1 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
6 on it early, rather than manual conversion late. For
7 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8 instead of performing manual conversion.
10 2017-09-12 Carl Love <cel@us.ibm.com>
12 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16 VMULOSW): Add definitions.
17 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
25 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
27 (movti_aarch64): Likewise.
28 (movdf_aarch64): Likewise.
29 (movtf_aarch64): Likewise.
30 (load_pairdi): Likewise.
31 (store_pairdi): Likewise.
32 (load_pairdf): Likewise.
33 (store_pairdf): Likewise.
34 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
35 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
36 (ldr_got_small_<mode>): Likewise.
37 (ldr_got_small_28k_<mode>): Likewise.
38 (ldr_got_tiny): Likewise.
39 * config/aarch64/iterators.md (ldst_sz): New.
40 (ldpstp_sz): Likewise.
41 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
43 (thunderx_load): Split load_8 to load_16.
44 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
46 (thunderx2t99_storepair_basic): Split store_8 to store_16.
47 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
48 (xgene1_store_pair): Split store_8 to store_16.
49 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
50 (falkor_st_0_st_sd): Split store_8 to store_16.
52 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
54 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
55 and store1/2/3/4 to store_4/8/12/16.
56 * config/aarch64/aarch64.md: Update for rename.
57 * config/arm/arm.md: Likewise.: Likewise.
58 * config/arm/arm.c: Likewise.
59 * config/arm/thumb1.md: Likewise.
60 * config/arm/thumb2.md: Likewise.
61 * config/arm/vfp.md: Likewise.
62 * config/arm/arm-generic.md: Likewise.
63 * config/arm/arm1020e.md: Likewise.
64 * config/arm/arm1026ejs.md: Likewise.
65 * config/arm/arm1136jfs.md: Likewise.
66 * config/arm/arm926ejs.md: Likewise.
67 * config/arm/cortex-a15.md: Likewise.
68 * config/arm/cortex-a17.md: Likewise.
69 * config/arm/cortex-a5.md: Likewise.
70 * config/arm/cortex-a53.md: Likewise.
71 * config/arm/cortex-a57.md: Likewise.
72 * config/arm/cortex-a7.md: Likewise.
73 * config/arm/cortex-a8.md: Likewise.
74 * config/arm/cortex-a9.md: Likewise.
75 * config/arm/cortex-m4.md: Likewise.
76 * config/arm/cortex-m7.md: Likewise.
77 * config/arm/cortex-r4.md: Likewise.
78 * config/arm/exynos-m1.md: Likewise.
79 * config/arm/fa526.md: Likewise.
80 * config/arm/fa606te.md: Likewise.
81 * config/arm/fa626te.md: Likewise.
82 * config/arm/fa726te.md: Likewise.
83 * config/arm/fmp626.md: Likewise.
84 * config/arm/iwmmxt.md: Likewise.
85 * config/arm/ldmstm.md: Likewise.
86 * config/arm/marvell-pj4.md: Likewise.
87 * config/arm/xgene1.md: Likewise.
88 * config/aarch64/thunderx.md: Likewise.
89 * config/aarch64/thunderx2t99.md: Likewise.
90 * config/aarch64/falkor.md: Likewise.
92 2017-09-12 Martin Liska <mliska@suse.cz>
94 * attribs.c (private_lookup_attribute): New function.
95 * attribs.h (private_lookup_attribute): Declared here.
96 (lookup_attribute): Called from this place.
98 2017-09-12 Richard Biener <rguenther@suse.de>
100 PR tree-optimization/82157
101 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
102 stmts with side-effects.
104 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
105 Alan Hayward <alan.hayward@arm.com>
106 David Sherwood <david.sherwood@arm.com>
108 * target.def (hard_regno_nregs): New hook.
109 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
110 * targhooks.h (default_hard_regno_nregs): Declare.
111 * targhooks.c (default_hard_regno_nregs): New function.
112 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
113 (TARGET_HARD_REGNO_NREGS): ...this hook.
114 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
115 (CLASS_MAX_NREGS): Likewise.
116 * doc/tm.texi: Regenerate.
117 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
118 instead of HARD_REGNO_NREGS.
119 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
120 HARD_REGNO_NREGS in the comment.
121 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
122 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
123 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
124 Return an unsigned int.
125 (TARGET_HARD_REGNO_NREGS): Redefine.
126 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
127 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
128 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
129 (arc_hard_regno_nregs): New function.
130 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
131 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
132 (arm_hard_regno_nregs): New function.
133 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
134 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
135 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
136 (TARGET_HARD_REGNO_NREGS): Redefine.
137 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
138 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
139 (HARD_REGNO_NREGS): Delete.
140 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
141 (cr16_hard_regno_nregs): New function.
142 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
143 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
144 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
145 (cris_hard_regno_nregs): New function.
146 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
147 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
148 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
149 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
150 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
151 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
152 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
153 (frv_hard_regno_nregs): Make static. Take and return an
155 (frv_class_max_nregs): Remove outdated copy of documentation.
156 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
157 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
158 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
159 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
160 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
161 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
162 (TARGET_HARD_REGNO_NREGS): Redefine.
163 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
164 (CLASS_MAX_NREGS): Update comment.
165 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
166 (ia64_hard_regno_nregs): New function.
167 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
168 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
169 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
170 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
171 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
173 (m32c_hard_regno_nregs): Likewise. Make static.
174 (TARGET_HARD_REGNO_NREGS): Redefine.
175 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
176 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
177 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
178 (m68k_hard_regno_nregs): New function.
179 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
180 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
181 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
182 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
183 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
184 Take and return an unsigned int.
185 (TARGET_HARD_REGNO_NREGS): Redefine.
186 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
187 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
188 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
189 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
190 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
191 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
192 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
193 (msp430_hard_regno_nregs): Make static. Take and return an
195 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
196 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
197 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
198 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
199 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
200 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
201 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
202 (TARGET_HARD_REGNO_NREGS): Redefine.
203 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
204 (PA_HARD_REGNO_NREGS): ...this.
205 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
206 (PA_HARD_REGNO_NREGS): ...this.
207 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
208 (pa_hard_regno_nregs): New function.
209 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
210 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
211 (pdp11_hard_regno_nregs): New function.
212 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
213 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
214 (rs6000_hard_regno_nregs_hook): New function.
215 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
216 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
217 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
218 Take and return an unsigned int. Move earlier in file.
219 (TARGET_HARD_REGNO_NREGS): Redefine.
220 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
221 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
222 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
223 (rl78_hard_regno_nregs): Make static. Take and return an
225 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
226 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
227 (rs6000_hard_regno_nregs_hook): New function.
228 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
229 * config/rx/rx.c (rx_hard_regno_nregs): New function.
230 (TARGET_HARD_REGNO_NREGS): Redefine.
231 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
232 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
233 instead of HARD_REGNO_NREGS.
234 (s390_hard_regno_nregs): New function.
235 (s390_hard_regno_mode_ok): Add comment from s390.h.
236 (TARGET_HARD_REGNO_NREGS): Redefine.
237 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
238 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
239 (sh_hard_regno_nregs): New function.
240 (sh_pass_in_reg_p): Use it.
241 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
242 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
243 (sparc_hard_regno_nregs): New function.
244 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
245 * config/spu/spu.c (spu_hard_regno_nregs): New function.
246 (spu_function_arg_advance): Use it, supplying a valid register number.
247 (TARGET_HARD_REGNO_NREGS): Redefine.
248 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
249 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
250 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
251 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
252 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
253 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
254 (CLASS_MAX_NREGS): Remove copy of old documentation.
255 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
256 (visium_hard_regno_nregs): New function.
257 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
258 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
259 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
260 xtensa_hard_regno_nregs): New function.
261 * system.h (HARD_REGNO_NREGS): Poison.
263 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
265 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
266 hard_regno_nregs instead of HARD_REGNO_NREGS.
267 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
268 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
269 (c6x_expand_epilogue): Likewise.
270 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
271 (frv_read_iacc_argument): Likewise.
272 * config/sh/sh.c: Include regs.h.
273 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
274 (regs_used): Likewise.
275 (output_stack_adjust): Likewise.
276 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
277 * expmed.c: Include regs.h.
278 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
279 * ree.c: Include regs.h.
280 (combine_reaching_defs): Use hard_regno_nregs instead of
282 (add_removable_extension): Likewise.
284 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
286 * regs.h (hard_regno_nregs): Turn into a function.
287 (end_hard_regno): Update accordingly.
288 * caller-save.c (setup_save_areas): Likewise.
289 (save_call_clobbered_regs): Likewise.
290 (replace_reg_with_saved_mem): Likewise.
291 (insert_restore): Likewise.
292 (insert_save): Likewise.
293 * combine.c (can_change_dest_mode): Likewise.
294 (move_deaths): Likewise.
295 (distribute_notes): Likewise.
296 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
297 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
298 (rs6000_split_multireg_move): Likewise.
299 (rs6000_register_move_cost): Likewise.
300 (rs6000_memory_move_cost): Likewise.
301 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
302 (rs6000_split_multireg_move): Likewise.
303 (rs6000_register_move_cost): Likewise.
304 (rs6000_memory_move_cost): Likewise.
305 * cselib.c (cselib_reset_table): Likewise.
306 (cselib_lookup_1): Likewise.
307 * emit-rtl.c (set_mode_and_regno): Likewise.
308 * function.c (aggregate_value_p): Likewise.
309 * ira-color.c (setup_profitable_hard_regs): Likewise.
310 (check_hard_reg_p): Likewise.
311 (calculate_saved_nregs): Likewise.
312 (assign_hard_reg): Likewise.
313 (improve_allocation): Likewise.
314 (calculate_spill_cost): Likewise.
315 * ira-emit.c (modify_move_list): Likewise.
316 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
317 (ira_hard_reg_in_set_p): Likewise.
318 * ira.c (setup_reg_mode_hard_regset): Likewise.
319 (clarify_prohibited_class_mode_regs): Likewise.
320 (check_allocation): Likewise.
321 * lra-assigns.c (find_hard_regno_for_1): Likewise.
322 (lra_setup_reg_renumber): Likewise.
323 (setup_try_hard_regno_pseudos): Likewise.
324 (spill_for): Likewise.
325 (assign_hard_regno): Likewise.
326 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
327 * lra-constraints.c (in_class_p): Likewise.
328 (lra_constraint_offset): Likewise.
329 (simplify_operand_subreg): Likewise.
330 (lra_constraints): Likewise.
331 (split_reg): Likewise.
332 (split_if_necessary): Likewise.
333 (invariant_p): Likewise.
334 (inherit_in_ebb): Likewise.
335 * lra-lives.c (process_bb_lives): Likewise.
336 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
337 (get_hard_regs): Likewise.
338 (do_remat): Likewise.
339 * lra-spills.c (assign_spill_hard_regs): Likewise.
340 * mode-switching.c (create_pre_exit): Likewise.
341 * postreload.c (reload_combine_recognize_pattern): Likewise.
342 * recog.c (peep2_find_free_register): Likewise.
343 * regcprop.c (kill_value_regno): Likewise.
344 (set_value_regno): Likewise.
345 (copy_value): Likewise.
346 (maybe_mode_change): Likewise.
347 (find_oldest_value_reg): Likewise.
348 (copyprop_hardreg_forward_1): Likewise.
349 * regrename.c (check_new_reg_p): Likewise.
350 (regrename_do_replace): Likewise.
351 * reload.c (push_reload): Likewise.
352 (combine_reloads): Likewise.
353 (find_dummy_reload): Likewise.
354 (operands_match_p): Likewise.
355 (find_reloads): Likewise.
356 (find_equiv_reg): Likewise.
357 (reload_adjust_reg_for_mode): Likewise.
358 * reload1.c (count_pseudo): Likewise.
359 (count_spilled_pseudo): Likewise.
360 (find_reg): Likewise.
361 (clear_reload_reg_in_use): Likewise.
362 (free_for_value_p): Likewise.
363 (allocate_reload_reg): Likewise.
364 (choose_reload_regs): Likewise.
365 (reload_adjust_reg_for_temp): Likewise.
366 (emit_reload_insns): Likewise.
367 (delete_output_reload): Likewise.
368 * rtlanal.c (subreg_get_info): Likewise.
369 * sched-deps.c (sched_analyze_reg): Likewise.
370 * sel-sched.c (init_regs_for_mode): Likewise.
371 (mark_unavailable_hard_regs): Likewise.
372 (choose_best_reg_1): Likewise.
373 (verify_target_availability): Likewise.
374 * valtrack.c (dead_debug_insert_temp): Likewise.
375 * var-tracking.c (track_loc_p): Likewise.
376 (emit_note_insn_var_location): Likewise.
377 * varasm.c (make_decl_rtl): Likewise.
378 * reginfo.c (choose_hard_reg_mode): Likewise.
379 (init_reg_modes_target): Refer directly to
380 this_target_regs->x_hard_regno_nregs.
382 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
384 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
385 instead of hard_regno_nregs.
387 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
389 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
390 end_hard_regno instead of hard_regno_nregs.
391 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
392 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
393 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
394 * ira-color.c (improve_allocation): Likewise.
395 * lra-assigns.c (find_hard_regno_for_1): Likewise.
396 * lra-lives.c (mark_regno_live): Likewise.
397 (mark_regno_dead): Likewise.
398 * lra-remat.c (operand_to_remat): Likewise.
399 * lra.c (collect_non_operand_hard_regs): Likewise.
400 * postreload.c (reload_combine_note_store): Likewise.
401 (move2add_valid_value_p): Likewise.
402 * reload.c (regno_clobbered_p): Likewise.
404 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
406 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
408 * config/v850/v850.c (v850_reorg): Likewise.
409 * reload.c (refers_to_regno_for_reload_p): Likewise.
410 (find_equiv_reg): Likewise.
411 * reload1.c (reload_reg_reaches_end_p): Likewise.
413 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
415 * caller-save.c (add_used_regs): Use REG_NREGS instead of
417 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
418 * config/arm/arm.c (output_move_neon): Likewise.
419 (arm_attr_length_move_neon): Likewise.
420 (neon_split_vcombine): Likewise.
421 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
422 (c6x_mark_reg_written): Likewise.
423 (c6x_dwarf_register_span): Likewise.
424 * config/i386/i386.c (ix86_save_reg): Likewise.
425 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
426 (rws_access_reg): Likewise.
427 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
428 * mode-switching.c (create_pre_exit): Likewise.
429 * ree.c (combine_reaching_defs): Likewise.
430 (add_removable_extension): Likewise.
431 * regcprop.c (find_oldest_value_reg): Likewise.
432 (copyprop_hardreg_forward_1): Likewise.
433 * reload.c (reload_inner_reg_of_subreg): Likewise.
434 (push_reload): Likewise.
435 (combine_reloads): Likewise.
436 (find_dummy_reload): Likewise.
437 (reload_adjust_reg_for_mode): Likewise.
438 * reload1.c (find_reload_regs): Likewise.
439 (forget_old_reloads_1): Likewise.
440 (reload_reg_free_for_value_p): Likewise.
441 (reload_adjust_reg_for_temp): Likewise.
442 (emit_reload_insns): Likewise.
443 (delete_output_reload): Likewise.
444 * sel-sched.c (choose_best_reg_1): Likewise.
445 (choose_best_pseudo_reg): Likewise.
447 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
448 Alan Hayward <alan.hayward@arm.com>
449 David Sherwood <david.sherwood@arm.com>
451 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
452 * target.def (slow_unaligned_access): New hook.
453 * targhooks.h (default_slow_unaligned_access): Declare.
454 * targhooks.c (default_slow_unaligned_access): New function.
455 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
456 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
457 * doc/tm.texi: Regenerate.
458 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
459 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
460 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
462 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
463 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
465 (rs6000_slow_unaligned_access): New function.
466 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
467 (expand_block_compare): Likewise.
468 (expand_strn_compare): Likewise.
469 (rs6000_rtx_costs): Likewise.
470 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
471 (riscv_slow_unaligned_access): Likewise.
472 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
473 (riscv_slow_unaligned_access_p): ...this and make static.
474 (riscv_option_override): Update accordingly.
475 (riscv_slow_unaligned_access): New function.
476 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
477 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
478 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
479 (rs6000_slow_unaligned_access): New function.
480 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
481 (rs6000_rtx_costs): Likewise.
482 * config/rs6000/rs6000-string.c (expand_block_compare)
483 (expand_strn_compare): Use targetm.slow_unaligned_access instead
484 of SLOW_UNALIGNED_ACCESS.
485 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
486 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
487 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
488 of SLOW_UNALIGNED_ACCESS.
489 * expmed.c (simple_mem_bitfield_p): Likewise.
490 * expr.c (alignment_for_piecewise_move): Likewise.
491 (emit_group_load_1): Likewise.
492 (emit_group_store): Likewise.
493 (copy_blkmode_from_reg): Likewise.
494 (emit_push_insn): Likewise.
495 (expand_assignment): Likewise.
496 (store_field): Likewise.
497 (expand_expr_real_1): Likewise.
498 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
499 * lra-constraints.c (simplify_operand_subreg): Likewise.
500 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
501 * gimple-ssa-store-merging.c: Likewise in block comment at start
503 * tree-ssa-strlen.c: Include target.h.
504 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
505 of SLOW_UNALIGNED_ACCESS.
506 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
508 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
510 PR rtl-optimization/82185
511 * expmed.c (emit_store_flag_int): Only test tem if it has been
514 2017-09-12 Richard Biener <rguenther@suse.de>
517 * match.pd ((FTYPE) N CMP CST): Fix typo.
519 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
521 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
523 (mips_near_type_p): Add 'short_call' attribute as a synonym
525 * doc/extend.texi (short_call): Document new function attribute.
527 2017-09-12 Jakub Jelinek <jakub@redhat.com>
530 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
531 assertion check that in the condition.
532 (get_atomic_generic_size): Likewise. Before testing if parameter
533 has pointer type, if it has array type, call for C++
534 default_conversion to perform array-to-pointer conversion.
536 2017-09-12 Richard Biener <rguenther@suse.de>
538 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
539 for operations we cannot scalarize.
541 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
543 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
544 vectors heap vectors. Clean up comments.
545 Make visited_bbs a reference.
546 (profitable_jump_thread_path): Make GC
547 vectors heap vectors. Clean up comments.
549 (convert_and_register_jump_thread_path): Make GC vectors heap
551 (check_subpath_and_update_thread_path): Same. Clean up comments.
552 Make visited_bbs a reference.
553 (handle_phi): Abstract common code to to
554 register_jump_thread_path_if_profitable.
555 Rename VAR_BB to DEF_BB.
557 Make GC vectors heap vectors.
558 Make visited_bbs a reference.
559 (handle_assignment): Same.
560 (register_jump_thread_path_if_profitable): New.
561 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
563 Make GC vectors heap vectors. Clean up comments.
564 Make visited_bbs a reference.
565 (find_jump_threads_backwards): Make visited_bbs live in the stack.
566 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
569 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
572 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
573 words of E_DImode object are reachable by xtensa_uimm8x4 access.
575 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
577 Revert r251800 and r251799.
579 2017-09-11 Martin Jambor <mjambor@suse.cz>
582 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
583 arguments in advance.
584 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
585 use it to find predecessor edges.
586 (naive_outof_ssa): Collect vector of predecessors.
588 2017-09-08 Jason Merrill <jason@redhat.com>
590 PR c++/70029 - ICE with ref-qualifier and -flto
591 * langhooks.h (struct lang_hooks_for_types): Add
592 copy_lang_qualifiers.
593 * attribs.c (build_type_attribute_qual_variant): Use it.
594 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
596 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
597 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
599 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
602 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
603 (*mulsi3_sp64): New instruction.
604 (mulsi3): New expander.
606 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
608 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
610 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
612 * sancov.c: Include memmodel.h.
614 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
617 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
620 2017-09-07 Carl Love <cel@us.ibm.com>
622 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
623 the sldi instruction.
625 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
627 * sancov.c: Include tm_p.h.
629 2017-09-07 Jakub Jelinek <jakub@redhat.com>
632 * output.h (switch_to_other_text_partition): New declaration.
633 * varasm.c (switch_to_other_text_partition): New function.
634 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
635 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
636 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
637 to the other text partition before emitting LCL label and switch back
638 after emitting the word after it.
640 2017-09-07 Richard Biener <rguenther@suse.de>
642 * passes.def (pass_split_crit_edges): Remove instance before PRE.
643 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
644 critical edges here, after loop init.
645 (pass_data_pre): Remove PROP_no_crit_edges flags.
646 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
647 for valueization of call args to avoid leaking VN_TOP.
648 (visit_use): Assert we do not visit default defs.
649 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
650 Use error_mark_node to more easily detect leaking VN_TOP.
651 All default-defs are varying, not VN_TOP. Mark them visited.
652 (run_scc_vn): Make code match comment.
654 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
656 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
657 OPTION_MASK_FLOAT128_KEYWORD.
658 (POWERPC_MASKS): Likewise.
659 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
660 support for the -mfloat128-type option, and make -mfloat128
661 default on PowerPC Linux systems. Define or undefine
662 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
663 Define __float128 to be __ieee128 if IEEE 128-bit support is
664 enabled, or undefine it.
665 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
666 Delete defining __FLOAT128_TYPE__.
667 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
668 -mfloat128-type option and make -mfloat128 default on PowerPC
670 (TARGET_FLOAT128_TYPE): Likewise.
671 (-mfloat128-type): Likewise.
672 * config/rs6000/rs6000.c (rs6000_option_override_internal):
673 Delete the -mfloat128-type option and make -mfloat128 default on
674 PowerPC Linux systems. Always use __ieee128 to be the keyword for
675 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
676 128-bit floating point is enabled. Change tests from using
677 -mfloat128-type to -mfloat128.
678 (rs6000_mangle_type): Use the correct mangling for the __float128
679 type even if normal long double is restricted to 64-bits.
680 (floatn_mode): Enable the _Float128 type by default on VSX Linux
682 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
683 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
684 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
686 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
687 documentation for -mfloat128.
689 2017-09-06 Olivier Hainque <hainque@adacore.com>
691 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
693 2017-09-06 Wish Wu <wishwu007@gmail.com>
694 Jakub Jelinek <jakub@redhat.com>
696 * asan.c (initialize_sanitizer_builtins): Add
697 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
698 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
699 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
700 BT_FN_VOID_UINT64_PTR variables.
701 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
702 (BT_FN_VOID_UINT16_UINT16): Likewise.
703 (BT_FN_VOID_UINT32_UINT32): Likewise.
704 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
705 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
706 (BT_FN_VOID_UINT64_PTR): Likewise.
707 * common.opt (flag_sanitize_coverage): New variable.
708 (fsanitize-coverage=trace-pc): Remove.
709 (fsanitize-coverage=): Add.
710 * flag-types.h (enum sanitize_coverage_code): New enum.
711 * fold-const.c (fold_range_test): Disable non-short-circuit
712 optimization if flag_sanitize_coverage.
713 (fold_truth_andor): Likewise.
714 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
715 * opts.c (COVERAGE_SANITIZER_OPT): Define.
716 (coverage_sanitizer_opts): New array.
717 (get_closest_sanitizer_option): Add OPTS argument, handle also
718 OPT_fsanitize_coverage_.
719 (parse_sanitizer_options): Adjusted to also handle
720 OPT_fsanitize_coverage_.
721 (common_handle_option): Add OPT_fsanitize_coverage_.
722 * sancov.c (instrument_comparison, instrument_switch): New function.
723 (sancov_pass): Add trace-cmp support.
724 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
725 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
726 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
727 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
728 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
729 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
730 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
731 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
732 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
733 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
735 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
737 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
738 error. Only quit immediately if parsing is complete.
739 (BEGIN): Initialize fatal_err and parse_done.
740 (begin fpu, end fpu): Check number of arguments.
741 (begin arch, end arch): Likewise.
742 (begin cpu, end cpu): Likewise.
743 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
744 (optalias): Likewise.
746 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
748 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
749 * config/arm/arm-isa.h: Delete. Move definitions to ...
750 * arm-cpus.in: ... here. Use new feature and fgroup values.
751 * config/arm/arm.c (arm_option_override): Use lower case for feature
753 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
754 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
755 * config/arm/parsecpu.awk (END): Add new command 'isa'.
757 (print_isa_bits_for): New function.
758 (gen_isa): New function.
759 (gen_comm_data): Use print_isa_bits_for.
760 (define feature): New keyword.
761 (define fgroup): New keyword.
762 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
763 (arm-isa.h): Add rule to generate file.
764 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
765 case for feature bit names.
767 2017-09-06 Richard Biener <rguenther@suse.de>
769 * tree-ssa-pre.c (NECESSARY): Remove.
770 (create_expression_by_pieces): Do not touch pass-local flags.
771 (insert_into_preds_of_block): Likewise.
772 (do_pre_regular_insertion): Likewise.
773 (eliminate_insert): Likewise.
774 (eliminate_dom_walker::before_dom_children): Likewise.
775 (fini_eliminate): Do not look at inserted_exprs.
776 (mark_operand_necessary): Remove.
777 (remove_dead_inserted_code): Replace with simple work-list
778 algorithm based on inserted_exprs and SSA uses.
779 (pass_pre::execute): Re-order fini_eliminate and
780 remove_dead_inserted_code.
782 2017-09-06 Olivier Hainque <hainque@adacore.com>
784 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
785 for VxWorks 7. Adjust surrounding comments.
787 2017-09-06 Richard Biener <rguenther@suse.de>
789 * gimple-ssa-strength-reduction.c
790 (find_candidates_dom_walker::before_dom_children): Also allow
793 2017-09-06 Richard Biener <rguenther@suse.de>
795 PR tree-optimization/82108
796 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
797 for gap in the non-permutation SLP case.
799 2017-09-06 Martin Jambor <mjambor@suse.cz>
801 PR tree-optimization/82078
802 * tree-sra.c (sort_and_splice_var_accesses): Move call to
803 add_access_to_work_queue...
804 (build_accesses_from_assign): ...here.
805 (propagate_all_subaccesses): Make sure racc is the group
806 representative, if there is one.
808 2017-09-06 Jakub Jelinek <jakub@redhat.com>
811 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
814 2017-09-06 Richard Biener <rguenther@suse.de>
816 * gimple-ssa-strength-reduction.c
817 (find_candidates_dom_walker::before_doom_children): Use a
818 type and not a mode check.
820 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
823 * config/arm/predicates.md (arm_general_adddi_operand): Create new
825 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
827 2017-09-05 Jeff Law <law@redhat.com>
829 PR tree-optimization/64910
830 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
831 cases where we have 3 or more operands.
833 2017-09-05 Jakub Jelinek <jakub@redhat.com>
836 * omp-low.c (lower_omp_for): Recompute tree invariant if
837 gimple_omp_for_initial/final is ADDR_EXPR.
840 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
841 into gimple val before gimplification fo the COND_EXPR.
843 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
845 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
846 REGION_COPY argument.
847 (thread_through_all_blocks): Remove unused argument to
848 duplicate_thread_path.
850 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
851 Alan Hayward <alan.hayward@arm.com>
852 David Sherwood <david.sherwood@arm.com>
854 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
855 Take a scalar_mode rather than a machine_mode.
856 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
857 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
858 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
859 (aarch64_gen_adjusted_ldpstp): Likewise.
860 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
862 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
863 Alan Hayward <alan.hayward@arm.com>
864 David Sherwood <david.sherwood@arm.com>
866 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
867 Take a scalar_int_mode instead of a machine_mode.
868 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
869 (aarch64_output_scalar_simd_mov_immediate): Likewise.
870 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
871 (aarch64_simd_attr_length_rglist): Delete.
872 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
873 a scalar_int_mode instead of a machine_mode.
874 (aarch64_add_offset): Likewise.
875 (aarch64_internal_mov_immediate): Likewise
876 (aarch64_add_constant_internal): Likewise.
877 (aarch64_add_constant): Likewise.
878 (aarch64_movw_imm): Likewise.
879 (aarch64_rtx_arith_op_extract_p): Likewise.
880 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
881 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
882 Remove assert that the mode isn't a vector.
883 (aarch64_output_scalar_simd_mov_immediate): Likewise.
884 (aarch64_expand_mov_immediate): Update calls after above changes.
885 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
886 (aarch64_and_bitmask_imm): Check for scalar integer modes.
887 (aarch64_move_imm): Likewise.
888 (aarch64_can_const_movi_rtx_p): Likewise.
889 (aarch64_strip_extend): Likewise.
890 (aarch64_extr_rtx_p): Likewise.
891 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
892 a CONST_INT when the mode parameter is VOIDmode.
893 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
895 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
897 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
898 * stor-layout.c (bitwise_mode_for_mode): Likewise.
899 (bitwise_type_for_mode): Update accordingly.
901 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
903 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
904 * stor-layout.c (mode_for_size_tree): Likewise.
905 (mode_for_array): Update accordingly.
906 (layout_decl): Likewise.
907 (compute_record_mode): Likewise. Only set the mode once.
909 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
911 * target.def (get_mask_mode): Change return type to opt_mode.
913 * doc/tm.texi: Regenerate.
914 * targhooks.h (default_get_mask_mode): Return an opt_mode.
915 * targhooks.c (default_get_mask_mode): Likewise.
916 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
917 * optabs-query.c (can_vec_mask_load_store_p): Update use of
918 targetm.get_mask_mode.
919 * tree.c (build_truth_vector_type): Likewise.
921 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
923 * machmode.h (mode_for_vector): Return an opt_mode.
924 * stor-layout.c (mode_for_vector): Likewise.
925 (mode_for_int_vector): Update accordingly.
926 (layout_type): Likewise.
927 * config/i386/i386.c (emit_memmov): Likewise.
928 (ix86_expand_set_or_movmem): Likewise.
929 (ix86_expand_vector_init): Likewise.
930 (ix86_get_mask_mode): Likewise.
931 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
933 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
934 * expmed.c (extract_bit_field_1): Likewise.
935 * expr.c (expand_expr_real_2): Likewise.
936 * optabs-query.c (can_vec_perm_p): Likewise.
937 (can_vec_mask_load_store_p): Likewise.
938 * optabs.c (expand_vec_perm): Likewise.
939 * targhooks.c (default_get_mask_mode): Likewise.
940 * tree-vect-stmts.c (vectorizable_store): Likewise.
941 (vectorizable_load): Likewise.
942 (get_vectype_for_scalar_type_and_size): Likewise.
944 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
946 * machmode.h (mode_for_int_vector): New function.
947 * stor-layout.c (mode_for_int_vector): Likewise.
948 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
949 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
950 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
951 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
952 (s390_expand_vcond): Likewise.
954 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
956 * machmode.h (opt_machine_mode): New type.
957 (opt_mode<T>): Allow construction from anything that can be
959 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
960 (mode_for_size): Return an opt_machine_mode.
961 * stor-layout.c (mode_for_size): Likewise.
962 (mode_for_size_tree): Update call accordingly.
963 (bitwise_mode_for_mode): Likewise.
964 (make_fract_type): Likewise.
965 (make_accum_type): Likewise.
966 * caller-save.c (replace_reg_with_saved_mem): Update call
968 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
969 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
970 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
971 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
972 * expmed.c (extract_bit_field_1): Likewise.
973 * reload.c (get_secondary_mem): Likewise.
974 * varasm.c (assemble_integer): Likewise.
975 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
978 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
980 * machmode.h (decimal_float_mode_for_size): New function.
981 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
982 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
983 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
984 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
985 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
986 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
990 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
991 (get_builtin_sync_mode): Likewise.
992 (expand_ifn_atomic_compare_exchange): Likewise.
993 (expand_builtin_atomic_clear): Likewise.
994 (expand_builtin_atomic_test_and_set): Likewise.
995 (fold_builtin_atomic_always_lock_free): Likewise.
996 * calls.c (compute_argument_addresses): Likewise.
997 (emit_library_call_value_1): Likewise.
998 (store_one_arg): Likewise.
999 * combine.c (combine_instructions): Likewise.
1000 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1001 * config/arm/arm.c (arm_function_value): Likewise.
1002 (aapcs_allocate_return_reg): Likewise.
1003 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1004 * config/i386/i386.c (construct_container): Likewise.
1005 (ix86_gimplify_va_arg): Likewise.
1006 (ix86_expand_sse_cmp): Likewise.
1007 (emit_memmov): Likewise.
1008 (emit_memset): Likewise.
1009 (expand_small_movmem_or_setmem): Likewise.
1010 (ix86_expand_pextr): Likewise.
1011 (ix86_expand_pinsr): Likewise.
1012 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1013 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1015 * config/mips/mips.c (mips_function_value_1) Likewise.
1016 (mips_block_move_straight): Likewise.
1017 (mips_expand_ins_as_unaligned_store): Likewise.
1018 * config/powerpcspe/powerpcspe.c
1019 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1020 (rs6000_darwin64_record_arg_flush): Likewise.
1021 * config/rs6000/rs6000.c
1022 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1023 (rs6000_darwin64_record_arg_flush): Likewise.
1024 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1025 (sparc_function_value_1): Likewise.
1026 * config/spu/spu.c (adjust_operand): Likewise.
1027 (spu_emit_branch_or_set): Likewise.
1028 (arith_immediate_p): Likewise.
1029 * emit-rtl.c (gen_lowpart_common): Likewise.
1030 * expr.c (expand_expr_real_1): Likewise.
1031 * function.c (assign_parm_setup_block): Likewise.
1032 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1033 * reload1.c (alter_reg): Likewise.
1034 * stor-layout.c (mode_for_vector): Likewise.
1035 (layout_type): Likewise.
1037 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1039 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1040 (spu_convert_move): Likewise.
1041 * lower-subreg.c (resolve_simple_move): Likewise.
1043 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1046 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1047 define_insn to a define_expand.
1048 (altivec_vsum2sws_direct): New define_insn.
1049 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1051 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1053 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1056 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1061 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1062 realign stack if stack alignment needed is less than incoming
1065 2017-09-05 Marek Polacek <polacek@redhat.com>
1068 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1071 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1073 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1075 2017-09-05 Richard Biener <rguenther@suse.de>
1077 PR tree-optimization/82084
1078 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1080 2017-09-05 Richard Biener <rguenther@suse.de>
1082 PR tree-optimization/82102
1083 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1085 2017-09-05 Martin Liska <mliska@suse.cz>
1087 PR tree-optimization/82032
1088 * tree-cfg.c (generate_range_test): New function.
1089 * tree-cfg.h (generate_range_test): Declared here.
1090 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1091 (cleanup_control_expr_graph): Use it.
1092 * tree-switch-conversion.c (try_switch_expansion): Remove
1094 (emit_case_nodes): Use generate_range_test.
1096 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1099 * config/i386/i386.md (*<btsc><mode>_mask): Add
1100 TARGET_USE_BT to insn constraint.
1101 (*btr<mode>_mask): Ditto.
1103 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1105 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1106 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1108 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1111 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1112 TARGET_NEON and TARGET_IWMMXT.
1113 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1114 TARGET_NEON and TARGET_IWMMXT.
1115 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1117 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1119 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1120 (ix86_rewrite_tls_address): Ditto.
1121 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1122 (ix86_rewrite_tls_address_1): Ditto.
1123 (ix86_rewrite_tls_address): Ditto.
1124 * config/i386/predicates.md (tls_address_pattern): New predicate.
1125 * config/i386/i386.md (TLS address splitter): New splitter.
1127 2017-09-04 Richard Biener <rguenther@suse.de>
1129 PR tree-optimization/82084
1130 * fold-const.h (can_native_encode_string_p): Declare.
1131 * fold-const.c (can_native_encode_string_p): Factor out from ...
1132 (native_encode_string): ... here.
1133 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1134 vectorizing stores from constants we later cannot handle.
1136 2017-09-04 Marek Polacek <polacek@redhat.com>
1139 * doc/invoke.texi: Update -Wtautological-compare documentation.
1141 2017-09-04 Jeff Law <law@redhat.com>
1143 PR tree-optimization/64910
1144 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1145 swap the first and last operand if the last is a constant.
1147 2017-09-04 Marek Polacek <polacek@redhat.com>
1150 * convert.c (do_narrow): When sanitizing signed integer overflows,
1151 bail out for signed types.
1152 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1154 2017-09-04 Richard Biener <rguenther@suse.de>
1156 PR tree-optimization/82060
1157 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1158 Move devirtualization after stmt folding and before EH/AB/noreturn
1159 cleanup to get the stmt refs canonicalized. Use a bool instead
1160 of gimple_modified_p since that doesn't work for NOPs. Schedule
1161 NOPs generated by folding for removal.
1163 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1164 Alan Hayward <alan.hayward@arm.com>
1165 David Sherwood <david.sherwood@arm.com>
1167 * coretypes.h (pad_direction): New enum.
1168 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1169 (FUNCTION_ARG_PADDING): Likewise.
1170 * target.def (function_arg_padding): New hook.
1171 * targhooks.h (default_function_arg_padding): Declare.
1172 * targhooks.c (default_function_arg_padding): New function.
1173 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1174 (TARGET_FUNCTION_ARG_PADDING): ...this.
1175 * doc/tm.texi: Regenerate.
1176 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1177 instead of direction.
1178 (compute_argument_addresses): Likewise.
1179 (load_register_parameters): Likewise.
1180 (emit_library_call_value_1): Likewise.
1181 (store_one_arg): Use targetm.calls.function_arg_padding instead
1182 of FUNCTION_ARG_PADDING.
1183 (must_pass_in_stack_var_size_or_pad): Likewise.
