79159
23735 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23736 (record_nonwrapping_iv): Improve boundary using above function if no
23737 value range information.
23739 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23740 Martin Jambor <mjambor@suse.cz>
23742 * brig-builtins.def: New file.
23743 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23744 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23745 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23746 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23747 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23748 * builtin-types.def (BT_INT8): New.
23749 (BT_INT16): Likewise.
23750 (BT_UINT8): Likewise.
23751 (BT_UINT16): Likewise.
23752 (BT_FN_ULONG): Likewise.
23753 (BT_FN_UINT_INT): Likewise.
23754 (BT_FN_UINT_ULONG): Likewise.
23755 (BT_FN_UINT_LONG): Likewise.
23756 (BT_FN_UINT_PTR): Likewise.
23757 (BT_FN_ULONG_PTR): Likewise.
23758 (BT_FN_INT8_FLOAT): Likewise.
23759 (BT_FN_INT16_FLOAT): Likewise.
23760 (BT_FN_UINT32_FLOAT): Likewise.
23761 (BT_FN_UINT16_FLOAT): Likewise.
23762 (BT_FN_UINT8_FLOAT): Likewise.
23763 (BT_FN_UINT64_FLOAT): Likewise.
23764 (BT_FN_UINT16_UINT32): Likewise.
23765 (BT_FN_UINT32_UINT16): Likewise.
23766 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23767 (BT_FN_INT_PTR_INT): Likewise.
23768 (BT_FN_UINT_PTR_UINT): Likewise.
23769 (BT_FN_LONG_PTR_LONG): Likewise.
23770 (BT_FN_ULONG_PTR_ULONG): Likewise.
23771 (BT_FN_VOID_UINT64_UINT64): Likewise.
23772 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23773 (BT_FN_INT8_INT8_INT8): Likewise.
23774 (BT_FN_INT16_INT16_INT16): Likewise.
23775 (BT_FN_INT_INT_INT): Likewise.
23776 (BT_FN_UINT_FLOAT_UINT): Likewise.
23777 (BT_FN_FLOAT_UINT_UINT): Likewise.
23778 (BT_FN_ULONG_UINT_UINT): Likewise.
23779 (BT_FN_ULONG_UINT_PTR): Likewise.
23780 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23781 (BT_FN_UINT_UINT_UINT): Likewise.
23782 (BT_FN_VOID_UINT_PTR): Likewise.
23783 (BT_FN_UINT_UINT_PTR: Likewise.
23784 (BT_FN_UINT32_UINT64_PTR): Likewise.
23785 (BT_FN_INT_INT_UINT_UINT): Likewise.
23786 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23787 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23788 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23789 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23790 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23791 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23792 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23793 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23794 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23795 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23796 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23797 * doc/frontends.texi: List BRIG FE.
23798 * doc/install.texi (Testing): Add BRIG tesring requirements.
23799 * doc/invoke.texi (Overall Options): Mention BRIG.
23800 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23802 2017-01-24 Richard Biener <rguenther@suse.de>
23804 PR translation/79208
23805 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23807 2017-01-24 Martin Jambor <mjambor@suse.cz>
23810 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23811 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23812 and known_contexts.
23814 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23816 PR middle-end/79123
23817 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23818 casts from signed to unsigned really don't have a range.
23820 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23822 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23823 GMP_RNDx for compatiblity.
23825 2017-01-24 Martin Liska <mliska@suse.cz>
23828 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23829 that would prevent us to call alloca with -1 as argument.
23831 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23833 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23834 Avoid -Wformat-security warning.
23836 2017-01-23 Andrew Pinski <apinski@cavium.com>
23838 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23841 2017-01-23 Martin Sebor <msebor@redhat.com>
23843 PR middle-end/78703
23844 * gimple-ssa-sprintf.c (warn_level): New global.
23845 (format_integer): Use it here and throughout the rest of the file.
23846 Use the same switch to compute sign as base.
23847 (maybe_warn): New function.
23848 (format_directive): Factor out warnings into maybe_warn.
23849 Add debugging output. Use warn_level.
23850 (add_bytes): Use warn_level.
23851 (pass_sprintf_length::compute_format_length): Add debugging output.
23852 (try_substitute_return_value): Same.
23853 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23855 PR middle-end/78703
23856 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23857 (struct fmtresult, format_integer, format_floating): Adjust.
23858 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23859 (get_string_length, format_string,format_directive): Same.
23860 (pass_sprintf_length::compute_format_length): Same.
23861 (try_substitute_return_value): Simplify slightly.
23863 PR middle-end/78703
23864 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23865 (fmtresult::operator+=): Outlined.
23866 (struct fmtresult): Add ctors.
23867 (struct conversion_spec): Rename...
23868 (struct directive): ...to this. Add and remove data members.
23869 (directive::set_width, directive::set_precision): New functions.
23870 (format_percent): Use fmtresult ctor.
23871 (get_width_and_precision): Remove.
23872 (format_integer): Make naming changes. Avoid computing width and
23874 (format_floating): Same. Adjust indentation.
23875 (format_character, format_none): New functions.
23876 (format_string): Moved character handling to format_character.
23877 (format_directive): Remove arguments, change return type.
23878 (parse_directive): New function.
23879 (pass_sprintf_length::compute_format_length): Move directive
23880 parsing to parse_directive.
23882 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23884 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23885 (assign_assembler_name_if_needed): ... this.
23886 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23887 (assign_assembler_name_if_needed): ... this.
23888 (free_lang_data_in_cgraph): Adjust callers.
23889 * cgraphunit.c (cgraph_node::analyze): Likewise.
23890 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23893 2017-01-23 Richard Biener <rguenther@suse.de>
23895 PR tree-optimization/79088
23896 PR tree-optimization/79188
23897 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23898 resetting loop bounds after last path deletion. Reset loop
23899 bounds of the target loop, make code match the comments.
23900 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23901 Make sure loops need no fixups.
23903 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23905 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23906 exponent support with double type for first argument.
23907 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23908 type returned by __builtin_vec_extract_sig,
23909 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23910 functions from "vector int" to "vector unsigned int" or from
23911 "vector long long int" to "vector unsigned long long int".
23912 Changed type returned by __builtin_vec_extract_exp,
23913 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23914 functions from "vector int" to "vector unsigned int" or from
23915 "vector long long int" to "vector unsigned long long int".
23916 Changed return type of __builtin_vec_test_data_class,
23917 __builtin_vec_test_data_class_sp, and
23918 __builtin_vec_test_data_class_dp from "vector int" to
23919 "vector bool int" or from "vector long long int" to "vector bool
23920 long long int" and changed second argument type from "unsigned
23921 int" to "int". Added new overloaded function forms "vector float
23922 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23923 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23924 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23925 double, vector unsigned long long int)" and "vector double
23926 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23927 long int)". Changed return type of
23928 __builtin_scalar_test_data_class and
23929 __builtin_scalar_test_data_class_sp and
23930 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23931 int" and changed second argument from "unsigned int" to "int".
23932 Changed type returned by __builtin_scalar_test_neg,
23933 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23934 from "int" to "bool int". Added new overloaded function form
23935 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23936 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23937 exponent double-precision with floating point first argument.
23938 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23939 documentation of scalar_test_data_class, scalar_test_neg,
23940 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23941 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23942 vec_test_data_class built-in functions to reflect refinements in
23943 their type signatures.
23945 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23947 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23949 (aarch64_elf_asm_destructor): Likewise.
23951 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23953 PR rtl-optimization/78634
23954 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23955 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23956 * ifcvt.c (noce_try_cmove): Add missing cost check.
23958 PR rtl-optimization/71724
23959 * combine.c (if_then_else_cond): Look for situations where it is
23960 beneficial to undo the work of one of the recursive calls.
23962 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23964 PR tree-optimization/70754
23965 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23966 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23967 combined stmt before it if not NULL.
23968 (combine_chains): Process refs reversely and compute dominance point
23971 2017-01-23 Martin Liska <mliska@suse.cz>
23973 PR tree-optimization/79196
23974 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23975 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23977 (strlen_optimize_stmt): Call the renamed function.
23979 2017-01-23 Michael Matz <matz@suse.de>
23981 PR tree-optimization/78384
23982 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23984 2017-01-23 Richard Biener <rguenther@suse.de>
23986 PR tree-optimization/79186
23987 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23988 both incoming edges before moving an assert.
23990 2017-01-23 Martin Jambor <mjambor@suse.cz>
23992 * ipa-prop.c (load_from_param_1): Removed.
23993 (load_from_unmodified_param): Bits from load_from_param_1 put back
23995 (load_from_param): Removed.
23996 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23997 with stmt. Reverted back to use of load_from_unmodified_param.
23999 2017-01-23 Martin Jambor <mjambor@suse.cz>
24002 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24003 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24004 field a pointer to garbage collected vector, mark lattices and
24005 ipcp_orig_node with GTY((skip)).
24006 (ipa_get_param_count): Adjust to descriptors being a pointer.
24007 (ipa_get_param): Likewise.
24008 (ipa_get_type): Likewise.
24009 (ipa_get_param_move_cost): Likewise.
24010 (ipa_set_param_used): Likewise.