1184 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1185 (emit_group_store): Likewise.
1186 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1187 instead of FUNCTION_ARG_PADDING.
1188 (emit_push_insn): Likewise, and propagate enum change throughout
1190 * function.h (direction): Delete.
1191 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1193 * function.c (assign_parm_find_stack_rtl): Likewise.
1194 (assign_parm_setup_block_p): Likewise.
1195 (assign_parm_setup_block): Likewise.
1196 (gimplify_parameters): Likewise.
1197 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1198 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1200 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1201 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1202 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1203 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1204 (aarch64_function_arg_padding): ...this new function.
1205 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1206 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1207 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1208 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1209 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1210 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1211 (arm_pad_arg_upward): Replace with...
1212 (arm_function_arg_padding): ...this new function.
1213 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1215 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1216 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1217 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1218 (ia64_hpux_function_arg_padding): Replace with...
1219 (ia64_function_arg_padding): ...this new function. Use pad_direction
1220 instead of direction. Check for TARGET_HPUX.
1221 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1222 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1223 (iq2000_function_arg_padding): New function.
1224 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1225 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1226 (mips_function_arg_padding): ...this new function.
1227 (mips_pad_reg_upward): Update accordingly.
1228 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1229 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1230 targetm.calls.function_arg_padding.
1231 (FUNCTION_ARG_PADDING): Delete.
1232 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1233 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1234 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1235 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1236 (nios2_block_reg_padding): Return pad_direction instead of direction.
1237 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1238 instead of direction.
1239 (nios2_function_arg_padding): Likewise. Make static.
1240 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1241 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1242 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1243 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1244 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1245 (pa_function_arg_padding): Make static. Return pad_direction instead
1247 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1248 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1249 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1250 instead of direction. Use targetm.calls.function_arg_padding.
1251 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1252 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1253 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1254 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1255 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1257 (function_arg_padding): Rename to...
1258 (rs6000_function_arg_padding): ...this. Make static. Return
1259 pad_direction instead of direction.
1260 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1261 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1262 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1263 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1264 instead of direction. Use targetm.calls.function_arg_padding.
1265 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1266 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1267 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1268 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1269 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1270 (function_arg_padding): Rename to...
1271 (rs6000_function_arg_padding): ...this. Make static. Return
1272 pad_direction instead of direction.
1273 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1274 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1275 * config/s390/s390.c (s390_function_arg_padding): New function.
1276 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1277 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1278 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1279 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1280 (function_arg_padding): Rename to...
1281 (sparc_function_arg_padding): ...this. Make static. Return
1282 pad_direction instead of direction.
1283 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1284 * config/spu/spu.c (spu_function_arg_padding): New function.
1285 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1286 * system.h (FUNCTION_ARG_PADDING): Poison.
1288 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1289 Alan Hayward <alan.hayward@arm.com>
1290 David Sherwood <david.sherwood@arm.com>
1292 * target.def (modes_tieable_p): New hook.
1293 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1294 (TARGET_MODES_TIEABLE_P): ...this.
1295 * doc/tm.texi.in: Regenerate.
1296 * hooks.h (hook_bool_mode_mode_true): Declare.
1297 * hooks.c (hook_bool_mode_mode_true): New function.
1298 * combine.c (subst): Use targetm.modes_tieable_p instead of
1300 * dse.c (find_shift_sequence): Likewise.
1301 * expmed.c (extract_low_bits): Likewise.
1302 * lower-subreg.c: Include target.h.
1303 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1305 * rtlanal.c (rtx_cost): Likewise.
1306 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1307 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1308 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1309 (TARGET_MODES_TIEABLE_P): Redefine.
1310 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1311 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1312 (TARGET_MODES_TIEABLE_P): Redefine.
1313 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1314 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1315 (arc_modes_tieable_p): New function.
1316 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1317 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1318 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1319 (arm_modes_tieable_p): Make static.
1320 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1321 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1322 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1323 (TARGET_MODES_TIEABLE_P): Redefine.
1324 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1325 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1326 (TARGET_MODES_TIEABLE_P): Redefine.
1327 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1328 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1329 (cr16_modes_tieable_p): New function.
1330 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1331 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1332 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1333 (TRULY_NOOP_TRUNCATION): Update comment.
1334 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1335 (TRULY_NOOP_TRUNCATION): Update comment.
1336 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1337 (frv_modes_tieable_p): New function.
1338 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1339 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1340 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1341 (TARGET_MODES_TIEABLE_P): Redefine.
1342 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1343 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1344 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1345 (TARGET_MODES_TIEABLE_P): Redefine.
1346 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1347 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1348 (ia64_modes_tieable_p): New function.
1349 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1350 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1351 (iq2000_modes_tieable_p): New function.
1352 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1353 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1354 (lm32_modes_tieable_p): New function.
1355 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1356 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1357 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1358 (TARGET_MODES_TIEABLE_P): Redefine.
1359 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1360 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1361 (m32r_modes_tieable_p): New function.
1362 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1363 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1364 (m68k_modes_tieable_p): New function.
1365 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1366 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1367 (mcore_modes_tieable_p): New function.
1368 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1369 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1371 (TARGET_MODES_TIEABLE_P): Redefine.
1372 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1373 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1374 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1375 (TARGET_MODES_TIEABLE_P): Redefine.
1376 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1377 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1378 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1379 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1380 (mn10300_modes_tieable_p): ...this and make static.
1381 (TARGET_MODES_TIEABLE_P): Redefine.
1382 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1383 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1384 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1385 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1386 (msp430_modes_tieable_p): Make static.
1387 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1388 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1389 (TARGET_MODES_TIEABLE_P): Redefine.
1390 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1391 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1392 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1393 (TARGET_MODES_TIEABLE_P): Redefine.
1394 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1395 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1396 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1397 (TARGET_MODES_TIEABLE_P): Redefine.
1398 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1399 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1400 (pdp11_modes_tieable_p): New function.
1401 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1402 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1403 (rs6000_modes_tieable_p): New function.
1404 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1405 * config/powerpcspe/powerpcspe.md: Update comment.
1406 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1407 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1408 (TARGET_MODES_TIEABLE_P): Redefine.
1409 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1410 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1411 (rl78_modes_tieable_p): New function.
1412 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1413 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1414 (rs6000_modes_tieable_p): New function.
1415 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1416 * config/rs6000/rs6000.md: Update comment.
1417 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1418 * config/rx/rx.c (rx_modes_tieable_p): New function.
1419 (TARGET_MODES_TIEABLE_P): Redefine.
1420 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1421 * config/s390/s390.c (s390_modes_tieable_p): New function.
1422 (TARGET_MODES_TIEABLE_P): Redefine.
1423 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1424 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1425 (sh_modes_tieable_p): New function.
1426 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1427 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1428 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1429 (sparc_modes_tieable_p): Make static.
1430 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1431 * config/spu/spu.c (spu_modes_tieable_p): New function.
1432 (TARGET_MODES_TIEABLE_P): Redefine.
1433 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1434 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1435 (TARGET_MODES_TIEABLE_P): Redefine.
1436 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1437 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1438 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1439 * config/v850/v850.c (v850_modes_tieable_p): New function.
1440 (TARGET_MODES_TIEABLE_P): Redefine.
1441 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1442 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1443 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1444 (visium_modes_tieable_p): New function.
1445 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1446 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1447 (xtensa_modes_tieable_p): New function.
1448 * system.h (MODES_TIEABLE_P): Poison.
1450 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1451 Alan Hayward <alan.hayward@arm.com>
1452 David Sherwood <david.sherwood@arm.com>
1454 * target.def (hard_regno_mode_ok): New hook.
1455 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1456 (TARGET_HARD_REGNO_MODE_OK): ...this.
1457 * doc/tm.texi.in: Regenerate.
1458 * hooks.h (hook_bool_uint_mode_true): Declare.
1459 * hooks.c (hook_bool_uint_mode_true): New function.
1460 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1462 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1463 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1464 instead of HARD_REGNO_MODE_OK.
1465 * caller-save.c: Include target.h.
1466 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1468 * combine.c (can_combine_p): Likewise.
1469 (combinable_i3pat): Likewise.
1470 (can_change_dest_mode): Likewise.
1471 * expr.c (init_expr_target): Likewise.
1472 (convert_move): Likewise.
1473 (convert_modes): Likewise.
1474 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1475 (setup_prohibited_mode_move_regs): Likewise.
1476 * ira.h (target_ira): Likewise.
1477 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1478 * lra-constraints.c (process_alt_operands): Likewise.
1479 (split_reg): Likewise.
1480 * recog.c (peep2_find_free_register): Likewise.
1481 * ree.c (combine_reaching_defs): Likewise.
1482 * regcprop.c (maybe_mode_change): Likewise.
1483 * reginfo.c (init_reg_sets_1): Likewise.
1484 (choose_hard_reg_mode): Likewise.
1485 (simplifiable_subregs): Likewise.
1486 * regrename.c (check_new_reg_p): Likewise.
1487 * reload.c (find_valid_class): Likewise.
1488 (find_valid_class_1): Likewise.
1489 (reload_inner_reg_of_subreg): Likewise.
1490 (push_reload): Likewise.
1491 (combine_reloads): Likewise.
1492 (find_dummy_reload): Likewise.
1493 (find_reloads): Likewise.
1494 * reload1.c (find_reg): Likewise.
1495 (set_reload_reg): Likewise.
1496 (allocate_reload_reg): Likewise.
1497 (choose_reload_regs): Likewise.
1498 (reload_adjust_reg_for_temp): Likewise.
1499 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1500 (simplify_subreg_regno): Likewise.
1501 * sel-sched.c (init_regs_for_mode): Likewise.
1502 * varasm.c (make_decl_rtl): Likewise.
1503 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1504 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1506 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1507 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1508 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1509 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1510 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1511 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1512 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1513 (arc_mode_class): Delete.
1514 (HARD_REGNO_MODE_OK): Delete.
1515 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1516 (arc_hard_regno_mode_ok): Rename old array to...
1517 (arc_hard_regno_mode_ok_modes): ...this.
1518 (arc_conditional_register_usage): Update accordingly.
1519 (arc_mode_class): Make static.
1520 (arc_hard_regno_mode_ok): New function.
1521 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1522 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1523 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1524 (arm_hard_regno_mode_ok): Make static.
1525 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1527 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1528 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1529 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1531 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1532 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1533 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1534 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1535 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1537 * config/bfin/predicates.md (valid_reg_operand): Use
1538 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1539 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1540 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1542 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1543 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1544 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1545 (cr16_hard_regno_mode_ok): Make static and return a bool.
1546 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1547 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1548 (cris_hard_regno_mode_ok): New function.
1549 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1550 (epiphany_mode_class): Delete.
1551 (HARD_REGNO_MODE_OK): Delete.
1552 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1553 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1554 (hard_regno_mode_ok): Rename to...
1555 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1556 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1557 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1559 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1560 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1561 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1562 (frv_hard_regno_mode_ok): Make static and return a bool.
1563 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1565 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1566 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1567 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1568 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1570 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1571 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1572 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1573 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1576 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1577 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1578 (ia64_hard_regno_mode_ok): New function.
1579 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1580 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1581 (iq2000_hard_regno_mode_ok): New function.
1582 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1583 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1584 (lm32_hard_regno_mode_ok): New function.
1585 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1586 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1587 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1588 instead of HARD_REGNO_MODE_OK.
1589 (m32c_hard_regno_ok): Rename to...
1590 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1591 (m32c_cannot_change_mode_class): Update accordingly.
1592 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1593 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1594 (m32r_mode_class): Delete.
1595 (HARD_REGNO_MODE_OK): Delete.
1596 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1597 (m32r_hard_regno_mode_ok): Rename to...
1598 (m32r_hard_regno_modes): ...this.
1599 (m32r_mode_class): Make static.
1600 (m32r_hard_regno_mode_ok): New function.
1601 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1602 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1603 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1604 (m68k_hard_regno_mode_ok): Make static.
1605 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1606 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1607 (mcore_hard_regno_mode_ok): New function.
1608 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1609 (HARD_REGNO_MODE_OK): Delete.
1610 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1612 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1613 (microblaze_hard_regno_mode_ok): New function.
1614 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1615 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1616 (mips_hard_regno_mode_ok): Delete.
1617 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1618 (mips_hard_regno_mode_ok_p): ...this and make static.
1619 (mips_hard_regno_mode_ok_p): Rename to...
1620 (mips_hard_regno_mode_ok_uncached): ...this.
1621 (mips_hard_regno_mode_ok): New function.
1622 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1623 of HARD_REGNO_MODE_OK.
1624 (mips_option_override): Update after above name changes.
1625 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1626 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1627 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1628 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1629 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1630 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1631 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1632 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1633 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1634 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1635 (msp430_hard_regno_mode_ok): Make static and return a bool.
1636 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1637 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1638 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1640 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1641 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1642 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1643 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1644 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1645 (PA_HARD_REGNO_MODE_OK): ...this
1646 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1647 (PA_HARD_REGNO_MODE_OK): ...this.
1648 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1649 (pa_hard_regno_mode_ok): New function.
1650 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1651 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1652 (pdp11_hard_regno_mode_ok): New function.
1653 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1654 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1656 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1658 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1659 (rs6000_hard_regno_mode_ok): Rename to...
1660 (rs6000_hard_regno_mode_ok_uncached): ...this.
1661 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1662 (rs6000_hard_regno_mode_ok): New function.
1663 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1664 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1665 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1666 (riscv_hard_regno_mode_ok): ...this and make static.
1667 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1668 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1669 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1670 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1671 (rl78_hard_regno_mode_ok): Make static and return bool.
1672 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1673 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1675 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1677 (rs6000_hard_regno_mode_ok): Rename to...
1678 (rs6000_hard_regno_mode_ok_uncached): ...this.
1679 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1680 (rs6000_hard_regno_mode_ok): New function.
1681 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1682 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1683 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1684 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1685 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1686 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1687 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1688 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1689 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1690 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1691 (sh_hard_regno_mode_ok): Make static.
1692 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1693 instead of HARD_REGNO_MODE_OK.
1694 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1695 (sparc_mode_class): Delete.
1696 (HARD_REGNO_MODE_OK): Delete.
1697 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1698 (hard_regno_mode_classes): Make static.
1699 (sparc_mode_class): Likewise.
1700 (sparc_hard_regno_mode_ok): New function.
1701 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1702 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1703 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1705 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1706 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1707 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1708 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1709 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1711 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1712 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1713 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1714 (visium_hard_regno_mode_ok): New function.
1715 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1716 instead of HARD_REGNO_MODE_OK.
1717 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1718 (HARD_REGNO_MODE_OK): Delete.
1719 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1720 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1721 (xtensa_option_override): Update accordingly.
1722 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1723 (xtensa_hard_regno_mode_ok): New function.
1724 * system.h (HARD_REGNO_MODE_OK): Poison.
1726 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1727 Alan Hayward <alan.hayward@arm.com>
1728 David Sherwood <david.sherwood@arm.com>
1730 * target.def (hard_regno_call_part_clobbered): New hook.
1731 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1732 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1733 * doc/tm.texi: Regenerate.
1734 * hooks.h (hook_bool_uint_mode_false): Declare.
1735 * hooks.c (hook_bool_uint_mode_false): New function.
1736 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1737 * cselib.c (cselib_process_insn): Use
1738 targetm.hard_regno_call_part_clobbered instead of
1739 HARD_REGNO_CALL_PART_CLOBBERED.
1740 * ira-conflicts.c (ira_build_conflicts): Likewise.
1741 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1742 * lra-constraints.c (need_for_call_save_p): Likewise.
1743 * lra-lives.c: Include target.h.
1744 (check_pseudos_live_through_calls): Use
1745 targetm.hard_regno_call_part_clobbered instead of
1746 HARD_REGNO_CALL_PART_CLOBBERED.
1747 * regcprop.c: Include target.h.
1748 (copyprop_hardreg_forward_1): Use
1749 targetm.hard_regno_call_part_clobbered instead of
1750 HARD_REGNO_CALL_PART_CLOBBERED.
1751 * reginfo.c (choose_hard_reg_mode): Likewise.
1752 * regrename.c (check_new_reg_p): Likewise.
1753 * reload.c (find_equiv_reg): Likewise.
1754 * reload1.c (emit_reload_insns): Likewise.
1755 * sched-deps.c (deps_analyze_insn): Likewise.
1756 * sel-sched.c (init_regs_for_mode): Likewise.
1757 (mark_unavailable_hard_regs): Likewise.
1758 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1759 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1760 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1762 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1763 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1764 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1766 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1768 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1769 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1770 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1772 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1773 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1774 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1776 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1777 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1779 * config/powerpcspe/powerpcspe.c
1780 (rs6000_hard_regno_call_part_clobbered): New function.
1781 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1782 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1783 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1785 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1786 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1787 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1789 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1790 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1791 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1793 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1794 Alan Hayward <alan.hayward@arm.com>
1795 David Sherwood <david.sherwood@arm.com>
1797 * rtl.h (subreg_memory_offset): Declare.
1798 * emit-rtl.c (subreg_memory_offset): New function.
1799 * expmed.c (store_bit_field_1): Use it.
1800 * expr.c (undefined_operand_subword_p): Likewise.
1801 * simplify-rtx.c (simplify_subreg): Likewise.
1803 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1805 PR rtl-optimization/57448
1808 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1809 using atomic_load pattern.
1810 (expand_atomic_store): Likewise.
1812 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1815 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1816 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1817 handling. Use replace_call_with_value with NULL instead of
1818 gsi_replace, unlink_stmt_vdef and release_defs.
1820 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1823 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1825 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1828 * rtl.h (emit_library_call_value_1): Declare.
1829 (emit_library_call): Replace declaration with a series of overloads.
1830 Remove the parameter count argument.
1831 (emit_library_call_value): Likewise.
1832 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1833 with an "rtx_mode_t *".
1834 (emit_library_call_value): Delete.
1835 (emit_library_call): Likewise.
1836 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1837 (asan_emit_allocas_unpoison): Likewise.
1838 * builtins.c (expand_builtin_powi): Likewise.
1839 (expand_asan_emit_allocas_unpoison): Likewise.
1840 * cfgexpand.c (expand_main_function): Likewise.
1841 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1842 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1843 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1844 * config/arm/arm.c (arm_trampoline_init): Likewise.
1845 (arm_call_tls_get_addr): Likewise.
1846 (arm_expand_divmod_libfunc): Likewise.
1847 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1848 (smulsi3_highpart): Likewise.
1849 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1850 (c6x_expand_compare): Likewise.
1851 (c6x_expand_movmem): Likewise.
1852 * config/frv/frv.c (frv_trampoline_init): Likewise.
1853 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1854 (ix86_expand_divmod_libfunc): Likewise.
1855 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1856 (ia64_expand_compare): Likewise.
1857 (ia64_profile_hook): Likewise.
1858 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1859 (nonlocal_goto): Likewise.
1860 (restore_stack_nonlocal): Likewise.
1861 * config/m32r/m32r.c (block_move_call): Likewise.
1862 (m32r_trampoline_init): Likewise.
1863 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1864 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1865 (m68k_call_m68k_read_tp): Likewise.
1866 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1867 (microblaze_expand_divide): Likewise.
1868 * config/mips/mips.h (mips_args): Likewise.
1869 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1870 (MIPS_ICACHE_SYNC): Likewise.
1871 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1872 (nios2_trampoline_init): Likewise.
1873 * config/pa/pa.c (hppa_tls_call): Likewise.
1874 (pa_trampoline_init): Likewise.
1875 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1876 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1877 (expand_strn_compare): Likewise.
1878 (rs6000_generate_compare): Likewise.
1879 (rs6000_expand_float128_convert): Likewise.
1880 (output_profile_hook): Likewise.
1881 (rs6000_trampoline_init): Likewise.
1882 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1883 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1884 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1885 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1886 (rs6000_generate_compare): Likewise.
1887 (rs6000_expand_float128_convert): Likewise.
1888 (output_profile_hook): Likewise.
1889 (rs6000_trampoline_init): Likewise.
1890 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1891 * config/sh/sh.c (sh_trampoline_init): Likewise.
1892 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1893 (sparc_emit_float_lib_cmp): Likewise.
1894 (sparc32_initialize_trampoline): Likewise.
1895 (sparc64_initialize_trampoline): Likewise.
1896 (sparc_profile_hook): Likewise.
1897 * config/spu/spu.c (ea_load_store): Likewise.
1898 * config/spu/spu.md (floatunssidf2): Likewise.
1899 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1900 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1901 * config/visium/visium.c (expand_block_move_4): Likewise.
1902 (expand_block_move_2): Likewise.
1903 (expand_block_move_1): Likewise.
1904 (expand_block_set_4): Likewise.
1905 (expand_block_set_2): Likewise.
1906 (expand_block_set_1): Likewise.
1907 (visium_trampoline_init): Likewise.
1908 (visium_profile_hook): Likewise.
1909 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1910 (xtensa_setup_frame_addresses): Likewise.
1911 (xtensa_trampoline_init): Likewise.
1912 * except.c (sjlj_emit_function_enter): Likewise.
1913 (sjlj_emit_function_exit): Likewise.
1914 * explow.c (allocate_dynamic_stack_space): Likewise.
1915 (probe_stack_range): Likewise.
1916 * expr.c (convert_mode_scalar): Likewise.
1917 * optabs.c (expand_binop): Likewise.
1918 (expand_twoval_binop_libfunc): Likewise.
1919 (expand_unop): Likewise.
1920 (prepare_cmp_insn): Likewise.
1921 (prepare_float_lib_cmp): Likewise.
1922 (expand_float): Likewise.
1923 (expand_fix): Likewise.
1924 (expand_fixed_convert): Likewise.
1925 (maybe_emit_sync_lock_test_and_set): Likewise.
1926 (expand_atomic_compare_and_swap): Likewise.
1927 (expand_mem_thread_fence): Likewise.
1928 (expand_atomic_fetch_op): Likewise.
1930 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1932 * doc/generic.texi (OpenACC): Adjust URL.
1933 * doc/invoke.texi (C Dialect Options): Ditto.
1935 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1937 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1938 predicate for operand 1. Add (m,<S>) constraint.
1939 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1940 Prevent memory operand 1 with register operand 2.
1942 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1944 PR rtl-optimization/82024
1945 * combine.c (try_combine): If the combination result is a PARALLEL,
1946 and we only need to retain the SET in there that would be placed
1947 at I2, check that we can place that at I3 instead, before doing so.
1949 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1952 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1954 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1955 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1958 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1959 Jeff Law <law@redhat.com>
1961 * varasm.c (bss_initializer_p): Do not put constants into .bss
1962 (categorize_decl_for_section): Handle bss_initializer_p returning
1963 false when DECL_INITIAL is NULL.
1965 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1968 * config/s390/s390.c (s390_can_inline_p): New function.
1970 2017-09-01 Jeff Law <law@redhat.com>
1972 PR tree-optimization/82052
1973 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1974 Always initialize the returned slot after a hash table miss
1975 when INSERT is true.
1977 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1979 * config/s390/s390.md (mem_signal_fence): Remove.
1980 * doc/md.texi (mem_signal_fence): Remove.
1981 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1983 * target-insns.def (mem_signal_fence): Remove.
1985 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1988 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1991 * asan.c (create_odr_indicator): Strip name encoding from assembler
1992 name before appending it after __odr_asan_.
1994 2017-09-01 Martin Liska <mliska@suse.cz>
1996 PR tree-optimization/82059
1997 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1998 frequency only when an edge is redirected.
2000 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2002 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2003 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2004 (arc_conditional_register_usage): Remove ARC600 lp_count
2006 (arc_file_start): Emit Tag_ARC_CPU_variation.
2007 (arc_can_use_doloop_p): New conditions to use ZOLs.
2008 (hwloop_fail): New function.
2009 (hwloop_optimize): Likewise.
2010 (hwloop_pattern_reg): Likewise.
2011 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2012 (arc_reorg_loops): New function, calls reorg_loops.
2013 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2014 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2016 (arc_loop_hazard): Remove function, functionality moved into
2018 (arc_hazard): Remove arc_loop_hazard call.
2019 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2020 into hwloop_optimize.
2021 (arc_label_align): Remove ZOL handling.
2022 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2023 * config/arc/arc.md (doloop_begin): Remove pattern.
2024 (doloop_begin_i): Likewise.
2025 (doloop_end_i): Likewise.
2026 (doloop_fallback): Likewise.
2027 (doloop_fallback_m): Likewise.
2028 (doloop_end): Reimplement expand.
2029 (arc_lp): New pattern for LP instruction.
2030 (loop_end): New pattern.
2031 (loop_fail): Likewise.
2032 (decrement_and_branch_until_zero): Likewise.
2033 * config/arc/arc.opt (mlpc-width): New option.
2034 * doc/invoke.texi (mlpc-width): Document option.
2036 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2038 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2039 (arc_ccfsm_advance): Fix checking for delay slots.
2040 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2042 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2044 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2046 (movhi_insn): Update store instruction constraint which are saving
2047 6-bit short immediates.
2048 (movsi_insn): Consider also short scaled load operations.
2049 (zero_extendhisi2_i): Use Usd constraint instead of T.
2050 (extendhisi2_i): Add q constraint.
2051 (arc_clzsi2): Add type and length attributes.
2052 (arc_ctzsi2): Likewise.
2053 * config/arc/constraints.md (Usc): Update constraint, the
2054 assembler can parse two relocations for a single instruction.
2056 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2058 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2059 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2061 2017-08-31 Olivier Hainque <hainque@adacore.com>
2063 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2064 match as powerpc-wrs-vxworks*.
2066 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2068 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2069 register constraint for by-element operand.
2070 (aarch64_mls_elt_merge<mode>): Likewise.
2072 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2074 * config/arc/arc.c (arc_can_follow_jump): Check for short
2077 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2079 * config.gcc: Use g.opt for arc.
2080 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2081 functionality moved to ...
2082 (legitimate_scaled_address_p): New function, ...here.
2083 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2084 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2085 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2087 (arc_override_options): Handle G option.
2088 (arc_output_pic_addr_const): Correct function definition.
2089 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2090 (arc_decl_anon_ns_mem_p): Delete.
2091 (arc_in_small_data_p): Overhaul this function to take into
2092 consideration the value given via G option.
2093 (arc_rewrite_small_data_1): Renamed and corrected old
2094 arc_rewrite_small_data function.
2095 (arc_rewrite_small_data): New function.
2096 (small_data_pattern): Don't use pic_offset_table_rtx.
2097 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2098 * config/arc/simdext.md (movmisalignv2hi): Use
2099 prepare_move_operands function.
2101 (movmisalign*): Likewise.
2102 * doc/invoke.texi (ARC options): Document -G option.
2104 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2106 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2108 * config/arc/arc.c (arc_print_operand): Output scalled address for
2109 sdata whenever is possible.
2110 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2111 load/stores are available.
2112 (compact_sda_memory_operand): Check for the alignment required by
2113 code density instructions.
2114 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2116 * config/arc/constraints.md (Usd): Update constraint.
2117 (Us0): New constraint.
2118 (Usc): Update constraint.
2120 2017-08-31 Richard Biener <rguenther@suse.de>
2123 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2126 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2128 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2129 Resize type_signature.
2131 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2132 Alan Hayward <alan.hayward@arm.com>
2133 David Sherwood <david.sherwood@arm.com>
2135 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2136 subregs whose inner modes can be stored in GPRs.
2137 (aarch64_classify_index): Likewise.
2139 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2140 Alan Hayward <alan.hayward@arm.com>
2141 David Sherwood <david.sherwood@arm.com>
2143 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2144 (V_INT_EQUIV): ...this.
2145 (v_cmp_result): Rename to...
2146 (v_int_equiv): ...this.
2147 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2148 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2149 (copysign<mode>3): Likewise.
2150 (aarch64_simd_bsl<mode>_internal): Likewise.
2151 (aarch64_simd_bsl<mode>): Likewise.
2152 (vec_cmp<mode><mode>): Likewise.
2153 (vcond<mode><mode>): Likewise.
2154 (vcond<v_cmp_mixed><mode>): Likewise.
2155 (vcondu<mode><v_cmp_mixed>): Likewise.
2156 (aarch64_cm<optab><mode>): Likewise.
2157 (aarch64_cmtst<mode>): Likewise.
2158 (aarch64_fac<optab><mode>): Likewise.
2159 (vec_perm_const<mode>): Likewise.
2160 (vcond_mask_<mode><v_cmp_result>): Rename to...
2161 (vcond_mask_<mode><v_int_equiv>): ...this.
2162 (vec_cmp<mode><v_cmp_result>): Rename to...
2163 (vec_cmp<mode><v_int_equiv>): ...this.
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-modes.def: Remove 32-, 48- and 64-byte
2171 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2172 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2173 (UNSPEC_LD4_DREG): New unspecs.
2174 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2175 (aarch64_ld2<mode>_dreg_be): Replace with...
2176 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2178 (aarch64_ld3<mode>_dreg_le)
2179 (aarch64_ld3<mode>_dreg_be): Replace with...
2180 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2182 (aarch64_ld4<mode>_dreg_le)
2183 (aarch64_ld4<mode>_dreg_be): Replace with...
2184 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2187 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2189 PR tree-optimization/81987
2190 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2191 insert an initializer in a location not dominated by the stride
2194 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2196 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2197 on the entire header of the finally block in the fallthru case.
2199 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2201 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2203 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2205 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2206 rs6000_emit_move_from_cr and call renamed function.
2207 (rs6000_emit_prologue): Call renamed functions.
2208 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2209 movesi_from_cr, remove volatile CRs.
2211 2017-08-30 Jon Beniston <jon@beniston.com>
2212 Richard Biener <rguenther@suse.de>
2214 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2215 of VECTOR_MODE_P check.
2216 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2217 element vector types.
2219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2221 * df.h (df_read_modify_subreg_p): Remove in favor of...
2222 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2223 const_rtx instead of an rtx.
2224 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2225 * df-problems.c (df_word_lr_mark_ref): Likewise.
2226 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2227 (mark_pseudo_reg_dead): Likewise.
2228 (mark_ref_dead): Likewise.
2229 * reginfo.c (init_subregs_of_mode): Likewise.
2230 * sched-deps.c (sched_analyze_1): Likewise.
2231 * df-scan.c (df_def_record_1): Likewise.
2232 (df_uses_record): Likewise.
2233 (df_read_modify_subreg_p): Remove in favor of...
2234 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2235 const_rtx instead of an rtx.
2237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2238 Alan Hayward <alan.hayward@arm.com>
2239 David Sherwood <david.sherwood@arm.com>
2241 * rtl.h (partial_subreg_p): New function.
2242 * caller-save.c (save_call_clobbered_regs): Use it.
2243 * calls.c (expand_call): Likewise.
2244 * combine.c (combinable_i3pat): Likewise.
2245 (simplify_set): Likewise.
2246 (make_extraction): Likewise.
2247 (make_compound_operation_int): Likewise.
2248 (gen_lowpart_or_truncate): Likewise.
2249 (force_to_mode): Likewise.
2250 (make_field_assignment): Likewise.
2251 (reg_truncated_to_mode): Likewise.
2252 (record_truncated_value): Likewise.
2253 (move_deaths): Likewise.
2254 * cse.c (record_jump_cond): Likewise.
2255 (cse_insn): Likewise.
2256 * cselib.c (cselib_lookup_1): Likewise.
2257 * expmed.c (extract_bit_field_using_extv): Likewise.
2258 * function.c (assign_parm_setup_reg): Likewise.
2259 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2260 * ira-build.c (create_insn_allocnos): Likewise.
2261 * lra-coalesce.c (merge_pseudos): Likewise.
2262 * lra-constraints.c (match_reload): Likewise.
2263 (simplify_operand_subreg): Likewise.
2264 (curr_insn_transform): Likewise.
2265 * lra-lives.c (process_bb_lives): Likewise.
2266 * lra.c (new_insn_reg): Likewise.
2267 (lra_substitute_pseudo): Likewise.
2268 * regcprop.c (mode_change_ok): Likewise.
2269 (maybe_mode_change): Likewise.
2270 (copyprop_hardreg_forward_1): Likewise.
2271 * reload.c (push_reload): Likewise.
2272 (find_reloads): Likewise.
2273 (find_reloads_subreg_address): Likewise.
2274 * reload1.c (alter_reg): Likewise.
2275 (eliminate_regs_1): Likewise.
2276 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2278 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2280 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2281 back to if statements, including unpack.
2283 2017-08-30 Martin Liska <mliska@suse.cz>
2286 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2288 (func_checker::compare_asm_inputs_outputs): ... this function.
2289 (func_checker::compare_gimple_asm): Use the function to compare
2290 also ASM constrains.
2291 * ipa-icf-gimple.h: Rename the function.
2293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2294 Alan Hayward <alan.hayward@arm.com>
2295 David Sherwood <david.sherwood@arm.com>
2297 * coretypes.h (complex_mode): New type.
2298 * gdbhooks.py (build_pretty_printer): Handle it.
2299 * machmode.h (complex_mode): New class.
2300 (complex_mode::includes_p): New function.
2301 (is_complex_int_mode): Likewise.
2302 (is_complex_float_mode): Likewise.
2303 * genmodes.c (get_mode_class): Handle complex mode classes.
2304 * function.c (expand_function_end): Use is_complex_int_mode.
2306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2307 Alan Hayward <alan.hayward@arm.com>
2308 David Sherwood <david.sherwood@arm.com>
2310 * coretypes.h (scalar_mode_pod): New typedef.
2311 * gdbhooks.py (build_pretty_printer): Handle it.
2312 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2313 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2314 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2315 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * machmode.h (mode_for_vector): Take a scalar_mode instead
2324 * stor-layout.c (mode_for_vector): Likewise.
2325 * explow.c (promote_mode): Use as_a <scalar_mode>.
2326 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2329 Alan Hayward <alan.hayward@arm.com>
2330 David Sherwood <david.sherwood@arm.com>
2332 * target.def (preferred_simd_mode): Take a scalar_mode
2333 instead of a machine_mode.
2334 * targhooks.h (default_preferred_simd_mode): Likewise.
2335 * targhooks.c (default_preferred_simd_mode): Likewise.
2336 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2337 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2338 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2339 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2340 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2341 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2342 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2343 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2345 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2346 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2347 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2348 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2349 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2350 * doc/tm.texi: Regenerate.
2351 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2355 Alan Hayward <alan.hayward@arm.com>
2356 David Sherwood <david.sherwood@arm.com>
2358 * target.def (scalar_mode_supported_p): Take a scalar_mode
2359 instead of a machine_mode.
2360 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2361 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2362 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2363 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2364 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2365 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2366 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2367 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2368 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2369 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2370 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2371 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2372 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2373 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2375 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2376 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2377 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2378 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2379 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2381 * doc/tm.texi: Regenerate.
2383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2384 Alan Hayward <alan.hayward@arm.com>
2385 David Sherwood <david.sherwood@arm.com>
2387 * coretypes.h (opt_scalar_mode): New typedef.
2388 * gdbhooks.py (build_pretty_printers): Handle it.
2389 * machmode.h (mode_iterator::get_2xwider): Add overload for
2391 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2393 * expr.c (convert_mode_scalar): Likewise.
2394 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2395 * optabs.c (expand_float): Likewise.
2396 (expand_fix): Likewise.
2397 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2400 Alan Hayward <alan.hayward@arm.com>
2401 David Sherwood <david.sherwood@arm.com>
2403 * optabs.c (expand_float): Explicitly check for scalars before
2404 using a branching expansion.
2405 (expand_fix): Likewise.
2407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2408 Alan Hayward <alan.hayward@arm.com>
2409 David Sherwood <david.sherwood@arm.com>
2411 * expr.c (convert_mode): Split scalar handling out into...
2412 (convert_mode_scalar): ...this new function. Treat the modes
2415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2416 Alan Hayward <alan.hayward@arm.com>
2417 David Sherwood <david.sherwood@arm.com>
2419 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2421 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2424 Alan Hayward <alan.hayward@arm.com>
2425 David Sherwood <david.sherwood@arm.com>
2427 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2428 rather than a machine_mode.
2429 (fixed_from_string): Likewise.
2430 (fixed_convert): Likewise.
2431 (fixed_convert_from_int): Likewise.
2432 (fixed_convert_from_real): Likewise.
2433 (real_convert_from_fixed): Likewise.
2434 * fixed-value.c (fixed_from_double_int): Likewise.
2435 (fixed_from_string): Likewise.
2436 (fixed_convert): Likewise.
2437 (fixed_convert_from_int): Likewise.
2438 (fixed_convert_from_real): Likewise.
2439 (real_convert_from_fixed): Likewise.
2440 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2443 Alan Hayward <alan.hayward@arm.com>
2444 David Sherwood <david.sherwood@arm.com>
2446 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2447 of separate mode class checks. Do not allow vector modes here.
2448 (immed_wide_int_const): Use as_a <scalar_mode>.
2449 * explow.c (trunc_int_for_mode): Likewise.
2450 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2451 (wi::shwi): Likewise.
2452 (wi::min_value): Likewise.
2453 (wi::max_value): Likewise.
2454 * dwarf2out.c (loc_descriptor): Likewise.
2455 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2459 Alan Hayward <alan.hayward@arm.com>
2460 David Sherwood <david.sherwood@arm.com>
2462 * tree.h (SCALAR_TYPE_MODE): New macro.
2463 * expr.c (expand_expr_addr_expr_1): Use it.
2464 (expand_expr_real_2): Likewise.
2465 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2466 (fold_convert_const_fixed_from_int): Likewise.
2467 (fold_convert_const_fixed_from_real): Likewise.
2468 (native_encode_fixed): Likewise
2469 (native_encode_complex): Likewise
2470 (native_encode_vector): Likewise.
2471 (native_interpret_fixed): Likewise.
2472 (native_interpret_real): Likewise.
2473 (native_interpret_complex): Likewise.
2474 (native_interpret_vector): Likewise.
2475 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2476 (simd_clone_adjust_argument_types): Likewise.
2477 (simd_clone_init_simd_arrays): Likewise.
2478 (simd_clone_adjust): Likewise.
2479 * stor-layout.c (layout_type): Likewise.
2480 * tree.c (build_minus_one_cst): Likewise.
2481 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2482 * tree-inline.c (estimate_move_cost): Likewise.
2483 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2484 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2485 (vectorizable_reduction): Likewise.
2486 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2487 (vect_recog_mixed_size_cond_pattern): Likewise.
2488 (check_bool_pattern): Likewise.
2489 (adjust_bool_pattern): Likewise.
2490 (search_type_for_mask_1): Likewise.
2491 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2492 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2493 (vectorizable_load): Likewise.
2494 (vectorizable_store): Likewise.
2495 * ubsan.c (ubsan_encode_value): Likewise.
2496 * varasm.c (output_constant): Likewise.
2498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2499 Alan Hayward <alan.hayward@arm.com>
2500 David Sherwood <david.sherwood@arm.com>
2502 * coretypes.h (scalar_mode): New class.
2503 * machmode.h (scalar_mode): Likewise.
2504 (scalar_mode::includes_p): New function.
2505 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2506 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2507 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2508 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2509 * expmed.c (store_bit_field_1): Likewise.
2510 (extract_bit_field_1): Likewise.
2511 * expr.c (write_complex_part): Likewise.
2512 (read_complex_part): Likewise.
2513 (emit_move_complex_push): Likewise.
2514 (expand_expr_real_2): Likewise.
2515 * function.c (assign_parm_setup_reg): Likewise.
2516 (assign_parms_unsplit_complex): Likewise.
2517 * optabs.c (expand_binop): Likewise.
2518 * rtlanal.c (subreg_get_info): Likewise.
2519 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2520 * varasm.c (output_constant_pool_2): Likewise.
2522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2523 Alan Hayward <alan.hayward@arm.com>
2524 David Sherwood <david.sherwood@arm.com>
2526 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2528 (expmed_mult_highpart_optab): Likewise.
2529 (expmed_mult_highpart): Likewise.
2531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2532 Alan Hayward <alan.hayward@arm.com>
2533 David Sherwood <david.sherwood@arm.com>
2535 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2536 instead of a machine_mode.
2537 (builtin_memset_read_str): Likewise.
2538 * builtins.c (c_readstr): Likewise.
2539 (builtin_memcpy_read_str): Likewise.
2540 (builtin_strncpy_read_str): Likewise.
2541 (builtin_memset_read_str): Likewise.
2542 (builtin_memset_gen_str): Likewise.
2543 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2544 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2545 instead of a machine_mode.
2546 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2548 (make_extraction): Likewise.
2549 (try_widen_shift_mode): Take and return scalar_int_modes instead
2551 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2552 a scalar_int_mode instead of a machine_mode.
2553 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2554 (avr_addr_space_pointer_mode): Likewise.
2555 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2556 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2557 (msp430_unwind_word_mode): Likewise.
2558 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2559 (spu_addr_space_pointer_mode): Likewise.
2560 (spu_addr_space_address_mode): Likewise.
2561 (spu_libgcc_cmp_return_mode): Likewise.
2562 (spu_libgcc_shift_count_mode): Likewise.
2563 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2564 (rl78_addr_space_pointer_mode): Likewise.
2565 (fl78_unwind_word_mode): Likewise.
2566 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2568 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2569 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2570 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2571 (mips_valid_pointer_mode): Likewise.
2572 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2573 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2574 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2576 (ft32_addr_space_address_mode): Likewise.
2577 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2578 scalar_int_mode instead of a machine_mode.
2579 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2581 (m32c_addr_space_address_mode): Likewise.
2582 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2583 (rs6000_eh_return_filter_mode): Likewise.
2584 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2585 (rs6000_eh_return_filter_mode): Likewise.
2586 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2587 (s390_libgcc_shift_count_mode): Likewise.
2588 (s390_unwind_word_mode): Likewise.
2589 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2591 * target.def (mode_rep_extended): Likewise.
2592 (valid_pointer_mode): Likewise.
2593 (addr_space.valid_pointer_mode): Likewise.
2594 (eh_return_filter_mode): Return a scalar_int_mode rather than
2596 (libgcc_cmp_return_mode): Likewise.
2597 (libgcc_shift_count_mode): Likewise.
2598 (unwind_word_mode): Likewise.
2599 (addr_space.pointer_mode): Likewise.
2600 (addr_space.address_mode): Likewise.
2601 * doc/tm.texi: Regenerate.
2602 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2604 (do_jump): Use scalar_int_mode for local variables.
2605 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2606 rather than a machine_mode.
2607 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2608 (scompare_loc_descriptor_wide): Likewise.
2609 (scompare_loc_descriptor_narrow): Likewise.
2610 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2612 * except.c (sjlj_emit_dispatch_table): Likewise.
2613 (expand_builtin_eh_copy_values): Likewise.
2614 * explow.c (convert_memory_address_addr_space_1): Likewise.
2615 Take a scalar_int_mode rather than a machine_mode.
2616 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2617 than a machine_mode.
2618 (memory_address_addr_space): Use scalar_int_mode for local variables.
2619 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2620 rather than a machine_mode.
2621 * expmed.c (mask_rtx): Likewise.
2622 (init_expmed_one_conv): Likewise.
2623 (expand_mult_highpart_adjust): Likewise.
2624 (extract_high_half): Likewise.
2625 (expmed_mult_highpart_optab): Likewise.
2626 (expmed_mult_highpart): Likewise.
2627 (expand_smod_pow2): Likewise.
2628 (expand_sdiv_pow2): Likewise.
2629 (emit_store_flag_int): Likewise.
2630 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2632 (extract_low_bits): Likewise.
2633 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2635 * expr.c (pieces_addr::adjust): Likewise.
2636 (can_store_by_pieces): Likewise.
2637 (store_by_pieces): Likewise.
2638 (clear_by_pieces_1): Likewise.
2639 (expand_expr_addr_expr_1): Likewise.
2640 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2641 (expand_expr_real_1): Likewise.
2642 (try_casesi): Likewise.
2643 * final.c (shorten_branches): Likewise.
2644 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2645 type of "mode" to machine_mode.
2646 * internal-fn.c (expand_arith_overflow_result_store): Take a
2647 scalar_int_mode rather than a machine_mode.
2648 (expand_mul_overflow): Use scalar_int_mode for local variables.
2649 * loop-doloop.c (doloop_modify): Likewise.
2650 (doloop_optimize): Likewise.
2651 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2652 than a machine_mode.
2653 (expand_doubleword_shift_condmove): Likewise.
2654 (expand_doubleword_shift): Likewise.
2655 (expand_doubleword_clz): Likewise.
2656 (expand_doubleword_popcount): Likewise.
2657 (expand_doubleword_parity): Likewise.
2658 (expand_absneg_bit): Use scalar_int_mode for local variables.
2659 (prepare_float_lib_cmp): Likewise.
2660 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2661 rather than a machine_mode.
2662 (convert_memory_address_addr_space): Likewise.
2663 (get_mode_bounds): Likewise.
2664 (get_address_mode): Return a scalar_int_mode rather than a
2666 * rtlanal.c (get_address_mode): Likewise.
2667 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2668 than a machine_mode.
2669 * targhooks.c (default_mode_rep_extended): Likewise.
2670 (default_valid_pointer_mode): Likewise.
2671 (default_addr_space_valid_pointer_mode): Likewise.
2672 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2673 than a machine_mode.
2674 (default_libgcc_cmp_return_mode): Likewise.
2675 (default_libgcc_shift_count_mode): Likewise.
2676 (default_unwind_word_mode): Likewise.
2677 (default_addr_space_pointer_mode): Likewise.
2678 (default_addr_space_address_mode): Likewise.
2679 * targhooks.h (default_eh_return_filter_mode): Likewise.
2680 (default_libgcc_cmp_return_mode): Likewise.
2681 (default_libgcc_shift_count_mode): Likewise.
2682 (default_unwind_word_mode): Likewise.
2683 (default_addr_space_pointer_mode): Likewise.
2684 (default_addr_space_address_mode): Likewise.
2685 (default_mode_rep_extended): Take a scalar_int_mode rather than
2687 (default_valid_pointer_mode): Likewise.
2688 (default_addr_space_valid_pointer_mode): Likewise.
2689 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2691 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2692 rather than a machine_mode.
2693 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2694 for local variables.
2695 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2696 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2697 than a machine_mode.
2699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2700 Alan Hayward <alan.hayward@arm.com>
2701 David Sherwood <david.sherwood@arm.com>
2703 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2704 the mode argument to scalar_int_mode.
2705 (do_jump_by_parts_zero_rtx): Likewise.
2706 (do_jump_by_parts_equality_rtx): Likewise.
2707 (do_jump_by_parts_greater): Take a mode argument.
2708 (do_jump_by_parts_equality): Likewise.
2709 (do_jump_1): Update calls accordingly.
2711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2712 Alan Hayward <alan.hayward@arm.com>
2713 David Sherwood <david.sherwood@arm.com>
2715 * is-a.h (safe_dyn_cast): New function.
2716 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2717 (jump_table_for_label): Likewise.
2718 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2719 instead of an rtx_insn *.
2720 (shorten_branches): Use dyn_cast instead of LABEL_P and
2721 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2722 rtx_jump_table_data::get_data_mode.
2723 (final_scan_insn): Likewise.
2725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2726 Alan Hayward <alan.hayward@arm.com>
2727 David Sherwood <david.sherwood@arm.com>
2729 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2730 trying to combine a full-register integer set with a subreg
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 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2738 that is always either address_mode or pointer_mode.
2740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2741 Alan Hayward <alan.hayward@arm.com>
2742 David Sherwood <david.sherwood@arm.com>
2744 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2745 when the two are known to be equal.
2747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * simplify-rtx.c (simplify_const_unary_operation): Use
2752 is_a <scalar_int_mode> instead of checking for a nonzero
2753 precision. Forcibly convert op_mode to a scalar_int_mode
2754 in that case. More clearly differentiate the operand and
2755 result modes and use the former when deciding what the value
2756 of a count-bits operation should be. Use is_int_mode instead
2757 of checking for a MODE_INT. Remove redundant check for whether
2758 this mode has a zero precision.
2760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2761 Alan Hayward <alan.hayward@arm.com>
2762 David Sherwood <david.sherwood@arm.com>
2764 * optabs.c (widen_leading): Change the type of the mode argument
2765 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2766 (widen_bswap): Likewise.
2767 (expand_parity): Likewise.
2768 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2769 (expand_ffs): Likewise.
2770 (epand_unop): Check for scalar integer modes before calling the
2773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2774 Alan Hayward <alan.hayward@arm.com>
2775 David Sherwood <david.sherwood@arm.com>
2777 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2779 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2786 value_mode arguments. Use scalar_int_mode internally.
2787 (store_bit_field_1): Rename the new integer mode from imode
2788 to op0_mode and use it instead of GET_MODE (op0). Update calls
2789 to store_split_bit_field, store_bit_field_using_insv and
2790 store_fixed_bit_field.
2791 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2792 Use scalar_int_mode internally. Use a bit count rather than a mode
2793 when calculating the largest bit size for get_best_mode.
2794 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2795 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2796 Remove assertion that OP0 has a scalar integer mode.
2797 (store_split_bit_field): Add op0_mode and value_mode arguments.
2798 Update calls to extract_fixed_bit_field.
2799 (extract_bit_field_using_extv): Add an op0_mode argument.
2800 Use scalar_int_mode internally.
2801 (extract_bit_field_1): Rename the new integer mode from imode to
2802 op0_mode and use it instead of GET_MODE (op0). Update calls to
2803 extract_split_bit_field, extract_bit_field_using_extv and
2804 extract_fixed_bit_field.
2805 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2806 to extract_split_bit_field and extract_fixed_bit_field_1.
2807 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2808 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2810 (extract_split_bit_field): Add an op0_mode argument. Update call
2811 to extract_fixed_bit_field.
2813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2814 Alan Hayward <alan.hayward@arm.com>
2815 David Sherwood <david.sherwood@arm.com>
2817 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2818 * explow.c (hard_function_value): Likewise.
2819 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2820 convert_to_mode call outside the loop.
2821 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2822 for the mode iterator. Require the mode specified by max_pieces
2824 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2826 (copy_blkmode_to_reg): Likewise.
2827 (set_storage_via_setmem): Likewise.
2828 * optabs.c (prepare_cmp_insn): Likewise.
2829 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2830 * stor-layout.c (finish_bitfield_representative): Likewise.
2832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2833 Alan Hayward <alan.hayward@arm.com>
2834 David Sherwood <david.sherwood@arm.com>
2836 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2837 * expr.c (convert_move): Use them.
2838 (convert_modes): Likewise.
2839 (store_expr_with_bounds): Likewise.
2841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2842 Alan Hayward <alan.hayward@arm.com>
2843 David Sherwood <david.sherwood@arm.com>
2845 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2846 parameter for the mode of "x". Remove the "known_x", "known_mode"
2847 and "known_ret" arguments. Change the type of the mode argument
2849 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2850 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2851 (reg_num_sign_bit_copies_for_combine): Likewise.
2852 * rtlanal.c (nonzero_bits1): Likewise.
2853 (num_sign_bit_copies1): Likewise.
2854 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2855 (reg_num_sign_bit_copies_general): Likewise.
2856 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2857 (reg_nonzero_bits_general): Likewise.
2859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2860 Alan Hayward <alan.hayward@arm.com>
2861 David Sherwood <david.sherwood@arm.com>
2863 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2864 than in subroutines. Return 1 for non-integer modes.
2865 (cached_num_sign_bit_copies): Change the type of the mode parameter
2867 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2868 classes. Handle CONST_INT_P first and then check whether X also
2869 has a scalar integer mode. Check the same thing for inner registers
2870 of a SUBREG and for values that are being extended or truncated.
2872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2873 Alan Hayward <alan.hayward@arm.com>
2874 David Sherwood <david.sherwood@arm.com>
2876 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2877 in subroutines. Return the mode mask for non-integer modes.
2878 (cached_nonzero_bits): Change the type of the mode parameter
2880 (nonzero_bits1): Likewise. Remove early exit for other mode
2881 classes. Handle CONST_INT_P first and then check whether X
2882 also has a scalar integer mode.
2884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2885 Alan Hayward <alan.hayward@arm.com>
2886 David Sherwood <david.sherwood@arm.com>
2888 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2889 Return a scalar_int_mode and assert that the size is greater than
2891 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2892 (op_by_pieces_d::op_by_pieces_d): Likewise.
2893 (op_by_pieces_d::run): Likewise.
2894 (can_store_by_pieces): Likewise.
2896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2897 Alan Hayward <alan.hayward@arm.com>
2898 David Sherwood <david.sherwood@arm.com>
2900 * combine.c (extract_left_shift): Add a mode argument and update
2902 (make_compound_operation_int): Change the type of the mode parameter
2903 to scalar_int_mode and update the call to extract_left_shift.
2905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2906 Alan Hayward <alan.hayward@arm.com>
2907 David Sherwood <david.sherwood@arm.com>
2909 * combine.c (simplify_and_const_int): Change the type of the mode
2910 parameter to scalar_int_mode.
2911 (simplify_and_const_int_1): Likewise. Update recursive call.
2913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2914 Alan Hayward <alan.hayward@arm.com>
2915 David Sherwood <david.sherwood@arm.com>
2917 * combine.c (simplify_compare_const): Check that the mode is a
2918 scalar_int_mode (rather than VOIDmode) before testing its
2920 (simplify_comparison): Move COMPARISON_P handling out of the
2921 loop and restrict the latter part of the loop to scalar_int_modes.
2922 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2923 and when considering SUBREG_REGs. Use is_int_mode instead of
2924 checking GET_MODE_CLASS against MODE_INT.
2926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2927 Alan Hayward <alan.hayward@arm.com>
2928 David Sherwood <david.sherwood@arm.com>
2930 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2931 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2932 shift_unit_mode and for modes involved in scalar shifts.
2934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2935 Alan Hayward <alan.hayward@arm.com>
2936 David Sherwood <david.sherwood@arm.com>
2938 * combine.c (force_int_to_mode): New function, split out from...
2939 (force_to_mode): ...here. Keep xmode up-to-date and use it
2940 instead of GET_MODE (x).
2942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2943 Alan Hayward <alan.hayward@arm.com>
2944 David Sherwood <david.sherwood@arm.com>
2946 * optabs-query.h (extraction_insn::struct_mode): Change type to
2947 opt_scalar_int_mode and update comment.
2948 (extraction_insn::field_mode): Change type to scalar_int_mode.
2949 (extraction_insn::pos_mode): Likewise.
2950 * combine.c (make_extraction): Update accordingly.
2951 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2952 (get_optab_extraction_insn): Likewise.
2953 * recog.c (simplify_while_replacing): Likewise.
2954 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2955 parameter to opt_scalar_int_mode.
2957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2958 Alan Hayward <alan.hayward@arm.com>
2959 David Sherwood <david.sherwood@arm.com>
2961 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2962 to a scalar_int_mode instead of a machine_mode.
2963 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2964 (get_best_mode): Return a boolean and use a pointer argument to store
2965 the selected mode. Replace the limit mode parameter with a bit limit.
2966 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2967 for the values returned by bit_field_mode_iterator::next_mode.
2968 (store_bit_field): Update call to get_best_mode.
2969 (store_fixed_bit_field): Likewise.
2970 (extract_fixed_bit_field): Likewise.
2971 * expr.c (optimize_bitfield_assignment_op): Likewise.
2972 * fold-const.c (optimize_bit_field_compare): Likewise.
2973 (fold_truth_andor_1): Likewise.
2974 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2975 Update for new type of m_mode.
2976 (get_best_mode): As above.
2978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2979 Alan Hayward <alan.hayward@arm.com>
2980 David Sherwood <david.sherwood@arm.com>
2982 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2983 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2984 (store_bit_field): Check is_a <scalar_int_mode> before calling
2985 strict_volatile_bitfield_p.
2986 (extract_bit_field): Likewise.
2988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2989 Alan Hayward <alan.hayward@arm.com>
2990 David Sherwood <david.sherwood@arm.com>
2992 * target.def (cstore_mode): Return a scalar_int_mode.
2993 * doc/tm.texi: Regenerate.
2994 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2995 * targhooks.h (default_cstore_mode): Likewise.
2996 * targhooks.c (default_cstore_mode): Likewise, using a forced
2998 * expmed.c (emit_cstore): Expect the target of the cstore to be
3001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3002 Alan Hayward <alan.hayward@arm.com>
3003 David Sherwood <david.sherwood@arm.com>
3005 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3007 (niter_desc): Likewise mode.
3008 (iv_analyze): Add a mode parameter.
3010 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3011 and change its type to scalar_int_mode.
3012 * loop-iv.c: Update commentary at head of file.
3013 (iv_constant): Pass the mode paraeter before the rtx it describes
3014 and change its type to scalar_int_mode. Remove VOIDmode handling.
3015 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3016 (iv_extend): Likewise.
3017 (shorten_into_mode): Likewise.
3018 (iv_add): Use scalar_int_mode.
3019 (iv_mult): Likewise.
3020 (iv_shift): Likewise.
3021 (canonicalize_iv_subregs): Likewise.
3022 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3023 it describes and change its mode to scalar_int_mode. Also change
3024 the type of the returned inner_mode to scalar_int_mode.
3025 (get_biv_step): Likewise, turning outer_mode from a pointer
3026 into a direct parameter. Update call to get_biv_step_1.
3027 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3028 iv_constant and get_biv_step.
3029 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3030 and change its type to scalar_int_mode. Don't initialise iv->mode
3031 to VOIDmode and remove later checks for its still being VOIDmode.
3032 Update calls to iv_analyze_op and iv_analyze_expr. Check
3033 is_a <scalar_int_mode> when changing the mode under consideration.
3034 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3035 Update call to iv_analyze_expr.
3036 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3037 inner register is not also a scalar_int_mode. Update call to
3039 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3040 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3041 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3042 separate mode class checks. Update calls to iv_analyze. Remove
3043 fix-up of VOIDmodes after iv_analyze_biv.
3044 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3045 don't have a scalar_int_mode. Update call to biv_p.
3047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3048 Alan Hayward <alan.hayward@arm.com>
3049 David Sherwood <david.sherwood@arm.com>
3051 * cfgexpand.c (convert_debug_memory_address): Use
3052 as_a <scalar_int_mode>.
3053 * combine.c (expand_compound_operation): Likewise.
3054 (make_extraction): Likewise.
3055 (change_zero_ext): Likewise.
3056 (simplify_comparison): Likewise.
3057 * cse.c (cse_insn): Likewise.
3058 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3059 (mem_loc_descriptor): Likewise.
3060 (loc_descriptor): Likewise.
3061 * expmed.c (init_expmed_one_mode): Likewise.
3062 (synth_mult): Likewise.
3063 (emit_store_flag_1): Likewise.
3064 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3065 of a comparison with size.
3066 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3067 (reduce_to_bit_field_precision): Likewise.
3068 * function.c (expand_function_end): Likewise.
3069 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3070 * loop-doloop.c (doloop_modify): Likewise.
3071 * optabs.c (expand_binop): Likewise.
3072 (expand_unop): Likewise.
3073 (expand_copysign_absneg): Likewise.
3074 (prepare_cmp_insn): Likewise.
3075 (maybe_legitimize_operand): Likewise.
3076 * recog.c (const_scalar_int_operand): Likewise.
3077 * rtlanal.c (get_address_mode): Likewise.
3078 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3079 (simplify_cond_clz_ctz): Likewise.
3080 * tree-nested.c (get_nl_goto_field): Likewise.
3081 * tree.c (build_vector_type_for_mode): Likewise.
3082 * var-tracking.c (use_narrower_mode): Likewise.
3084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3085 Alan Hayward <alan.hayward@arm.com>
3086 David Sherwood <david.sherwood@arm.com>
3088 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3089 * builtins.c (expand_builtin_signbit): Use it.
3090 * cfgexpand.c (expand_debug_expr): Likewise.
3091 * dojump.c (do_jump): Likewise.
3092 (do_compare_and_jump): Likewise.
3093 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3094 * expmed.c (make_tree): Likewise.
3095 * expr.c (expand_expr_real_2): Likewise.
3096 (expand_expr_real_1): Likewise.
3097 (try_casesi): Likewise.
3098 * fold-const-call.c (fold_const_call_ss): Likewise.
3099 * fold-const.c (unextend): Likewise.
3100 (extract_muldiv_1): Likewise.
3101 (fold_single_bit_test): Likewise.
3102 (native_encode_int): Likewise.
3103 (native_encode_string): Likewise.
3104 (native_interpret_int): Likewise.
3105 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3106 * internal-fn.c (expand_addsub_overflow): Likewise.
3107 (expand_neg_overflow): Likewise.
3108 (expand_mul_overflow): Likewise.
3109 (expand_arith_overflow): Likewise.
3110 * match.pd: Likewise.
3111 * stor-layout.c (layout_type): Likewise.
3112 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3113 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3114 * tree-ssanames.c (get_range_info): Likewise.
3115 * tree-switch-conversion.c (array_value_type) Likewise.
3116 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3117 (vect_recog_divmod_pattern): Likewise.
3118 (vect_recog_mixed_size_cond_pattern): Likewise.
3119 * tree-vrp.c (extract_range_basic): Likewise.
3120 (simplify_float_conversion_using_ranges): Likewise.
3121 * tree.c (int_fits_type_p): Likewise.
3122 * ubsan.c (instrument_bool_enum_load): Likewise.
3123 * varasm.c (mergeable_string_section): Likewise.
3124 (narrowing_initializer_constant_valid_p): Likewise.
3125 (output_constant): Likewise.
3127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3128 Alan Hayward <alan.hayward@arm.com>
3129 David Sherwood <david.sherwood@arm.com>
3131 * machmode.h (NARROWEST_INT_MODE): New macro.
3132 * expr.c (alignment_for_piecewise_move): Use it instead of
3133 GET_CLASS_NARROWEST_MODE (MODE_INT).
3134 (push_block): Likewise.
3135 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3137 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3140 Alan Hayward <alan.hayward@arm.com>
3141 David Sherwood <david.sherwood@arm.com>
3143 * postreload.c (move2add_valid_value_p): Change the type of the
3144 mode parameter to scalar_int_mode.
3145 (move2add_use_add2_insn): Add a mode parameter and use it instead
3147 (move2add_use_add3_insn): Likewise.
3148 (reload_cse_move2add): Update accordingly.
3150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3151 Alan Hayward <alan.hayward@arm.com>
3152 David Sherwood <david.sherwood@arm.com>
3154 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3156 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3157 * optabs.c (expand_unop): Likewise.
3159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3160 Alan Hayward <alan.hayward@arm.com>
3161 David Sherwood <david.sherwood@arm.com>
3163 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3164 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3165 (popcount_loc_descriptor): Likewise.
3166 (bswap_loc_descriptor): Likewise.
3167 (rotate_loc_descriptor): Likewise.
3168 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3169 calling the functions above.
3171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3172 Alan Hayward <alan.hayward@arm.com>
3173 David Sherwood <david.sherwood@arm.com>
3175 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3177 (try_combine): Likewise.
3178 (simplify_if_then_else): Likewise.
3179 * cse.c (cse_insn): Likewise.
3180 * dwarf2out.c (mem_loc_descriptor): Likewise.
3181 * emit-rtl.c (gen_lowpart_common): Likewise.
3182 * simplify-rtx.c (simplify_truncation): Likewise.
3183 (simplify_binary_operation_1): Likewise.
3184 (simplify_const_relational_operation): Likewise.
3185 (simplify_ternary_operation): Likewise.
3186 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3189 Alan Hayward <alan.hayward@arm.com>
3190 David Sherwood <david.sherwood@arm.com>
3192 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3193 * reload.c (push_reload): Likewise.
3194 (find_reloads): Likewise.
3196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3197 Alan Hayward <alan.hayward@arm.com>
3198 David Sherwood <david.sherwood@arm.com>
3200 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3201 (make_compound_operation_int): Likewise.
3202 (change_zero_ext): Likewise.
3203 * expr.c (convert_move): Likewise.
3204 (convert_modes): Likewise.
3205 * fwprop.c (forward_propagate_subreg): Likewise.
3206 * loop-iv.c (get_biv_step_1): Likewise.
3207 * optabs.c (widen_operand): Likewise.
3208 * postreload.c (move2add_valid_value_p): Likewise.
3209 * recog.c (simplify_while_replacing): Likewise.
3210 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3211 (simplify_binary_operation_1): Likewise. Remove redundant
3212 mode equality check.
3214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3215 Alan Hayward <alan.hayward@arm.com>
3216 David Sherwood <david.sherwood@arm.com>
3218 * combine.c (combine_simplify_rtx): Add checks for
3219 is_a <scalar_int_mode>.
3220 (simplify_if_then_else): Likewise.
3221 (make_field_assignment): Likewise.
3222 (simplify_comparison): Likewise.
3223 * ifcvt.c (noce_try_bitop): Likewise.
3224 * loop-invariant.c (canonicalize_address_mult): Likewise.
3225 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3228 Alan Hayward <alan.hayward@arm.com>
3229 David Sherwood <david.sherwood@arm.com>
3231 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3232 is_a <scalar_int_mode> instead of != BLKmode.
3234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3235 Alan Hayward <alan.hayward@arm.com>
3236 David Sherwood <david.sherwood@arm.com>
3238 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3239 instead of != VOIDmode.
3240 * combine.c (if_then_else_cond): Likewise.
3241 (change_zero_ext): Likewise.
3242 * dwarf2out.c (mem_loc_descriptor): Likewise.
3243 (loc_descriptor): Likewise.
3244 * rtlanal.c (canonicalize_condition): Likewise.
3245 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3248 Alan Hayward <alan.hayward@arm.com>
3249 David Sherwood <david.sherwood@arm.com>
3251 * simplify-rtx.c (simplify_binary_operation_1): Use
3252 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3255 Alan Hayward <alan.hayward@arm.com>
3256 David Sherwood <david.sherwood@arm.com>
3258 * wide-int.h (int_traits<unsigned char>) New class.
3259 (int_traits<unsigned short>) Likewise.
3260 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3261 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3263 * combine.c (set_nonzero_bits_and_sign_copies): Use
3264 is_a <scalar_int_mode>.
3265 (find_split_point): Likewise.
3266 (combine_simplify_rtx): Likewise.
3267 (simplify_logical): Likewise.
3268 (expand_compound_operation): Likewise.
3269 (expand_field_assignment): Likewise.
3270 (make_compound_operation): Likewise.
3271 (extended_count): Likewise.
3272 (change_zero_ext): Likewise.
3273 (simplify_comparison): Likewise.
3274 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3275 (ucompare_loc_descriptor): Likewise.
3276 (minmax_loc_descriptor): Likewise.
3277 (mem_loc_descriptor): Likewise.
3278 (loc_descriptor): Likewise.
3279 * expmed.c (init_expmed_one_mode): Likewise.
3280 * lra-constraints.c (lra_constraint_offset): Likewise.
3281 * optabs.c (prepare_libcall_arg): Likewise.
3282 * postreload.c (move2add_note_store): Likewise.
3283 * reload.c (operands_match_p): Likewise.
3284 * rtl.h (load_extend_op): Likewise.
3285 * rtlhooks.c (gen_lowpart_general): Likewise.
3286 * simplify-rtx.c (simplify_truncation): Likewise.
3287 (simplify_unary_operation_1): Likewise.
3288 (simplify_binary_operation_1): Likewise.
3289 (simplify_const_binary_operation): Likewise.
3290 (simplify_const_relational_operation): Likewise.
3291 (simplify_subreg): Likewise.
3292 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3293 * var-tracking.c (adjust_mems): Likewise.
3294 (prepare_call_arguments): Likewise.
3296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3297 Alan Hayward <alan.hayward@arm.com>
3298 David Sherwood <david.sherwood@arm.com>
3300 * machmode.h (is_int_mode): New fuction.
3301 * combine.c (find_split_point): Use it.
3302 (combine_simplify_rtx): Likewise.
3303 (simplify_if_then_else): Likewise.
3304 (simplify_set): Likewise.
3305 (simplify_shift_const_1): Likewise.
3306 (simplify_comparison): Likewise.
3307 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3308 * cse.c (notreg_cost): Likewise.
3309 (cse_insn): Likewise.
3310 * cselib.c (cselib_lookup_1): Likewise.
3311 * dojump.c (do_jump_1): Likewise.
3312 (do_compare_rtx_and_jump): Likewise.
3313 * dse.c (get_call_args): Likewise.
3314 * dwarf2out.c (rtl_for_decl_init): Likewise.
3315 (native_encode_initializer): Likewise.
3316 * expmed.c (emit_store_flag_1): Likewise.
3317 (emit_store_flag): Likewise.
3318 * expr.c (convert_modes): Likewise.
3319 (store_field): Likewise.
3320 (expand_expr_real_1): Likewise.
3321 * fold-const.c (fold_read_from_constant_string): Likewise.
3322 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3323 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3324 * optabs.c (expand_binop): Likewise.
3325 (expand_unop): Likewise.
3326 (expand_abs_nojump): Likewise.
3327 (expand_one_cmpl_abs_nojump): Likewise.
3328 * simplify-rtx.c (mode_signbit_p): Likewise.
3329 (val_signbit_p): Likewise.
3330 (val_signbit_known_set_p): Likewise.
3331 (val_signbit_known_clear_p): Likewise.
3332 (simplify_relational_operation_1): Likewise.
3333 * tree.c (vector_type_mode): Likewise.
3335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3336 Alan Hayward <alan.hayward@arm.com>
3337 David Sherwood <david.sherwood@arm.com>
3339 * machmode.h (smallest_mode_for_size): Fix formatting.
3340 (smallest_int_mode_for_size): New function.
3341 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3342 instead of smallest_mode_for_size.
3343 * combine.c (make_extraction): Likewise.
3344 * config/arc/arc.c (arc_expand_movmem): Likewise.
3345 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3346 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3347 * config/s390/s390.c (s390_expand_insv): Likewise.
3348 * config/sparc/sparc.c (assign_int_registers): Likewise.
3349 * config/spu/spu.c (spu_function_value): Likewise.
3350 (spu_function_arg): Likewise.
3351 * coverage.c (get_gcov_type): Likewise.
3352 (get_gcov_unsigned_t): Likewise.
3353 * dse.c (find_shift_sequence): Likewise.
3354 * expmed.c (store_bit_field_1): Likewise.
3355 * expr.c (convert_move): Likewise.
3356 (store_field): Likewise.
3357 * internal-fn.c (expand_arith_overflow): Likewise.
3358 * optabs-query.c (get_best_extraction_insn): Likewise.
3359 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3360 * stor-layout.c (layout_type): Likewise.
3361 (initialize_sizetypes): Likewise.
3362 * targhooks.c (default_get_mask_mode): Likewise.
3363 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3366 Alan Hayward <alan.hayward@arm.com>
3367 David Sherwood <david.sherwood@arm.com>
3369 * machmode.h (opt_mode::else_blk): New function.
3370 (int_mode_for_mode): Declare.
3371 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3372 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3374 * cfgexpand.c (expand_debug_expr): Likewise.
3375 * combine.c (gen_lowpart_or_truncate): Likewise.
3376 (gen_lowpart_for_combine): Likewise.
3377 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3378 * config/avr/avr.c (avr_to_int_mode): Likewise.
3379 (avr_out_plus_1): Likewise.
3380 (avr_out_plus): Likewise.
3381 (avr_out_round): Likewise.
3382 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3383 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3384 (s390_expand_vcond): Likewise.
3385 * config/spu/spu.c (spu_split_immediate): Likewise.
3386 (spu_expand_mov): Likewise.
3387 * dse.c (get_stored_val): Likewise.
3388 * expmed.c (store_bit_field_1): Likewise.
3389 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3391 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3392 (extract_low_bits): Likewise.
3393 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3394 handling rather than repeating the check.
3395 (emit_group_store): Likewise.
3396 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3397 * optabs.c (expand_absneg_bit): Likewise.
3398 (expand_copysign_absneg): Likewise.
3399 (expand_copysign_bit): Likewise.
3400 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3401 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3402 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3403 * var-tracking.c (prepare_call_arguments): Likewise.
3404 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3405 int_mode_for_mode instead of mode_for_size.
3406 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3409 Alan Hayward <alan.hayward@arm.com>
3410 David Sherwood <david.sherwood@arm.com>
3412 * machmode.h (int_mode_for_size): New function.
3413 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3414 instead of mode_for_size.
3415 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3417 * combine.c (expand_field_assignment): Use int_mode_for_size
3418 instead of mode_for_size.
3419 (make_extraction): Likewise.
3420 (simplify_shift_const_1): Likewise.
3421 (simplify_comparison): Likewise.
3422 * dojump.c (do_jump): Likewise.
3423 * dwarf2out.c (mem_loc_descriptor): Likewise.
3424 * emit-rtl.c (init_derived_machine_modes): Likewise.
3425 * expmed.c (flip_storage_order): Likewise.
3426 (convert_extracted_bit_field): Likewise.
3427 * expr.c (copy_blkmode_from_reg): Likewise.
3428 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3429 * internal-fn.c (expand_mul_overflow): Likewise.
3430 * lower-subreg.c (simple_move): Likewise.
3431 * optabs-libfuncs.c (init_optabs): Likewise.
3432 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3433 * tree.c (vector_type_mode): Likewise.
3434 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3435 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3436 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3437 * tree-vect-stmts.c (vectorizable_load): Likewise.
3438 (vectorizable_store): Likewise.
3440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3441 Alan Hayward <alan.hayward@arm.com>
3442 David Sherwood <david.sherwood@arm.com>
3444 * coretypes.h (pod_mode): New type.
3445 (scalar_int_mode_pod): New typedef.
3446 * machmode.h (pod_mode): New class.
3447 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3448 * genmodes.c (emit_mode_int_n): Update accordingly.
3449 * lower-subreg.h (target_lower_subreg): Change type to
3450 scalar_int_mode_pod.
3451 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3452 scalar_int_mode_pod.
3454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3455 Alan Hayward <alan.hayward@arm.com>
3456 David Sherwood <david.sherwood@arm.com>
3458 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3459 machine_mode to scalar_int_mode.
3460 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3461 (rs6000_option_override_internal): Remove cast to int.
3462 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3463 machine_mode to scalar_int_mode.
3464 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3465 (rs6000_option_override_internal): Remove cast to int.
3466 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3467 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3469 * config/s390/s390.c (s390_expand_builtin): Likewise.
3470 * coretypes.h (scalar_int_mode): New type.
3471 (opt_scalar_int_mode): New typedef.
3472 * machmode.h (scalar_int_mode): New class.
3473 (scalar_int_mode::includes_p): New function.
3474 (byte_mode): Change type to scalar_int_mode.
3475 (word_mode): Likewise.
3476 (ptr_mode): Likewise.