24011 (ipa_get_controlled_uses): Likewise.
24012 (ipa_set_controlled_uses): Likewise.
24013 (ipa_is_param_used): Likewise.
24014 (ipa_node_params_t): Move into garbage collector. New methods insert
24016 (ipa_node_params_sum): Annotate wth GTY(()).
24017 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24019 (ipa_load_from_parm_agg): Adjust declaration.
24020 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24021 * ipa-profile.c (ipa_profile): Likewise.
24022 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24023 (ipa_populate_param_decls): Make descriptors parameter garbage
24025 (ipa_dump_param): Adjust to descriptors being a pointer.
24026 (ipa_alloc_node_params): Likewise.
24027 (ipa_initialize_node_params): Likewise.
24028 (load_from_param_1): Make descriptors parameter garbage collected.
24029 (load_from_unmodified_param): Likewise.
24030 (load_from_param): Likewise.
24031 (ipa_load_from_parm_agg): Likewise.
24032 (ipa_node_params::~ipa_node_params): Removed.
24033 (ipa_free_all_node_params): Remove call to delete operator.
24034 (ipa_node_params_t::insert): New.
24035 (ipa_node_params_t::remove): Likewise.
24036 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24037 copy known_csts and known_contexts vectors.
24038 (ipa_read_node_info): Adjust to descriptors being a pointer.
24039 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24041 (ipcp_transform_function): Make descriptors variable garbage
24044 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24046 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24047 * config/i386/avx512dqintrin.h: Ditto.
24048 * config/i386/avx512fintrin.h: Ditto.
24049 * config/i386/i386.c: Handle new builtins.
24050 * config/i386/i386-builtin.def: Add new builtins.
24051 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24052 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24054 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24055 Martin Liska <mliska@suse.cz>
24057 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24058 * asan.c (asan_expand_poison_ifn): Support stores and use
24059 appropriate ASAN report function.
24060 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24061 * internal-fn.def (ASAN_POISON_USE): Declare.
24062 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24063 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24064 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24065 ASAN_POISON calls w/o LHS.
24066 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24067 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24068 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24069 * gimplify.c (asan_poison_variables): Add attribute
24070 use_after_scope_memory to variables that really needs to live
24072 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24073 having the attribute.
24075 2017-01-23 Martin Liska <mliska@suse.cz>
24077 * asan.c (create_asan_shadow_var): New function.
24078 (asan_expand_poison_ifn): Likewise.
24079 * asan.h (asan_expand_poison_ifn): New declaration.
24080 * internal-fn.c (expand_ASAN_POISON): Likewise.
24081 * internal-fn.def (ASAN_POISON): New builtin.
24082 * sanopt.c (pass_sanopt::execute): Expand
24083 asan_expand_poison_ifn.
24084 * tree-inline.c (copy_decl_for_dup_finish): Make function
24086 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24087 * tree-ssa.c (is_asan_mark_p): New function.
24088 (execute_update_addresses_taken): Rewrite local variables
24089 (identified just by use-after-scope as addressable) into SSA.
24091 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24093 * doc/install.texi (Specific): opensource.apple.com uses https
24094 now. Remove trailing slash.
24096 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24098 * README.Portability: Remove note on an Irix compatibility issue.
24100 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24102 * gcov.c (INCLUDE_ALGORITHM): Define.
24103 (INCLUDE_VECTOR): Define.
24104 No longer include <vector> and <algorithm> directly.
24106 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24108 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24110 * doc/invoke.texi (Code Gen Options): Ditto.
24112 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24115 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24117 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24119 rtl-optimization/79125
24120 * cprop.c (local_cprop_pass): Handle cases where we make an
24121 unconditional trap.
24123 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24127 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24128 read from, for big endian.
24130 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24132 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24133 register pauth builtins for LP64 only.
24135 2017-01-20 Marek Polacek <polacek@redhat.com>
24138 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24141 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24143 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24145 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24146 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24147 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24149 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24152 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24153 in big-endian mode when they are not a single duplicated value.
24155 2017-01-20 Richard Biener <rguenther@suse.de>
24157 * BASE-VER: Bump to 7.0.1.
24159 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24161 * omp-low.c (omplow_simd_context): New struct. Use it...
24162 (lower_rec_simd_input_clauses): ...here and...
24163 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24164 references to idx, lane, max_vf, is_simt.
24166 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24168 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24171 2017-01-20 Martin Jambor <mjambor@suse.cz>
24173 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24174 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24176 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24177 (GTFILES): Rename hsa.c to hsa-common.c.
24178 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24179 * hsa-dump.c: Likewise.
24180 * hsa-gen.c: Likewise.
24181 * hsa-regalloc.c: Likewise.
24182 * ipa-hsa.c: Likewise.
24183 * omp-expand.c: Likewise.
24184 * omp-low.c: Likewise.
24185 * toplev.c: Likewise.
24187 2017-01-20 Marek Polacek <polacek@redhat.com>
24190 * doc/invoke.texi: Document -Wduplicated-branches.
24191 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24192 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24193 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24194 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24195 return 0 only when not OEP_LEXICOGRAPHIC.
24196 (fold_build_cleanup_point_expr): Use the expression
24197 location when building CLEANUP_POINT_EXPR.
24198 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24199 * tree.c (add_expr): Handle error_mark_node.
24201 2017-01-20 Martin Liska <mliska@suse.cz>
24204 * tree-profile.c (init_ic_make_global_vars): Do not call
24206 (gimple_init_gcov_profiler): Likewise.
24208 2017-01-20 Martin Liska <mliska@suse.cz>
24211 * cgraph.h (maybe_create_reference): Remove argument and
24213 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24215 * ipa-cp.c (create_specialized_node): Likewise.
24216 * symtab.c (symtab_node::maybe_create_reference): Handle
24217 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24219 2017-01-20 Martin Liska <mliska@suse.cz>
24221 * read-rtl-function.c (function_reader::create_function): Use
24222 build_decl instread of build_decl_stat.
24224 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24226 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24227 * config/i386/avx512dqintrin.h: Ditto.
24228 * config/i386/avx512fintrin.h: Ditto.
24229 * config/i386/i386-builtin-types.def: Add new types.
24230 * config/i386/i386.c: Handle new types.
24231 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24232 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24233 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24234 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24235 (__builtin_ia32_kshiftridi): New.
24236 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24238 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24242 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24243 define to rs6000_init_stack_protect_guard.
24244 (rs6000_init_stack_protect_guard): New function.
24246 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24247 Yunqiang Su <yunqiang.su@imgtec.com>
24249 * config.gcc (supported_defaults): Add madd4.
24250 (with_madd4): Add validation.
24251 (all_defaults): Add madd4.
24252 * config/mips/mips.opt (mmadd4): New option.
24253 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24255 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24257 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24258 (ISA_HAS_FUSED_MADD4): Likewise.
24259 * doc/invoke.texi (-mmadd4): Document the new option.
24260 * doc/install.texi (--with-madd4): Document the new option.
24262 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24264 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24265 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24266 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24267 (aarch64_init_pauth_hint_builtins): New.
24268 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24269 (aarch64_expand_builtin): Expand new builtins.
24271 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24273 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24274 * combine-stack-adj.c (no_unhandled_cfa): Handle
24275 REG_CFA_TOGGLE_RA_MANGLE.
24276 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24277 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24278 info for return address signing.
24279 (aarch64_expand_epilogue): Likewise.
24281 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24283 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24284 * config/aarch64/aarch64-protos.h
24285 (aarch64_return_address_signing_enabled): New declaration.
24286 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24288 (aarch64_expand_prologue): Sign return address before it's pushed onto
24290 (aarch64_expand_epilogue): Authenticate return address fetched from
24292 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24293 (aarch64_attributes): New function attributes for "sign-return-address".
24294 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24295 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24296 ("*do_return"): Generate combined instructions according to key index.
24297 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24298 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24300 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24301 * config/aarch64/aarch64.opt (msign-return-address=): New.
24302 * doc/extend.texi (AArch64 Function Attributes): Documents
24303 "sign-return-address=".
24304 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24306 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24308 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24309 overall option summary.
24311 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24313 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24314 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24315 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24316 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24318 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24320 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24321 -mpower9-minmax by default for -mcpu=power9.
24322 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24323 128-bit floating point.
24325 2017-01-20 Alan Modra <amodra@gmail.com>
24327 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24328 optimizing for size.
24330 2017-01-20 Alan Modra <amodra@gmail.com>
24333 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24334 for strcmp and strncmp from corresponding builtin decl.
24336 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24338 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24339 instead of i386/rtems-64.h.
24340 * config/i386/rtems-64.h: Remove.
24342 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24346 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24348 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24350 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24352 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24353 Change int to HOST_WIDE_INT.
24354 * config/aarch64/aarch64-protos.h
24355 (aarch64_simd_gen_const_vector_dup): Likewise.
24356 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24358 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24360 * langhooks-def.h (lhd_type_for_size): New decl.
24361 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24362 * langhooks.c (lhd_type_for_size): New function, taken from
24365 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24367 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24368 define_bypass for CR latency.
24369 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24370 (power9-alu2): Add define_bypass for CR latency.