3477 * emit-rtl.c (byte_mode): Likewise.
3478 (word_mode): Likewise.
3479 (ptr_mode): Likewise.
3480 (init_derived_machine_modes): Update accordingly.
3481 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3482 and MODE_PARTIAL_INT.
3483 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3484 opt_scalar_int_mode.
3486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3487 Alan Hayward <alan.hayward@arm.com>
3488 David Sherwood <david.sherwood@arm.com>
3490 * target.def (libgcc_floating_mode_supported_p): Take a
3492 * doc/tm.texi: Regenerate.
3493 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3495 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3496 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3500 Alan Hayward <alan.hayward@arm.com>
3501 David Sherwood <david.sherwood@arm.com>
3503 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3504 * doc/tm.texi: Regenerate.
3505 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3506 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3507 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3508 * targhooks.h (default_floatn_mode): Likewise.
3509 * targhooks.c (default_floatn_mode): Likewise.
3510 * tree.c (build_common_tree_nodes): Update accordingly.
3512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3513 Alan Hayward <alan.hayward@arm.com>
3514 David Sherwood <david.sherwood@arm.com>
3516 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3517 (mode_iterator::iterate_p): Likewise.
3518 (mode_iterator::get_wider): Likewise.
3519 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3522 Alan Hayward <alan.hayward@arm.com>
3523 David Sherwood <david.sherwood@arm.com>
3525 * coretypes.h (opt_scalar_float_mode): New typedef.
3526 * machmode.h (float_mode_for_size): New function.
3527 * emit-rtl.c (double_mode): Delete.
3528 (init_emit_once): Use float_mode_for_size.
3529 * stor-layout.c (layout_type): Likewise.
3530 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3533 Alan Hayward <alan.hayward@arm.com>
3534 David Sherwood <david.sherwood@arm.com>
3536 * output.h (assemble_real): Take a scalar_float_mode.
3537 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3538 * config/arm/arm.md (consttable_4): Likewise.
3539 (consttable_8): Likewise.
3540 (consttable_16): Likewise.
3541 * config/mips/mips.md (consttable_float): Likewise.
3542 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3543 * varasm.c (assemble_real): Take a scalar_float_mode.
3544 (output_constant_pool_2): Update accordingly.
3545 (output_constant): Likewise.
3547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3548 Alan Hayward <alan.hayward@arm.com>
3549 David Sherwood <david.sherwood@arm.com>
3551 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3552 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3553 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3554 (native_encode_real): Likewise.
3555 (native_interpret_real): Likewise.
3556 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3557 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3560 Alan Hayward <alan.hayward@arm.com>
3561 David Sherwood <david.sherwood@arm.com>
3563 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3564 <scalar_float_mode>. Simplify.
3565 (gen_extend_conv_libfunc): Likewise.
3567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3568 Alan Hayward <alan.hayward@arm.com>
3569 David Sherwood <david.sherwood@arm.com>
3571 * coretypes.h (scalar_float_mode): New type.
3572 * machmode.h (mode_traits::from_int): Use machine_mode if
3573 USE_ENUM_MODES is defined.
3574 (is_a): New function.
3576 (dyn_cast): Likewise.
3577 (scalar_float_mode): New class.
3578 (scalar_float_mode::includes_p): New function.
3579 (is_float_mode): Likewise.
3580 * gdbhooks.py (MachineModePrinter): New class.
3581 (build_pretty_printer): Use it for scalar_float_mode.
3582 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3583 (format_helper::format_helper): Turn into a template.
3584 * genmodes.c (get_mode_class): New function.
3585 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3586 or machine_mode if none.
3587 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3588 as_a <scalar_float_mode>.
3589 * dwarf2out.c (mem_loc_descriptor): Likewise.
3590 (insert_float): Likewise.
3591 (add_const_value_attribute): Likewise.
3592 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3593 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3594 (expand_unop): Update accordingly.
3595 (expand_abs_nojump): Likewise.
3596 (expand_copysign_absneg): Take a scalar_float_mode.
3597 (expand_copysign_bit): Likewise.
3598 (expand_copysign): Update accordingly.
3600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3601 Alan Hayward <alan.hayward@arm.com>
3602 David Sherwood <david.sherwood@arm.com>
3604 * coretypes.h (opt_mode): New class.
3605 * machmode.h (opt_mode): Likewise.
3606 (opt_mode::else_void): New function.
3607 (opt_mode::require): Likewise.
3608 (opt_mode::exists): Likewise.
3609 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3610 (GET_MODE_2XWIDER_MODE): Likewise.
3611 (mode_iterator::get_wider): Update accordingly.
3612 (mode_iterator::get_2xwider): Likewise.
3613 (mode_iterator::get_known_wider): Likewise, turning into a template.
3614 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3615 forcing a wider mode to exist.
3616 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3617 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3618 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3619 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3620 * lower-subreg.c (init_lower_subreg): Likewise.
3621 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3622 on the final iteration.
3623 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3624 a wider mode exists before asking for a move pattern.
3625 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3626 forcing a wider mode to exist.
3627 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3628 returning false if no such mode exists.
3629 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3630 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3631 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3632 Avoid checking for a MODE_INT if we already know the mode is not a
3634 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3635 forcing a wider mode to exist.
3636 (expmed_mult_highpart_optab): Likewise.
3637 (expmed_mult_highpart): Likewise.
3638 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3640 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3641 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3642 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3643 * internal-fn.c (expand_mul_overflow): Update use of
3644 GET_MODE_2XWIDER_MODE.
3645 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3646 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3647 GET_MODE_WIDER_MODE.
3648 (convert_plusminus_to_widen): Likewise.
3649 * tree-switch-conversion.c (array_value_type): Likewise.
3650 * var-tracking.c (emit_note_insn_var_location): Likewise.
3651 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3652 Return false inside rather than outside the loop if no wider mode
3654 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3655 and GET_MODE_2XWIDER_MODE
3656 (can_compare_p): Use else_void.
3657 * gdbhooks.py (OptMachineModePrinter): New class.
3658 (build_pretty_printer): Use it for opt_mode.
3660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3661 Alan Hayward <alan.hayward@arm.com>
3662 David Sherwood <david.sherwood@arm.com>
3664 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3665 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3668 Alan Hayward <alan.hayward@arm.com>
3669 David Sherwood <david.sherwood@arm.com>
3671 * machmode.h (mode_traits): New structure.
3672 (get_narrowest_mode): New function.
3673 (mode_iterator::start): Likewise.
3674 (mode_iterator::iterate_p): Likewise.
3675 (mode_iterator::get_wider): Likewise.
3676 (mode_iterator::get_known_wider): Likewise.
3677 (mode_iterator::get_2xwider): Likewise.
3678 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3679 (FOR_EACH_MODE): Likewise.
3680 (FOR_EACH_MODE_FROM): Likewise.
3681 (FOR_EACH_MODE_UNTIL): Likewise.
3682 (FOR_EACH_WIDER_MODE): Likewise.
3683 (FOR_EACH_2XWIDER_MODE): Likewise.
3684 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3685 * combine.c (simplify_comparison): Likewise
3686 * config/i386/i386.c (type_natural_mode): Likewise.
3687 * cse.c (cse_insn): Likewise.
3688 * dse.c (find_shift_sequence): Likewise.
3689 * emit-rtl.c (init_derived_machine_modes): Likewise.
3690 (init_emit_once): Likewise.
3691 * explow.c (hard_function_value): Likewise.
3692 * expmed.c (extract_fixed_bit_field_1): Likewise.
3693 (extract_bit_field_1): Likewise.
3694 (expand_divmod): Likewise.
3695 (emit_store_flag_1): Likewise.
3696 * expr.c (init_expr_target): Likewise.
3697 (convert_move): Likewise.
3698 (alignment_for_piecewise_move): Likewise.
3699 (widest_int_mode_for_size): Likewise.
3700 (emit_block_move_via_movmem): Likewise.
3701 (copy_blkmode_to_reg): Likewise.
3702 (set_storage_via_setmem): Likewise.
3703 (compress_float_constant): Likewise.
3704 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3705 * optabs-query.c (get_best_extraction_insn): Likewise.
3706 * optabs.c (expand_binop): Likewise.
3707 (expand_twoval_unop): Likewise.
3708 (expand_twoval_binop): Likewise.
3709 (widen_leading): Likewise.
3710 (widen_bswap): Likewise.
3711 (expand_parity): Likewise.
3712 (expand_unop): Likewise.
3713 (prepare_cmp_insn): Likewise.
3714 (prepare_float_lib_cmp): Likewise.
3715 (expand_float): Likewise.
3716 (expand_fix): Likewise.
3717 (expand_sfix_optab): Likewise.
3718 * postreload.c (move2add_use_add2_insn): Likewise.
3719 * reg-stack.c (reg_to_stack): Likewise.
3720 * reginfo.c (choose_hard_reg_mode): Likewise.
3721 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3722 * stor-layout.c (mode_for_size): Likewise.
3723 (smallest_mode_for_size): Likewise.
3724 (mode_for_vector): Likewise.
3725 (finish_bitfield_representative): Likewise.
3726 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3727 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3728 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3729 * var-tracking.c (prepare_call_arguments): Likewise.
3731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3732 Alan Hayward <alan.hayward@arm.com>
3733 David Sherwood <david.sherwood@arm.com>
3735 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3736 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3737 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3738 * machmode.h (mode_size): Move earlier in file.
3739 (mode_precision): Likewise.
3740 (mode_inner): Likewise.
3741 (mode_nunits): Likewise.
3742 (mode_unit_size): Likewise.
3743 (unit_unit_precision): Likewise.
3744 (mode_wider): Likewise.
3745 (mode_2xwider): Likewise.
3746 (machine_mode): New class.
3747 (mode_to_bytes): New function.
3748 (mode_to_bits): Likewise.
3749 (mode_to_precision): Likewise.
3750 (mode_to_inner): Likewise.
3751 (mode_to_unit_size): Likewise.
3752 (mode_to_unit_precision): Likewise.
3753 (mode_to_nunits): Likewise.
3754 (GET_MODE_SIZE): Use mode_to_bytes.
3755 (GET_MODE_BITSIZE): Use mode_to_bits.
3756 (GET_MODE_PRECISION): Use mode_to_precision.
3757 (GET_MODE_INNER): Use mode_to_inner.
3758 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3759 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3760 (GET_MODE_NUNITS): Use mode_to_nunits.
3761 * system.h (ALWAYS_INLINE): New macro.
3762 * config/powerpcspe/powerpcspe-c.c
3763 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3764 int for arg1_mode and arg2_mode.
3766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3767 Alan Hayward <alan.hayward@arm.com>
3768 David Sherwood <david.sherwood@arm.com>
3770 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3771 Prefix mode names with E_ in case statements.
3772 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3773 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3774 (aarch64_split_simd_move): Likewise.
3775 (aarch64_gen_storewb_pair): Likewise.
3776 (aarch64_gen_loadwb_pair): Likewise.
3777 (aarch64_gen_store_pair): Likewise.
3778 (aarch64_gen_load_pair): Likewise.
3779 (aarch64_get_condition_code_1): Likewise.
3780 (aarch64_constant_pool_reload_icode): Likewise.
3781 (get_rsqrte_type): Likewise.
3782 (get_rsqrts_type): Likewise.
3783 (get_recpe_type): Likewise.
3784 (get_recps_type): Likewise.
3785 (aarch64_gimplify_va_arg_expr): Likewise.
3786 (aarch64_simd_container_mode): Likewise.
3787 (aarch64_emit_load_exclusive): Likewise.
3788 (aarch64_emit_store_exclusive): Likewise.
3789 (aarch64_expand_compare_and_swap): Likewise.
3790 (aarch64_gen_atomic_cas): Likewise.
3791 (aarch64_emit_bic): Likewise.
3792 (aarch64_emit_atomic_swap): Likewise.
3793 (aarch64_emit_atomic_load_op): Likewise.
3794 (aarch64_evpc_trn): Likewise.
3795 (aarch64_evpc_uzp): Likewise.
3796 (aarch64_evpc_zip): Likewise.
3797 (aarch64_evpc_ext): Likewise.
3798 (aarch64_evpc_rev): Likewise.
3799 (aarch64_evpc_dup): Likewise.
3800 (aarch64_gen_ccmp_first): Likewise.
3801 (aarch64_gen_ccmp_next): Likewise.
3802 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3803 (alpha_emit_xfloating_libcall): Likewise.
3804 (emit_insxl): Likewise.
3805 (alpha_arg_type): Likewise.
3806 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3807 (arc_preferred_simd_mode): Likewise.
3808 (arc_secondary_reload): Likewise.
3809 (get_arc_condition_code): Likewise.
3810 (arc_print_operand): Likewise.
3811 (arc_legitimate_constant_p): Likewise.
3812 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3813 * config/arc/arc.md (casesi_load): Likewise.
3814 (casesi_compact_jump): Likewise.
3815 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3816 (cc_use_register): Likewise.
3817 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3818 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3819 (arm_init_iwmmxt_builtins): Likewise.
3820 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3821 (neon_expand_vector_init): Likewise.
3822 (arm_attr_length_move_neon): Likewise.
3823 (maybe_get_arm_condition_code): Likewise.
3824 (arm_emit_vector_const): Likewise.
3825 (arm_preferred_simd_mode): Likewise.
3826 (arm_output_iwmmxt_tinsr): Likewise.
3827 (thumb1_output_casesi): Likewise.
3828 (thumb2_output_casesi): Likewise.
3829 (arm_emit_load_exclusive): Likewise.
3830 (arm_emit_store_exclusive): Likewise.
3831 (arm_expand_compare_and_swap): Likewise.
3832 (arm_evpc_neon_vuzp): Likewise.
3833 (arm_evpc_neon_vzip): Likewise.
3834 (arm_evpc_neon_vrev): Likewise.
3835 (arm_evpc_neon_vtrn): Likewise.
3836 (arm_evpc_neon_vext): Likewise.
3837 (arm_validize_comparison): Likewise.
3838 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3839 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3840 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3841 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3842 (c6x_preferred_simd_mode): Likewise.
3843 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3844 (epiphany_rtx_costs): Likewise.
3845 * config/epiphany/predicates.md (proper_comparison_operator):
3847 * config/frv/frv.c (condexec_memory_operand): Likewise.
3848 (frv_emit_move): Likewise.
3849 (output_move_single): Likewise.
3850 (output_condmove_single): Likewise.
3851 (frv_hard_regno_mode_ok): Likewise.
3852 (frv_matching_accg_mode): Likewise.
3853 * config/h8300/h8300.c (split_adds_subs): Likewise.
3854 (h8300_rtx_costs): Likewise.
3855 (h8300_print_operand): Likewise.
3856 (compute_mov_length): Likewise.
3857 (output_logical_op): Likewise.
3858 (compute_logical_op_length): Likewise.
3859 (compute_logical_op_cc): Likewise.
3860 (h8300_shift_needs_scratch_p): Likewise.
3861 (output_a_shift): Likewise.
3862 (compute_a_shift_length): Likewise.
3863 (compute_a_shift_cc): Likewise.
3864 (expand_a_rotate): Likewise.
3865 (output_a_rotate): Likewise.
3866 * config/i386/i386.c (classify_argument): Likewise.
3867 (function_arg_advance_32): Likewise.
3868 (function_arg_32): Likewise.
3869 (function_arg_64): Likewise.
3870 (function_value_64): Likewise.
3871 (ix86_gimplify_va_arg): Likewise.
3872 (ix86_legitimate_constant_p): Likewise.
3873 (put_condition_code): Likewise.
3874 (split_double_mode): Likewise.
3875 (ix86_avx256_split_vector_move_misalign): Likewise.
3876 (ix86_expand_vector_logical_operator): Likewise.
3877 (ix86_split_idivmod): Likewise.
3878 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3879 (ix86_build_const_vector): Likewise.
3880 (ix86_build_signbit_mask): Likewise.
3881 (ix86_match_ccmode): Likewise.
3882 (ix86_cc_modes_compatible): Likewise.
3883 (ix86_expand_branch): Likewise.
3884 (ix86_expand_sse_cmp): Likewise.
3885 (ix86_expand_sse_movcc): Likewise.
3886 (ix86_expand_int_sse_cmp): Likewise.
3887 (ix86_expand_vec_perm_vpermi2): Likewise.
3888 (ix86_expand_vec_perm): Likewise.
3889 (ix86_expand_sse_unpack): Likewise.
3890 (ix86_expand_int_addcc): Likewise.
3891 (ix86_split_to_parts): Likewise.
3892 (ix86_vectorize_builtin_gather): Likewise.
3893 (ix86_vectorize_builtin_scatter): Likewise.
3894 (avx_vpermilp_parallel): Likewise.
3895 (inline_memory_move_cost): Likewise.
3896 (ix86_tieable_integer_mode_p): Likewise.
3897 (x86_maybe_negate_const_int): Likewise.
3898 (ix86_expand_vector_init_duplicate): Likewise.
3899 (ix86_expand_vector_init_one_nonzero): Likewise.
3900 (ix86_expand_vector_init_one_var): Likewise.
3901 (ix86_expand_vector_init_concat): Likewise.
3902 (ix86_expand_vector_init_interleave): Likewise.
3903 (ix86_expand_vector_init_general): Likewise.
3904 (ix86_expand_vector_set): Likewise.
3905 (ix86_expand_vector_extract): Likewise.
3906 (emit_reduc_half): Likewise.
3907 (ix86_emit_i387_round): Likewise.
3908 (ix86_mangle_type): Likewise.
3909 (ix86_expand_round_sse4): Likewise.
3910 (expand_vec_perm_blend): Likewise.
3911 (canonicalize_vector_int_perm): Likewise.
3912 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3913 (expand_vec_perm_1): Likewise.
3914 (expand_vec_perm_interleave3): Likewise.
3915 (expand_vec_perm_even_odd_pack): Likewise.
3916 (expand_vec_perm_even_odd_1): Likewise.
3917 (expand_vec_perm_broadcast_1): Likewise.
3918 (ix86_vectorize_vec_perm_const_ok): Likewise.
3919 (ix86_expand_vecop_qihi): Likewise.
3920 (ix86_expand_mul_widen_hilo): Likewise.
3921 (ix86_expand_sse2_abs): Likewise.
3922 (ix86_expand_pextr): Likewise.
3923 (ix86_expand_pinsr): Likewise.
3924 (ix86_preferred_simd_mode): Likewise.
3925 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3926 * config/i386/sse.md (*andnot<mode>3): Likewise.
3927 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3928 (*<code><mode>3): Likewise.
3929 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3930 (ia64_expand_atomic_op): Likewise.
3931 (ia64_arg_type): Likewise.
3932 (ia64_mode_to_int): Likewise.
3933 (ia64_scalar_mode_supported_p): Likewise.
3934 (ia64_vector_mode_supported_p): Likewise.
3935 (expand_vec_perm_broadcast): Likewise.
3936 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3937 (iq2000_function_arg_advance): Likewise.
3938 (iq2000_function_arg): Likewise.
3939 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3940 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3941 (m68k_libcall_value): Likewise.
3942 (m68k_function_value): Likewise.
3943 (sched_attr_op_type): Likewise.
3944 * config/mcore/mcore.c (mcore_output_move): Likewise.
3945 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3947 (microblaze_function_arg): Likewise.
3948 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3949 (mips_print_operand): Likewise.
3950 (mips_mode_ok_for_mov_fmt_p): Likewise.
3951 (mips_vector_mode_supported_p): Likewise.
3952 (mips_preferred_simd_mode): Likewise.
3953 (mips_expand_vpc_loongson_even_odd): Likewise.
3954 (mips_expand_vec_unpack): Likewise.
3955 (mips_expand_vi_broadcast): Likewise.
3956 (mips_expand_vector_init): Likewise.
3957 (mips_expand_vec_reduc): Likewise.
3958 (mips_expand_msa_cmp): Likewise.
3959 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3960 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3961 (cc_flags_for_mode): Likewise.
3962 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3963 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3964 (nds32_output_casesi_pc_relative): Likewise.
3965 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3966 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3967 (nvptx_gen_unpack): Likewise.
3968 (nvptx_gen_pack): Likewise.
3969 (nvptx_gen_shuffle): Likewise.
3970 (nvptx_gen_wcast): Likewise.
3971 (nvptx_preferred_simd_mode): Likewise.
3972 * config/pa/pa.c (pa_secondary_reload): Likewise.
3973 * config/pa/predicates.md (base14_operand): Likewise.
3974 * config/powerpcspe/powerpcspe-c.c
3975 (altivec_resolve_overloaded_builtin): Likewise.
3976 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3978 (rs6000_preferred_simd_mode): Likewise.
3979 (output_vec_const_move): Likewise.
3980 (rs6000_expand_vector_extract): Likewise.
3981 (rs6000_split_vec_extract_var): Likewise.
3982 (reg_offset_addressing_ok_p): Likewise.
3983 (rs6000_legitimate_offset_address_p): Likewise.
3984 (rs6000_legitimize_address): Likewise.
3985 (rs6000_emit_set_const): Likewise.
3986 (rs6000_const_vec): Likewise.
3987 (rs6000_emit_move): Likewise.
3988 (spe_build_register_parallel): Likewise.
3989 (rs6000_darwin64_record_arg_recurse): Likewise.
3990 (swap_selector_for_mode): Likewise.
3991 (spe_init_builtins): Likewise.
3992 (paired_init_builtins): Likewise.
3993 (altivec_init_builtins): Likewise.
3994 (do_load_for_compare): Likewise.
3995 (rs6000_generate_compare): Likewise.
3996 (rs6000_expand_float128_convert): Likewise.
3997 (emit_load_locked): Likewise.
3998 (emit_store_conditional): Likewise.
3999 (rs6000_output_function_epilogue): Likewise.
4000 (rs6000_handle_altivec_attribute): Likewise.
4001 (rs6000_function_value): Likewise.
4002 (emit_fusion_gpr_load): Likewise.
4003 (emit_fusion_p9_load): Likewise.
4004 (emit_fusion_p9_store): Likewise.
4005 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4006 (fusion_gpr_mem_load): Likewise.
4007 (fusion_addis_mem_combo_load): Likewise.
4008 (fusion_addis_mem_combo_store): Likewise.
4009 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4010 (fusion_gpr_mem_load): Likewise.
4011 (fusion_addis_mem_combo_load): Likewise.
4012 (fusion_addis_mem_combo_store): Likewise.
4013 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4015 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4016 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4017 (rs6000_preferred_simd_mode): Likewise.
4018 (output_vec_const_move): Likewise.
4019 (rs6000_expand_vector_extract): Likewise.
4020 (rs6000_split_vec_extract_var): Likewise.
4021 (reg_offset_addressing_ok_p): Likewise.
4022 (rs6000_legitimate_offset_address_p): Likewise.
4023 (rs6000_legitimize_address): Likewise.
4024 (rs6000_emit_set_const): Likewise.
4025 (rs6000_const_vec): Likewise.
4026 (rs6000_emit_move): Likewise.
4027 (rs6000_darwin64_record_arg_recurse): Likewise.
4028 (swap_selector_for_mode): Likewise.
4029 (paired_init_builtins): Likewise.
4030 (altivec_init_builtins): Likewise.
4031 (rs6000_expand_float128_convert): Likewise.
4032 (emit_load_locked): Likewise.
4033 (emit_store_conditional): Likewise.
4034 (rs6000_output_function_epilogue): Likewise.
4035 (rs6000_handle_altivec_attribute): Likewise.
4036 (rs6000_function_value): Likewise.
4037 (emit_fusion_gpr_load): Likewise.
4038 (emit_fusion_p9_load): Likewise.
4039 (emit_fusion_p9_store): Likewise.
4040 * config/rx/rx.c (rx_gen_move_template): Likewise.
4041 (flags_from_mode): Likewise.
4042 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4043 (s390_slb_comparison): Likewise.
4044 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4045 (s390_vector_mode_supported_p): Likewise.
4046 (s390_cc_modes_compatible): Likewise.
4047 (s390_match_ccmode_set): Likewise.
4048 (s390_canonicalize_comparison): Likewise.
4049 (s390_emit_compare_and_swap): Likewise.
4050 (s390_branch_condition_mask): Likewise.
4051 (s390_rtx_costs): Likewise.
4052 (s390_secondary_reload): Likewise.
4053 (__SECONDARY_RELOAD_CASE): Likewise.
4054 (s390_expand_cs): Likewise.
4055 (s390_preferred_simd_mode): Likewise.
4056 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4057 * config/sh/sh.c (sh_print_operand): Likewise.
4058 (dump_table): Likewise.
4059 (sh_secondary_reload): Likewise.
4060 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4061 * config/sh/sh.md (casesi_worker_1): Likewise.
4062 (casesi_worker_2): Likewise.
4063 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4064 (fcc_comparison_operator): Likewise.
4065 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4066 (emit_soft_tfmode_cvt): Likewise.
4067 (sparc_preferred_simd_mode): Likewise.
4068 (output_cbranch): Likewise.
4069 (sparc_print_operand): Likewise.
4070 (sparc_expand_vec_perm_bmask): Likewise.
4071 (vector_init_bshuffle): Likewise.
4072 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4073 (spu_vector_mode_supported_p): Likewise.
4074 (spu_expand_insv): Likewise.
4075 (spu_emit_branch_or_set): Likewise.
4076 (spu_handle_vector_attribute): Likewise.
4077 (spu_builtin_splats): Likewise.
4078 (spu_builtin_extract): Likewise.
4079 (spu_builtin_promote): Likewise.
4080 (spu_expand_sign_extend): Likewise.
4081 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4082 (tilegx_simd_int): Likewise.
4083 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4084 (tilepro_simd_int): Likewise.
4085 * config/v850/v850.c (const_double_split): Likewise.
4086 (v850_print_operand): Likewise.
4087 (ep_memory_offset): Likewise.
4088 * config/vax/vax.c (vax_rtx_costs): Likewise.
4089 (vax_output_int_move): Likewise.
4090 (vax_output_int_add): Likewise.
4091 (vax_output_int_subtract): Likewise.
4092 * config/visium/predicates.md (visium_branch_operator): Likewise.
4093 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4094 (visium_print_operand_address): Likewise.
4095 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4096 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4097 (xtensa_expand_conditional_branch): Likewise.
4098 (xtensa_copy_incoming_a7): Likewise.
4099 (xtensa_output_literal): Likewise.
4100 * dfp.c (decimal_real_maxval): Likewise.
4101 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4104 Alan Hayward <alan.hayward@arm.com>
4105 David Sherwood <david.sherwood@arm.com>
4107 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4108 (mode_nunits_inline): Likewise.
4109 (mode_inner_inline): Likewise.
4110 (mode_unit_size_inline): Likewise.
4111 (mode_unit_precision_inline): Likewise.
4112 (emit_insn_modes_h): Likewise. Also emit a #define of the
4114 (emit_mode_wider): Add an E_ prefix to mode names.
4115 (emit_mode_complex): Likewise.
4116 (emit_mode_inner): Likewise.
4117 (emit_mode_adjustments): Likewise.
4118 (emit_mode_int_n): Likewise.
4119 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4120 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4121 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4122 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4123 (CRC32_BUILTIN, ENTRY): Likewise.
4124 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4125 (aarch64_pop_regs): Likewise.
4126 (aarch64_process_components): Likewise.
4127 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4128 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4129 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4130 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4131 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4132 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4134 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4135 (pcmpistr): Likewise.
4136 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4137 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4138 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4140 * config/rl78/rl78.c (mduc_regs): Likewise.
4141 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4142 (htm_expand_builtin): Likewise.
4143 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4144 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4145 * config/xtensa/xtensa.c (print_operand): Likewise.
4146 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4147 (NUM_MODE_VECTOR_INT): Likewise.
4148 * genoutput.c (null_operand): Likewise.
4149 (output_operand_data): Likewise.
4150 * genrecog.c (print_parameter_value): Likewise.
4151 * lra.c (debug_operand_data): Likewise.
4153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4155 * dwarf2out.c (scompare_loc_descriptor_wide)
4156 (scompare_loc_descriptor_narrow): New functions, split out from...
4157 (scompare_loc_descriptor): ...here.
4158 * expmed.c (emit_store_flag_int): New function, split out from...
4159 (emit_store_flag): ...here.
4161 2017-08-30 Richard Biener <rguenther@suse.de>
4163 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4164 (dwarf2out_early_finish): Move setting of AT_pubnames from
4165 early debug output to early finish.
4167 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4169 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4170 and -mdata-region to the assembler.
4172 2017-08-30 Richard Biener <rguenther@suse.de>
4174 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4176 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4177 (dwarf2out_early_global_decl): Only generate a DIE for the
4178 abstract origin if it doesn't already exist or is a declaration DIE.
4179 (resolve_addr): Do not add the linkage name twice when
4180 generating a stub DIE for the DW_TAG_GNU_call_site target.
4182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4184 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4185 Use machine_mode rather than int for arg1_mode.
4187 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4190 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4191 that the second argument of the built-in functions to unpack
4192 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4193 switch statement instead a lot of if statements.
4194 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4195 Allow 64-bit values to be in Altivec registers as well as
4196 traditional floating point registers.
4197 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4199 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4201 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4202 MAX_REGS_PER_ADDRESS == 1.
4204 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4206 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4207 * config/i386/i386.c (ix86_option_override_internal): Simplify
4208 setting of opts->x_flag_entry.
4210 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4211 Jakub Jelinek <jakub@redhat.com>
4212 Richard Biener <rguenther@suse.de>
4214 PR tree-optimization/81503
4215 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4216 folded constant fits in the target type; reorder tests for clarity.
4218 2017-08-29 Martin Liska <mliska@suse.cz>
4220 * passes.def: Include pass_lower_switch.
4221 * stmt.c (dump_case_nodes): Remove and move to
4222 tree-switch-conversion.
4223 (case_values_threshold): Likewise.
4224 (expand_switch_as_decision_tree_p): Likewise.
4225 (emit_case_decision_tree): Likewise.
4226 (expand_case): Likewise.
4227 (balance_case_nodes): Likewise.
4228 (node_has_low_bound): Likewise.
4229 (node_has_high_bound): Likewise.
4230 (node_is_bounded): Likewise.
4231 (emit_case_nodes): Likewise.
4232 (struct simple_case_node): New struct.
4233 (add_case_node): Remove.
4234 (emit_case_dispatch_table): Use vector instead of case_list.
4235 (reset_out_edges_aux): Remove.
4236 (compute_cases_per_edge): Likewise.
4237 (expand_case): Build list of simple_case_node.
4238 (expand_sjlj_dispatch_table): Use it.
4239 * tree-switch-conversion.c (struct case_node): Moved from
4240 stmt.c and adjusted.
4241 (emit_case_nodes): Likewise.
4242 (node_has_low_bound): Likewise.
4243 (node_has_high_bound): Likewise.
4244 (node_is_bounded): Likewise.
4245 (case_values_threshold): Likewise.
4246 (reset_out_edges_aux): Likewise.
4247 (compute_cases_per_edge): Likewise.
4248 (add_case_node): Likewise.
4249 (dump_case_nodes): Likewise.
4250 (balance_case_nodes): Likewise.
4251 (expand_switch_as_decision_tree_p): Likewise.
4252 (emit_jump): Likewise.
4253 (emit_case_decision_tree): Likewise.
4254 (try_switch_expansion): Likewise.
4255 (do_jump_if_equal): Likewise.
4256 (emit_cmp_and_jump_insns): Likewise.
4257 (fix_phi_operands_for_edge): New function.
4258 (record_phi_operand_mapping): Likewise.
4259 (class pass_lower_switch): New pass.
4260 (pass_lower_switch::execute): New function.
4261 (make_pass_lower_switch): Likewise.
4262 (conditional_probability):
4263 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4264 * tree-pass.h: Add make_pass_lower_switch.
4266 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4269 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4272 2017-08-29 Richard Biener <rguenther@suse.de>
4274 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4275 we do not add a DW_AT_inline attribute twice.
4276 (gen_subprogram_die): Remove code setting DW_AT_inline on
4277 DECL_ABSTRACT_P nodes.
4279 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4281 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4282 calls to internal functions.
4283 (gimplify_modify_expr): Likewise.
4284 * tree-call-cdce.c (use_internal_fn): Likewise.
4285 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4286 (convert_to_divmod): Set the nothrow flag.
4287 * tree-if-conv.c (predicate_mem_writes): Likewise.
4288 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4289 (vectorizable_call): Likewise.
4290 (vectorizable_store): Likewise.
4291 (vectorizable_load): Likewise.
4292 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4293 (vect_recog_mask_conversion_pattern): Likewise.
4295 2017-08-29 Martin Liska <mliska@suse.cz>
4298 * gcc.c (driver_handle_option): Add new argument.
4299 * opts-common.c (handle_option): Pass
4300 target_option_override_hook.
4301 * opts-global.c (lang_handle_option): Add new option.
4302 (set_default_handlers): Add new argument.
4303 (decode_options): Likewise.
4304 * opts.c (target_handle_option): Likewise.
4305 (common_handle_option): Call target_option_override_hook.
4306 * opts.h (struct cl_option_handler_func): Add hook for
4307 target option override.
4308 (struct cl_option_handlers): Likewise.
4309 (set_default_handlers): Add new argument.
4310 (decode_options): Likewise.
4311 (common_handle_option): Likewise.
4312 (target_handle_option): Likewise.
4313 * toplev.c (toplev::main): Pass targetm.target_option.override
4316 2017-08-29 Richard Biener <rguenther@suse.de>
4317 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4319 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4320 life to the active subtree.
4322 2017-08-28 Jeff Law <law@redhat.com>
4324 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4325 derive_equivalences.
4326 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4328 (edge_info::derive_equivalences): New private member function
4330 * tree-ssa-dom.c (class edge_info): Changed from a struct
4331 to a class. Add ctor/dtor, methods and data members.
4332 (edge_info::edge_info): Renamed from allocate_edge_info.
4333 Initialize additional members.
4334 (edge_info::~edge_info): New.
4335 (free_dom_edge_info): Delete the edge info.
4336 (record_edge_info): Use new class & associated member functions.
4337 Tighten forms for testing for edge equivalences.
4338 (record_temporary_equivalences): Iterate over the simple
4339 equivalences rather than assuming there's only one per edge.
4340 (cprop_into_successor_phis): Iterate over the simple
4341 equivalences rather than assuming there's only one per edge.
4342 (optimize_stmt): Use operand_equal_p rather than pointer
4343 equality for mini-DSE code.
4345 2017-08-28 Nathan Sidwell <nathan@acm.org>
4347 * gcc.c (execute): Fold SIGPIPE handling into switch
4348 statement. Adjust internal error message.
4350 2017-08-28 Richard Biener <rguenther@suse.de>
4353 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4354 Do nothing for removed DIEs.
4356 2017-08-28 Richard Biener <rguenther@suse.de>
4358 PR tree-optimization/81977
4359 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4362 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4365 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4366 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4367 using targetm.gen_mem_thread_fence.
4369 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4372 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4373 predicate to register_operand. Reorder operands.
4374 (*btr<mode>): Ditto.
4375 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4376 (*btr<mode>_mask): Ditto.
4378 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4380 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4381 * config/rs6000/xmmintrin.h: New file.
4382 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4384 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4387 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4388 parameter and_insn and return it.
4389 (recombine_lvx_pattern): Insert a copy to ensure availability of
4390 the base register of the copied masking operation at the point of
4391 the instruction replacement.
4392 (recombine_stvx_pattern): Likewise.
4394 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4396 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4397 undocumented switches.
4398 (-mpower9-dform-vector): Likewise.
4399 (-mpower9-dform): Likewise.
4400 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4401 comments to delete references to -mpower9-dform* switches.
4402 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4403 Delete reference to -mpower9-dform* switches, test for
4404 -mpower9-vector instead.
4405 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4406 (OTHER_P9_VECTOR_MASKS): Likewise.
4407 (POWERPC_MASKS): Likewise.
4408 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4409 tests against -mpower9-dform* to -mpower9-vector. Delete code
4410 that checked for -mpower9-dform* consistancy with other options.
4411 Add test for -mpower9-misc to enable other power9 switches.
4412 (rs6000_init_hard_regno_mode_ok): Likewise.
4413 (rs6000_option_override_internal): Likewise.
4414 (rs6000_emit_prologue): Likewise.
4415 (rs6000_emit_epilogue): Likewise.
4416 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4417 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4418 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4420 (emit_fusion_p9_store): Likewise.
4421 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4422 resetting these macros if the assembler does not support ISA 3.0
4424 (TARGET_P9_DFORM_VECTOR): Likewise.
4425 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4426 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4428 2017-08-25 Alan Modra <amodra@gmail.com>
4430 PR rtl-optimization/81747
4431 * cse.c (cse_extended_basic_block): Don't attempt to record
4432 equivalences for degenerate conditional branches that branch
4433 to their fall-through.
4435 2017-08-24 Martin Sebor <msebor@redhat.com>
4438 * gimple-fold.c (size_must_be_zero_p): New function.
4439 (gimple_fold_builtin_memory_op): Call it.
4441 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4443 * config/rs6000/mm_malloc.h: New file.
4445 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4447 PR tree-optimization/81913
4448 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4449 analysis when either IVs in condition can wrap.
4451 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4453 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4454 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4456 2017-08-24 Richard Biener <rguenther@suse.de>
4459 * targhooks.c (default_target_can_inline_p): Properly
4460 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4461 is present and always compare.
4462 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4463 infer -mfpmath=sse from TARGET_SSE_P.
4464 (ix86_can_inline_p): Properly use target_option_default_node when
4465 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4467 2017-08-24 Richard Biener <rguenther@suse.de>
4470 * dwarf2out.c (output_die): Handle flag_generate_offload like
4472 (output_comp_unit): Likewise.
4473 (gen_array_type_die): Likewise.
4474 (dwarf2out_early_finish): Likewise.
4475 (note_variable_value_in_expr): Likewise.
4476 (dwarf2out_finish): Likewise. Adjust assert.
4477 * cgraphunit.c (symbol_table::compile): Move setting of
4478 flag_generate_offload earlier ...
4479 (symbol_table::finalize_compilation_unit): ... here, before
4480 early debug finalization.
4482 2017-08-24 Richard Biener <rguenther@suse.de>
4484 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4485 and ipa-fnsummary.h.
4486 (ix86_can_inline_p): When ix86_fpmath flags do not match
4487 check whether the callee uses FP math at all.
4489 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4492 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4493 instead of TYPE_PRECISION.
4495 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4496 Alan Hayward <alan.hayward@arm.com>
4497 David Sherwood <david.sherwood@arm.com>
4499 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4501 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4503 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4505 (ix86_frame::stack_realign_allocate): New field.
4506 (struct machine_frame_state): Modify comments.
4507 (machine_frame_state::sp_realigned_fp_end): New field.
4508 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4510 (sp_valid_at): Add assertion to assure no attempt to access invalid
4511 offset of a realigned stack.
4512 (fp_valid_at): Likewise.
4513 (choose_baseaddr): Modify comments.
4514 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4515 ix86_expand_prologue.
4516 (ix86_expand_prologue): Modify stack realignment and allocation.
4517 (ix86_expand_epilogue): Modify comments.
4518 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4519 avx2_runtime, avx512f, and avx512f_runtime.
4521 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4523 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4524 (mstackrealign): Do not init to -1.
4525 * config/i386/i386.c (ix86_option_override_internal):
4526 Check opts_set, not opts when setting default value of
4527 opts->x_ix86_force_align_arg_pointer.
4529 2017-08-23 Richard Biener <rguenther@suse.de>
4531 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4532 lang_hooks.decl_printable_name.
4533 * print-rtl-function.c (print_rtx_function): Likewise.
4534 * tree-pretty-print.c (dump_function_header): Likewise.
4536 2017-08-23 Richard Biener <rguenther@suse.de>
4539 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4540 -g0 at compile-time.
4542 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4545 * doc/sourcebuild.texi (Other hardware attributes):
4548 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4551 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4552 Fix single-use check.
4554 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4556 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4558 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4560 * doc/install.texi: Modify to add more details on running selected
4563 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4565 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4566 is combined with -mabi=ms.
4567 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4570 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4572 PR tree-optimization/81488
4573 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4574 and cached_basis fields.
4575 (MAX_SPREAD): New constant.
4576 (alloc_cand_and_find_basis): Initialize new fields.
4577 (clear_visited): New function.
4578 (create_phi_basis_1): Rename from create_phi_basis, set visited
4579 and cached_basis fields.
4580 (create_phi_basis): New wrapper function.
4581 (phi_add_costs_1): Rename from phi_add_costs, add spread
4582 parameter, set visited field, short-circuit when limits reached.
4583 (phi_add_costs): New wrapper function.
4584 (record_phi_increments_1): Rename from record_phi_increments, set
4586 (record_phi_increments): New wrapper function.
4587 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4588 (phi_incr_cost): New wrapper function.
4589 (all_phi_incrs_profitable_1): Rename from
4590 all_phi_incrs_profitable, set visited field.
4591 (all_phi_incrs_profitable): New wrapper function.
4593 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4594 Alan Hayward <alan.hayward@arm.com>
4595 David Sherwood <david.sherwood@arm.com>
4597 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4598 that takes the outer and inner modes.
4599 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4600 comparison as the canonical test for a paradoxical subreg.
4601 * combine.c (simplify_set): Use paradoxical_subreg_p.
4602 (make_extraction): Likewise.
4603 (force_to_mode): Likewise.
4604 (rtx_equal_for_field_assignment_p): Likewise.
4605 (gen_lowpart_for_combine): Likewise.
4606 (simplify_comparison): Likewise.
4607 * cse.c (equiv_constant): Likewise.
4608 * expmed.c (store_bit_field_1): Likewise.
4609 * final.c (alter_subreg): Likewise.
4610 * fwprop.c (propagate_rtx): Likewise.
4611 (forward_propagate_subreg): Likewise.
4612 * ira-conflicts.c (ira_build_conflicts): Likewise.
4613 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4614 * lra-constraints.c (curr_insn_transform): Likewise.
4615 (split_reg): Likewise.
4616 * lra-eliminations.c (move_plus_up): Likewise.
4617 (lra_eliminate_regs_1): Likewise.
4618 * recog.c (general_operand): Likewise.
4619 * ree.c (combine_reaching_defs): Likewise.
4620 * reload.c (push_reload): Likewise.
4621 (find_reloads): Likewise.
4622 * reload1.c (elimination_effects): Likewise.
4623 (compute_reload_subreg_offset): Likewise.
4624 (choose_reload_regs): Likewise.
4625 * rtlanal.c (subreg_lsb_1): Likewise.
4626 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4627 (simplify_subreg): Likewise.
4628 * var-tracking.c (track_loc_p): Likewise.
4629 * emit-rtl.c (byte_lowpart_offset): Likewise.
4630 (paradoxical_subreg_p): Delete out-of-line definition.
4632 2017-08-22 Jeff Law <law@redhat.com>
4634 PR tree-optimization/81741
4635 PR tree-optimization/71947
4636 * tree-ssa-dom.c: Include tree-inline.h.
4637 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4638 equivalences if one is more expensive to compute than the other.
4639 * tree-ssa-scopedtables.h (class const_or_copies): Make
4640 record_const_or_copy_raw method private.
4641 (class avail_exprs_stack): New method simplify_binary_operation.
4642 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4643 avail_exprs_stack::simplify_binary_operation as needed.
4644 (avail_exprs_stack::simplify_binary_operation): New function.
4646 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4648 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4649 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4650 (DOT_SYMBOLS): Likewise.
4651 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4652 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4653 (RS6000_ABI_NAME): Likewise.
4654 (TARGET_CMODEL): Likewise.
4655 (TOC_SECTION_ASM_OP): Likewise.
4656 (SET_CMODEL): New macro.
4657 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4659 2017-08-22 Richard Biener <rguenther@suse.de>
4661 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4662 to allow for free-lang-data replacements similar to verify_type_variant.
4664 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4665 Alan Hayward <alan.hayward@arm.com>
4666 David Sherwood <david.sherwood@arm.com>
4668 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4669 VOIDmode for the LABEL_REF.
4671 2017-08-22 Richard Biener <rguenther@suse.de>
4673 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4675 2017-08-22 Richard Biener <rguenther@suse.de>
4677 * common.opt (feliminate-dwarf2-dups): Ignore.
4678 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4679 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4680 same_die_p_wrap, compute_section_prefix,
4681 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4682 (comdat_symbol_id, comdat_symbol_number): Likewise.
4683 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4685 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4686 (output_die): Mark unreachable path unreachable.
4687 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4688 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4689 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4690 (dwarf2out_early_finish): Likewise.
4692 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4694 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4696 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4699 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4700 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4701 (avr_attribute_table) <io, io_low, address>: Initialize
4702 .decl_required with true.
4704 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4706 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4707 undocumented debugging options.
4708 (-mvsx-scalar-double): Likewise.
4709 (-mallow-df-permute): Likewise.
4710 (-mvectorize-builtins): Likewise.
4711 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4712 (rs6000_builtin_vectorized_function): Likewise.
4713 (rs6000_builtin_md_vectorized_function): Likewise.
4714 (rs6000_opt_vars): Likewise.
4716 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4719 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4720 (*btrq_imm): Rename from *btrq.
4721 (*btcq_imm): Rename from *btcq.
4722 (btsc): New code attribute.
4723 (*<btsc><mode>): New insn pattern.
4724 (*btr<mode>): Ditto.
4725 (*<btsc><mode>_mask): New insn_and_split pattern.
4726 (*btr<mode>_mask): Ditto.
4728 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4729 Alan Hayward <alan.hayward@arm.com>
4730 David Sherwood <david.sherwood@arm.com>
4732 * function.c (pad_below): Simplify padding calculation.
4734 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4735 Alan Hayward <alan.hayward@arm.com>
4736 David Sherwood <david.sherwood@arm.com>
4738 * target.def (function_prologue): Remove frame size argument.
4739 (function_epilogue): Likewise.
4740 * doc/tm.texi: Regenerate.
4741 * final.c (final_start_function): Update call to function_prologue.
4742 (final_end_function): Update call to function_epilogue.
4743 (default_function_pro_epilogue): Remove frame size argument.
4744 * output.h (default_function_pro_epilogue): Likewise.
4745 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4746 (arm_output_function_prologue): Likewise.
4747 * config/frv/frv.c (frv_function_prologue): Likewise.
4748 (frv_function_epilogue): Likewise.
4749 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4750 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4751 (ia64_output_function_epilogue): Likewise.
4752 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4753 (m32r_output_function_epilogue): Likewise.
4754 * config/microblaze/microblaze.c (microblaze_function_prologue)
4755 (microblaze_function_epilogue): Likewise.
4756 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4757 (mips_output_function_epilogue): Likewise.
4758 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4759 (mmix_target_asm_function_epilogue): Likewise.
4760 * config/msp430/msp430.c (msp430_start_function): Likewise.
4761 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4762 (nds32_asm_function_epilogue): Likewise.
4763 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4764 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4765 (pa_output_function_epilogue): Likewise.
4766 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4767 (rs6000_output_function_epilogue): Likewise.
4768 * config/rl78/rl78.c (rl78_start_function): Likewise.
4769 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4770 (rs6000_output_function_epilogue): Likewise.
4771 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4772 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4773 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4774 (sparc_asm_function_epilogue): Likewise.
4776 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4778 * tree.h (type_has_mode_precision_p): New function.
4779 * convert.c (convert_to_integer_1): Use it.
4780 * expr.c (expand_expr_real_2): Likewise.
4781 (expand_expr_real_1): Likewise.
4782 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4783 * match.pd: Likewise.
4784 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4785 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4786 * tree-tailcall.c (process_assignment): Likewise.
4787 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4788 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4789 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4790 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4791 (vectorizable_assignment): Likewise.
4792 (vectorizable_shift): Likewise.
4793 (vectorizable_operation): Likewise.
4794 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4796 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4798 * match.pd: Add pow (C, x) simplification.
4800 2017-08-21 Richard Biener <rguenther@suse.de>
4802 PR tree-optimization/81900
4803 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4804 for blocks with abnormal predecessors.
4805 (compute_antic): Do not set visited flag prematurely.
4807 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4810 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4812 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4814 * stor-layout.h (vector_type_mode): Move to...
4815 * tree.h (vector_type_mode): ...here.
4816 * stor-layout.c (vector_type_mode): Move to...
4817 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4819 2017-08-21 Richard Biener <rguenther@suse.de>
4821 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4822 register_external_die hooks.
4823 (debug_false_tree_charstarstar_uhwistar): Declare.
4824 (debug_nothing_tree_charstar_uhwi): Likewise.
4825 * debug.c (do_nothing_debug_hooks): Adjust.
4826 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4827 (debug_nothing_tree_charstar_uhwi): Likewise.
4828 * dbxout.c (dbx_debug_hooks): Adjust.
4829 (xcoff_debug_hooks): Likewise.
4830 * sdbout.c (sdb_debug_hooks): Likewise.
4831 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4832 * dwarf2out.c (macinfo_label_base): New global.
4833 (dwarf2out_register_external_die): New function for the
4834 register_external_die hook.
4835 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4836 (dwarf2_debug_hooks): Use them.
4837 (dwarf2_lineno_debug_hooks): Adjust.
4838 (struct die_struct): Add with_offset flag.
4839 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4840 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4841 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4842 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4843 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4844 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4845 defining section names for the early LTO debug variants.
4846 (reset_indirect_string): New helper.
4847 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4848 (print_dw_val): Add support for offsetted symbol references.
4849 (get_ultimate_context): Split out from is_cxx.
4850 (is_cxx): Use get_ultimate_context.
4851 (is_fortran): Add decl overload.
4852 (compute_comp_unit_symbol): Split out worker from
4853 compute_section_prefix.
4854 (compute_section_prefix): Call compute_comp_unit_symbol and
4855 set comdat_type_p here.
4856 (output_die): Skip DIE symbol output for the LTO added one.
4857 Handle DIE symbol references with offset.
4858 (output_comp_unit): Guard section name mangling properly.
4859 For LTO debug sections emit a symbol at the section beginning
4860 which we use to refer to its DIEs.
4861 (add_abstract_origin_attribute): For DIEs registered via
4862 dwarf2out_register_external_die directly refer to the early
4863 DIE rather than indirectly through the shadow one we created.
4864 Remove obsolete call to dwarf2out_abstract_function for
4865 non-function/block origins.
4866 (gen_array_type_die): When generating early LTO debug do
4867 not emit DW_AT_string_length.
4868 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4869 late when in LTO. As suggested place a gcc_unreachable for
4870 the DECL_ABSTRACT_P case.
4871 (gen_subprogram_die): Avoid another specification DIE
4872 for early built declarations/definitions for the late LTO case.
4873 (gen_variable_die): Add type references for late duplicated VLA dies
4875 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4876 we have the abstract instance already.
4877 (process_scope_var): Adjust decl DIE contexts in LTO which
4878 first puts them in limbo.
4879 (gen_decl_die): Do not generate type DIEs late apart from
4880 types for VLAs or for decls we do not yet have a DIE. Do not
4881 call dwarf2out_abstract_function late.
4882 (dwarf2out_early_global_decl): Make sure to create DIEs
4883 for abstract instances of a decl first.
4884 (dwarf2out_late_global_decl): Adjust comment.
4885 (output_macinfo_op): With multiple macro sections use
4886 macinfo_label_base to distinguish labels.
4887 (output_macinfo): Likewise. Update macinfo_label_base.
4888 Pass in the line info label.
4889 (note_variable_value_in_expr): When generating LTO resolve
4890 all variable values here by generating DIEs as needed.
4891 (init_sections_and_labels): Add early LTO debug flag parameter
4892 and generate different sections and names if set. Add generation
4893 counter for the labels so we can have multiple of them.
4894 (reset_dies): Helper to allow DIEs to be output multiple times.
4895 (dwarf2out_finish): When outputting DIEs to the fat part of an
4896 LTO object first reset DIEs.
4897 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4898 (modified_type_die): Check for decl_ultimate_origin being self
4900 (gen_type_die_with_usage): Likewise.
4901 (gen_typedef_die): Allow decl_ultimate_origin being self.
4902 (set_decl_abstract_flags): Remove.
4903 (set_block_abstract_flags): Likewise.
4904 (dwarf2out_abstract_function): Treat the early generated DIEs
4905 as the abstract copy and only add DW_AT_inline and
4906 DW_AT_artificial here and call set_decl_origin_self.
4907 If the DIE has an abstract origin don't do anything.
4908 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4909 if we have none yet (Go fails to build one, PR78628).
4910 (variably_modified_type_p): Prevent endless recursion for Ada
4911 cyclic pointer types.
4912 * lto-streamer-in.c: Include debug.h.
4913 (dref_queue): New global.
4914 (lto_read_tree_1): Stream in DIE references.
4915 (lto_input_tree): Register DIE references.
4916 (input_function): Stream DECL_DEBUG_ARGS.
4917 * lto-streamer-out.c: Include debug.h.
4918 (lto_write_tree_1): Output DIE references.
4919 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4920 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4921 (output_function): Stream DECL_DEBUG_ARGS.
4922 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4923 Stream DECL_ABSTRACT_ORIGIN.
4924 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4925 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4926 DECL_CONTEXT for file-scope decls.
4927 * lto-streamer.h (struct dref_entry): Declare.
4928 (dref_queue): Likewise.
4929 * cfgexpand.c (pass_expand::execute): Do not call the
4930 outlining_inline_function hook here.
4931 * lto-wrapper.c (debug_obj): New global.
4932 (tool_cleanup): Unlink it if required.
4933 (debug_objcopy): New function.
4934 (run_gcc): Handle early debug sections in the IL files by
4935 extracting them to separate files, partially linkin them and
4936 feeding the result back as result to the linker.
4937 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4938 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4939 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4940 sections into a separate segment.
4941 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4943 (darwin_asm_dwarf_section): Likewise.
4944 (darwin_asm_output_dwarf_offset): Likewise.
4945 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4947 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4948 Alan Hayward <alan.hayward@arm.com>
4949 David Sherwood <david.sherwood@arm.com>
4951 * read-md.h (md_reader::record_potential_iterator_use): Replace
4952 pointer argument with an rtx and an index.
4953 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4954 (apply_mode_iterator): Likewise.
4955 (apply_code_iterator): Likewise.
4956 (apply_int_iterator): Likewise.
4957 (apply_subst_iterator): Likewise.
4958 (record_iterator_use): Likewise.
4959 (record_attribute_use): Likewise.
4960 (md_reader::record_potential_iterator_use): Likewise. Update calls
4961 to record_iterator_use and apply_iterator.
4962 (iterator_use): Replace ptr with x and index.
4963 (attribute_use): Likewise.
4964 (apply_attribute_uses): Update calls to apply_iterator.
4965 (apply_iterators): Likewise. Update initialization of iterator_use.
4966 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4967 and record_potential_iterator_use.
4968 (rtx_reader::read_rtx_operand): Likewise.
4970 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4971 Alan Hayward <alan.hayward@arm.com>
4972 David Sherwood <david.sherwood@arm.com>
4974 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4977 2017-08-21 Richard Biener <rguenther@suse.de>
4980 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4981 at struct end conservatively when comparing common bases.
4983 2017-08-21 Richard Biener <rguenther@suse.de>
4985 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4986 (mem_ref_in_stmt): Remove.
4987 (determine_max_movement): Use ref index to get at the reference.
4988 (invariantness_dom_walker::before_dom_children): Deal with
4989 lim data already initialized.
4990 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4992 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4994 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4995 (TARGET_ISA_ROUND): Ditto.
4996 (TARGET_ROUND): Ditto.
4997 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4998 * config/i386/i386.md: Ditto.
4999 * config/i386/sse.md: Ditto.
5000 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5001 with OPTION_MASK_ISA_SSE4_1.
5003 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5006 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5007 __builtin_ia32_lzcnt_u16.
5009 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5012 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5013 (rs6000_set_current_function): Rewrite function to use it.
5015 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5018 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5019 and TYPE_WARN_IF_NOT_ALIGN.
5020 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5021 (handle_warn_if_not_align): New.
5022 (place_union_field): Call handle_warn_if_not_align.
5023 (place_field): Call handle_warn_if_not_align.
5024 Copy TYPE_WARN_IF_NOT_ALIGN.
5025 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5026 (layout_type): Likewise.
5027 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5029 (tree_decl_common): Add warn_if_not_align.
5030 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5031 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5032 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5033 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5034 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5035 * doc/extend.texi: Document warn_if_not_aligned attribute.
5036 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5038 2017-08-17 Martin Liska <mliska@suse.cz>
5041 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5042 (get_data_dependence): Use it as pointer type.
5043 (distribute_loop): Likewise.
5045 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5047 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5048 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5049 (p8_vmrgow_<mode>_direct): New define_insn.
5050 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5051 handle endianness for vmrgew and vmrgow permute patterns.
5053 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5055 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5056 * config/rs6000/rs6000-cpus.def: Remove comment.
5057 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5058 (POWERPC_MASKS): Likewise.
5059 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5060 use of TARGET_VSX_TIMODE.
5061 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5062 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5063 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5064 (rs6000_option_override_internal): Remove dead code.
5065 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5066 (rs6000_legitimize_reload_address): Likewise.
5067 (rs6000_legitimate_address_p): Likewise.
5068 (rs6000_opt_masks): Delete "vsx-timode".
5069 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5070 from function comment.
5071 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5072 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5073 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5075 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5076 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5077 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5080 (define_peephole2): Likewise.
5082 2017-08-17 Martin Sebor <msebor@redhat.com>
5085 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5086 past the end of an array.
5087 (test_pp_format): Add test cases.
5089 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5091 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5092 missing ECF_NOTHROW flags.
5094 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5097 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5098 operands residing in integer registers.
5099 (*vsx_le_perm_load_<mode>): Likewise.
5100 (*vsx_le_perm_store_<mode>): Likewise.
5101 (define_peephole2): Add peepholes to optimize the above.
5103 2017-08-17 Marek Polacek <polacek@redhat.com>
5106 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5107 to mimic what shorten_compare did. Change the return type to bool.
5108 (fold_cond_expr_with_comparison): Update call to
5109 operand_equal_for_comparison_p.
5110 (fold_ternary_loc): Likewise.
5112 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5114 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5116 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5118 2017-08-17 Richard Biener <rguenther@suse.de>
5120 * tree-ssa-structalias.c (solve_graph): When propagating
5121 to successors update the graphs succ edges and avoid duplicate work.
5123 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5126 * config/i386/i386.c (ix86_option_override_internal): Save target
5127 specific options after ix86_stack_protector_guard_reg was changed.
5129 2017-08-17 Richard Biener <rguenther@suse.de>
5131 PR tree-optimization/81827
5132 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5133 (new_var_info): Initialize it conservatively.
5134 (get_call_vi): Mark register vars.
5135 (new_scalar_tmp_constraint_exp): Likewise.
5136 (handle_rhs_call): Likewise.
5137 (handle_const_call): Likewise.
5138 (create_function_info_for): Likewise.
5139 (solve_constraints): Sort varinfos to separate register from
5140 non-register vars to pack points-to solution bitmaps during
5143 2017-08-17 Marek Polacek <polacek@redhat.com>
5145 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5147 2017-08-17 Richard Biener <rguenther@suse.de>
5149 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5150 to true when overflow is undefined and we saturated the result.
5152 2017-08-17 Alan Modra <amodra@gmail.com>
5155 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5156 Don't use store multiple if only one reg needs saving.
5157 (interesting_frame_related_regno): New function.
5158 (rs6000_frame_related): Don't emit frame info for regs that
5160 (rs6000_emit_epilogue): Likewise.
5162 2017-08-16 Nathan Sidwell <nathan@acm.org>
5164 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5165 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5166 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5167 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5168 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5169 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5170 (verify_type): Adjust for TYPE_BINFO move.
5171 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5172 process TYPE_BINFO directly.
5173 (hash_tree): Likewise.
5174 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5176 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5179 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5181 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5183 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5186 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5187 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5188 Add ix86_binary_operator_ok to insn constraint.
5189 (*iordi_1_bts): Ditto.
5190 (*xordi_1_btc): Ditto.
5191 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5192 Update corresponding peephole2 pattern.
5196 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5198 PR tree-optimization/81832
5199 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5200 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5202 2017-08-16 Marek Polacek <polacek@redhat.com>
5205 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5206 regarding size_zero_node.
5208 2017-08-16 Martin Liska <mliska@suse.cz>
5211 * config.gcc: Respect previously set extra_objs in case
5214 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5216 PR tree-optimization/81835
5217 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5218 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5219 not depend on the phi.
5221 2017-08-16 Alan Modra <amodra@gmail.com>
5223 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5226 2017-08-16 Alan Modra <amodra@gmail.com>
5228 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5229 (save_reg_p): ..into this. Update all callers.
5230 (first_reg_to_save): Simplify.
5232 2017-08-16 Alan Modra <amodra@gmail.com>
5234 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5237 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5241 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5242 constant count if that count is less than 32.
5244 2017-08-15 Nathan Sidwell <nathan@acm.org>
5246 * gcc.c (execute): Emit friendlier message if inferior is killed
5247 by an external cause.
5249 2017-08-15 Richard Biener <rguenther@suse.de>
5251 PR tree-optimization/81790
5252 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5253 CONSTRUCTORs from simplifying and VN.
5255 2017-08-14 Martin Sebor <msebor@redhat.com>
5257 * builtin-attrs.def: Add comments.
5259 2017-08-14 Martin Sebor <msebor@redhat.com>
5262 * doc/extend.texi (attribute nonstring): Document new attribute.
5264 2017-08-14 Martin Sebor <msebor@redhat.com>
5267 * tree-diagnostic.c (default_tree_printer): Handle %G.
5268 * gimple-pretty-print.h (percent_G_format): Declare new function.
5269 * gimple-pretty-print.c (percent_G_format): Define.
5270 * tree-pretty-print.c (percent_K_format): Add argument.
5272 2017-08-14 Martin Sebor <msebor@redhat.com>
5274 PR translation/79998
5275 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5276 Remove a stray space.
5278 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5281 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5282 (*iordi_1_bts): Ditto.
5283 (*xordi_1_btc): Ditto.
5285 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5288 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5289 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5291 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5292 quoted strings, and make more translator-friendly.
5293 (darwin_rs6000_override_options): Likewise.
5294 (rs6000_option_override_internal): Likewise.
5295 (rs6000_return_in_memory): Fix overlong line.
5296 (init_cmulative_args): Use quoted strings, and make more
5297 translator-friendly.
5298 (rs6000_pass_by_reference): Fix overlong line.
5299 (def_builtin): Use quoted strings.
5300 (altivec_expand_predicate_builtin): Use quoted strings, and make
5301 more translator-friendly.
5302 (htm_expand_builtin): Use quoted strings.
5303 (cpu_expand_builtin): Use quoted strings, and make more
5304 translator-friendly.
5305 (altivec_expand_builtin): Likewise.
5306 (paired_expand_predicate_builtin): Likewise.
5307 (rs6000_invalid_builtin): Likewise.
5308 (builtin_function_type): Use quoted strings.
5309 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5310 more translator-friendly.
5311 (rs6000_trampoline_init): Likewise.
5312 (rs6000_handle_altivec_attribute): Likewise.
5313 (rs6000_inner_target_options): Use quoted strings.
5314 (rs6000_disable_incompatible_switches): Likewise.
5315 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5316 strings, and make more translator-friendly.
5317 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5319 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5321 PR tree-optimization/81799
5322 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5323 cond_expr to simple gimple operand.
5325 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5328 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5330 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5334 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5335 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5336 TARGET_GASISR_PROLOGUES.
5337 * config/avr/avr.c (avr_option_override): Same.
5338 (avr_pass_pre_proep::execute): Same.
5340 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5343 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5344 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5346 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5348 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5349 define_insn_and_split. Split to a memory load from 0 in
5350 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5351 using PTR mode iterator.
5352 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5353 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5354 (*add_tp_<mode>): Redefine as define_insn_and_split.
5355 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5356 address space. Merge with *add_tp_x32 using PTR mode iterator.
5357 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5358 Split to an add with a memory load from 0 in
5359 DEFAULT_TLS_SEG_REG address space.
5361 2017-08-12 Andrew Pinski <apinski@cavium.com>
5363 * config/aarch64/aarch64-option-extensions.def (rdma):
5364 Fix feature string to what Linux prints out in /proc/cpuinfo.
5366 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5369 * dwarf2out.c (modified_type_die): For C typedef types that have
5370 an ultimate origin, process the ultimate origin instead of the
5372 (gen_typedef_die): Assert that input DECLs have no ultimate
5374 (gen_type_die_with_usage): For typedef variants that have an
5375 ultimate origin, just call gen_decl_die on the original DECL.
5376 (process_scope_var): Avoid creating DIEs for local typedefs and
5377 concrete static variables.
5379 2017-08-12 Alan Modra <amodra@gmail.com>
5383 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5384 match gnu-user.h startfile.
5385 (ENDFILE_LINUX_SPEC): Similarly.
5387 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5390 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5392 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5394 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5395 * config/aarch64/aarch64.md (mov<mode>): Change.
5396 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5397 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5398 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5400 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5402 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5403 for storage order barriers.
5405 2017-08-11 Martin Liska <mliska@suse.cz>
5408 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5409 variables of void type.
5411 2017-08-11 Martin Liska <mliska@suse.cz>
5413 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5414 TARGET_SUPPORTS_ALIASES.
5415 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5416 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5417 (optimize_weakref): Likewise.
5418 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5419 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5420 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5422 2017-08-11 Martin Liska <mliska@suse.cz>
5425 * config/i386/i386.c (make_resolver_func): Do complete
5426 refactoring of the function.
5428 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5431 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5432 * config/i386/i386.c (ix86_stack_protect_guard): Use
5433 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5434 * doc/invoke.texi (x86 Options): Document
5435 -mstack-protector-guard-symbol= option.
5437 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5439 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5440 * config/i386/i386.c (ix86_split_stack_guard): New function.
5441 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5442 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5443 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5444 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5445 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5446 (split_stack_space_check): Call ix86_split_stack_guard.
5448 2017-08-10 Martin Sebor <msebor@redhat.com>
5450 * print-tree.c (print_node): Print location using the established
5452 Replace spaces with colons.
5453 (debug_raw, debug): Ditto.
5455 2017-08-10 Martin Sebor <msebor@redhat.com>
5458 * pretty-print.c (pp_format): Correct the handling of %s precision.
5460 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5463 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5465 (ix86_finalize_stack_frame_flags): This. Also clear
5466 frame_pointer_needed if -fno-omit-frame-pointer is used without
5468 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5469 with ix86_finalize_stack_frame_flags.
5470 (ix86_expand_epilogue): Likewise.
5471 (ix86_expand_split_stack_prologue): Likewise.
5472 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5474 2017-08-10 Martin Liska <mliska@suse.cz>
5477 * c-attribs.c (handle_target_attribute):
5478 Report warning for an empty string argument of target attribute.
5480 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5483 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5485 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5486 or DECL_NONLOCAL labels.
5487 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5488 or DECL_NONLOCAL labels here.
5490 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5492 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5493 to indicate when early gimple folding has been disabled.
5494 (rs6000_gimple_fold_builtin): Add debug content.
5495 (rs6000_invalid_builtin): Fix whitespace.
5496 (rs6000_expand_builtin): Fix whitespace.
5497 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5499 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5502 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5503 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5505 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5507 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5508 (qdf24xx): Likewise.
5509 * config/aarch64/aarch64.md: Include falkor.md.
5510 * config/aarch64/falkor.md: New.
5512 2017-08-09 Marek Polacek <polacek@redhat.com>
5515 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5516 * diagnostic.c (emit_diagnostic): Add a comment.
5517 (emit_diagnostic_valist): New function.
5519 2017-08-09 Marek Polacek <polacek@redhat.com>
5522 * input.c (make_location): New overload.
5523 * input.h (make_location): Declare.
5525 2017-08-08 Alan Modra <amodra@gmail.com>
5526 H.J. Lu <hongjiu.lu@intel.com>
5529 * gcc.c (NO_PIE_SPEC): Delete.
5530 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5532 (LINK_PIE_SPEC): ..to here.
5533 (LINK_COMMAND_SPEC): Support -no-pie.
5534 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5535 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5537 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5538 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5539 (ENDFILE_CRTEND_SPEC): Similarly.
5541 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5544 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5545 (mstack-protector-guard-offset=): Ditto.
5546 * config/i386/i386.c (ix86_option_override): Handle
5547 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5549 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5550 ix86_stack_protect_guard_offset variables.
5551 (TARGET_STACK_PROTECT_GUARD): Always define.
5552 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5553 and -mstack-protector-guard-offset= options.
5555 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5557 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5558 boundary case for the last candidate.
5560 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5562 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5563 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5565 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5568 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5569 * config/aarch64/aarch64-builtins.c
5570 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5571 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5572 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5574 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5575 Andrew Pinski <pinskia@gmail.com>
5578 * internal-fn.def (XORSIGN): New.
5579 * optabs.def (xorsign_optab): New.
5580 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5581 (convert_expand_mult_copysign): New.
5582 (pass_optimize_widening_mul::execute): Call
5583 convert_expand_mult_copysign.
5585 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5587 PR tree-optimization/81354
5588 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5589 Insert on edges rather than explicitly creating landing pads.
5590 (analyze_candidates_and_replace): Commit edge inserts.
5592 2017-08-08 Richard Biener <rguenther@suse.de>
5595 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5596 (expand_simple_operations): Also look through ADDR_EXPRs with
5597 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5599 2017-08-08 Richard Biener <rguenther@suse.de>
5601 PR tree-optimization/81723
5602 * tree-vect-slp.c (struct bst_traits): New hash traits.
5603 (bst_fail): New global.
5604 (vect_build_slp_tree_2): New worker, split out from ...
5605 (vect_build_slp_tree): ... this now wrapping it with using
5606 bst_fail set to cache SLP tree build fails. Properly handle
5608 (vect_analyze_slp_instance): Allocate and free bst_fail.
5610 2017-08-08 Martin Liska <mliska@suse.cz>
5613 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5614 LABEL_DECLs that can be from a different function.
5616 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5618 PR tree-optimization/81744
5619 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5620 loop's number of iterations.
5622 2017-08-08 Martin Liska <mliska@suse.cz>
5624 * asan.c: Include header files.
5625 * attribs.c (build_decl_attribute_variant): New function moved
5627 (build_type_attribute_qual_variant): Likewise.
5628 (cmp_attrib_identifiers): Likewise.
5629 (simple_cst_list_equal): Likewise.
5630 (omp_declare_simd_clauses_equal): Likewise.
5631 (attribute_value_equal): Likewise.
5632 (comp_type_attributes): Likewise.
5633 (build_type_attribute_variant): Likewise.
5634 (lookup_ident_attribute): Likewise.
5635 (remove_attribute): Likewise.
5636 (merge_attributes): Likewise.
5637 (merge_type_attributes): Likewise.
5638 (merge_decl_attributes): Likewise.
5639 (merge_dllimport_decl_attributes): Likewise.
5640 (handle_dll_attribute): Likewise.
5641 (attribute_list_equal): Likewise.
5642 (attribute_list_contained): Likewise.
5643 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5644 (lookup_attribute_by_prefix): Likewise.
5645 * bb-reorder.c: Include header files.
5646 * builtins.c: Likewise.
5647 * calls.c: Likewise.
5648 * cfgexpand.c: Likewise.
5649 * cgraph.c: Likewise.
5650 * cgraphunit.c: Likewise.
5651 * convert.c: Likewise.
5652 * dwarf2out.c: Likewise.
5653 * final.c: Likewise.
5654 * fold-const.c: Likewise.
5655 * function.c: Likewise.
5656 * gimple-expr.c: Likewise.
5657 * gimple-fold.c: Likewise.
5658 * gimple-pretty-print.c: Likewise.
5659 * gimple.c: Likewise.
5660 * gimplify.c: Likewise.
5661 * hsa-common.c: Likewise.
5662 * hsa-gen.c: Likewise.
5663 * internal-fn.c: Likewise.
5664 * ipa-chkp.c: Likewise.
5665 * ipa-cp.c: Likewise.
5666 * ipa-devirt.c: Likewise.
5667 * ipa-fnsummary.c: Likewise.
5668 * ipa-inline.c: Likewise.
5669 * ipa-visibility.c: Likewise.
5671 * lto-cgraph.c: Likewise.
5672 * omp-expand.c: Likewise.
5673 * omp-general.c: Likewise.
5674 * omp-low.c: Likewise.
5675 * omp-offload.c: Likewise.
5676 * omp-simd-clone.c: Likewise.
5677 * opts-global.c: Likewise.
5678 * passes.c: Likewise.
5679 * predict.c: Likewise.
5680 * sancov.c: Likewise.
5681 * sanopt.c: Likewise.
5682 * symtab.c: Likewise.
5683 * toplev.c: Likewise.
5684 * trans-mem.c: Likewise.
5685 * tree-chkp.c: Likewise.
5686 * tree-eh.c: Likewise.
5687 * tree-into-ssa.c: Likewise.
5688 * tree-object-size.c: Likewise.
5689 * tree-parloops.c: Likewise.
5690 * tree-profile.c: Likewise.
5691 * tree-ssa-ccp.c: Likewise.
5692 * tree-ssa-live.c: Likewise.
5693 * tree-ssa-loop.c: Likewise.
5694 * tree-ssa-sccvn.c: Likewise.
5695 * tree-ssa-structalias.c: Likewise.
5696 * tree-ssa.c: Likewise.
5697 * tree-streamer-in.c: Likewise.
5698 * tree-vectorizer.c: Likewise.
5699 * tree-vrp.c: Likewise.
5701 * ubsan.c: Likewise.
5702 * varasm.c: Likewise.
5703 * varpool.c: Likewise.
5704 * tree.c: Remove functions moved to attribs.[ch].
5706 * config/aarch64/aarch64.c: Add attrs.h header file.
5707 * config/alpha/alpha.c: Likewise.
5708 * config/arc/arc.c: Likewise.
5709 * config/arm/arm.c: Likewise.
5710 * config/avr/avr.c: Likewise.
5711 * config/bfin/bfin.c: Likewise.
5712 * config/c6x/c6x.c: Likewise.
5713 * config/cr16/cr16.c: Likewise.
5714 * config/cris/cris.c: Likewise.
5715 * config/darwin.c: Likewise.
5716 * config/epiphany/epiphany.c: Likewise.
5717 * config/fr30/fr30.c: Likewise.
5718 * config/frv/frv.c: Likewise.
5719 * config/ft32/ft32.c: Likewise.
5720 * config/h8300/h8300.c: Likewise.
5721 * config/i386/winnt.c: Likewise.