24372 (power9-mul): Update insn latency.
24373 (power9-mul-compare): Update insn latency, bypass latency and remove
24376 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24378 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24380 * config/aarch64/aarch64.md
24381 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24382 aarch64_nopcrelative_literal_loads.
24383 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24385 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24387 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24388 TARGET_LOONGSON_3A.
24389 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24391 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24394 * config.gcc (supported_defaults): Add lxc1-sxc1.
24395 (with_lxc1_sxc1): Add validation.
24396 (all_defaults): Add lxc1-sxc1.
24397 * config/mips/mips.opt (mlxc1-sxc1): New option.
24398 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24400 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24401 __mips_no_lxc1_sxc1.
24402 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24403 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24404 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24406 2017-01-19 Richard Biener <rguenther@suse.de>
24408 PR tree-optimization/72488
24409 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24410 sure to restore SSA info.
24411 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24413 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24415 PR rtl-optimization/79121
24416 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24417 of the inner type when shifting an extended value.
24419 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24422 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24423 interposable aliases.
24425 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24428 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24429 Use the evmergelohi instruction.
24430 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24431 (mov_sitf_e500_subreg8_2_be): Likewise.
24432 (mov_sitf_e500_subreg12_2_le): Likewise.
24433 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24434 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24435 (mov_sitf_e500_subreg8_2_le): Likewise.
24436 (mov_sitf_e500_subreg12_2_be): Likewise.
24438 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24440 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24441 attribute from vecsimple to vecperm.
24442 (altivec_vbpermq2): Likewise.
24444 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24447 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24449 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24450 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24451 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24452 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24453 case where N arg is SIZE_MAX.
24454 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24455 (cmpstrsi): Add pattern.
24457 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24459 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24460 __builtin_vec_revb builtins.
24461 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24462 built-in functions to support generation of the ISA 3.0 XXBR<x>
24463 vector byte reverse instructions.
24464 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24465 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24466 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24467 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24468 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24469 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24470 (P9V_BUILTIN_VEC_REVB): Likewise.
24471 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24472 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24473 (p9_xxbrq_v16qi): Likewise.
24474 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24475 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24476 (p9_xxbrh_v8hi): Likewise.
24477 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24478 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24479 vec_revb built-in functions.
24481 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24483 PR rtl-optimization/78952
24484 * config/i386/i386.md (any_extract): New code iterator.
24485 (*insvqi_2): Use any_extract for source operand.
24486 (*insvqi_3): Use any_shiftrt for source operand.
24488 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24490 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24492 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24494 2017-01-18 Matthias Klose <doko@ubuntu.com>
24496 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24498 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24500 * config/rs6000/altivec.h (vec_bperm): Change #define.
24501 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24502 (altivec_vbpermq2): New define_insn.
24503 (altivec_vbpermd): Likewise.
24504 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24505 function interface.
24506 (VBPERMD): Likewise.
24507 (VBPERM): New polymorphic function interface.
24508 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24509 Add entries for P9V_BUILTIN_VEC_VBPERM.
24510 * doc/extend.texi: Add interfaces for vec_bperm.
24512 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24514 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24515 first letter of error messages.
24516 (s390_resolve_overloaded_builtin): Likewise.
24517 * config/s390/s390.c (s390_expand_builtin): Likewise.
24518 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24519 (s390_valid_target_attribute_inner_p): Likewise.
24520 * config/s390/s390.md ("tabort"): Likewise.
24522 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24524 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24525 (ISA_AVOID_DIV_HILO): New macro.
24526 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24527 (ISA_HAS_DDIV): Likewise.
24529 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24531 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24533 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24535 * doc/invoke.texi (fabi-version): Spelling fix.
24537 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24540 * doc/invoke.texi (fabi-version): Mention mangling fix for
24543 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24546 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24548 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24551 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24552 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24554 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24556 (stack_protect_test): Ditto.
24557 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24558 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24560 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24561 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24562 -mstack-protector-guard-offset=.
24563 (RS/6000 and PowerPC Options): Ditto.
24565 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24567 * config/i386/i386.h (MASK_CLASS_P): New define.
24568 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24569 there are no registers from different register sets also when
24570 mask registers are used. Update function comment.
24571 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24572 to (*k/*r) and (*k/*km) alternatives.
24574 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24576 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24577 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24578 (EH_RETURN_HANDLER_RTX): New define.
24579 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24580 Force frame pointer in EH return functions.
24581 (aarch64_expand_epilogue): Add barrier for eh_return.
24582 (aarch64_final_eh_return_addr): Remove.
24583 (aarch64_eh_return_handler_rtx): New function.
24584 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24586 (aarch64_eh_return_handler_rtx): New prototype.
24588 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24590 * config/rs6000/altivec.h (vec_rlmi): New #define.
24591 (vec_vrlnm): Likewise.
24592 (vec_rlnm): Likewise.
24593 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24594 (UNSPEC_VRLNM): Likewise.
24595 (VIlong): New mode iterator.
24596 (altivec_vrl<VI_char>mi): New define_insn.
24597 (altivec_vrl<VI_char>nm): Likewise.
24598 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24600 (VRLDNM): Likewise.
24601 (RLNM): New polymorphic function entry.
24602 (VRLWMI): New monomorphic function entry.
24603 (VRLDMI): Likewise.
24604 (RLMI): New polymorphic function entry.
24605 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24606 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24607 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24610 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24613 * dwarf2out.c (field_byte_offset): Restore the
24614 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24615 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24616 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24619 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24622 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24624 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24627 * dwarf2out.c (add_data_member_location_attribute): For constant
24628 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24629 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24630 DW_AT_byte_size attributes.
24632 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24634 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24635 after forcing to constant memory when the code model is medium.
24637 2017-01-17 Julia Koval <julia.koval@intel.com>
24640 * config/i386/avx512fintrin.h
24641 (_mm512_i32gather_ps): Change __addr type to void const*.
24642 (_mm512_mask_i32gather_ps): Ditto.
24643 (_mm512_i32gather_pd): Ditto.
24644 (_mm512_mask_i32gather_pd): Ditto.
24645 (_mm512_i64gather_ps): Ditto.
24646 (_mm512_mask_i64gather_ps): Ditto.
24647 (_mm512_i64gather_pd): Ditto.
24648 (_mm512_mask_i64gather_pd): Ditto.
24649 (_mm512_i32gather_epi32): Ditto.
24650 (_mm512_mask_i32gather_epi32): Ditto.
24651 (_mm512_i32gather_epi64): Ditto.
24652 (_mm512_mask_i32gather_epi64): Ditto.
24653 (_mm512_i64gather_epi32): Ditto.
24654 (_mm512_mask_i64gather_epi32): Ditto.
24655 (_mm512_i64gather_epi64): Ditto.
24656 (_mm512_mask_i64gather_epi64): Ditto.
24657 (_mm512_i32scatter_ps): Change __addr type to void*.
24658 (_mm512_mask_i32scatter_ps): Ditto.
24659 (_mm512_i32scatter_pd): Ditto.
24660 (_mm512_mask_i32scatter_pd): Ditto.
24661 (_mm512_i64scatter_ps): Ditto.
24662 (_mm512_mask_i64scatter_ps): Ditto.
24663 (_mm512_i64scatter_pd): Ditto.
24664 (_mm512_mask_i64scatter_pd): Ditto.
24665 (_mm512_i32scatter_epi32): Ditto.
24666 (_mm512_mask_i32scatter_epi32): Ditto.
24667 (_mm512_i32scatter_epi64): Ditto.
24668 (_mm512_mask_i32scatter_epi64): Ditto.
24669 (_mm512_i64scatter_epi32): Ditto.
24670 (_mm512_mask_i64scatter_epi32): Ditto.
24671 (_mm512_i64scatter_epi64): Ditto.
24672 (_mm512_mask_i64scatter_epi64): Ditto.
24673 * config/i386/avx512pfintrin.h
24674 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24675 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24676 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24677 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24678 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24679 (_mm512_prefetch_i32scatter_ps): Ditto.
24680 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24681 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24682 (_mm512_prefetch_i64scatter_pd): Ditto.
24683 (_mm512_prefetch_i64scatter_ps): Ditto.
24684 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24685 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24686 * config/i386/avx512vlintrin.h
24687 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24688 (_mm_mmask_i32gather_ps): Ditto.
24689 (_mm256_mmask_i32gather_pd): Ditto.
24690 (_mm_mmask_i32gather_pd): Ditto.
24691 (_mm256_mmask_i64gather_ps): Ditto.
24692 (_mm_mmask_i64gather_ps): Ditto.
24693 (_mm256_mmask_i64gather_pd): Ditto.
24694 (_mm_mmask_i64gather_pd): Ditto.
24695 (_mm256_mmask_i32gather_epi32): Ditto.
24696 (_mm_mmask_i32gather_epi32): Ditto.
24697 (_mm256_mmask_i32gather_epi64): Ditto.
24698 (_mm_mmask_i32gather_epi64): Ditto.
24699 (_mm256_mmask_i64gather_epi32): Ditto.
24700 (_mm_mmask_i64gather_epi32): Ditto.
24701 (_mm256_mmask_i64gather_epi64): Ditto.