5722 * config/ia64/ia64.c: Likewise.
5723 * config/iq2000/iq2000.c: Likewise.
5724 * config/lm32/lm32.c: Likewise.
5725 * config/m32c/m32c.c: Likewise.
5726 * config/m32r/m32r.c: Likewise.
5727 * config/m68k/m68k.c: Likewise.
5728 * config/mcore/mcore.c: Likewise.
5729 * config/microblaze/microblaze.c: Likewise.
5730 * config/mips/mips.c: Likewise.
5731 * config/mmix/mmix.c: Likewise.
5732 * config/mn10300/mn10300.c: Likewise.
5733 * config/moxie/moxie.c: Likewise.
5734 * config/msp430/msp430.c: Likewise.
5735 * config/nds32/nds32-isr.c: Likewise.
5736 * config/nds32/nds32.c: Likewise.
5737 * config/nios2/nios2.c: Likewise.
5738 * config/nvptx/nvptx.c: Likewise.
5739 * config/pa/pa.c: Likewise.
5740 * config/pdp11/pdp11.c: Likewise.
5741 * config/powerpcspe/powerpcspe.c: Likewise.
5742 * config/riscv/riscv.c: Likewise.
5743 * config/rl78/rl78.c: Likewise.
5744 * config/rx/rx.c: Likewise.
5745 * config/s390/s390.c: Likewise.
5746 * config/sh/sh.c: Likewise.
5747 * config/sol2.c: Likewise.
5748 * config/sparc/sparc.c: Likewise.
5749 * config/spu/spu.c: Likewise.
5750 * config/stormy16/stormy16.c: Likewise.
5751 * config/tilegx/tilegx.c: Likewise.
5752 * config/tilepro/tilepro.c: Likewise.
5753 * config/v850/v850.c: Likewise.
5754 * config/vax/vax.c: Likewise.
5755 * config/visium/visium.c: Likewise.
5756 * config/xtensa/xtensa.c: Likewise.
5758 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5761 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5762 constraints since the -mupper-regs-* switches have been
5764 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5765 into a vector from a double word element that was extracted from
5766 another vector, and eliminate extra XXPERMDI instructions.
5767 (vsx_concat_<mode>_2): Likewise.
5768 (vsx_concat_<mode>_3): Likewise.
5769 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5770 concat to allow optimizing inserts from previous extracts.
5772 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5774 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5775 memory reference to a SSP offset in TLS address space.
5776 (ix86_print_operand) <case '@'>: Remove.
5777 (ix86_print_operand_punct_valid_p): Remove '@' code.
5778 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5780 (stack_tls_protect_set_<mode>): Remove.
5781 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5782 (stack_tls_protect_test_<mode>): Remove.
5783 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5785 2017-08-07 Olivier Hainque <hainque@adacore.com>
5788 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5790 2017-08-07 Douglas Rupp <rupp@adacore.com>
5792 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5793 variable was referenced as multidir in command.
5795 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5798 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5801 2017-08-07 Martin Liska <mliska@suse.cz>
5803 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5804 * config/rl78/rl78.c: Add include of attribs.h.
5805 * config/sh/sh.c: Likewise.
5806 * config/v850/v850.c: Likewise.
5808 2017-08-07 Tom de Vries <tom@codesourcery.com>
5811 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5813 2017-08-07 Martin Liska <mliska@suse.cz>
5815 * config/mips/mips.c: Include attribs.h.
5817 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5820 * doc/invoke.texi: Document change in behvaior for -Ofast for
5823 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5825 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5827 (aarch64_pop_regs): Likewise.
5828 (aarch64_gen_load_pair): Likewise.
5829 (aarch64_save_callee_saves): Likewise.
5830 (aarch64_restore_callee_saves): Likewise.
5832 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5834 * config/i386/i386.c: Revert the last change.
5836 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5839 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5841 (ix86_finalize_stack_frame_flags): This. Also clear
5842 frame_pointer_needed if -fno-omit-frame-pointer is used without
5844 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5845 with ix86_finalize_stack_frame_flags.
5846 (ix86_expand_epilogue): Likewise.
5847 (ix86_expand_split_stack_prologue): Likewise.
5849 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5852 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5853 to P_AES for Westmere.
5855 2017-08-07 Jonathan Yong <10walls@gmail.com>
5857 * config/i386/mingw.opt (fset-stack-executable): Removed.
5858 * config/i386/cygming.opt (fset-stack-executable): Moved
5860 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5862 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5864 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5866 2017-08-07 Marek Polacek <polacek@redhat.com>
5869 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5871 2017-08-07 Martin Liska <mliska@suse.cz>
5873 * attribs.h (canonicalize_attr_name): New function.
5874 (cmp_attribs): Move from c-format.c and adjusted.
5875 (is_attribute_p): Moved from tree.h.
5876 * tree-inline.c: Add new includes.
5877 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5878 (private_is_attribute_p): Remove.
5879 (private_lookup_attribute): Likewise.
5880 (private_lookup_attribute_by_prefix): Simplify.
5881 (remove_attribute): Use is_attribute_p.
5882 * tree.h: Remove removed declarations.
5884 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5887 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5888 instead of computing it in the function. Formatting fix.
5889 (expand_case): Don't rely on default_edge being the first edge,
5890 clear it if removing it, pass default_edge to
5891 emit_case_dispatch_table.
5892 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5895 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5897 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5898 insn in the function, emit NOP after the insn.
5900 2017-08-06 Tom de Vries <tom@codesourcery.com>
5902 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5905 2017-08-06 Tom de Vries <tom@codesourcery.com>
5907 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5910 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5912 PR tree-optimization/57371
5913 * match.pd: New pattern.
5915 2017-08-04 Marek Polacek <polacek@redhat.com>
5918 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5919 perform the computation in offset_int.
5921 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5923 PR tree-optimization/81136
5924 * tree-vectorizer.h: Include tree-hash-traits.h.
5925 (vec_base_alignments): New typedef.
5926 (vec_info): Add a base_alignments field.
5927 (vect_record_base_alignments): Declare.
5928 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5930 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5931 (create_data_ref): Add an is_conditional_in_stmt argument.
5932 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5933 the is_conditional_in_stmt field.
5934 (data_ref_loc): Add an is_conditional_in_stmt field.
5935 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5936 (find_data_references_in_stmt): Update call to create_data_ref.
5937 (graphite_find_data_references_in_stmt): Likewise.
5938 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5939 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5940 (vect_record_base_alignment): New function.
5941 (vect_record_base_alignments): Likewise.
5942 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5943 for nested statements even if we fail to compute a misalignment.
5944 Use pooled base alignments for unconditional references.
5945 (vect_find_same_alignment_drs): Compare base addresses instead
5947 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5948 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5950 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5952 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5953 Add an explicit name for the enum. Use auto_vec for slp_instances
5955 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5957 (_bb_vec_info): Add a constructor and destructor.
5958 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5959 (destroy_loop_vec_info): Delete.
5960 (vect_destroy_datarefs): Likewise.
5961 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5962 (vec_info::vec_info): New function.
5963 (vec_info::~vec_info): Likewise.
5964 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5965 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5966 destroy_loop_vec_info.
5967 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5968 (_loop_vec_info::_loop_vec_info): ...this.
5969 (destroy_loop_vec_info): Replace with...
5970 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5971 the stmt_vec_infos. Leave handling of vec_info information to its
5972 destructor. Remove explicit vector releases.
5973 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5974 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5975 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5976 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5977 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5978 (destroy_bb_vec_info): Replace with...
5979 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5980 information to its destructor.
5981 (vect_slp_analyze_bb_1): Use new and delete instead of
5982 new_bb_vec_info and destroy_bb_vec_info.
5983 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5986 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5988 * tree-data-ref.h (subscript): Add access_fn field.
5989 (data_dependence_relation): Add could_be_independent_p.
5990 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5991 (same_access_functions): Move to tree-data-ref.c.
5992 * tree-data-ref.c (ref_contains_union_access_p): New function.
5993 (access_fn_component_p): Likewise.
5994 (access_fn_components_comparable_p): Likewise.
5995 (dr_analyze_indices): Add a reference to access_fn_component_p.
5996 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5998 (constant_access_functions): Likewise.
5999 (add_other_self_distances): Likewise.
6000 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6001 (initialize_data_dependence_relation): Use XCNEW and remove
6002 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6003 of access functions that have the same type. Allow the
6004 subsequence to end with different bases in some circumstances.
6005 Record the chosen access functions in SUB_ACCESS_FN.
6006 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6007 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6008 (subscript_dependence_tester_1): Likewise dra and drb.
6009 (build_classic_dist_vector): Update calls accordingly.
6010 (subscript_dependence_tester): Likewise.
6011 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6012 DDR_COULD_BE_INDEPENDENT_P.
6013 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6014 comp_alias_ddrs instead of may_alias_ddrs.
6015 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6017 (vect_analyze_data_ref_dependence): Use it if
6018 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6019 distance vectors if that fails.
6020 (dependence_distance_ge_vf): New function.
6021 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6022 LOOP_VINFO_MAY_ALIAS_DDRS.
6024 2017-08-04 Richard Biener <rguenther@suse.de>
6027 * fold-const.c (fold_binary_loc): Properly restrict
6028 minus_var0 && minus_var1 case when associating undefined overflow
6031 2017-08-04 Tom de Vries <tom@codesourcery.com>
6033 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6035 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6037 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6038 Don't start diagnostic messages with a capital letter.
6039 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6041 (rs6000_invalid_builtin): Likewise.
6042 (rs6000_trampoline_init): Likewise.
6044 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6047 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6048 after setting changeable df flags.
6050 2017-08-03 Richard Biener <rguenther@suse.de>
6052 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6053 up if the use is in USE - X.
6055 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6057 * toplev.c (dumpfile.h): New include.
6058 (internal_error_reentered): New static function. Use it...
6059 (internal_error_function): ...here to handle reentered internal_error.
6061 2017-08-03 Richard Biener <rguenther@suse.de>
6064 * fold-const.c (split_tree): Add minus_var and minus_con
6065 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6066 here but always use minus_*.
6067 (associate_trees): Assert we never associate with MINUS_EXPR
6068 and NULL first operand. Do not recurse for PLUS_EXPR operands
6069 when associating as MINUS_EXPR either.
6070 (fold_binary_loc): Track minus_var and minus_con.
6072 2017-08-03 Tom de Vries <tom@codesourcery.com>
6075 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6076 ACCEL_COMPILER, apply finish_options on
6077 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6079 2017-08-03 Tom de Vries <tom@codesourcery.com>
6082 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6083 function_entry_patch_area_size > 0.
6085 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6088 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6089 instead of 10??LU, perform unit multiplication in wide_int,
6090 don't change alloc_object_size_limit if the limit is larger
6093 PR tree-optimization/81655
6094 PR tree-optimization/81588
6095 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6096 the case when ranges[i].low and high are 1 for unsigned type with
6100 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6101 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6103 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6105 * tree-vrp.h: Add include guard.
6107 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6110 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6111 (ud2): New insn pattern.
6112 * config/i386/i386.c (ix86_expand_epilogue):
6113 For naked functions, generate ud2 instead of trap insn.
6115 2017-08-02 Marek Polacek <polacek@redhat.com>
6118 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6120 2017-08-02 Tom de Vries <tom@codesourcery.com>
6121 Cesar Philippidis <cesar@codesourcery.com>
6123 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6124 Add missing edge probabilities.
6126 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6128 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6131 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6134 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6135 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6136 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6137 prologue_seq sequences - if any.
6139 2017-08-02 Richard Biener <rguenther@suse.de>
6141 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6142 via vectors if supported, integer extracts via punning if supported
6143 or otherwise vector extracts.
6145 2017-08-02 Richard Biener <rguenther@suse.de>
6147 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6149 (bitmap_insert_into_set): ... this.
6151 2017-08-02 Richard Biener <rguenther@suse.de>
6153 PR tree-optimization/81633
6155 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6157 PR tree-optimization/71752
6158 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6160 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6162 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6163 (machine_function::call_ms2sysv_pad_out): Remove field.
6164 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6165 (ix86_compute_frame_layout): Likewise.
6167 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6170 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6171 attribute with interrupt attribute.
6173 2017-08-01 Andrew Pinski <apinski@cavium.com>
6175 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6176 BIT_INSERT_EXPR's operand 1
6177 to see if the types precision matches.
6179 2017-08-01 Martin Liska <mliska@suse.cz>
6182 * builtins.c (expand_builtin_memcpy_args): Remove.
6183 (expand_builtin_memcpy): Call newly added function
6184 expand_builtin_memory_copy_args.
6185 (expand_builtin_memcpy_with_bounds): Likewise.
6186 (expand_builtin_mempcpy): Remove last argument.
6187 (expand_builtin_mempcpy_with_bounds): Likewise.
6188 (expand_builtin_memory_copy_args): New function created from
6189 expand_builtin_mempcpy_args with small modifications.
6190 (expand_builtin_mempcpy_args): Remove.
6191 (expand_builtin_stpcpy): Remove unused argument.
6192 (expand_builtin): Likewise.
6193 (expand_builtin_with_bounds): Likewise.
6195 2017-08-01 Martin Liska <mliska@suse.cz>
6198 Make mempcpy more optimal (PR middle-end/70140).
6200 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6203 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6204 __builtin_vec_cmpne verify both arguments are compatible vectors
6205 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6206 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6207 move computation of aligned to after checking the argument types.
6211 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6212 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6215 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6216 Doug Rupp <rupp@adacore.com>
6217 Olivier Hainque <hainque@adacore.com>
6219 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6220 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6222 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6223 for TARGET_OS_CPP_BUILTIN.
6224 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6225 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6227 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6228 passing required abi options to the assembler for EABI configurations.
6229 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6230 of .text.hot and .text.unlikely sections for kernel modules when
6231 using ARM style exceptions.
6232 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6233 options. Add EXTRA_CC1_SPEC.
6234 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6236 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6238 (ARM_TARGET2_DWARF_FORMAT): Define.
6239 * config/arm/t-vxworks: Adjust multilib control to removal of the
6240 Diab command line options.
6242 2017-08-01 Martin Liska <mliska@suse.cz>
6244 PR gcov-profile/81561
6245 * gcov.c (unblock): Make unblocking safe as we need to preserve
6246 index correspondence of blocks and block_lists.
6248 2017-08-01 Richard Biener <rguenther@suse.de>
6250 PR tree-optimization/81181
6251 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6252 (compute_antic): ... end of iteration here.
6254 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6256 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6257 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6258 (ftree-slp-vectorize): Likewise.
6259 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6260 can no longer be set independent of flag_tree_loop_vectorize.
6261 * omp-general.c (emp_max_vf): Likewise.
6262 * opts.c (enable_fdo_optimizations): Remove references to
6263 flag_tree_vectorize, these are now implicit.
6264 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6265 and leave it for the options machinery.
6267 2017-08-01 Martin Liska <mliska@suse.cz>
6270 * builtins.c (expand_builtin_memcpy_args): Remove.
6271 (expand_builtin_memcpy): Call newly added function
6272 expand_builtin_memory_copy_args.
6273 (expand_builtin_memcpy_with_bounds): Likewise.
6274 (expand_builtin_mempcpy): Remove last argument.
6275 (expand_builtin_mempcpy_with_bounds): Likewise.
6276 (expand_builtin_memory_copy_args): New function created from
6277 expand_builtin_mempcpy_args with small modifications.
6278 (expand_builtin_mempcpy_args): Remove.
6279 (expand_builtin_stpcpy): Remove unused argument.
6280 (expand_builtin): Likewise.
6281 (expand_builtin_with_bounds): Likewise.
6283 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6286 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6287 print "ds:" only for immediates in generic address space.
6289 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6292 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6293 (ix86_function_ok_for_sibcall): Return false for naked functions.
6295 2017-08-01 Richard Biener <rguenther@suse.de>
6297 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6298 (compute_antic): Seed worklist with exit block predecessors.
6299 * cfganal.c (dfs_find_deadend): For a cycle return the source
6300 of the edge closing it.
6302 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6304 * config/aarch64/aarch64.c
6305 (aarch64_can_const_movi_rtx_p): Move 0 check.
6307 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6309 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6310 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6312 * match.pd: Ditto in address comparison pattern.
6314 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6316 PR tree-optimization/81627
6317 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6318 closed ssa form for store-store chain.
6320 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6322 PR tree-optimization/81620
6323 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6324 for store-store chain.
6326 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6328 PR tree-optimization/81588
6329 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6330 ranges[i].in_p, invert comparison code ccode. For >/>=,
6331 swap rhs1 and rhs2 and comparison code unconditionally,
6332 for </<= don't do that. Don't swap rhs1/rhs2 again if
6333 ranges[i].in_p, instead invert comparison code ccode if
6334 opcode or oe->rank is BIT_IOR_EXPR.
6337 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6338 a direct optab to conversion optab.
6339 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6340 with GET_MODE_INNER as last argument instead of optab_handler.
6341 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6342 vector extraction if possible and optab is available.
6343 * expr.c (store_constructor): Use convert_optab_handler instead
6344 of optab_handler. Use vector initialization from smaller
6345 vectors if possible and optab is available.
6346 * tree-vect-stmts.c (vectorizable_load): Likewise.
6347 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6349 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6350 of vec_init from half-sized vectors with the same element mode.
6351 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6352 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6353 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6354 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6355 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6356 after mode in gen_vec_extract* calls.
6357 (vec_extract<mode>): Renamed to ...
6358 (vec_extract<mode><ssescalarmodelower>): ... this.
6359 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6360 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6361 element mode after mode in gen_vec_init* calls.
6362 (VEC_INIT_HALF_MODE): New mode iterator.
6363 (vec_init<mode>): Renamed to ...
6364 (vec_init<mode><ssescalarmodelower>): ... this.
6365 (vec_init<mode><ssehalfvecmodelower>): New expander.
6366 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6367 (vec_extractv2sfsf): ... this.
6368 (vec_initv2sf): Renamed to ...
6369 (vec_initv2sfsf): ... this.
6370 (vec_extractv2si): Renamed to ...
6371 (vec_extractv2sisi): ... this.
6372 (vec_initv2si): Renamed to ...
6373 (vec_initv2sisi): ... this.
6374 (vec_extractv4hi): Renamed to ...
6375 (vec_extractv4hihi): ... this.
6376 (vec_initv4hi): Renamed to ...
6377 (vec_initv4hihi): ... this.
6378 (vec_extractv8qi): Renamed to ...
6379 (vec_extractv8qiqi): ... this.
6380 (vec_initv8qi): Renamed to ...
6381 (vec_initv8qiqi): ... this.
6382 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6383 (vec_init<mode>): Renamed to ...
6384 (vec_init<mode><VEC_base_l>): ... this.
6385 (vec_extract<mode>): Renamed to ...
6386 (vec_extract<mode><VEC_base_l>): ... this.
6387 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6388 (vec_initv2sfsf): ... this.
6389 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6390 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6391 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6392 element mode after mode in gen_vec_init* calls.
6393 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6394 (vec_init<mode><Vel>): ... this.
6395 (vec_extract<mode>): Renamed to ...
6396 (vec_extract<mode><Vel>): ... this.
6397 * config/aarch64/iterators.md (Vel): New mode attribute.
6398 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6399 Add element mode after mode in gen_vec_extract* calls.
6400 * config/s390/vector.md (non_vec_l): New mode attribute.
6401 (vec_extract<mode>): Renamed to ...
6402 (vec_extract<mode><non_vec_l>): ... this.
6403 (vec_init<mode>): Renamed to ...
6404 (vec_init<mode><non_vec_l>): ... this.
6405 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6406 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6408 * config/arm/iterators.md (V_elem_l): New mode attribute.
6409 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6410 (vec_extract<mode><V_elem_l>): ... this.
6411 (vec_extractv2di): Renamed to ...
6412 (vec_extractv2didi): ... this.
6413 (vec_init<mode>): Renamed to ...
6414 (vec_init<mode><V_elem_l>): ... this.
6415 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6416 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6417 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6418 Add element mode after gen_vec_extract* calls.
6419 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6420 (vec_init<mode><unitmode>): ... this.
6421 (vec_extract<mode>): Renamed to ...
6422 (vec_extract<mode><unitmode>): ... this.
6423 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6424 (vec_init<mode><unitmode>): ... this.
6425 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6426 (vec_initv2sfsf): ... this.
6427 (vec_extractv2sf): Renamed to ...
6428 (vec_extractv2sfsf): ... this.
6429 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6430 Add element mode after gen_vec_extract* calls.
6431 * config/mips/mips.md (unitmode): New mode iterator.
6432 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6433 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6434 * config/spu/spu.md (inner_l): New mode attribute.
6435 (vec_init<mode>): Renamed to ...
6436 (vec_init<mode><inner_l>): ... this.
6437 (vec_extract<mode>): Renamed to ...
6438 (vec_extract<mode><inner_l>): ... this.
6439 * config/sparc/sparc.md (veltmode): New mode iterator.
6440 (vec_init<VMALL:mode>): Renamed to ...
6441 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6442 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6443 (vec_initv2sisi): ... this.
6444 (vec_initv2sf): Renamed to ...
6445 (vec_initv2sfsf): ... this.
6446 (vec_extractv2sf): Renamed to ...
6447 (vec_extractv2sfsf): ... this.
6448 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6449 (vec_init<mode>): Renamed to ...
6450 (vec_init<mode><VEC_base_l>): ... this.
6451 (vec_extract<mode>): Renamed to ...
6452 (vec_extract<mode><VEC_base_l>): ... this.
6453 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6454 (vec_initv2sfsf): ... this.
6455 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6456 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6457 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6458 gen_vec_init* calls.
6460 2017-08-01 Richard Biener <rguenther@suse.de>
6462 PR tree-optimization/81297
6463 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6464 TREE_OVERFLOW from INTEGER_CSTs.
6466 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6468 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6470 2017-07-31 Carl Love <cel@us.ibm.com>
6472 * config/rs6000/rs6000-c: Add support for built-in functions
6473 vector signed char vec_xl_be (signed long long, signed char *);
6474 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6475 vector signed int vec_xl_be (signed long long, signed int *);
6476 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6477 vector signed long long vec_xl_be (signed long long, signed long long *);
6478 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6479 vector signed short vec_xl_be (signed long long, signed short *);
6480 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6481 vector double vec_xl_be (signed long long, double *);
6482 vector float vec_xl_be (signed long long, float *);
6483 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6484 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6485 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6487 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6488 (altivec_expand_builtin): Add switch statement to call
6489 altivec_expand_xl_be for each builtin.
6490 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6491 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6492 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6493 __builtin_vsx_le_be_v16qi.
6494 * doc/extend.texi: Update the built-in documentation file for the
6495 new built-in functions.
6497 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6500 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6502 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6504 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6506 * config.gcc: Add z14.
6507 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6508 CPU model numbers for z13s and z14.
6509 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6511 * config/s390/s390-opts.h (enum processor_type): Rename
6512 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6513 * config/s390/s390.c (processor_table): Add field for CPU name to
6514 be passed to Binutils.
6515 (s390_asm_output_machine_for_arch): Use the new field in
6516 processor_table for Binutils.
6517 (s390_expand_builtin): Replace arch12 with z14.
6518 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6519 (s390_get_sched_attrmask): Likewise.
6520 (s390_get_unit_mask): Likewise.
6521 * config/s390/s390.opt: Add z14 to processor_type enum.
6523 2017-07-31 Martin Jambor <mjambor@suse.cz>
6526 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6527 regardless of optimization level.
6529 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6530 Martin Liska <mliska@suse.cz>
6532 * predict.def: Remove old comment and adjust probability.
6533 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6536 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6539 * config/i386/i386.c (ix86_function_naked): New function.
6540 (ix86_can_use_return_insn_p): Return false for naked functions.
6541 (ix86_expand_prologue): Skip prologue for naked functions.
6542 (ix86_expand_epilogue): Skip epilogue for naked functions
6543 and emit trap instruction.
6544 (ix86_warn_func_return): New function.
6545 (ix86_attribute_table): Add "naked" attribute specification.
6546 (TARGET_WARN_FUNC_RETURN): Define.
6547 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6549 2017-07-31 Martin Liska <mliska@suse.cz>
6551 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6552 (dump_gimple_bb_header): Always dump BB info.
6553 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6555 2017-07-31 Martin Liska <mliska@suse.cz>
6558 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6559 also with current_function_decl non-null equality.
6561 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6564 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6565 change type to the element type, instead add eltype variable and
6566 use it where we are interested in the element type.
6568 PR tree-optimization/81603
6569 * ipa-polymorphic-call.c
6570 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6571 offset arithmetic in offset_int, bail out if the resulting bit offset
6572 doesn't fit into shwi.
6574 2017-07-31 Martin Liska <mliska@suse.cz>
6576 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6577 (gimplify_save_expr): Fix comment.
6579 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6582 * config/i386/i386.c (ix86_function_arg): Update arguments for
6584 (ix86_compute_frame_layout): Set the initial stack offset to
6585 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6586 INCOMING_FRAME_SP_OFFSET.
6587 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6588 stack before exception handler returns.
6589 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6590 the 'ERROR_CODE' for exception handler.
6592 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6594 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6595 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6596 (ASM_OUTPUT_REG_POP): Ditto.
6597 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6598 instead of asm_fprintf to output pure string.
6600 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6602 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6603 to imported_module_or_decl hook.
6604 (debug_nothing_tree_tree_tree_bool): Remove.
6605 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6606 * debug.c (do_nothing_debug_hooks): Use
6607 debug_nothing_tree_tree_tree_bool_bool instead of
6608 debug_nothing_tree_tree_tree_bool.
6609 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6610 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6611 * sdbout.c (sdb_debug_hooks): Likewise.
6612 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6613 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6615 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6616 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6617 attribute, don't add anything.
6619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6621 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6622 (fold_build2_stat_loc): Likewise.
6623 (fold_build3_stat_loc): Likewise.
6624 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6625 (fold_build1_loc): Remove macro.
6626 (fold_build2_loc): Likewise.
6627 (fold_build3_loc): Likewise.
6629 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6631 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6632 (gimple_build_debug_bind_source_stat): Likewise.
6633 * gimple.h (gimple_build_debug_bind): Remove macro.
6634 (gimple_build_debug_bind_source): Likewise.
6636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6638 * bitmap.c (bitmap_alloc): Adjust.
6639 (bitmap_gc_alloc): Likewise.
6640 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6642 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6644 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6645 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6646 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6647 (bitmap_gc_alloc_stat): Likewise.
6648 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6650 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6652 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6653 * rtl.h (shallow_copy_rtx): Remove macro.
6655 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6657 * emit-rtl.c (gen_raw_REG): Adjust.
6658 * gengenrtl.c (gendef): Likewise.
6659 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6660 * rtl.h (rtx_alloc): Remove macro.
6662 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6664 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6665 (build_tree_list_stat): Likewise.
6666 * tree.h (build_tree_list): Remove macro.
6667 (build_tree_list_vec): Likewise.
6669 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6671 * tree.c (make_vector_stat): Remove _stat from name.
6672 (build_vector_stat): Likewise.
6673 * tree.h (make_vector_stat): Remove macro.
6674 (build_vector_stat): Likewise.
6676 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6678 * tree.h (build_var_debug_value): Remove prototype.
6680 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6682 * tree.c (tree_cons_stat): Remove _stat from name.
6683 * tree.h (tree_cons): Remove macro.
6685 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6687 * tree.c (build_vl_exp_stat): Remove _stat from name.
6688 * tree.h (build_vl_exp): Remove macro.
6690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6692 * tree.c (build_decl_stat): Remove _stat from name.
6693 * tree.h (build_decl): Remove macro.
6695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6697 * gimple.c (gimple_build_with_ops_stat): Adjust.
6698 (gimple_alloc_stat): Remove _stat from name.
6699 * gimple.h (gimple_alloc): Remove macro.
6701 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6703 * tree.c (make_tree_vec_stat): Remove _stat from name.
6704 (grow_tree_vec_stat): Likewise.
6705 * tree.h (make_tree_vec_stat): Adjust prototype.
6706 (grow_tree_vec_stat): Likewise.
6707 (make_tree_vec): Remove macro.
6708 (grow_tree_vec): Likewise.
6710 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6712 * fold-const.c (fold_build1_stat_loc): Adjust.
6713 (fold_build2_stat_loc): Likewise.
6714 (fold_build3_stat_loc): Likewise.
6715 * tree.c (build0_stat): Remove _stat from name.
6716 (build1_stat): Likewise.
6717 (build2_stat): Likewise.
6718 (build3_stat): Likewise.
6719 (build4_stat): Likewise.
6720 (build5_stat): Likewise.
6721 * tree.h (build1_loc): Remove macro, and rename _stat function
6723 (build2_loc): Likewise.
6724 (build3_loc): Likewise.
6725 (build4_loc): Likewise.
6726 (build5_loc): Likewise.
6728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6730 * tree.c (make_int_cst_stat): Remove _stat from name.
6731 * tree.h (make_int_cst_stat): Adjust prototype.
6732 (make_int_cst): Remove macro.
6734 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6736 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6737 * tree.h (make_tree_binfo_stat): Adjust prototype.
6738 (make_tree_binfo): Remove.
6740 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6742 * tree.c (copy_node_stat): Rename to copy_node.
6743 (build_distinct_type_copy): Adjust.
6744 * tree.h (copy_node_stat): Adjust prototype.
6745 (copy_node): Remove macro.
6747 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6749 * tree.c (make_node_stat): rename to make_node.
6750 (build_tree_list_stat): Adjust.
6751 (build0_stat): Likewise.
6752 (build2_stat): Likewise.
6753 (build3_stat): Likewise.
6754 (build4_stat): Likewise.
6755 (build5_stat): Likewise.
6756 (build_decl_stat): Likewise.
6757 * tree.h (make_node_stat): Adjust prototype.
6758 (make_node): remove macro.
6760 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6762 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6763 (PPC_FEATURE2_SCV): Likewise.
6764 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6766 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6768 * config/aarch64/aarch64.c
6769 (aarch64_internal_mov_immediate): Add new special pattern.
6770 * config/aarch64/aarch64.md (*movdi_aarch64):
6771 Add reg/32bit const mov case.
6773 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6774 Richard Sandiford <richard.sandiford@linaro.org>
6776 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6777 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6778 Add integer and movi cases.
6779 (movi-split-hf-df-sf split, fp16): New.
6780 (enabled): Added TARGET_FP_F16INST.
6781 * config/aarch64/iterators.md (GPF_HF): New.
6782 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6784 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6786 * config/aarch64/aarch64.c
6787 (aarch64_simd_container_mode): Add prototype.
6788 (aarch64_expand_mov_immediate): Add HI support.
6789 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6790 (aarch64_can_const_movi_rtx_p): New.
6791 (aarch64_preferred_reload_class):
6792 Remove restrictions of using FP registers for certain SIMD operations.
6793 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6794 (aarch64_valid_floating_const): Add integer move validation.
6795 (aarch64_simd_imm_scalar_p): Remove.
6796 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6797 (aarch64_legitimate_constant_p): Expand list of supported cases.
6798 * config/aarch64/aarch64-protos.h
6799 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6800 (aarch64_reinterpret_float_as_int): New.
6801 (aarch64_simd_imm_scalar_p): Remove.
6802 * config/aarch64/constraints.md (Uvi): New.
6803 (Dd): Split into Ds and new Dd.
6804 * config/aarch64/aarch64.md (*movsi_aarch64):
6806 (*movdi_aarch64): Add SIMD mov case.
6808 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6810 * tree-predcom.c: (struct chain): Handle store-store chain in which
6811 stores for elimination only store loop invariant values.
6812 (execute_pred_commoning_chain): Ditto.
6813 (prepare_initializers_chain_store_elim): Ditto.
6814 (prepare_finalizers): Ditto.
6815 (is_inv_store_elimination_chain): New function.
6816 (initialize_root_vars_store_elim_1): New function.
6818 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6820 * tree-predcom.c: Revise general description of the pass.
6821 (enum chain_type): New enum type for store elimination.
6822 (struct chain): New field supporting store elimination.
6823 (struct component): Ditto.
6824 (dump_chain): Dump store-stores chain.
6825 (release_chain): Release resources.
6826 (split_data_refs_to_components): Compute and create component
6827 contains only stores for elimination.
6828 (get_chain_last_ref_at): New function.
6829 (make_invariant_chain): Initialization.
6830 (make_rooted_chain): Specify chain type in parameter and record it.
6831 (add_looparound_copies): Skip for store-stores chain.
6832 (determine_roots_comp): Compute type of chain and pass it to
6834 (initialize_root_vars_store_elim_2): New function.
6835 (finalize_eliminated_stores): New function.
6836 (remove_stmt): Handle store for elimination.
6837 (execute_pred_commoning_chain): Execute predictive commoning on
6839 (determine_unroll_factor): Skip unroll for store-stores chain.
6840 (prepare_initializers_chain_store_elim): New function.
6841 (prepare_initializers_chain): Hanlde store-store chain.
6842 (prepare_finalizers_chain, prepare_finalizers): New function.
6843 (tree_predictive_commoning_loop): Return integer value indicating
6844 if loop is unrolled or lcssa form is corrupted.
6845 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6847 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6849 * tree-predcom.c (initialize_root): Delete.
6850 (execute_pred_commoning_chain): Initialize root vars and replace
6851 reference of non-combined chain directly, rather than call above
6854 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6856 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6857 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6859 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6861 * tree-predcom.c (struct chain): New field init_seq.
6862 (release_chain): Release init_seq.
6863 (prepare_initializers_chain): Record intialization stmts in above
6865 (insert_init_seqs): New function.
6866 (tree_predictive_commoning_loop): Call insert_init_seqs.
6868 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6870 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6872 2017-07-28 Richard Biener <rguenther@suse.de>
6874 * match.pd: Remove superfluous :c.
6875 * genmatch.c (simplify::id): Add member.
6876 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6878 (current_id): New global.
6879 (dt_node::parent): Move from ...
6880 (dt_operand::parent): ... here. Add for_id member.
6881 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6882 (decision_tree::find_node): Relax order requirement when
6883 merging DT_TRUE nodes to ones inbetween the current simplify
6884 and the one we try to merge with. Add diagnostic whenever
6885 we need to enforce pattern order by not merging.
6886 (decision_tree::insert): Set current_id.
6887 (decision_tree::print_node): Dump parent node and for_id.
6888 (parser::last_id): Add member.
6889 (parser::push_simplify): Assign unique id.
6890 (parser::parser): Initialize last_id.
6892 2017-07-28 Martin Liska <mliska@suse.cz>
6895 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6896 gimple_build_debug_bind.
6898 2017-07-28 Richard Biener <rguenther@suse.de>
6900 PR tree-optimization/81502
6901 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6903 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6905 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6906 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6907 for BIT_FIELD_REF args.
6908 * fold-const.c (make_bit_field_ref): Likewise.
6909 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6911 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6914 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6915 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6916 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6917 Or it into SANITIZER_UNDEFINED.
6918 * ubsan.c: Include gimple-fold.h and varasm.h.
6919 (ubsan_expand_ptr_ifn): New function.
6920 (instrument_pointer_overflow): New function.
6921 (maybe_instrument_pointer_overflow): New function.
6922 (instrument_object_size): Formatting fix.
6923 (pass_ubsan::execute): Call instrument_pointer_overflow
6924 and maybe_instrument_pointer_overflow.
6925 * internal-fn.c (expand_UBSAN_PTR): New function.
6926 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6927 * sanitizer.def (__ubsan_handle_pointer_overflow,
6928 __ubsan_handle_pointer_overflow_abort): New builtins.
6929 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6930 * internal-fn.def (UBSAN_PTR): New internal function.
6931 * opts.c (sanitizer_opts): Add pointer-overflow.
6932 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6933 * fold-const.c (build_range_check): Compute pointer range check in
6934 integral type if pointer arithmetics would be needed. Formatting
6937 2017-07-28 Martin Liska <mliska@suse.cz>
6940 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6941 parameters that are of a variable-length.
6943 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6945 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6947 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6948 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6949 (CRT_CALL_STATIC_FUNCTION): Likewise.
6950 (ASM_DEFAULT_SPEC): New define.
6951 (ASM_SPEC32): Likewise.
6952 (ASM_SPEC64): Likewise.
6953 (ASM_SPEC_COMMON): Likewise.
6954 (ASM_SPEC): Likewise.
6955 (INVALID_64BIT): Likewise.
6956 (LINK_OS_DEFAULT_SPEC): Likewise.
6957 (LINK_OS_SPEC32): Likewise.
6958 (LINK_OS_SPEC64): Likewise.
6959 (POWERPC_LINUX): Likewise.
6960 (PTRDIFF_TYPE): Likewise.
6961 (RESTORE_FP_PREFIX): Likewise.
6962 (RESTORE_FP_SUFFIX): Likewise.
6963 (SAVE_FP_PREFIX): Likewise.
6964 (SAVE_FP_SUFFIX): Likewise.
6965 (SIZE_TYPE): Likewise.
6966 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6967 (TARGET_64BIT): Likewise.
6968 (TARGET_64BIT): Likewise.
6969 (TARGET_AIX): Likewise.
6970 (WCHAR_TYPE_SIZE): Likewise.
6971 (WCHAR_TYPE): Undefine.
6972 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6973 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6974 (CPP_OS_RTEMS_SPEC): Delete.
6975 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6976 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6978 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6980 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6982 PR tree-optimization/81578
6983 * tree-parloops.c (build_new_reduction): Bail out if
6984 reduction_code isn't one of the standard OpenMP reductions.
6985 Move the details printing after that decision.
6987 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6989 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6990 related to reload_in_progress.
6991 (splat_input_operand): Likewise.
6992 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6994 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6996 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6997 (TARGET_INSTANTIATE_DECLS): Likewise.
6998 (legitimate_indexed_address_p): Delete reload_in_progress code.
6999 (rs6000_debug_legitimate_address_p): Likewise.
7000 (rs6000_eliminate_indexed_memrefs): Likewise.
7001 (rs6000_emit_le_vsx_store): Likewise.
7002 (rs6000_emit_move_si_sf_subreg): Likewise.
7003 (rs6000_emit_move): Likewise.
7004 (register_to_reg_type): Likewise.
7005 (rs6000_pre_atomic_barrier): Likewise.
7006 (rs6000_machopic_legitimize_pic_address): Likewise.
7007 (rs6000_allocate_stack_temp): Likewise.
7008 (rs6000_address_for_fpconvert): Likewise.
7009 (rs6000_address_for_altivec): Likewise.
7010 (rs6000_secondary_memory_needed_rtx): Delete function.
7011 (rs6000_check_sdmode): Likewise.
7012 (rs6000_alloc_sdmode_stack_slot): Likewise.
7013 (rs6000_instantiate_decls): Likewise.
7014 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7015 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7016 Delete reload_in_progress.
7017 (*vec_reload_and_plus_<mptrsize>): Likewise.
7018 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7019 (vsx_div_v2di): Likewise.
7020 (vsx_udiv_v2di): Likewise.
7022 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7024 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7025 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7026 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7027 (rs6000_debug_reg_global): Delete print of LRA status.
7028 (rs6000_option_override_internal): Delete dead LRA related code.
7029 (rs6000_lra_p): Delete function.
7030 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7032 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7034 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7035 * config/riscv/rtems.h: New file.
7037 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7038 Sudakshina Das <sudi.das@arm.com>
7040 * config/aarch64/aarch64.md
7041 (define_split for and<mode>3nr_compare): Move
7042 non aarch64_logical_operand to a register.
7043 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7044 register immediate operand to a register.
7045 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7047 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7050 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7052 2017-07-27 Richard Biener <rguenther@suse.de>
7054 PR tree-optimization/81573
7055 PR tree-optimization/81494
7056 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7057 multi defuse cycle case.
7059 2017-07-27 Richard Biener <rguenther@suse.de>
7061 PR tree-optimization/81571
7062 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7065 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7067 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7068 earlier and only if MASK_FPU is set. Adjust formatting.
7070 2017-07-27 Martin Liska <mliska@suse.cz>
7072 * opt-functions.awk: Add validation of value of Init.
7073 * optc-gen.awk: Pass new argument.
7075 2017-07-27 Martin Liska <mliska@suse.cz>
7077 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7078 Fix wrong condition.
7080 2017-07-27 Martin Liska <mliska@suse.cz>
7082 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7083 BBs and edges seen by autoFDO.
7085 2017-07-27 Richard Biener <rguenther@suse.de>
7087 PR tree-optimization/81502
7088 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7089 with incompatible but same sized type.
7090 (execute_update_addresses_taken): Likewise.
7092 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7094 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7095 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7097 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7100 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7101 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7102 Change s_operand to memory_operand.
7104 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7106 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7107 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7108 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7109 Emit instructions rather than returning an expression. Handle TFmode
7110 and KFmode by casting to TImode.
7111 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7112 (rs6000_emit_le_vsx_store): Likewise.
7113 * config/rs6000/vsx.md (VSX_TI): New iterator.
7114 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7115 (*vsx_le_undo_permute_<mode>): Likewise.
7116 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7117 emit the split sequence.
7118 (*vsx_le_perm_store_<mode>): Likewise.
7120 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7122 PR tree-optimization/81555
7123 PR tree-optimization/81556
7124 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7125 if true, force CHANGED for the recursive invocation.
7126 (reassociate_bb): Remember original length of ops array, pass
7127 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7129 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7130 attributes for noipa attribute. For naked attribute use
7131 lookup_attribute first before lookup_attribute_spec.
7132 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7134 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7135 for functions with noipa attribute.
7136 (cgraph_externally_visible_p): Return true for functions with noipa
7138 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7139 for functions with noipa attribute.
7140 * doc/extend.texi: Document noipa function attribute.
7141 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7142 also for functions with noipa attribute.
7143 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7145 2017-07-26 Andrew Pinski <apinski@cavium.com>
7147 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7148 vec_unalign_load_cost and vec_unalign_store_cost.
7150 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7152 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7153 -mvsx-small-integer option.
7154 (ISA_3_0_MASKS_IEEE): Likewise.
7155 (OTHER_VSX_VECTOR_MASKS): Likewise.
7156 (POWERPC_MASKS): Likewise.
7157 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7158 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7159 code, only testing for DImode being allowed in non-VSX floating
7161 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7162 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7164 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7165 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7166 TARGET_P8_VECTOR test.
7167 (rs6000_secondary_reload_simple_move): Likewise.
7168 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7169 since TARGET_P9_VECTOR was already tested.
7170 (rs6000_opt_masks): Remove -mvsx-small-integer.
7171 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7172 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7174 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7175 test for TARGET_VEXTRACTUB was used, and that uses
7177 (p9 extract splitter): Likewise.
7178 (vsx_extract_<mode>_di_p9): Likewise.
7179 (vsx_extract_<mode>_store_p9): Likewise.
7180 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7181 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7182 the elimination of TARGET_VSX_SMALL_INTEGER.
7183 (vsx_extract_<mode>_p8): Likewise.
7184 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7185 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7186 (vsx_set_<mode>_p9): Likewise.
7187 (vsx_set_v4sf_p9): Likewise.
7188 (vsx_set_v4sf_p9_zero): Likewise.
7189 (vsx_insert_extract_v4sf_p9): Likewise.
7190 (vsx_insert_extract_v4sf_p9_2): Likewise.
7191 * config/rs6000/rs6000.md (sign extend splitter): Change
7192 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7193 (floatsi<mode>2_lfiwax_mem): Likewise.
7194 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7195 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7196 since a test for TARGET_P9_VECTOR was used.
7197 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7198 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7199 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7200 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7201 TARGET_P8_VECTOR test.
7202 (fix_trunc<mode>si2_stfiwx): Likewise.
7203 (fix_trunc<mode>si2_internal): Likewise.
7204 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7205 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7207 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7208 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7209 TARGET_P8_VECTOR test.
7210 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7211 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7212 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7214 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7215 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7216 since a test for TARGET_P9_VECTOR was used.
7217 (splitter for loading small constants): Likewise.
7219 2017-07-26 Andrew Pinski <apinski@cavium.com>
7221 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7224 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7227 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7228 (fp_valid_at): Likewise.
7230 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7232 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7233 (qdf24xx_addrcost_table): Likewise.
7234 (cortexa57_tunings): Update to use generic_branch_cost.
7235 (cortexa72_tunings): Likewise.
7236 (cortexa73_tunings): Likewise.
7237 (qdf24xx_tunings): Likewise.
7239 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7241 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7242 (thunderx2t99_branch_cost): Likewise.
7243 (cortexa35_tunings): Update to use generic_branch_cost.
7244 (cortexa53_tunings): Likewise.
7245 (cortexa57_tunings): Likewise.
7246 (cortexa72_tunings): Likewise.
7247 (cortexa73_tunings): Likewise.
7248 (thunderx2t99_tunings): Likewise.
7250 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7252 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7253 (sparc_option_override): Honour MASK_FSMULD.
7254 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7255 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7256 * config/sparc/sparc.opt (mfsmuld): New option.
7257 * doc/invoke.texi (mfsmuld): Document option.
7259 2017-07-26 Marek Polacek <polacek@redhat.com>
7262 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7264 2017-07-26 Richard Biener <rguenther@suse.de>
7266 * gimple-match-head.c (do_valueize): Return OP if valueize
7268 (get_def): New helper to get at the def stmt of a SSA name
7270 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7271 do_valueize to get at the def stmt.
7272 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7274 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7277 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7279 2017-07-26 Martin Liska <mliska@suse.cz>
7282 * function.c (expand_function_start): Make expansion of
7283 nonlocal_goto_save_area after parm_birth_insn.
7285 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7287 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7288 from all CPU target flags enable members.
7290 2017-07-26 Richard Biener <rguenther@suse.de>
7292 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7293 (decision_tree::gen): Make 'type' const.
7294 (write_predicate): Likewise.
7296 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7298 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7299 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7300 (rs6000_option_override_internal): Likewise.
7301 (rs6000_expand_vector_set): Likewise.
7302 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7303 (TARGET_UPPER_REGS_SF): Likewise.
7304 (TARGET_UPPER_REGS_DI): Likewise.
7305 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7306 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7307 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7308 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7309 (Splitters for DI constants in Altivec registers): Likewise.
7310 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7311 (vsx_set_v4sf_p9): Likewise.
7312 (vsx_set_v4sf_p9_zero): Likewise.
7313 (vsx_insert_extract_v4sf_p9): Likewise.
7314 (vsx_insert_extract_v4sf_p9_2): Likewise.
7316 2017-07-25 Carl Love <cel@us.ibm.com>
7318 * doc/extend.texi: Update the built-in documentation file for the
7319 existing built-in functions
7320 vector signed char vec_cnttz (vector signed char);
7321 vector unsigned char vec_cnttz (vector unsigned char);
7322 vector signed short vec_cnttz (vector signed short);
7323 vector unsigned short vec_cnttz (vector unsigned short);
7324 vector signed int vec_cnttz (vector signed int);
7325 vector unsigned int vec_cnttz (vector unsigned int);
7326 vector signed long long vec_cnttz (vector signed long long);
7327 vector unsigned long long vec_cnttz (vector unsigned long long);
7329 2017-07-25 Andrew Pinski <apinski@cavium.com>
7331 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7332 accesses where the use is for the first operand of a BIT_INSERT.
7334 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7337 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7338 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7340 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7342 * config/i386/gstabs.h: Delete.
7343 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7345 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7347 * config/i386/i386.c (ix86_decompose_address): Do not check for
7348 register RTX when looking at index_reg or base_reg.
7349 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7351 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7353 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7354 to update EH info here.
7356 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7358 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7360 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7362 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7364 2017-07-25 Torsten Duwe <duwe@suse.de>
7366 * common.opt: Introduce -fpatchable-function-entry
7367 command line option, and its variables function_entry_patch_area_size
7368 and function_entry_patch_area_start.
7369 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7370 including a two-value parser.
7371 * target.def (print_patchable_function_entry): New target hook.
7372 * targhooks.h (default_print_patchable_function_entry): New function.
7373 * targhooks.c (default_print_patchable_function_entry): Likewise.
7374 * toplev.c (process_options): Switch off IPA-RA if
7375 patchable function entries are being generated.
7376 * varasm.c (assemble_start_function): Look at the
7377 patchable-function-entry command line switch and current
7378 function attributes and maybe generate NOP instructions by
7379 calling the print_patchable_function_entry hook.
7380 * doc/extend.texi: Document patchable_function_entry attribute.
7381 * doc/invoke.texi: Document -fpatchable_function_entry
7382 command line option.
7383 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7385 * doc/tm.texi: Re-generate.
7387 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7390 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7391 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7393 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7395 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7397 2017-07-25 Richard Biener <rguenther@suse.de>
7399 PR tree-optimization/81455
7400 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7401 not walk in cycles when looking for guards.
7403 2017-07-25 Richard Biener <rguenther@suse.de>
7405 PR tree-optimization/81529
7406 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7407 when optimizing backedge uses.
7409 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7411 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7413 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7414 to dl_section_ref. On AIX, append an expression to subtract
7415 the size of the section length to dl_section_ref.
7417 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7419 * configure.ac: If any of the config.* scripts fail, exit 1.
7420 * configure: Regenerate.
7422 2017-07-25 Richard Biener <rguenther@suse.de>
7425 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7426 of immediate uses, be more verbose on errors.
7428 2017-07-25 Richard Biener <rguenther@suse.de>
7430 PR tree-optimization/81510
7431 * tree-vect-loop.c (vect_is_simple_reduction): When the
7432 reduction stmt is not inside the loop bail out.
7434 2017-07-25 Richard Biener <rguenther@suse.de>
7436 PR tree-optimization/81303
7437 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7438 profitability check against LOOP_VINFO_NITERSM1.
7440 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7442 * domwalk.c (cmp_bb_postorder): Simplify.
7443 (sort_bbs_postorder): New function. Use it...
7444 (dom_walker::walk): ...here to optimize common cases.
7446 2017-07-25 Martin Liska <mliska@suse.cz>
7449 * ipa-visibility.c (function_and_variable_visibility): Make the
7450 redirection just on target that supports aliasing.
7451 Fix GNU coding style.
7453 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7456 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7458 * config/arm/rtems.h: Likewise.
7459 * config/bfin/rtems.h: Likewise.
7460 * config/i386/rtemself.h: Likewise.
7461 * config/lm32/rtems.h: Likewise.
7462 * config/m32c/rtems.h: Likewise.
7463 * config/m68k/rtemself.h: Likewise.
7464 * config/microblaze/rtems.h: Likewise.
7465 * config/mips/rtems.h: Likewise.
7466 * config/moxie/rtems.h: Likewise.
7467 * config/nios2/rtems.h: Likewise.
7468 * config/powerpcspe/rtems.h: Likewise.
7469 * config/rs6000/rtems.h: Likewise.
7470 * config/rtems.h: Likewise.
7471 * config/sh/rtems.h: Likewise.
7472 * config/sh/rtemself.h: Likewise.
7473 * config/sparc/rtemself.h: Likewise.
7475 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7478 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7479 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7480 * tree-ssa-structalias.c (alias_get_name): Same.
7482 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7485 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7486 instructions if no du chain is found.
7488 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7490 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7492 2017-07-25 Richard Biener <rguenther@suse.de>
7495 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7498 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7500 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7502 (ISA_2_7_MASKS_SERVER): Likewise.
7503 (ISA_3_0_MASKS_IEEE): Likewise.
7504 (OTHER_P8_VECTOR_MASKS): Likewise.
7505 (OTHER_VSX_VECTOR_MASKS): Likewise.
7506 (POWERPC_MASKS): Likewise.
7507 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7508 duplicate list of options.
7509 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7510 explicit -mupper-regs options.
7511 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7512 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7513 alias for -mupper-regs-df.
7514 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7515 (rs6000_init_hard_regno_mode_ok): Likewise.
7516 (rs6000_option_override_internal): Likewise.
7517 (rs6000_opt_masks): Likewise.
7518 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7519 options in terms of whether -mvsx or -mpower8-vector was used.
7520 (TARGET_UPPER_REGS_DI): Likewise.
7521 (TARGET_UPPER_REGS_SF): Likewise.
7522 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7523 -mupper-regs-* options.
7525 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7527 * passes.c (emergency_dump_function): Print some empty lines and a
7528 header before the RTL dump.
7530 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7532 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7534 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7537 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7538 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7540 2017-07-24 Carl Love <cel@us.ibm.com>
7542 * config/rs6000/rs6000-c.c: Add support for built-in functions
7543 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7544 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7545 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7546 vec_extract_fp_from_shortl): Add defines for the two builtins.
7547 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7548 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7550 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7551 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7552 * doc/extend.texi: Update the built-in documentation file for the
7553 new built-in function.
7555 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7558 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7560 * doc/generic.texi: Likewise.
7561 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7562 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7564 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7566 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7567 (aarch64_mls_elt_merge<mode>): Likewise.
7569 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7571 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7572 having __cxa_atexit.
7574 2017-07-23 Michael Collison <michael.collison@arm.com>
7576 * config/arm/arm.c (arm_option_override): Deprecate
7577 use of -mstructure-size-boundary.
7578 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7579 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7581 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7584 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7585 Reduce cost estimate for direct moves.
7587 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7590 * config/i386/i386.c (ix86_option_override_internal): Disable
7591 BMI, BMI2 and TBM instructions for -m16.
7593 2017-07-21 Carl Love <cel@us.ibm.com>
7595 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7596 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7597 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7598 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7599 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7600 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7601 VMULOSW): New enum "unspec" values.
7602 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7603 altivec_vmulosw): New patterns.
7604 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7605 VMULOSW): Add definitions.
7607 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7609 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7610 (qdf24xx): Likewise.
7611 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7612 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7613 (AARCH64_FL_V8_1): Renumber.
7614 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7615 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7616 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7617 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7618 rdma to feature modifiers list.
7620 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7623 * ipa-visibility (function_and_variable_visibility): Convert
7624 recursive PLT call to direct call if appropriate.
7626 2017-07-21 Andrew Pinski <apinski@cavium.com>
7628 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7629 operand 1 to see if the types precision matches.
7630 * fold-const.c (operand_equal_p): Likewise.
7632 2017-07-21 Richard Biener <rguenther@suse.de>
7634 PR tree-optimization/81303
7635 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7636 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7637 (vect_peeling_hash_get_lowest_cost): Adjust.
7638 (vect_enhance_data_refs_alignment): Likewise. Use
7639 vect_get_peeling_costs_all_drs to compute the penalty for no
7640 peeling to match up costs.
7642 2017-07-21 Richard Biener <rguenther@suse.de>
7644 PR tree-optimization/81500
7645 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7646 we didn't identify a reduction path.
7648 2017-07-21 Tom de Vries <tom@codesourcery.com>
7649 Cesar Philippidis <cesar@codesourcery.com>
7651 PR gcov-profile/81442
7652 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7655 2017-07-21 Tom de Vries <tom@codesourcery.com>
7658 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7660 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7661 nvptx_override_options_after_change.
7663 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7665 * dwarf2out.c (output_file_names): Avoid double testing for
7668 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7670 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7672 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7674 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7676 (try_crossjump_to_edge): Do not punt on partitioned functions.
7678 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7680 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7681 Put all BBs reachable only via paths crossing cold region to cold
7683 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7685 2016-07-21 Richard Biener <rguenther@suse.de>
7687 PR tree-optimization/81303
7688 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7689 into account prologue and epilogue iterations when raising
7690 min_profitable_iters to sth at least covering one vector iteration.
7692 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7694 * config/arm/arm.c (arm_test_cpu_arch_dat):
7697 2017-07-20 Nathan Sidwell <nathan@acm.org>
7699 Remove TYPE_METHODS.
7700 * tree.h (TYPE_METHODS): Delete.
7701 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7702 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7703 (dbxout_type_methods): Scan TYPE_FIELDS.
7704 (dbxout_type): Don't check TYPE_METHODS here.
7705 * function.c (use_register_for_decl): Always ignore register for
7706 class types when not optimizing.
7707 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7708 * tree.c (free_lang_data_in_type): Stitch out member functions and
7709 templates from TYPE_FIELDS.
7710 (build_distinct_type_copy, verify_type_variant,
7711 verify_type): Member fns are on TYPE_FIELDS.
7712 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7713 * tree-pretty-print.c (dump_generic_node): Likewise.
7715 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7718 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7719 V2TImode and V4TImode.
7720 (ix86_expand_vector_extract): Likewise.
7721 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7722 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7723 (ssescalarmode): Handle V4TImode and V2TImode.
7724 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7725 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7726 (VEXTRACTI128_MODE): New mode iterator.
7727 (splitter for *vec_extractv?ti first element): New.
7728 (VEC_INIT_MODE): New mode iterator.
7729 (vec_init<mode>): Consolidate 3 expanders into one using
7730 VEC_INIT_MODE mode iterator.
7732 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7734 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7735 non_spilled_static_chain_regno_p.
7737 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7739 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7741 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7743 * bb-reorder.c (connect_traces): Allow copying of blocks within
7746 2017-07-20 Richard Biener <rguenther@suse.de>
7748 * gimple.h (gimple_phi_result): Add gphi * overload.
7749 (gimple_phi_result_ptr): Likewise.
7750 (gimple_phi_arg): Likewise. Adjust index assert to only
7751 allow actual argument accesses rather than all slots available
7753 (gimple_phi_arg_def): Add gphi * overload.
7754 * tree-phinodes.c (make_phi_node): Initialize only actual
7756 (resize_phi_node): Clear memory not covered by old node,
7757 do not initialize excess argument slots.
7758 (reserve_phi_args_for_new_edge): Initialize new argument slot
7761 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7763 PR tree-optimization/81388
7765 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7767 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7768 by removing computation of may_be_zero.
7770 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7771 Tom de Vries <tom@codesourcery.com>
7774 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7775 when gimple level profile disagrees with what RTL expander did.
7777 2017-07-20 Richard Biener <rguenther@suse.de>
7779 PR tree-optimization/61171
7780 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7781 (vect_analyze_stmt): Add slp instance parameter.
7782 (vectorizable_reduction): Likewise.
7783 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7784 (vect_is_simple_reduction): Deal with chains not detected
7785 as SLP reduction chain, specifically not properly associated
7786 chains containing a mix of plus/minus.
7787 (get_reduction_op): Remove.
7788 (get_initial_defs_for_reduction): Simplify, pass in whether
7789 this is a reduction chain, pass in the SLP node for the PHIs.
7790 (vect_create_epilog_for_reduction): Get the SLP instance as
7792 (vectorizable_reduction): Get the SLP instance as arg.
7793 During analysis remember the SLP node with the PHIs in the
7794 instance. Simplify getting at the vectorized reduction PHIs.
7795 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7796 through SLP instance.
7797 (vect_slp_analyze_operations): Likewise.
7798 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7799 (vect_transform_stmt): Likewise.
7801 2017-07-20 Tom de Vries <tom@codesourcery.com>
7803 PR tree-optimization/81489
7804 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7805 read of phi arg location to before loop that modifies phi.
7807 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7809 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7812 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7815 * except.c (execute): Fix ordering issue.
7817 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7819 PR rtl-optimization/81423
7820 * combine.c (make_compound_operation_int): Don't try to optimize
7821 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7823 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7825 PR rtl-optimization/81423
7826 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7827 with a constant that is -1 in the truncated to mode.
7829 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7831 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7832 (determine_unlikely_bbs): ... here.
7833 * predict.h (propagate_unlikely_bbs_forward): Declare.
7834 * cfgexpand.c (pass_expand::execute): Use it.
7835 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7837 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7838 propagate_unlikely_bbs_forward.
7840 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7843 * except.c (maybe_add_nop_after_section_switch): New function.
7846 2017-07-19 Tom de Vries <tom@codesourcery.com>
7848 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7850 2017-07-19 Tom de Vries <tom@codesourcery.com>
7852 * gimple.h (gimple_phi_arg): Make assert more strict.
7854 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7856 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7857 * config/rs6000/mmintrin.h: New file.
7858 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7860 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7862 PR tree-optimization/81346
7863 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7865 2017-07-19 Tom de Vries <tom@codesourcery.com>
7867 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7869 2017-07-19 Tom de Vries <tom@codesourcery.com>
7871 * config/nvptx/nvptx-modes.def: Add V2DImode.
7872 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7873 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7874 (nvptx_output_mov_insn): Handle lack of mov.b128.
7875 (nvptx_print_operand): Handle 'H' and 'L' codes.
7876 (nvptx_vector_mode_supported): Allow V2DImode.
7877 (nvptx_preferred_simd_mode): New function.
7878 (nvptx_data_alignment): New function.
7879 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7880 nvptx_preferred_simd_mode.
7881 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7883 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7885 2017-07-19 Tom de Vries <tom@codesourcery.com>
7887 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7888 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7889 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7890 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7891 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7892 (mov<VECIM>_insn): New define_insn.
7893 (define_expand "mov<VECIM>): New define_expand.
7895 2017-07-19 Tom de Vries <tom@codesourcery.com>
7897 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7899 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7901 PR tree-optimization/81346
7902 * fold-const.h (fold_div_compare, range_check_type): Declare.
7903 * fold-const.c (range_check_type): New function.
7904 (build_range_check): Use range_check_type.
7905 (fold_div_compare): No longer static, rewritten into
7906 a match.pd helper function.
7907 (fold_comparison): Don't call fold_div_compare here.
7908 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7911 2017-07-19 Nathan Sidwell <nathan@acm.org>
7913 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7914 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7915 * tree.c (find_decls_types_r, verify_type): Use
7916 TYPE_{MIN,MAX}_VALUE_RAW.
7917 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7918 (hash_tree): Likewise.
7919 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7921 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7924 2017-07-18 Tom de Vries <tom@codesourcery.com>
7927 * omp-expand.c (expand_omp_for_static_chunk): Handle
7928 equal-argument loop exit phi.
7930 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7933 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7934 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7935 operand 2 predicate.
7936 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7937 operand 2 predicate.
7938 (ror,rol -> rorx splitters): Use const_int_operand as
7939 operand 2 predicate.
7941 2017-06-18 Richard Biener <rguenther@suse.de>
7943 PR tree-optimization/81410
7944 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7945 the gap in the ! slp_perm SLP case after each group.
7947 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7950 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7953 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7956 * predict.c (set_even_probabilities): Cleanup; do not affect
7957 probabilities that are already known.
7958 (combine_predictions_for_bb): Call even when count is set.
7960 2017-07-18 Nathan Sidwell <nathan@acm.org>
7962 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7965 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7968 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7969 optimization for loop niter analysis.
7971 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7974 * config/avr/avr.c (avr_optimize_casesi): Don't use
7975 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7977 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7979 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7980 body_cost_vec from _vect_peel_extended_info.
7981 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7982 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7985 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7987 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7989 2017-07-18 Richard Biener <rguenther@suse.de>
7991 PR tree-optimization/80620
7992 PR tree-optimization/81403
7993 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7994 info when re-using a VN table entry.
7996 2017-07-18 Richard Biener <rguenther@suse.de>
7998 PR tree-optimization/81418
7999 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8000 vectype_in. Verify that with lane-reducing reduction operations
8001 we have a single def-use cycle.
8003 2017-07-17 Carl Love <cel@us.ibm.com>
8005 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8007 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8008 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8009 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8010 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8011 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8012 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8013 VMULOSW): New enum "unspec" values.
8014 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8015 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8016 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8017 altivec_vmulosw): New patterns.
8018 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8019 VMULOSW): Add definitions.
8020 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8022 * config/alpha/alpha.c: Include predict.h.
8024 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8026 * tree-vrp.c (compare_assert_loc): Fix comparison function
8027 to return predictable results.
8029 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8031 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8034 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8035 * doc/invoke.texi (mexpand-adddi): Update text.
8037 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8039 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8040 that also clobbers the CC register. The old expand code is moved
8042 (*arc_clzsi2): ... here.
8043 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8044 the CC register. The old expand code is moved to ...
8045 (arc_ctzsi2): ... here.
8047 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8049 * config/arc/arc.opt (mindexed-loads): Use initial value
8050 TARGET_INDEXED_LOADS_DEFAULT.
8051 (mauto-modify-reg): Use initial value
8052 TARGET_AUTO_MODIFY_REG_DEFAULT.
8053 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8054 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8055 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8056 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8058 2017-07-17 Martin Liska <mliska@suse.cz>
8061 * opts.c (finish_options): Do not allow -fgnu-tm
8062 w/ -fsanitize={kernel-,}address. Say sorry.
8064 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8067 * tree-loop-distribution.c (classify_partition): Only assert on
8068 numer of iterations.
8069 (merge_dep_scc_partitions): Delete prameter. Update function call.
8070 (distribute_loop): Remove code handling loop with unknown niters.
8071 (pass_loop_distribution::execute): Skip loop with unknown niters.
8073 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8076 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8077 function sort_partitions_by_post_order.
8079 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8081 PR tree-optimization/81374
8082 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8083 the max index of basic blocks, rather than number of basic blocks.
8085 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8087 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8089 (arc_legitimate_pic_operand_p): Likewise.
8090 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8092 (arc_needs_pcl_p): Likewise.
8093 (arc_legitimate_pc_offset_p): Likewise.
8094 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8095 function is also used in constrains.md.
8096 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8097 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8098 PLUS. Only return true/false in known cases, otherwise assert.
8099 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8100 is already called in arc_legitimate_constant_p.
8101 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8103 (LEGITIMATE_PIC_OPERAND_P): Use
8104 arc_raw_symbolic_reference_mentioned_p function.
8105 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8110 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8111 Andrew Burgess <andrew.burgess@embecosm.com>
8113 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8114 (arc_return_address_register): New function.
8115 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8116 (arc_handle_fndecl_attribute): Add naked attribute.
8117 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8118 (TARGET_WARN_FUNC_RETURN): Likewise.
8119 (arc_allocate_stack_slots_for_args): New function.
8120 (arc_warn_func_return): Likewise.
8121 (machine_function): Change type fn_type.
8122 (arc_compute_function_type): Consider new naked function type,
8123 change function return type.
8124 (arc_must_save_register): Adapt to handle new
8125 arc_compute_function_type's return type.
8126 (arc_expand_prologue): Likewise.
8127 (arc_expand_epilogue): Likewise.
8128 (arc_return_address_regs): Delete.
8129 (arc_return_address_register): New function.
8130 (arc_epilogue_uses): Use above function.
8131 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8132 (arc_function_type): Change encoding, add naked type.
8133 (ARC_INTERRUPT_P): Change to handle the new encoding.
8134 (ARC_FAST_INTERRUPT_P): Likewise.
8135 (ARC_NORMAL_P): Define.
8136 (ARC_NAKED_P): Likewise.
8137 (arc_compute_function_type): Delete prototype.
8138 * config/arc/arc.md (in_ret_delay_slot): Use
8139 arc_return_address_register function.
8140 (simple_return): Likewise.
8141 (p_return_i): Likewise.
8143 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8145 PR tree-optimization/81428
8146 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8147 can't be built for those types.
8149 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8151 Remove stuff dead since r239246.
8153 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8154 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8155 (avr_inform_devices): Remove dead stuff.
8157 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8159 * config/arm/arm_neon.h: Fix softp typo.
8161 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8163 PR tree-optimization/81365
8164 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8165 aggregate moves onto bb predecessor edges, make sure there are no
8166 loads that could alias the lhs in between the start of bb and the
8169 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8172 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8173 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8174 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8176 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8178 PR tree-optimization/81396
8179 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8180 (init_symbolic_number): Initialize it to 1.
8181 (perform_symbolic_merge): Add n_ops from both operands into the new
8183 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8184 without base_addr as useless if they need more than one operation.
8185 (bswap_replace): Handle !bswap case for NULL base_addr.
8187 2017-07-17 Tom de Vries <tom@codesourcery.com>
8190 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8193 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8195 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8196 conditional builtin define __FIX_LEON3FT_B2BST.
8198 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8200 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8201 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8204 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8206 PR rtl-optimization/81424
8207 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8208 to remove potential trapping from operands if -fnon-call-exceptions.
8210 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8212 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8213 profile_proability for scalling.
8214 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8216 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8218 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8220 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8222 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8223 fixpoint arithmetics.
8225 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8227 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8228 fixpoint arithmetics.
8230 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8232 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8233 fixpoint arithmetics.
8235 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8237 * profile-count.h (profile_probability::from_reg_br_prob_note,
8238 profile_probability::to_reg_br_prob_note): New functions.
8239 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8240 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8241 * predict.c (probability_reliable_p): Update.
8242 (edge_probability_reliable_p): Update.
8243 (br_prob_note_reliable_p): Update.
8244 (invert_br_probabilities): Update.
8245 (add_reg_br_prob_note): New function.
8246 (combine_predictions_for_insn): Update.
8247 * asan.c (asan_clear_shadow): Update.
8248 * cfgbuild.c (compute_outgoing_frequencies): Update.
8249 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8250 (update_br_prob_note): Update.
8251 (rtl_verify_edges): Update.
8252 (purge_dead_edges): Update.
8253 (fixup_reorder_chain): Update.
8254 * emit-rtl.c (try_split): Update.
8255 * ifcvt.c (cond_exec_process_insns): Update.
8256 (cond_exec_process_if_block): Update.
8257 (dead_or_predicable): Update.
8258 * internal-fn.c (expand_addsub_overflow): Update.
8259 (expand_neg_overflow): Update.
8260 (expand_mul_overflow): Update.
8261 * loop-doloop.c (doloop_modify): Update.
8262 * loop-unroll.c (compare_and_jump_seq): Update.
8263 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8264 * predict.h: Update.
8265 * reorg.c (mostly_true_jump): Update.
8267 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8268 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8269 * config/arc/arc.c: (emit_unlikely_jump): Update.
8270 * config/arm/arm.c: (emit_unlikely_jump): Update.
8271 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8272 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8273 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8274 (ix86_print_operand): Update.
8275 (ix86_split_fp_branch): Update.
8276 (predict_jump): Update.
8277 * config/ia64/ia64.c (ia64_print_operand): Update.
8278 * config/mmix/mmix.c (mmix_print_operand): Update.
8279 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8280 (rs6000_expand_split_stack_prologue): Update.
8281 * config/rs6000/rs6000.c: Update.
8282 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8283 (s390_expand_vec_movstr): Update.
8284 (s390_expand_cs_tdsi): Update.
8285 (s390_expand_split_stack_prologue): Update.
8286 * config/sh/sh.c (sh_print_operand): Update.
8287 (expand_cbranchsi4): Update.
8288 (expand_cbranchdi4): Update.
8289 * config/sparc/sparc.c (output_v9branch): Update.
8290 * config/spu/spu.c (get_branch_target): Update.
8291 (ea_load_store_inline): Update.
8292 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8293 * config/tilepro/tilepro.c: Update.
8295 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8297 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8298 (gimplify_save_expr): Likewise.
8300 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8302 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8304 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8306 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8308 * timevar.def (TV_IPA_FNSUMMARY): Define.
8310 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8312 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8313 to back store errata sensitive sequence from being generated.
8314 (sqrtdf2_fix): Likewise.
8316 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8318 * tree-ssa-threadupdate.c (compute_path_counts,
8319 update_joiner_offpath_counts): Use profile_probability.
8321 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8324 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8326 * config/arm/arm-c.c (arm_cpu_builtins): Define
8327 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8329 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8332 array entries to represent __ieee128 versions of the
8333 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8334 scalar_extract_sig, and scalar_insert_exp built-in functions.
8335 (altivec_resolve_overloaded_builtin): Add special case handling
8336 for the __builtin_scalar_insert_exp function, as represented by
8337 the P9V_BUILTIN_VEC_VSIEDP constant.
8338 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8339 exponent support for __ieee128 argument.
8340 (VSESQP): Add scalar extract signature support for __ieee128
8342 (VSTDCNQP): Add scalar test negative support for __ieee128
8344 (VSIEQP): Add scalar insert exponent support for __int128 argument
8345 with __ieee128 result.
8346 (VSIEQPF): Add scalar insert exponent support for __ieee128
8347 argument with __ieee128 result.
8348 (VSTDCQP): Add scalar test data class support for __ieee128
8350 (VSTDCNQP): Add overload support for scalar test negative with
8352 (VSTDCQP): Add overload support for scalar test data class
8354 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8356 (UNSPEC_VSX_SIEXPQP): New constant.
8357 (xsxexpqp): New insn for VSX scalar extract exponent quad
8359 (xsxsigqp): New insn for VSX scalar extract significand quad
8361 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8362 precision with floating point argument.
8363 (xststdcqp): New expand for VSX scalar test data class quad
8365 (xststdcnegqp): New expand for VSX scalar test negative quad
8367 (xststdcqp): New insn to match expansions for VSX scalar test data
8368 class quad precision and VSX scalar test negative quad precision.
8369 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8370 special case operand checking to enforce that second operand of
8371 VSX scalar test data class with quad precision argument is a 7-bit
8373 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8374 prototypes and descriptions of __ieee128 versions of
8375 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8376 scalar_test_data_class, and scalar_test_neg built-in functions.
8378 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8380 PR tree-optimization/81162
8381 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8382 replace a negate with an add.
8384 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8386 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8388 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8390 * config/arm/arm-c.c (arm_cpu_builtins): Define
8391 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8393 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8395 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8396 (armv8-r): Set ARM Cortex-R52 as default CPU.
8397 * config/arm/arm-tables.opt: Regenerate.
8398 * config/arm/arm-tune.md: Regenerate.
8399 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8401 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8402 extension for -mcpu=cortex-r52.
8404 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8406 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8407 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8408 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8409 (fp-armv8): Define it as FP_ARMv8 only.
8410 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8411 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8413 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8414 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8415 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8416 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8417 than TARGET_FPU_ARMV8.
8418 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8419 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8420 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8422 * config/arm/neon.md (neon_vrint): Likewise.
8423 (neon_vcvt): Likewise.
8424 (neon_<fmaxmin_op><mode>): Likewise.
8425 (<fmaxmin><mode>3): Likewise.
8426 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8427 * config/arm/predicates.md (arm_cond_move_operator): Check against
8428 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8430 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8432 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8435 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8437 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8438 loop in comment with memset.
8440 2017-07-14 Martin Liska <mliska@suse.cz>
8442 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8443 * dwarf2out.c (is_java): Remove the function.
8444 (output_pubname): Remove usage of the function.
8445 (lower_bound_default): Remove usage of DW_LANG_Java.
8446 (gen_compile_unit_die): Likewise.
8447 * gcc.c: Remove compiler defaults for .java and .zip files.
8448 * gimple-expr.c (remove_suffix): Change as there's no longer
8449 extension than 4-letter one.
8450 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8451 (gimplify_save_expr): Likewise.
8452 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8453 as it's possible even for other languages than Java.