24702 (_mm_mmask_i64gather_epi64): Ditto.
24703 (_mm256_i32scatter_ps): Change __addr type to void*.
24704 (_mm256_mask_i32scatter_ps): Ditto.
24705 (_mm_i32scatter_ps): Ditto.
24706 (_mm_mask_i32scatter_ps): Ditto.
24707 (_mm256_i32scatter_pd): Ditto.
24708 (_mm256_mask_i32scatter_pd): Ditto.
24709 (_mm_i32scatter_pd): Ditto.
24710 (_mm_mask_i32scatter_pd): Ditto.
24711 (_mm256_i64scatter_ps): Ditto.
24712 (_mm256_mask_i64scatter_ps): Ditto.
24713 (_mm_i64scatter_ps): Ditto.
24714 (_mm_mask_i64scatter_ps): Ditto.
24715 (_mm256_i64scatter_pd): Ditto.
24716 (_mm256_mask_i64scatter_pd): Ditto.
24717 (_mm_i64scatter_pd): Ditto.
24718 (_mm_mask_i64scatter_pd): Ditto.
24719 (_mm256_i32scatter_epi32): Ditto.
24720 (_mm256_mask_i32scatter_epi32): Ditto.
24721 (_mm_i32scatter_epi32): Ditto.
24722 (_mm_mask_i32scatter_epi32): Ditto.
24723 (_mm256_i32scatter_epi64): Ditto.
24724 (_mm256_mask_i32scatter_epi64): Ditto.
24725 (_mm_i32scatter_epi64): Ditto.
24726 (_mm_mask_i32scatter_epi64): Ditto.
24727 (_mm256_i64scatter_epi32): Ditto.
24728 (_mm256_mask_i64scatter_epi32): Ditto.
24729 (_mm_i64scatter_epi32): Ditto.
24730 (_mm_mask_i64scatter_epi32): Ditto.
24731 (_mm256_i64scatter_epi64): Ditto.
24732 (_mm256_mask_i64scatter_epi64): Ditto.
24733 (_mm_i64scatter_epi64): Ditto.
24734 (_mm_mask_i64scatter_epi64): Ditto.
24735 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24736 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24737 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24738 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24739 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24740 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24741 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24742 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24743 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24744 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24745 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24746 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24747 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24748 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24749 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24750 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24751 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24752 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24753 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24754 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24755 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24756 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24757 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24758 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24759 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24760 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24761 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24762 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24763 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24764 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24765 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24766 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24767 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24768 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24769 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24770 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24771 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24772 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24773 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24774 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24775 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24776 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24777 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24778 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24779 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24780 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24781 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24782 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24783 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24784 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24785 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24786 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24787 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24788 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24789 definitions accordingly.
24791 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24792 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24795 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24798 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24801 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24802 hard regs for inner regno.
24804 2017-01-17 Martin Liska <mliska@suse.cz>
24807 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24808 assumption and add comment.
24810 2017-01-17 Nathan Sidwell <nathan@acm.org>
24812 * ipa-visibility.c (localize_node): New function, broken out of ...
24813 (function_and_variable_visibility): ... here. Call it.
24815 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24817 PR middle-end/77445
24818 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24819 correctly set frequency of oudgoing edge.
24820 (duplicate_thread_path): Fix profile updating.
24822 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24825 * configure.ac: Add GCC_BASE_VER.
24826 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24827 version from BASE-VER file.
24828 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24829 (gcc.o): Depend on $(BASEVER).
24830 * common.opt (dumpfullversion): New option.
24831 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24832 * doc/invoke.texi: Document -dumpfullversion.
24833 * doc/install.texi: Document --with-gcc-major-version-only.
24834 * configure: Regenerated.
24836 2017-01-17 Richard Biener <rguenther@suse.de>
24838 PR tree-optimization/71433
24839 * tree-vrp.c (register_new_assert_for): Merge same asserts
24840 on all incoming edges.
24841 (process_assert_insertions_for): Handle insertions at the
24844 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24846 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24847 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24849 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24852 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24855 2017-01-17 Alan Modra <amodra@gmail.com>
24858 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24859 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24860 symbolic stack limit when pic.
24862 2017-01-16 Martin Sebor <msebor@redhat.com>
24864 PR tree-optimization/78608
24865 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24867 2017-01-16 Jeff Law <law@redhat.com>
24870 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24871 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24872 for several include directories that may be relative to sysroot.
24873 * config/i386/x-mingw32 (gplus_includedir): Define.
24874 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24875 (native_system_includedir): Likewise.
24876 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24877 override if TARGET_SYSTEM_ROOT is defined.
24878 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24880 PR tree-optimization/79090
24881 PR tree-optimization/33562
24882 PR tree-optimization/61912
24883 PR tree-optimization/77485
24884 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24885 and computed trims into the dump file.
24887 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24889 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24891 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24894 * gimplify.c (gimplify_init_constructor): If want_value and
24895 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24899 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24900 sequence. Formatting fixes.
24901 (doloop_optimize): Formatting fixes.
24904 * gcc.c (debug_level_greater_than_spec_func): New function.
24905 (static_spec_functions): Add debug-level-gt spec function.
24906 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24908 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24909 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24910 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24911 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24912 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24913 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24915 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24917 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24918 QImode fixups to general and mask registers only.
24920 2017-01-16 Carl Love <cel@us.ibm.com>
24922 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24923 for built-in functions
24924 vector signed char vec_nabs (vector signed char)
24925 vector signed short vec_nabs (vector signed short)
24926 vector signed int vec_nabs (vector signed int)
24927 vector signed long long vec_nabs (vector signed long long)
24928 vector float vec_nabs (vector float)
24929 vector double vec_nabs (vector double)
24930 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24932 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24933 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24934 * doc/extend.texi: Update the documentation file for the new built-in
24937 2017-01-16 Martin Sebor <msebor@redhat.com>
24939 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24942 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24944 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24945 UNSPEC_VSX__XXSPLTD to require special splat handling.
24947 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24950 * system.h: Poison strndup.
24952 2017-01-16 Alan Modra <amodra@gmail.com>
24955 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24958 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24960 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24962 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24964 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24965 call recog here. Assert that INSN_CODE (insn) is non-negative.
24967 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24970 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24972 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24973 in the currently scheduled RTL fragment.
24975 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24977 PR rtl-optimization/78751
24978 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24981 2017-01-14 Jeff Law <law@redhat.com>
24983 PR tree-optimization/79090
24984 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24985 variable length stores.
24986 (compute_trims): Delete dead assignment to *trim_tail.
24987 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24990 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24992 PR rtl-optimization/78626
24993 PR rtl-optimization/78727
24994 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24995 of a block, and split such blocks after everything else is finished.
24997 2017-01-14 Alan Modra <amodra@gmail.com>
25000 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25001 target legitimate_combined_insn.
25002 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25003 (rs6000_legitimate_combined_insn): New function.
25004 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25006 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25007 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25008 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25010 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25012 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25014 2017-01-13 Jeff Law <law@redhat.com>
25016 PR tree-optimization/33562
25017 PR tree-optimization/61912
25018 PR tree-optimization/77485
25019 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25021 (delete_dead_assignment): Likewise.
25022 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25023 statement to delete_dead_call and delete_dead_assignment.
25025 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25028 * substring-locations.c (format_warning_va): Strengthen case 1 so
25029 that both endpoints of the substring must be within the format
25030 range for just the substring to be printed.
25032 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25034 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25035 * config/i386/i386.c (ix86_target_string): Add missing options
25036 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25037 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25038 flags_other and ix86_target_other to flags2_other. Display unknown
25040 (ix86_valid_target_attribute_inner_p): Add missing options and
25041 reorder options by implied ISAs, as in ix86_target_string.
25043 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25045 * hash-table.h (hash_table::too_empty_p): New function.
25046 (hash_table::expand): Use it.
25047 (hash_table::traverse): Likewise.
25048 (hash_table::empty_slot): Use sizeof (value_type) instead of
25049 sizeof (PTR) to convert bytes to elements. Shrink the table
25050 if the current size is excessive for the current number of
25053 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25055 * ira-costs.c (record_reg_classes): Break from the inner loop
25056 early once alt_fail is known to be true. Update outer loop
25057 handling accordingly.
25059 2017-01-13 Jeff Law <law@redhat.com>
25061 * tree-ssa-dse.c (decrement_count): New function.
25062 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25063 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25064 when we know the partially dead statement is a mem* function.
25066 PR tree-optimization/61912
25067 PR tree-optimization/77485
25068 * tree-ssa-dse.c: Include expr.h.
25069 (maybe_trim_constructor_store): New function.
25070 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25072 PR tree-optimization/33562
25073 PR tree-optimization/61912
25074 PR tree-optimization/77485
25075 * doc/invoke.texi: Document new dse-max-object-size param.
25076 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25077 * tree-ssa-dse.c: Include params.h.
25078 (dse_store_status): New enum.
25079 (initialize_ao_ref_for_dse): New, partially extracted from
25081 (valid_ao_ref_for_dse, normalize_ref): New.
25082 (setup_live_bytes_from_ref, compute_trims): Likewise.
25083 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25084 (maybe_trim_partially_dead_store): Likewise.
25085 (maybe_trim_complex_store): Likewise.
25086 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25087 Track what bytes live from the original store. Return tri-state
25088 for dead, partially dead or live.
25089 (dse_dom_walker): Add constructor, destructor and new private members.
25090 (delete_dead_call, delete_dead_assignment): New extracted from
25092 (dse_optimize_stmt): Make a member of dse_dom_walker.
25093 Use initialize_ao_ref_for_dse.
25095 PR tree-optimization/33562
25096 PR tree-optimization/61912
25097 PR tree-optimization/77485
25098 * sbitmap.h (bitmap_count_bits): Prototype.
25099 (bitmap_clear_range, bitmap_set_range): Likewise.
25100 * sbitmap.c (bitmap_clear_range): New function.
25101 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25103 2017-01-13 Martin Liska <mliska@suse.cz>
25106 * function.c (set_cfun): Add new argument force.
25107 * function.h (set_cfun): Likewise.
25108 * ipa-inline-transform.c (inline_call): Use the function when
25109 strict alising from is dropped for function we inline to.
25111 2017-01-13 Richard Biener <rguenther@suse.de>
25113 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25114 for dumping GIMPLE INTEGER_CSTs.
25116 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25118 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25119 to 201112L since C++17.
25121 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25124 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25125 if -fsanitize=kernel-address is present.
25127 2017-01-13 Richard Biener <rguenther@suse.de>
25129 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25130 as _Literal ( type ) number in case usual suffixes do not
25131 preserve all information.
25133 2017-01-13 Richard Biener <rguenther@suse.de>
25135 PR tree-optimization/77283
25136 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25137 and ssa-iterators.h.
25138 (is_feasible_trace): Implement a cost model based on joiner
25141 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25144 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25145 char or short to __float128/_Float128 directly.
25147 2017-01-12 Martin Sebor <msebor@redhat.com>
25149 to -Wformat-overflow.
25150 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25151 (min_bytes_remaining): Same.
25152 (get_string_length): Same.
25153 (format_string): Same.
25154 (format_directive): Same.
25156 (pass_sprintf_length::handle_gimple_call): Same.
25158 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25160 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25161 info.nowrite calls with no lhs that can't throw. Return bool
25162 whether gsi_remove has been called or not.
25163 (pass_sprintf_length::handle_gimple_call): Return bool whether
25164 try_substitute_return_value called gsi_remove. Formatting fix.
25165 (pass_sprintf_length::execute): Don't use gsi_remove if
25166 handle_gimple_call returned true.
25169 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25170 be removed due to side-effects, don't remove following barrier nor
25171 turn the successor edge into fallthru edge.
25173 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25176 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25177 element-reversing loads and stores as not swappable.
25179 2017-01-12 Nathan Sidwell <nathan@acm.org>
25180 Nicolai Stange <nicstange@gmail.com>
25182 * combine.c (try_combine): Don't ignore result of overlap checking
25183 loop. Combine overlap & asm check into single loop.
25185 2017-01-12 Richard Biener <rguenther@suse.de>
25187 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25188 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25190 2017-01-12 Richard Biener <rguenther@suse.de>
25192 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25193 and TS_TARGET_OPTION directly derive from TS_BASE.
25194 * tree-core.h (tree_optimization_option): Derive from tree_base.
25195 (tree_target_option): Likewise.
25197 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25199 * config/i386/i386.c (memory_address_length): Increase len
25200 only when rip_relative_addr_p returns false.
25202 2017-01-11 Julia Koval <julia.koval@intel.com>
25204 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25205 (OPTION_MASK_ISA_SGX_SET): New.
25206 (ix86_handle_option): Handle OPT_msgx.
25207 * config.gcc: Added sgxintrin.h.
25208 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25209 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25210 * config/i386/i386.c (ix86_target_string): Add -msgx.
25212 (ix86_option_override_internal): Handle new options.
25213 (ix86_valid_target_attribute_inner_p): Add sgx.
25214 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25215 * config/i386/i386.opt: Add msgx.
25216 * config/i386/sgxintrin.h: New file.
25217 * config/i386/x86intrin.h: Add sgxintrin.h.
25219 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25222 * fold-const.c (maybe_nonzero_address): Return 1 for function
25224 (tree_single_nonzero_warnv_p): Don't handle function local objects
25228 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25231 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25234 * opts.c: Include "spellcheck.h"
25235 (struct string_fragment): New struct.
25236 (struct edit_distance_traits<const string_fragment &>): New
25238 (get_closest_sanitizer_option): New function.
25239 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25241 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25243 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25245 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25246 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25247 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25248 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25249 for initial die_offset if dwarf_split_debug_info.
25250 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25251 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25253 (output_skeleton_debug_sections): Formatting fix. Use
25254 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25255 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25257 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25259 * config/arm/cortex-a53.md: Add bypasses for
25260 cortex_a53_r2f_cvt.
25261 (cortex_a53_r2f): Only use for transfers.
25262 (cortex_a53_f2r): Likewise.
25263 (cortex_a53_r2f_cvt): Add reservation for conversions.
25264 (cortex_a53_f2r_cvt): Likewise.
25266 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25268 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25269 to all inlined functions, change static to extern.
25271 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25274 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25276 (arm_assemble_integer): Likewise.
25278 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25280 * config.gcc: Use new awk script to check CPU, FPU and architecture
25281 parameters for --with-... options.
25282 * config/arm/parsecpu.awk: New file
25283 * config/arm/arm-cpus.in: New file.
25284 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25286 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25288 * config/arm/t-arm: Update dependency rules.
25289 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25290 of processing .def files.
25291 * config/arm/genopt.sh: Deleted.
25292 * config/arm/gentune.sh: Deleted.
25293 * config/arm/arm-cores.def: Deleted.
25294 * config/arm/arm-arches.def: Deleted.
25295 * config/arm/arm-fpus.def: Deleted.
25296 * config/arm/arm-tune.md: Regenerated.
25297 * config/arm/arm-tables.opt: Regenerated.
25298 * config/arm/arm-cpu.h: New generated file.
25299 * config/arm/arm-cpu-data.h: New generated file.
25300 * config/arm/arm-cpu-cdata.h: New generated file.
25302 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25305 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25307 (input_varpool_node): Unpack dynamically_initialized bit.
25309 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25311 PR rtl-optimization/79032
25312 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25313 the alignment of the adjusted memory reference against that of MODE,
25314 instead of the alignment of the original memory reference.
25316 2017-01-11 Martin Jambor <mjambor@suse.cz>
25318 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25320 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25321 decorated functions.
25323 2017-01-11 Richard Biener <rguenther@suse.de>
25325 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25326 set range/nonnull info for PHI results. Do not set it on
25327 stmts marked for removal.
25329 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25331 * expr.c (store_field): In the bitfield case, fetch the return value
25332 from the registers before applying a single big-endian adjustment.
25333 Always do a final load for a BLKmode value not larger than a word.
25335 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25338 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25339 that we correctly handle column numbers greater than
25340 LINE_MAP_MAX_COLUMN_NUMBER.
25342 2017-01-10 Martin Sebor <msebor@redhat.com>
25344 PR middle-end/78245
25345 * gimple-ssa-sprintf.c (get_destination_size): Call
25346 {init,fini}object_sizes.
25347 * tree-object-size.c (addr_object_size): Adjust.
25348 (pass_through_call): Adjust.
25349 (pass_object_sizes::execute): Adjust.
25350 * tree-object-size.h (fini_object_sizes): Declare.
25352 2017-01-10 Martin Sebor <msebor@redhat.com>
25354 PR tree-optimization/78775
25355 * builtins.c (get_size_range): Move...
25356 * calls.c: ...to here.
25357 (alloc_max_size): Accept zero argument.
25358 (operand_signed_p): Remove.
25359 (maybe_warn_alloc_args_overflow): Call get_size_range.
25360 * calls.h (get_size_range): Declare.
25362 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25364 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25365 from TI's devices.csv file as of September 2016.
25366 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25368 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25370 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25371 * doc/invoke.texi: Likewise.
25372 * doc/md.texi: Likewise.
25373 * doc/objc.texi: Likewise.
25375 2017-01-10 Joshua Conner <joshconner@google.com>
25377 * config/arm/fuchsia-elf.h: New file.
25378 * config/fuchsia.h: New file.
25379 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25380 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25382 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25384 2016-01-10 Richard Biener <rguenther@suse.de>
25386 PR tree-optimization/79034
25387 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25388 Propagate out degenerate PHIs in the joiner.
25390 2017-01-10 Martin Liska <mliska@suse.cz>
25392 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25393 (sort_congruence_classes_by_decl_uid): Likewise.
25394 (sort_congruence_class_groups_by_decl_uid): Likewise.
25395 (sem_item_optimizer::merge_classes): Sort class, groups in these
25396 classes and members in the groups by DECL_UID of declarations.
25397 This would make merge operations stable.
25399 2017-01-10 Martin Liska <mliska@suse.cz>
25401 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25402 usage of m_classes_vec.
25403 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25404 (sem_item_optimizer::get_group_by_hash): Likewise.
25405 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25406 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25407 (sem_item_optimizer::verify_classes): Likewise.
25408 (sem_item_optimizer::process_cong_reduction): Likewise.
25409 (sem_item_optimizer::dump_cong_classes): Likewise.
25410 (sem_item_optimizer::merge_classes): Likewise.
25411 * ipa-icf.h (congruence_class_hash): Rename from
25412 congruence_class_group_hash. Remove declaration of m_classes_vec.
25414 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25416 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25417 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25418 * config.gcc: Add avx512vpopcntdqintrin.h.
25419 * config/i386/avx512vpopcntdqintrin.h: New.
25420 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25421 * config/i386/i386-builtin-types.def: Add new types.
25422 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25423 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25424 __builtin_ia32_vpopcountq_v8di_mask): New.
25425 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25426 __AVX512VPOPCNTDQ__.
25427 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25428 (PTA_AVX512VPOPCNTDQ): Define.
25429 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25430 TARGET_AVX512VPOPCNTDQ_P): Define.
25431 * config/i386/i386.opt: Add mavx512vpopcntdq.
25432 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25433 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25435 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25437 PR middle-end/77484
25438 * predict.def (PRED_CALL): Set to 67.
25440 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25442 * expr.c (store_field): In the bitfield case, if the value comes from
25443 a function call and is of an aggregate type returned in registers, do
25444 not modify the field mode; extract the value in all cases if the mode
25445 is BLKmode and the size is not larger than a word.
25447 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25450 * config/i386/cpuid.h: Fix undefined behavior.
25452 2017-01-04 Jeff Law <law@redhat.com>
25454 PR tree-optimization/79007
25455 PR tree-optimization/67955
25456 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25457 conservative for pt.null when flag_non_call_exceptions is on.
25459 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25461 PR translation/79019
25462 PR translation/79020
25463 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25464 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25465 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25467 * config/avr/avr.opt (maccumulate-args): Likewise.
25468 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25469 * common.opt (freport-bug): Likewise.
25470 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25471 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25472 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25473 translatable string.
25474 * config/i386/i386.c (function_value_32): Likewise.
25475 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25476 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25478 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25479 * common/config/msp430/msp430-common.c (msp430_handle_option):
25481 * symtab.c (symtab_node::verify_base): Likewise.
25482 * opts.c (set_debug_level): Likewise.
25483 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25484 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25485 missing whitespace to translatable strings.
25486 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25487 * config/sh/superh.h: Likewise.
25488 * config/i386/xopintrin.h: Likewise.
25489 * config/i386/znver1.md: Likewise.
25490 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25491 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25492 * double-int.h (struct double_int): Likewise.
25493 * double-int.c (div_and_round_double): Likewise.
25494 * wide-int.cc: Likewise.
25495 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25496 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25497 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25498 (crossjumps_occurred): ... this.
25499 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25502 PR tree-optimization/78899
25503 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25504 returning bool return struct loop *, NULL for failure and the new
25506 (versionable_outer_loop_p): Don't version outer loop if it has
25507 dont_vectorized bit set.
25508 (tree_if_conversion): When versioning outer loop, ensure
25509 tree_if_conversion is performed also on the inner loop of the
25510 non-vectorizable outer loop copy.
25511 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25512 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25513 prevent vectorization of it.
25514 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25515 the outer loop vectorization of the non-scalar version is attempted
25516 before vectorization of the inner loop in scalar version. If
25517 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25518 vectorization of its inner loop.
25519 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25520 has 2 inner loops, rename also on edges from bb whose single pred
25521 is outer_loop->header. Fix typo in function comment.
25523 2017-01-09 Martin Sebor <msebor@redhat.com>
25526 * asan.c (asan_emit_stack_protection): Increase local buffer size
25527 to avoid snprintf truncation warning.
25529 2017-01-09 Andrew Pinski <apinski@cavium.com>
25531 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25532 to reference thunderx2t99 for the tuning structure
25533 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25535 (thunderx2t99_extra_costs): This.
25536 * config/aarch64/aarch64-tune.md: Regenerate.
25537 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25538 (vulcan_addrcost_table): This.
25539 (vulcan_regmove_cost): Rename to ...
25540 (thunderx2t99_regmove_cost): This.
25541 (vulcan_vector_cost): Rename to ...
25542 (thunderx2t99_vector_cost): this.
25543 (vulcan_branch_cost): Rename to ...
25544 (thunderx2t99_branch_cost): This.
25545 (vulcan_tunings): Rename to ...
25546 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25547 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25549 2017-01-09 Martin Jambor <mjambor@suse.cz>
25553 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25554 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25555 (propagate_vr_accross_jump_function): Use the above function for all
25556 value range computations for pass-through jump functions and type
25557 converasion from explicit value range values.
25558 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25559 parameters from TYPE_ARG_TYPES.
25560 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25561 (ipa_write_node_info): Stream type of the actual argument.
25562 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25564 2017-01-09 Martin Liska <mliska@suse.cz>
25567 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25568 (lookup_compiler): Do not show error message with have_E.
25570 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25572 PR tree-optimization/78938
25573 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25574 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25575 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25576 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25579 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25581 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25584 2017-01-09 Richard Biener <rguenther@suse.de>
25586 PR tree-optimization/78997
25587 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25588 name condition properly.
25590 2017-01-09 Richard Biener <rguenther@suse.de>
25593 * dwarf2out.c (is_cxx): New overload with context.
25594 (is_naming_typedef_decl): Use it.
25596 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25598 * invoke.texi (Option Summary): Correct spacing in option lists
25599 and add line breaks to fix over-long lines.
25601 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25603 PR middle-end/17660
25605 * extend.texi (Common Variable Attributes): Add xref to GCC
25606 Internals manual to explain mode attribute keywords.
25608 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25611 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25612 and Preprocessor Options.
25613 (Options for Linking): Document -pthread here....
25614 (RS/6000 and PowerPC Options): ...not here.
25615 (Solaris 2 Options): ...or here.
25616 * doc/cppopts.texi: Document -pthread.
25618 2017-01-08 Martin Sebor <msebor@redhat.com>
25620 PR middle-end/77708
25621 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25622 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25623 New member functions.
25624 (format_directive): Used them.
25626 (pass_sprintf_length::handle_gimple_call): Same.
25627 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25628 to avoid truncation for any argument.
25629 (extract_affine_mul): Same.
25630 * tree.c (get_file_function_name): Same.
25632 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25634 PR middle-end/77484
25635 * predict.def (PRED_INDIR_CALL): Set to 86.
25637 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25639 PR preprocessor/54124
25640 * doc/cppopts.texi: Reformat -d subtable to list the full name
25641 of the options. Add cross-reference to the docs for the general
25642 compiler -d options.
25643 * doc/invoke.texi (Developer Options): Add cross-reference to the
25644 preprocessor-specific -d option documentation.
25646 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25648 PR preprocessor/13498
25649 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25650 redudant material, and reflect new command-line options.
25651 (System Headers): Likewise.
25653 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25655 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25656 -isystem, and -idirafter. Copy-edit.
25657 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25658 default for -ftrack-macro-expansion. Delete obsolete and
25659 badly-formatted implementation details about -fdebug-cpp output.
25660 * doc/cppwarnopts.texi: Copy-edit.
25662 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25665 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25666 that the transition from a max line width >= 1<<10 to narrower
25667 lines works correctly.
25669 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25671 * doc/options.texi (PerFunction): New.
25672 * opt-functions.awk (switch_flags): Map both Optimization and
25673 PerFunction to CL_OPTIMIZATION.
25674 * opth-gen.awk: Test for PerFunction flag along with
25676 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25677 it only when the latter is present. Skip those that don't in
25678 the hash function generator.
25679 * common.opt (fvar-tracking): Mark as PerFunction instead of
25681 (fvar-tracking-assignments): Likewise.
25682 (fvar-tracking-assignments-toggle): Likewise.
25683 (fvar-tracking-uninit): Likewise.
25685 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25687 PR translation/79018
25688 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25691 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25694 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25695 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25696 TARGET_LONG_JUMP_TABLE_OFFSETS.
25697 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25698 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25699 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25700 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25701 * config/m68k/m68k.md (tablejump expander): Likewise.
25702 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25703 TARGET_LONG_JUMP_TABLE_OFFSETS.
25704 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25705 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25707 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25708 David Holsgrove <david.holsgrove@xilinx.com>
25710 * common/config/microblaze/microblaze-common.c
25711 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25712 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25714 * config/microblaze/microblaze.c (microblaze_must_save_register)
25715 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25717 (microblaze_eh_return): New function.
25718 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25719 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25720 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25721 * config/microblaze/microblaze.md (eh_return): New pattern.
25723 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25725 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25726 GCC_DIAGNOSTIC_STRINGIFY): Define.
25728 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25730 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25732 * config/arm/arm.md (<mcrr>): New.
25734 * config/arm/arm.c (arm_arch5te): New.
25735 (arm_option_override): Set arm_arch5te.
25736 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25738 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25739 (arm_mcrr_qualifiers): ... this. New.
25740 (MRRC_QUALIFIERS): Define to...
25741 (arm_mrrc_qualifiers): ... this. New.
25742 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25744 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25745 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25746 (MRRCI, mrrc, MRRC): New.
25747 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25748 VUNSPEC_MRRC2): New.
25750 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25752 * config/arm/arm.md (<mcr>): New.
25754 * config/arm/arm.c (arm_coproc_builtin_available): Add
25755 support for mcr, mrc, mcr2 and mrc2.
25756 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25757 (arm_mcr_qualifiers): ... this. New.
25758 (MRC_QUALIFIERS): Define to ...
25759 (arm_mrc_qualifiers): ... this. New.
25760 (MCR_QUALIFIERS): Define to ...
25761 (arm_mcr_qualifiers): ... this. New.
25762 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25764 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25765 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25766 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25767 VUNSPEC_MRC2): New.
25769 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25771 * config/arm/arm.md (*ldc): New.
25775 * config/arm/arm.c (arm_coproc_builtin_available): Add
25776 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25777 (arm_coproc_ldc_stc_legitimate_address): New.
25778 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25779 'qualifier_const_pointer'.
25780 (LDC_QUALIFIERS): Define to...
25781 (arm_ldc_qualifiers): ... this. New.
25782 (STC_QUALIFIERS): Define to...
25783 (arm_stc_qualifiers): ... this. New.
25784 * config/arm/arm-protos.h
25785 (arm_coproc_ldc_stc_legitimate_address): New.
25786 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25787 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25788 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25789 stc2, stcl, stc2l): New.
25790 * config/arm/constraints.md (Uz): New.
25791 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25792 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25793 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25794 VUNSPEC_STC2L): New.
25796 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25798 * config/arm/arm.md (<cdp>): New.
25799 * config/arm/arm.c (neon_const_bounds): Rename this ...
25800 (arm_const_bounds): ... this.
25801 (arm_coproc_builtin_available): New.
25802 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25803 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25804 (CDP_QUALIFIERS): Define to...
25805 (arm_cdp_qualifiers): ... this. New.
25807 (arm_expand_builtin_args): Add case for 6 arguments.
25808 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25809 (arm_const_bounds): ... this.
25810 (arm_coproc_builtin_available): New.
25811 * config/arm/arm_acle.h (__arm_cdp): New.
25813 * config/arm/arm_acle_builtins.def (cdp): New.
25815 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25816 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25817 'arm_const_bounds'.
25818 * config/arm/types.md (coproc): New.
25819 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25820 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25821 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25822 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25824 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25826 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25827 (UBINOP_QUALIFIERS): New.
25829 (acle_builtin_data): New. Change comment.
25830 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25831 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25832 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25833 arm_acle_builtins.def.
25834 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25835 (arm_init_acle_builtins): New.
25836 (CRC32_BUILTIN): Remove.
25837 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25838 crc32cb, crc32ch and crc32cw.
25839 (arm_init_crc32_builtins): Remove.
25840 (arm_init_builtins): Use arm_init_acle_builtins rather
25841 than arm_init_crc32_builtins.
25842 (arm_expand_acle_builtin): New.
25843 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25844 * config/arm/arm_acle_builtins.def: New.
25846 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25848 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25849 (arm_builtin_datum): ... this.
25850 (arm_init_neon_builtin): Rename to ...
25851 (arm_init_builtin): ... this. Add a new parameters PREFIX
25852 and USE_SIG_IN_NAME.
25853 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25854 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25855 'arm_builtin_datum'.
25856 (arm_init_vfp_builtins): Likewise.
25857 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25858 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25859 (arm_expand_neon_args): Rename to ...
25860 (arm_expand_builtin_args): ... this. Rename builtin_arg
25861 enum values and differentiate between ARG_BUILTIN_MEMORY
25862 and ARG_BUILTIN_NEON_MEMORY.
25863 (arm_expand_neon_builtin_1): Rename to ...
25864 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25865 values, arm_expand_builtin_args and add bool parameter NEON.
25866 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25867 (arm_expand_vfp_builtin): Likewise.
25868 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25870 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25872 PR middle-end/77484
25873 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25874 * predict.c (tree_estimate_probability_bb): Reverse direction of
25875 polymorphic call predictor.
25877 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25879 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25880 (determine_pass_name_match): ...this new function and...
25881 (should_skip_pass_p): ...this new function.
25883 2017-01-06 Nathan Sidwell <nathan@acm.org>
25885 * ipa-visibility.c (function_and_variable_visibility): Reformat
25886 comments and long lines. Remove extrneous if.
25887 * symtab.c (symtab_node::make_decl_local): Fix code format.
25888 (symtab_node::set_section_for_node): Fix comment typo.
25890 2017-01-06 Martin Liska <mliska@suse.cz>
25893 * lra-constraints.c: Rename invariant to lra_invariant.
25894 * predict.c (set_even_probabilities): Initialize e to NULL.
25896 2017-01-05 Martin Sebor <msebor@redhat.com>
25898 PR tree-optimization/78910
25899 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25900 (format_integer): Correct off-by-one error in the handling
25901 of precision with negative numbers in signed conversions..
25903 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25905 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25907 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25909 PR tree-optimization/71016
25910 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25911 factor_out_conditional_conversion. Formatting fix.
25912 (factor_out_conditional_conversion): Add cond_stmt argument.
25913 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25914 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25917 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25919 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25920 read-rtl-function.o, and selftest-rtl.o.
25921 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25922 (selftest::aarch64_test_loading_full_dump): New function.
25923 (selftest::aarch64_run_selftests): New function.
25924 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25925 selftest::aarch64_run_selftests.
25926 * config/i386/i386.c
25927 (selftest::ix86_test_loading_dump_fragment_1): New function.
25928 (selftest::ix86_test_loading_call_insn): New function.
25929 (selftest::ix86_test_loading_full_dump): New function.
25930 (selftest::ix86_test_loading_unspec): New function.
25931 (selftest::ix86_run_selftests): Call the new functions.
25932 * emit-rtl.c (maybe_set_max_label_num): New function.
25933 * emit-rtl.h (maybe_set_max_label_num): New decl.
25934 * function.c (instantiate_decls): Guard call to
25935 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25936 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25938 * gensupport.c (gen_reader::gen_reader): Pass "false"
25939 for new "compact" param of rtx_reader.
25940 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25941 rather than an empty string for NULL strings.
25942 * read-md.c: Potentially include config.h rather than bconfig.h.
25943 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25944 (have_error): New global, copied from errors.c.
25945 (md_reader::read_name): Rename to...
25946 (md_reader::read_name_1): ...this, adding "out_loc" param,
25947 and converting "missing name or number" to returning false, rather
25949 (md_reader::read_name): Reimplement in terms of read_name_1.
25950 (md_reader::read_name_or_nil): New function.
25951 (md_reader::read_string): Handle "(nil)" by returning NULL.
25952 (md_reader::md_reader): Add new param "compact".
25953 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25954 (md_reader::read_file): New method.
25955 * read-md.h (md_reader::md_reader): Add new param "compact".
25956 (md_reader::read_file): New method.
25957 (md_reader::is_compact): New accessor.
25958 (md_reader::read_name): Convert return type from void to file_location.
25959 (md_reader::read_name_or_nil): New decl.
25960 (md_reader::read_name_1): New decl.
25961 (md_reader::m_compact): New field.
25962 (noop_reader::noop_reader): Pass "false" for new "compact" param
25964 (rtx_reader::rtx_reader): Add new "compact" param.
25965 (rtx_reader::read_rtx_operand): Make virtual and convert return
25966 type from void to rtx.
25967 (rtx_reader::read_until): New decl.
25968 (rtx_reader::handle_any_trailing_information): New virtual function.
25969 (rtx_reader::postprocess): New virtual function.
25970 (rtx_reader::finalize_string): New virtual function.
25971 (rtx_reader::m_in_call_function_usage): New field.
25972 (rtx_reader::m_reuse_rtx_by_id): New field.
25973 * read-rtl-function.c: New file.
25974 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25975 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25976 (selftest::verify_three_block_rtl_cfg): New decl.
25977 * read-rtl-function.h: New file.
25978 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25979 For host, include function.h, memmodel.h, and emit-rtl.h.
25980 (one_time_initialization): New function.
25981 (struct compact_insn_name): New struct.
25982 (compact_insn_names): New array.
25983 (find_code): Handle insn codes in compact dumps.
25984 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25985 (bind_subst_iter_and_attr): Likewise.
25986 (add_condition_to_string): Likewise.
25987 (add_condition_to_rtx): Likewise.
25988 (apply_attribute_uses): Likewise.
25989 (add_current_iterators): Likewise.
25990 (apply_iterators): Likewise.
25991 (initialize_iterators): Guard usage of apply_subst_iterator with
25992 #ifdef GENERATOR_FILE.
25993 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25994 (md_reader::read_mapping): Likewise.
25995 (add_define_attr_for_define_subst): Likewise.
25996 (add_define_subst_attr): Likewise.
25997 (read_subst_mapping): Likewise.
25998 (check_code_iterator): Likewise.
25999 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26001 (one_time_initialization): New function.
26002 (rtx_reader::read_until): New method.
26003 (read_flags): New function.
26004 (parse_reg_note_name): New function.
26005 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26006 Handle reuse_rtx ids.
26007 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26008 Add parsing support for RTL dumps, mirroring the special-cases in
26009 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26010 values, and calling handle_any_trailing_information.
26011 (rtx_reader::read_rtx_operand): Convert return type from void
26012 to rtx, returning return_rtx. Handle case 'e'. Call
26013 finalize_string on XSTR and XTMPL fields.
26014 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26015 "(nil)" values were omitted. Call the postprocess vfunc on the
26017 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26018 class ctor. Initialize m_in_call_function_usage. Call
26019 one_time_initialization.
26020 * rtl-tests.c (selftest::test_uncond_jump): Call
26021 set_new_first_and_last_insn.
26022 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26023 * selftest-rtl.c: New file.
26024 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26025 (selftest::get_insn_by_uid): New decl.
26026 * selftest-run-tests.c (selftest::run_tests): Call
26027 read_rtl_function_c_tests.
26028 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26029 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26032 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26034 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26035 operands in a special way. Assert that pos+len <= mode precision.
26037 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26039 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26040 3 argument Alias with unlimited for the negative form.
26041 (fno-vect-cost-model): Removed.
26043 2017-01-05 Martin Liska <mliska@suse.cz>
26045 * hsa-gen.c (gen_hsa_divmod): New function.
26046 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26048 2017-01-05 Martin Liska <mliska@suse.cz>
26051 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26054 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26056 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26057 small constant length operands.
26059 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26061 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26062 between loop iterations.
26064 2017-01-05 Martin Liska <mliska@suse.cz>
26067 * gimplify.c (gimplify_decl_expr): Compare to
26068 asan_poisoned_variables instread of checking flags.
26069 (gimplify_target_expr): Likewise.
26070 (gimplify_expr): Likewise.
26071 (gimplify_function_tree): Conditionally initialize
26072 asan_poisoned_variables.
26074 2017-01-04 Jeff Law <law@redhat.com>
26076 PR tree-optimizatin/78812
26077 * rtl.h (contains_mem_rtx_p): Prototype.
26078 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26079 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26080 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26081 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26082 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26084 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26086 * input.c (assert_char_at_range): Default-initialize actual_range.
26088 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26090 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26091 to match the caller.
26093 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26095 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26096 insns after final jump in test to emit dummy move.
26098 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26100 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26101 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26103 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26105 * multiple_target.c (create_dispatcher_calls): Init e_next.
26106 * tree-ssa-loop-split.c (split_loop): Init border.
26107 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26110 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26115 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26116 (altivec_register_operand): Do not return true if the operand
26117 contains a SUBREG mixing SImode and SFmode.
26118 (vsx_register_operand): Likewise.
26119 (vsx_reg_sfsubreg_ok): New predicate.
26120 (vfloat_operand): Do not return true if the operand contains a
26121 SUBREG mixing SImode and SFmode.
26122 (vint_operand): Likewise.
26123 (vlogical_operand): Likewise.
26124 (gpc_reg_operand): Likewise.
26125 (int_reg_operand): Likewise.
26126 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26127 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26128 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26130 (rs6000_emit_move_si_sf_subreg): New helper function.
26131 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26132 fixup SUBREGs involving SImode and SFmode.
26133 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26134 numbers for the new peephole2 optimization.
26135 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26136 the GLIBC math library that do AND/IOR/XOR operations on single
26137 precision floating point.
26138 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26139 target macros to say whether we need to avoid SUBREGs mixing
26141 (TARGET_ALLOW_SF_SUBREG): Likewise.
26142 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26143 (UNSPEC_SI_FROM_SF): Likewise.
26144 (iorxor): Change spacing.
26145 (and_ior_xor): New iterator for AND, IOR, and XOR.
26146 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26147 (movdi_from_sf_zero_ext): Likewise.
26148 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26149 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26150 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26151 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26152 (fms<mode>4): Likewise.
26153 (fnma<mode>4): Likewise.
26154 (fnms<mode>4): Likewise.
26155 (nfma<mode>4): Likewise.
26156 (nfms<mode>4): Likewise.
26158 2017-01-04 Marek Polacek <polacek@redhat.com>
26161 * doc/invoke.texi: Document -Wpointer-compare.
26163 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26165 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26168 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26169 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26172 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26175 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26176 documentation of the powerpc_popcntb_ok attribute.
26177 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26178 code to issue warning messages if a requested CPU configuration is
26179 not supported by the binary (assembler and loader) toolchain.
26180 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26181 made to define a built-in function that has been disabled.
26182 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26183 made to define a built-in function that has been disabled.
26184 (altivec_init_builtins): Add comment explaining why definition
26185 of the DST built-in functions is not preceded by an assertion
26186 check. Add assertions to prevent ICE if attempts are made to
26187 define an altivec predicate or an abs* built-in function that has
26189 (htm_init_builtins): Add comment explaining why definition of the
26190 htm built-in functions is not preceded by an assertion check.
26192 2017-01-04 Jeff Law <law@redhat.com>
26194 PR tree-optimizatin/67955
26195 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26196 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26197 the points-to solution does not include pt_null. Use DECL_PT_UID
26200 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26202 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26203 Use gen_int_mode instead of gen_lopwart for const_int operands.
26205 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26207 PR tree-optimization/71563
26208 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26209 out of range value - has low bits known to be zero.
26211 2017-01-04 Alan Modra <amodra@gmail.com>
26213 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26214 * configure: Regenerate.
26215 * config.in: Regenerate.
26217 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26220 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26221 a substring of the message, but strcmp with the whole message. Ifdef
26222 ENABLE_NLS, translate the message first using dgettext.
26224 2017-01-03 Jeff Law <law@redhat.com>
26226 PR tree-optimizatin/78856
26227 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26228 (mark_threaded_blocks): Remove code to truncate thread paths that
26229 cross multiple loop headers. Instead invalidate the cached loop
26230 iteration information and handle case of a thread path walking
26231 into an irreducible region.
26233 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26236 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26237 assertions. Add support for doing the signbit if the IEEE 128-bit
26238 floating point value is in a GPR.
26239 * config/rs6000/rs6000.md (Fsignbit): Delete.
26240 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26241 Update the length attribute if the value is in a GPR.
26242 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26243 the sign or zero extension instruction, since the value is always 0/1.
26244 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26247 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26248 extracting SImode to a GPR register so that we can generate a
26249 store, limit the vector to be in a traditional Altivec register
26250 for the vextuwrx instruction.
26252 2017-01-03 Ian Lance Taylor <iant@google.com>
26254 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26256 2017-01-03 Martin Sebor <msebor@redhat.com>
26258 PR tree-optimization/78696
26259 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26260 precision. Use MPFR for %f for greater fidelity. Correct handling
26262 (pass_sprintf_length::compute_format_length): Set width and precision
26263 specified by asrerisk to void_node for vararg functions.
26264 (try_substitute_return_value): Adjust dump output.
26266 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26268 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26270 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26272 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26273 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26274 -mlra/-mno-lra was passed to the compiler.
26276 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26278 PR rtl-optimization/65618
26279 * emit-rtl.c (try_split): Move initialization of "before" and
26280 "after" to just before the call to emit_insn_after_setloc.
26282 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26284 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26286 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26288 * dwarf2out.c (gen_enumeration_type_die): When
26289 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26291 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26293 PR tree-optimization/78965
26294 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26295 Change first argument from const call_info & to call_info &. For %n
26296 set info.nowrite to false.
26298 PR middle-end/78901
26299 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26300 possibly throwing calls.
26302 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26303 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26304 and fns handling, rather than in a separate case SSA_NAME.
26306 2017-01-02 Jeff Law <law@redhat.com>
26308 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26309 fixes for first_period and second_period variables.
26311 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26314 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26315 (*insvqi_1): New insn pattern.
26316 (*insvqi_1_mem_rex64): Ditto.
26317 (*insvqi_2): Ditto.
26318 (*insvqi_3): Rename from *insvqi.
26320 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26322 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26324 * doc/cfg.texi (Edges): Remove reference to Java.
26325 (Maintaining the CFG): Ditto.
26327 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26329 PR middle-end/77674
26330 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26331 transparent aliases.
26333 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26335 PR middle-end/77484
26336 * predict.def (PRED_CALL): Update hitrate.
26337 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26338 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26339 into direct/indirect/polymorphic variants.
26341 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26343 Update copyright years.
26345 * gcc.c (process_command): Update copyright notice dates.
26346 * gcov-dump.c (print_version): Ditto.
26347 * gcov.c (print_version): Ditto.
26348 * gcov-tool.c (print_version): Ditto.
26349 * gengtype.c (create_file): Ditto.
26350 * doc/cpp.texi: Bump @copying's copyright year.
26351 * doc/cppinternals.texi: Ditto.
26352 * doc/gcc.texi: Ditto.
26353 * doc/gccint.texi: Ditto.
26354 * doc/gcov.texi: Ditto.
26355 * doc/install.texi: Ditto.
26356 * doc/invoke.texi: Ditto.
26358 Copyright (C) 2017 Free Software Foundation, Inc.
26360 Copying and distribution of this file, with or without modification,
26361 are permitted in any medium without royalty provided the copyright
26362 notice and this notice are preserved.