8454 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8455 * lto-opts.c (lto_write_options): Remove reference to Java.
8456 * opts.c (strip_off_ending): Update file extension handling.
8457 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8458 * tree-eh.c (lower_resx): Likewise.
8459 * tree.c (free_lang_data_in_type): Remove dead code.
8460 (find_decls_types_r): Likewise.
8461 (build_common_builtin_nodes): Remove Java from a comment.
8462 (verify_type): Remove dead code.
8463 * varasm.c (assemble_external): Remove Java from a comment.
8465 2017-07-14 Martin Liska <mliska@suse.cz>
8467 * opts.c (finish_options): Add quotes.
8468 (common_handle_option): Likewise.
8470 2017-07-14 Martin Liska <mliska@suse.cz>
8472 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8473 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8475 * dwarf2out.c (lower_bound_default): Do not handle
8477 (gen_compile_unit_die): Likewise.
8478 * gcc.c: Remove default extension binding for GNU Pascal.
8479 * stmt.c: Remove Pascal language from a comment.
8480 * xcoffout.c: Likewise.
8482 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8485 * diagnostic-show-locus.c (fixit_cmp): New function.
8486 (layout::layout): Sort m_fixit_hints.
8487 (column_range::column_range): Assert that the values are valid.
8488 (struct char_span): New struct.
8489 (correction::overwrite): New method.
8490 (struct source_line): New struct.
8491 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8492 calls in terms of classes source_line and char_span, and
8493 correction::overwrite.
8494 (selftest::test_overlapped_fixit_printing_2): New function.
8495 (selftest::diagnostic_show_locus_c_tests): Call it.
8497 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8499 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8500 early if there is no lhs.
8502 2017-07-13 Martin Liska <mliska@suse.cz>
8504 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8505 (gen_reference_type_die): Likewise.
8506 * stor-layout.c: Remove Pascal-related comment.
8508 2017-07-13 Martin Liska <mliska@suse.cz>
8510 * opts.c (finish_options): Add quotes to error messages.
8511 (parse_sanitizer_options): Likewise.
8513 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8515 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8517 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8519 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8521 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8523 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8525 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8527 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8530 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8531 provides the hardware capability bits, define the macro
8532 __BUILTIN_CPU_SUPPORTS__.
8533 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8534 if GLIBC does not provide the hardware capability bits. Add a
8535 gcc_unreachable call if the built-in cpu function is neither
8536 __builtin_cpu_is nor __builtin_cpu_supports.
8537 (rs6000_get_function_versions_dispatcher): Change the warning
8538 that an old GLIBC is used which does not export the capability
8539 bits to be an error.
8540 * doc/extend.texi (target_clones attribute): Document the
8541 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8542 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8543 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8544 the macros defined by GCC if the newer GLIBC is available.
8546 2017-07-12 Jeff Law <law@redhat.com>
8548 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8549 remaining includes slightly.
8550 * config/riscv/riscv-builtins.c: Include profile-count.h.
8552 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8555 * config/avr/avr.c (avr_set_current_function): In diagnostic
8556 messages: Quote keywords and (parts of) identifiers.
8557 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8560 2017-07-12 Carl Love <cel@us.ibm.com>
8562 * config/rs6000/rs6000-c.c: Add support for built-in functions
8563 vector bool char vec_revb (vector bool char);
8564 vector bool short vec_revb (vector short char);
8565 vector bool int vec_revb (vector bool int);
8566 vector bool long long vec_revb (vector bool long long);
8567 * doc/extend.texi: Update the built-in documentation file for the
8568 new built-in functions.
8570 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8572 * config/s390/s390.md: Remove movcc splitter.
8574 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8576 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8577 load/store on condition.
8579 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8582 * config/avr/avr.c (avr_encode_section_info)
8583 [progmem && !TREE_READONLY]: Error if progmem object needs
8586 2017-07-11 Michael Collison <michael.collison@arm.com>
8588 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8591 2017-07-11 Carl Love <cel@us.ibm.com>
8593 * config/rs6000/rs6000-c.c: Add support for builtins
8594 vector unsigned int vec_parity_lsbb (vector signed int);
8595 vector unsigned int vec_parity_lsbb (vector unsigned int);
8596 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8597 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8598 vector unsigned long long vec_parity_lsbb (vector signed long long);
8599 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8600 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8601 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8602 * doc/extend.texi: Update the built-in documentation file for the
8603 new built-in functions.
8605 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8607 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8608 (layout::m_primary_loc): New field.
8609 (layout::layout): Initialize new field. Move location filtering
8610 logic from here to...
8611 (layout::maybe_add_location_range): ...this new method. Add
8612 support for filtering to just the lines already specified by other
8614 (layout::will_show_line_p): New method.
8615 (gcc_rich_location::add_location_if_nearby): New method.
8616 (selftest::test_add_location_if_nearby): New test function.
8617 (selftest::diagnostic_show_locus_c_tests): Call it.
8618 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8621 2017-07-11 Tom de Vries <tom@codesourcery.com>
8623 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8624 (bb_first_real_insn): New function.
8625 (nvptx_single): Add extra initialization of broadcasted condition
8628 2017-07-11 Nathan Sidwell <nathan@acm.org>
8630 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8632 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8634 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8635 Binutils doc as TEXI will mess them up.
8636 * doc/invoke.texi (AVR Options): Same here.
8638 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8640 * config/sparc/sparc.opt (mfix-ut700): New option.
8641 (mfix-gr712rc): Likewise.
8642 (sparc_fix_b2bst): New variable.
8643 * doc/invoke.texi (SPARC options): Document them.
8644 (ARM options): Fix warnings.
8645 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8646 instructions to prevent sequences that can trigger the store-store
8647 errata for certain LEON3FT processors.
8648 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8649 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8650 * config/sparc/sparc.md (fix_b2bst): New attribute.
8651 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8653 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8656 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8658 (*rsqrtsf2_sse): Ditto.
8660 (div<mode>3): Macroize insn from divdf3 and divsf3
8661 using MODEF mode iterator.
8663 2017-07-10 Martin Sebor <msebor@redhat.com>
8665 PR tree-optimization/80397
8666 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8667 instead of testing for equality to INTEGER_TYPE.
8669 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8671 * config.gcc: Remove uclibc from arc target spec.
8673 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8675 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8677 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8680 * lto-wrapper.c (remove_option): New function.
8681 (merge_and_complain): Merge PIC/PIE options more realistically.
8683 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8685 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8689 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8690 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8691 * config.in: Regenerate.
8692 * configure: Regenerate.
8693 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8694 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8695 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8696 (TARGET_GASISR_PROLOGUES): ...target mask.
8697 * common/config/avr/avr-common.c
8698 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8699 Set -mgas-isr-prologues.
8700 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8701 INSERT_PASS_BEFORE for it.
8702 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8703 * config/avr/avr.c (avr_option_override)
8704 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8705 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8706 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8707 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8708 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8710 (make_avr_pass_pre_proep): New function.
8711 (emit_push_sfr) <treg>: Add argument to function and use it
8713 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8714 and set machine->gasisr.yes.
8715 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8716 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8717 __gcc_isr.n_pushed to .L__stack_usage.
8718 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8719 (avr_asm_final_postscan_insn): ...this new static function.
8720 * config/avr/avr.h (machine_function)
8721 <is_no_gccisr, use_L__stack_usage>: New fields.
8722 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8723 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8724 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8725 (gasisr, *gasisr): New expander and insn.
8726 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8727 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8728 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8730 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8732 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8735 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8737 Move jump-tables out of .text again.
8740 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8741 (ASM_OUTPUT_ADDR_VEC): New function.
8742 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8743 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8744 INSN_ADDRESSes as asm comment.
8745 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8746 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8747 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8748 * config/avr/avr.md (*tablejump): Adjust comment.
8749 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8750 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8752 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8753 (avr_output_addr_vec): New proto.
8754 (avr_log_t) <insn_addresses>: New field.
8756 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8759 * config/i386/i386.c (ix86_function_arg_advance): Set
8760 outgoing_args_on_stack to true if there are outgoing arguments
8762 (ix86_function_arg): Likewise.
8763 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8764 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8765 * config/i386/i386.h (machine_function): Add
8766 outgoing_args_on_stack.
8768 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8770 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8772 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8774 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8776 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8777 (REAL_H): Remove $(MACHMODE_H).
8778 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8780 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8781 $(MACHMODE_H) and double-int.h.
8782 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8784 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8787 2017-07-07 Andrew Pinski <apinski@cavium.com>
8789 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8790 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8792 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8794 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8795 Add warning if GCC was not configured to link against a GLIBC that
8796 exports the hardware capability bits.
8797 (make_resolver_func): Make resolver function private and not a
8798 COMDAT function. Create the name with clone_function_name instead
8799 of make_unique_name.
8802 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8803 correct operand in doing the split.
8805 2017-07-07 Carl Love <cel@us.ibm.com>
8807 * config/rs6000/rs6000-c: Add support for built-in function
8808 vector unsigned short vec_pack_to_short_fp32 (vector float,
8810 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8811 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8812 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8813 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8814 (convert_4f32_8i16): Add define_expand.
8815 * doc/extend.texi: Update the built-in documentation file for the
8816 new built-in function.
8818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8820 * config/sparc/m8.md: New file.
8821 * config/sparc/sparc.md: Include m8.md.
8823 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8825 * config/sparc/sparc.opt: New option -mvis4b.
8826 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8827 (sparc_option_override): Handle VIS4B.
8828 (enum sparc_builtins): Define
8829 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8830 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8831 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8832 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8833 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8834 (check_constant_argument): New function.
8835 (sparc_vis_init_builtins): Define builtins
8836 __builtin_vis_dictunpack{8,16,32},
8837 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8838 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8839 __builtin_vis_fpcmpde{8,16,32}shl and
8840 __builtin_vis_fpcmpur{8,16,32}shl.
8841 (sparc_expand_builtin): Check that the constant operands to
8842 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8843 constant and in range.
8844 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8846 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8847 (SPARC_IMM5_P): Likewise.
8848 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8849 (enabled): Handle vis4b.
8850 (UNSPEC_DICTUNPACK): New unspec.
8851 (UNSPEC_FPCMPSHL): Likewise.
8852 (UNSPEC_FPUCMPSHL): Likewise.
8853 (UNSPEC_FPCMPDESHL): Likewise.
8854 (UNSPEC_FPCMPURSHL): Likewise.
8855 (cpu_feature): New CPU feature `vis4b'.
8856 (dictunpack{8,16,32}): New insns.
8857 (FPCSMODE): New mode iterator.
8858 (fpcscond): New code iterator.
8859 (fpcsucond): Likewise.
8860 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8861 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8862 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8863 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8864 * config/sparc/constraints.md: Define constraints `q' for unsigned
8865 2-bit integer constants and `t' for unsigned 5-bit integer
8867 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8869 (imm5_operand_dictunpack16): Likewise.
8870 (imm5_operand_dictunpack32): Likewise.
8871 (imm2_operand): Likewise.
8872 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8873 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8874 ditunpack* and fpcmp*shl builtins.
8876 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8878 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8879 * config.in: Add HAVE_AS_SPARC6 define.
8880 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8882 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8884 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8885 (CPP_CPU_SPEC): Handle m8.
8886 (ASM_CPU_SPEC): Likewise.
8887 * config/sparc/sparc-opts.h (enum processor_type): Add
8889 * config/sparc/sparc.c (m8_costs): New struct.
8890 (sparc_option_override): Handle TARGET_CPU_m8.
8891 (sparc32_initialize_trampoline): Likewise.
8892 (sparc64_initialize_trampoline): Likewise.
8893 (sparc_issue_rate): Likewise.
8894 (sparc_register_move_cost): Likewise.
8895 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8896 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8897 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8898 (CPP_CPU_SPEC): Handle M8.
8899 (ASM_CPU_SPEC): Likewise.
8900 (AS_M8_FLAG): Define.
8901 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8902 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8903 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8905 * configure: Regenerate.
8906 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8909 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8911 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8913 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8914 ("*movdi_insn_sp32"): Do not set v3pipe.
8915 ("*movsi_insn"): Likewise.
8916 ("*movdi_insn_sp64"): Likewise.
8917 ("*movsf_insn"): Likewise.
8918 ("*movdf_insn_sp32"): Likewise.
8919 ("*movdf_insn_sp64"): Likewise.
8920 ("*zero_extendsidi2_insn_sp64"): Likewise.
8921 ("*sign_extendsidi2_insn"): Likewise.
8922 ("*mov<VM32:mode>_insn"): Likewise.
8923 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8924 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8925 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8926 ("<vlop:code><VL:mode>3"): Likewise.
8927 ("*not_<vlop:code><VL:mode>3"): Likewise.
8928 ("*nand<VL:mode>_vis"): Likewise.
8929 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8930 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8931 ("one_cmpl<VL:mode>2"): Likewise.
8932 ("faligndata<VM64:mode>_vis"): Likewise.
8933 ("alignaddrsi_vis"): Likewise.
8934 ("alignaddrdi_vis"): Likweise.
8935 ("alignaddrlsi_vis"): Likewise.
8936 ("alignaddrldi_vis"): Likewise.
8937 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8938 ("bmaskdi_vis"): Likewise.
8939 ("bmasksi_vis"): Likewise.
8940 ("bshuffle<VM64:mode>_vis"): Likewise.
8941 ("cmask8<P:mode>_vis"): Likewise.
8942 ("cmask16<P:mode>_vis"): Likewise.
8943 ("cmask32<P:mode>_vis"): Likewise.
8944 ("pdistn<P:mode>_vis"): Likewise.
8945 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8947 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8949 * config/sparc/sparc.md ("subtype"): New insn attribute.
8950 ("*wrgsr_sp64"): Set insn subtype.
8951 ("*rdgsr_sp64"): Likewise.
8952 ("alignaddrsi_vis"): Likewise.
8953 ("alignaddrdi_vis"): Likewise.
8954 ("alignaddrlsi_vis"): Likewise.
8955 ("alignaddrldi_vis"): Likewise.
8956 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8957 ("fexpand_vis"): Likewise.
8958 ("fpmerge_vis"): Likewise.
8959 ("faligndata<VM64:mode>_vis"): Likewise.
8960 ("bshuffle<VM64:mode>_vis"): Likewise.
8961 ("cmask8<P:mode>_vis"): Likewise.
8962 ("cmask16<P:mode>_vis"): Likewise.
8963 ("cmask32<P:mode>_vis"): Likewise.
8964 ("fchksm16_vis"): Likewise.
8965 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8966 ("fmean16_vis"): Likewise.
8967 ("fp<plusminus_insn>64_vis"): Likewise.
8968 ("<plusminus_insn>v8qi3"): Likewise.
8969 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8970 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8971 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8972 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8973 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8974 ("*movqi_insn"): Likewise.
8975 ("*movhi_insn"): Likewise.
8976 ("*movsi_insn"): Likewise.
8977 ("movsi_pic_gotdata_op"): Likewise.
8978 ("*movdi_insn_sp32"): Likewise.
8979 ("*movdi_insn_sp64"): Likewise.
8980 ("movdi_pic_gotdata_op"): Likewise.
8981 ("*movsf_insn"): Likewise.
8982 ("*movdf_insn_sp32"): Likewise.
8983 ("*movdf_insn_sp64"): Likewise.
8984 ("*zero_extendhisi2_insn"): Likewise.
8985 ("*zero_extendqihi2_insn"): Likewise.
8986 ("*zero_extendqisi2_insn"): Likewise.
8987 ("*zero_extendqidi2_insn"): Likewise.
8988 ("*zero_extendhidi2_insn"): Likewise.
8989 ("*zero_extendsidi2_insn_sp64"): Likewise.
8990 ("ldfsr"): Likewise.
8991 ("prefetch_64"): Likewise.
8992 ("prefetch_32"): Likewise.
8993 ("tie_ld32"): Likewise.
8994 ("tie_ld64"): Likewise.
8995 ("*tldo_ldub_sp32"): Likewise.
8996 ("*tldo_ldub1_sp32"): Likewise.
8997 ("*tldo_ldub2_sp32"): Likewise.
8998 ("*tldo_ldub_sp64"): Likewise.
8999 ("*tldo_ldub1_sp64"): Likewise.
9000 ("*tldo_ldub2_sp64"): Likewise.
9001 ("*tldo_ldub3_sp64"): Likewise.
9002 ("*tldo_lduh_sp32"): Likewise.
9003 ("*tldo_lduh1_sp32"): Likewise.
9004 ("*tldo_lduh_sp64"): Likewise.
9005 ("*tldo_lduh1_sp64"): Likewise.
9006 ("*tldo_lduh2_sp64"): Likewise.
9007 ("*tldo_lduw_sp32"): Likewise.
9008 ("*tldo_lduw_sp64"): Likewise.
9009 ("*tldo_lduw1_sp64"): Likewise.
9010 ("*tldo_ldx_sp64"): Likewise.
9011 ("*mov<VM32:mode>_insn"): Likewise.
9012 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9013 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9015 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9017 * config/sparc/sparc.md ("type"): New insn type viscmp.
9018 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9020 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9021 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9022 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9023 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9025 ("n7_vis_logical_11cycle"): Likewise.
9026 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9027 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9028 * config/sparc/niagara.md ("niag_vis"): Likewise.
9029 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9030 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9032 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9034 * config/sparc/sparc.md: New instruction type `bmask'.
9035 (bmaskdi_vis): Use the `bmask' type.
9036 (bmasksi_vis): Likewise.
9037 * config/sparc/ultra3.md (us3_array): Likewise.
9038 * config/sparc/niagara7.md (n7_array): Likewise.
9039 * config/sparc/niagara4.md (n4_array): Likewise.
9040 * config/sparc/niagara2.md (niag2_vis): Likewise.
9041 (niag3_vis): Likewise.
9042 * config/sparc/niagara.md (niag_vis): Likewise.
9044 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9046 * ipa-comdats.c: Remove optimize check from gate.
9047 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9048 for functions not optimized.
9049 (ipa_fn_summary_read): Skip optimize check.
9050 (ipa_fn_summary_write): Likewise.
9051 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9053 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9055 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9056 (check_callers): Give up on caller which is not optimized.
9057 (inline_small_functions): Likewise.
9058 (ipa_inline): Do not give up when not optimizing.
9059 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9060 away unoptimizes cdtors.
9061 (whole_program_function_and_variable_visibility): Do
9062 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9063 * ipa.c (process_references): Do not check optimize.
9064 (symbol_table::remove_unreachable_nodes): Update optimize check.
9065 (set_writeonly_bit): Update optimize check.
9066 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9067 (pass_ipa_single_use::gate): Remove.
9069 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9071 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9072 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9073 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9074 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9075 permute_load, permute_store, adjust_extract, adjust_splat,
9076 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9077 replace_swap_with_copy, dump_swap_insn_table,
9078 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9079 recombine_lvx_pattern, recombine_stvx_pattern,
9080 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9081 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9082 to file rs6000-p8swap.c.
9083 * config/rs6000/rs6000-p8swap.c: New file.
9084 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9085 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9086 and rs6000*-*-* targets.
9088 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9090 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9092 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9094 * lto-wrapper.c (merge_and_complain): Do not merge
9095 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9096 fsigned_zeros, ftrapping_math, fwrapv.
9097 (append_compiler_options): Do not track these options.
9098 (append_linker_options): Likewie
9100 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9102 * cgraphunit.c (cgraph_node::finalize_function): When
9103 !flag_toplevel_reorde set no_reorder flag.
9104 (varpool_node::finalize_decl): Likewise.
9105 (symbol_table::compile): Drop no toplevel reorder path.
9107 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9109 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9110 edges; zero probability is not better than uninitialized.
9112 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9114 * asan.h (asan_sanitize_allocas_p): Declare.
9115 * asan.c (asan_sanitize_allocas_p): New function.
9116 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9117 (handle_builtin_alloca): Likewise.
9118 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9119 if !asan_sanitize_allocas_p.
9120 * params.def (asan-instrument-allocas): Add new option.
9121 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9122 * opts.c (common_handle_option): Disable allocas sanitization for
9125 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9127 * asan.c: Include gimple-fold.h.
9128 (get_last_alloca_addr): New function.
9129 (handle_builtin_stackrestore): Likewise.
9130 (handle_builtin_alloca): Likewise.
9131 (asan_emit_allocas_unpoison): Likewise.
9132 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9133 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9134 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9135 (instrument_builtin_call): Pass gimple iterator to
9136 get_mem_refs_of_builtin_call.
9137 (last_alloca_addr): New global.
9138 * asan.h (asan_emit_allocas_unpoison): Declare.
9139 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9140 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9141 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9142 if function calls alloca.
9143 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9144 * gimple-fold.h (replace_call_with_value): Declare.
9145 * internal-fn.c: Include asan.h.
9146 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9147 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9149 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9151 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9152 (C_SELFTEST_FLAGS): New.
9153 (CPP_SELFTEST_FLAGS): New.
9154 (SELFTEST_DEPS): New, from deps of s-selftest.
9155 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9156 (CPP_SELFTEST_DEPS): New.
9157 (selftest): Add dependency on s-selftest-c++.
9158 (s-selftest): Rename to...
9159 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9160 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9161 than SELFTEST_FLAGS.
9162 (selftest-gdb): Rename to...
9163 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9165 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9166 (selftest-valgrind): Rename to...
9167 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9169 (selftest-valgrind): Reintroduce as an alias for
9170 selftest-c-valgrind.
9171 (s-selftest-c++): New.
9172 (selftest-c++-gdb): New.
9173 (selftest-c++-valgrind): New.
9175 2017-07-06 Olivier Hainque <hainque@adacore.com>
9177 * gcc.c (process_command): When deciding if undefined variables
9178 should be ignored when processing specs, accept "gcc -v" as well.
9180 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9182 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9183 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9185 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9187 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9188 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9189 * config/arm/arm-tables.opt: Regenerate.
9190 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9192 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9194 2017-07-06 Carl Love <cel@us.ibm.com>
9196 * ChangeLog: Clean up from mid air collision
9198 2017-07-06 Carl Love <cel@us.ibm.com>
9200 * config/rs6000/rs6000-c.c: Add support for built-in functions
9201 vector signed int vec_subc (vector signed int, vector signed int);
9202 vector signed __int128 vec_subc (vector signed __int128,
9203 vector signed __int128);
9204 vector unsigned __int128 vec_subc (vector unsigned __int128,
9205 vector unsigned __int128);
9206 vector signed int vec_sube (vector signed int, vector signed int,
9208 vector unsigned int vec_sube (vector unsigned int,
9209 vector unsigned int,
9210 vector unsigned int);
9211 vector signed __int128 vec_sube (vector signed __int128,
9212 vector signed __int128,
9213 vector signed__int128);
9214 vector unsigned __int128 vec_sube (vector unsigned __int128,
9215 vector unsigned __int128,
9216 vector unsigned __int128);
9217 vector signed int vec_subec (vector signed int, vector signed int,
9219 vector unsigned int vec_subec (vector unsigned int,
9220 vector unsigned int,
9221 vector unsigned int);
9222 vector signed __int128 vec_subec (vector signed __int128,
9223 vector signed __int128,
9224 vector signed__int128);
9225 vector unsigned __int128 vec_subec (vector unsigned __int128,
9226 vector unsigned __int128,
9227 vector unsigned __int128);
9228 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9229 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9230 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9231 BU_ALTIVEC_OVERLOAD_X definitions.
9232 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9233 * doc/extend.texi: Update the built-in documentation file for the new
9236 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9239 * diagnostic-show-locus.c (layout::layout): Use start and finish
9240 spelling location for the start and finish of each range.
9241 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9242 Add unused aspect param.
9243 * input.c (expand_location_1): Add "aspect" param, and use it
9244 to access the correct part of the location.
9245 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9247 (expand_location_to_spelling_point): Likewise.
9248 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9249 param, and pass it to expand_location_1.
9251 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9253 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9254 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9255 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9256 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9257 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9258 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9259 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9260 _mm_maskz_getmant_ss): New intrinsics.
9261 (__builtin_ia32_getexpss128_mask): Changed to ...
9262 __builtin_ia32_getexpss128_round ... this.
9263 (__builtin_ia32_getexpsd128_mask): Changed to ...
9264 __builtin_ia32_getexpsd128_round ... this.
9265 * config/i386/i386-builtin-types.def
9266 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9267 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9268 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9269 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9270 __builtin_ia32_getmantss_mask_round): New builtins.
9271 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9272 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9273 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9274 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9275 * config/i386/sse.md
9276 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9277 avx512f_sgetexp<mode><mask_scalar_name>
9278 <round_saeonly_scalar_name> ... this.
9279 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9280 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9281 vgetexp<ssescalarmodesuffix>
9282 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9283 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9284 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9285 avx512f_vgetmant<mode><mask_scalar_name>
9286 <round_saeonly_scalar_name> ... this.
9287 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9288 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9289 vgetmant<ssescalarmodesuffix>
9290 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9291 %0<mask_scalar_operand4>, %1, %2
9292 <round_saeonly_scalar_mask_op4>, %3} ... this.
9293 * config/i386/subst.md (mask_scalar_operand4,
9294 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9295 round_saeonly_scalar_nimm_predicate): New subst attributes.
9297 2017-07-06 Julia Koval <julia.koval@intel.com>
9299 * config/i386/i386.c (ix86_erase_embedded_rounding):
9300 Remove code for old rounding pattern.
9302 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9304 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9306 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9308 * doc/sourcebuild.texi (Test Directives, Variants of
9309 dg-require-support): Add documentation for dg-require-stack-check.
9311 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9313 * config/i386/subst.md (mask_scalar, round_scalar,
9314 round_saeonly_scalar): New meta-templates.
9315 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9316 round_scalar_mask_operand3, round_scalar_mask_op3,
9317 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9318 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9319 round_saeonly_scalar_constraint,
9320 round_saeonly_scalar_prefix): New subst attribute.
9321 * config/i386/sse.md
9322 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9323 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9324 <round_scalar_name> ... this.
9325 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9326 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9327 <round_scalar_name> ... this.
9328 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9329 <sse>_vm<code><mode>3<mask_scalar_name>
9330 <round_saeonly_scalar_name> ... this.
9331 (v<plusminus_mnemonic><ssescalarmodesuffix>
9332 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9333 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9334 v<plusminus_mnemonic><ssescalarmodesuffix>
9335 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9336 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9337 (v<multdiv_mnemonic><ssescalarmodesuffix>
9338 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9339 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9340 v<multdiv_mnemonic><ssescalarmodesuffix>
9341 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9342 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9343 (v<maxmin_float><ssescalarmodesuffix>
9344 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9345 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9346 v<maxmin_float><ssescalarmodesuffix>
9347 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9348 %0<mask_scalar_operand3>, %1, %<iptr>2
9349 <round_saeonly_scalar_mask_op3>} ... this.
9351 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9353 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9354 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9356 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9357 Alan Hayward <alan.hayward@arm.com>
9358 David Sherwood <david.sherwood@arm.com>
9360 * combine.c (simplify_if_then_else): Remove "enum" before
9362 * compare-elim.c (can_eliminate_compare): Likewise.
9363 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9365 (aarch64_lookup_simd_builtin_type): Likewise.
9366 (aarch64_simd_builtin_type): Likewise.
9367 (aarch64_init_simd_builtin_types): Likewise.
9368 (aarch64_simd_expand_args): Likewise.
9369 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9371 (aarch64_reverse_mask): Likewise.
9372 (aarch64_simd_emit_reg_reg_move): Likewise.
9373 (aarch64_gen_adjusted_ldpstp): Likewise.
9374 (aarch64_ccmp_mode_to_code): Likewise.
9375 (aarch64_operands_ok_for_ldpstp): Likewise.
9376 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9377 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9379 (aarch64_min_divisions_for_recip_mul): Likewise.
9380 (aarch64_reassociation_width): Likewise.
9381 (aarch64_get_condition_code_1): Likewise.
9382 (aarch64_simd_emit_reg_reg_move): Likewise.
9383 (aarch64_simd_attr_length_rglist): Likewise.
9384 (aarch64_reverse_mask): Likewise.
9385 (aarch64_operands_ok_for_ldpstp): Likewise.
9386 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9387 (aarch64_gen_adjusted_ldpstp): Likewise.
9388 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9390 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9391 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9392 (arm_lookup_simd_builtin_type): Likewise.
9393 (arm_simd_builtin_type): Likewise.
9394 (arm_init_simd_builtin_types): Likewise.
9395 (arm_expand_builtin_args): Likewise.
9396 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9397 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9398 (ft32_setup_incoming_varargs): Likewise.
9399 (ft32_function_arg): Likewise.
9400 (ft32_function_arg_advance): Likewise.
9401 (ft32_pass_by_reference): Likewise.
9402 (ft32_arg_partial_bytes): Likewise.
9403 (ft32_valid_pointer_mode): Likewise.
9404 (ft32_addr_space_pointer_mode): Likewise.
9405 (ft32_addr_space_legitimate_address_p): Likewise.
9406 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9408 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9409 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9410 (iamcu_alignment): Likewise.
9411 (canonicalize_vector_int_perm): Likewise.
9412 (ix86_noce_conversion_profitable_p): Likewise.
9413 (ix86_mpx_bound_mode): Likewise.
9414 (ix86_operands_ok_for_move_multiple): Likewise.
9415 * config/microblaze/microblaze-protos.h
9416 (microblaze_expand_conditional_branch_reg): Likewise.
9417 * config/microblaze/microblaze.c
9418 (microblaze_expand_conditional_branch_reg): Likewise.
9419 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9421 (rs6000_reassociation_width): Likewise.
9422 (rs6000_invalid_binary_op): Likewise.
9423 (fusion_p9_p): Likewise.
9424 (emit_fusion_p9_load): Likewise.
9425 (emit_fusion_p9_store): Likewise.
9426 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9428 (riscv_hard_regno_mode_ok_p): Likewise.
9429 (riscv_address_insns): Likewise.
9430 (riscv_split_symbol): Likewise.
9431 (riscv_legitimize_move): Likewise.
9432 (riscv_function_value): Likewise.
9433 (riscv_hard_regno_nregs): Likewise.
9434 (riscv_expand_builtin): Likewise.
9435 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9436 (riscv_build_integer): Likewise.
9437 (riscv_split_integer): Likewise.
9438 (riscv_legitimate_constant_p): Likewise.
9439 (riscv_cannot_force_const_mem): Likewise.
9440 (riscv_regno_mode_ok_for_base_p): Likewise.
9441 (riscv_valid_base_register_p): Likewise.
9442 (riscv_valid_offset_p): Likewise.
9443 (riscv_valid_lo_sum_p): Likewise.
9444 (riscv_classify_address): Likewise.
9445 (riscv_legitimate_address_p): Likewise.
9446 (riscv_address_insns): Likewise.
9447 (riscv_load_store_insns): Likewise.
9448 (riscv_force_binary): Likewise.
9449 (riscv_split_symbol): Likewise.
9450 (riscv_force_address): Likewise.
9451 (riscv_legitimize_address): Likewise.
9452 (riscv_move_integer): Likewise.
9453 (riscv_legitimize_const_move): Likewise.
9454 (riscv_legitimize_move): Likewise.
9455 (riscv_address_cost): Likewise.
9456 (riscv_subword): Likewise.
9457 (riscv_output_move): Likewise.
9458 (riscv_canonicalize_int_order_test): Likewise.
9459 (riscv_emit_int_order_test): Likewise.
9460 (riscv_function_arg_boundary): Likewise.
9461 (riscv_pass_mode_in_fpr_p): Likewise.
9462 (riscv_pass_fpr_single): Likewise.
9463 (riscv_pass_fpr_pair): Likewise.
9464 (riscv_get_arg_info): Likewise.
9465 (riscv_function_arg): Likewise.
9466 (riscv_function_arg_advance): Likewise.
9467 (riscv_arg_partial_bytes): Likewise.
9468 (riscv_function_value): Likewise.
9469 (riscv_pass_by_reference): Likewise.
9470 (riscv_setup_incoming_varargs): Likewise.
9471 (riscv_print_operand): Likewise.
9472 (riscv_elf_select_rtx_section): Likewise.
9473 (riscv_save_restore_reg): Likewise.
9474 (riscv_for_each_saved_reg): Likewise.
9475 (riscv_register_move_cost): Likewise.
9476 (riscv_hard_regno_mode_ok_p): Likewise.
9477 (riscv_hard_regno_nregs): Likewise.
9478 (riscv_class_max_nregs): Likewise.
9479 (riscv_memory_move_cost): Likewise.
9480 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9481 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9482 (rl78_addr_space_address_mode): Likewise.
9483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9485 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9486 (rs6000_reassociation_width): Likewise.
9487 (rs6000_invalid_binary_op): Likewise.
9488 (fusion_p9_p): Likewise.
9489 (emit_fusion_p9_load): Likewise.
9490 (emit_fusion_p9_store): Likewise.
9491 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9492 (ok_for_simple_move_operands): Likewise.
9493 (ok_for_simple_move_strict_operands): Likewise.
9494 (ok_for_simple_arith_logic_operands): Likewise.
9495 (visium_legitimize_reload_address): Likewise.
9496 (visium_select_cc_mode): Likewise.
9497 (output_cbranch): Likewise.
9498 (visium_split_double_move): Likewise.
9499 (visium_expand_copysign): Likewise.
9500 (visium_expand_int_cstore): Likewise.
9501 (visium_expand_fp_cstore): Likewise.
9502 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9503 (visium_function_arg): Likewise.
9504 (visium_function_arg_advance): Likewise.
9505 (visium_libcall_value): Likewise.
9506 (visium_setup_incoming_varargs): Likewise.
9507 (visium_legitimate_constant_p): Likewise.
9508 (visium_legitimate_address_p): Likewise.
9509 (visium_legitimize_address): Likewise.
9510 (visium_secondary_reload): Likewise.
9511 (visium_register_move_cost): Likewise.
9512 (visium_memory_move_cost): Likewise.
9513 (prepare_move_operands): Likewise.
9514 (ok_for_simple_move_operands): Likewise.
9515 (ok_for_simple_move_strict_operands): Likewise.
9516 (ok_for_simple_arith_logic_operands): Likewise.
9517 (visium_function_value_1): Likewise.
9518 (rtx_ok_for_offset_p): Likewise.
9519 (visium_legitimize_reload_address): Likewise.
9520 (visium_split_double_move): Likewise.
9521 (visium_expand_copysign): Likewise.
9522 (visium_expand_int_cstore): Likewise.
9523 (visium_expand_fp_cstore): Likewise.
9524 (visium_split_cstore): Likewise.
9525 (visium_select_cc_mode): Likewise.
9526 (visium_split_cbranch): Likewise.
9527 (output_cbranch): Likewise.
9528 (visium_print_operand_address): Likewise.
9529 * expmed.c (flip_storage_order): Likewise.
9530 * expmed.h (emit_cstore): Likewise.
9531 (flip_storage_order): Likewise.
9532 * genrecog.c (validate_pattern): Likewise.
9533 * hsa-gen.c (gen_hsa_addr): Likewise.
9534 * internal-fn.c (expand_arith_overflow): Likewise.
9535 * ira-color.c (allocno_copy_cost_saving): Likewise.
9536 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9537 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9538 (process_invariant_for_inheritance): Likewise.
9539 * lra-eliminations.c (move_plus_up): Likewise.
9540 * omp-low.c (lower_oacc_reductions): Likewise.
9541 * simplify-rtx.c (simplify_subreg): Likewise.
9542 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9543 (TARGET_CHKP_BOUND_MODE): Likewise..
9544 * targhooks.c (default_chkp_bound_mode): Likewise.
9545 (default_setup_incoming_vararg_bounds): Likewise.
9546 * targhooks.h (default_chkp_bound_mode): Likewise.
9547 (default_setup_incoming_vararg_bounds): Likewise.
9548 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9549 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9550 (have_whole_vector_shift): Likewise.
9551 * tree-vect-stmts.c (vectorizable_load): Likewise.
9552 * doc/tm.texi: Regenerate.
9554 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9556 Graceful degrade if Binutils PR21472 is not available.
9559 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9560 .rodata in flash test fails.
9561 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9562 * confgure: Regenerate.
9563 * config.in: Regenerate.
9564 * config/avr/avr.c (avr_asm_named_section)
9565 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9566 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9567 (avr_asm_init_sections): Same.
9569 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9571 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9572 (fma<VH:mode>4_intrinsic): Likewise.
9573 (*fmsub<VCVTF:mode>4): Likewise.
9574 (*fmsub<VH:mode>4_intrinsic): Likewise.
9576 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9579 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9580 Don't depend on "optimize > 0".
9581 (out_movhi_r_mr, out_movqi_mr_r): Same.
9582 (out_movhi_mr_r, out_movqi_r_mr): Same.
9583 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9584 io_address_operand on "optimize > 0".
9586 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9588 * tree-loop-distribution.c: Add general explanantion on the pass.
9589 (generate_loops_for_partition): Mark distributed loop.
9590 (pg_add_dependence_edges): New parameter. Handle alias data
9591 dependence specially and record it in the parameter if asked.
9592 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9593 (init_partition_graph_vertices, add_partition_graph_edge): New.
9594 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9595 (free_partition_graph_vdata, build_partition_graph): New.
9596 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9597 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9598 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9599 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9600 (version_for_distribution_p, finalize_partitions): New.
9601 (distribute_loop): Handle alias data dependence specially. Factor
9602 out loop fusion code as functions and call these functions.
9604 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9606 * tree-loop-distribution.c (classify_partition): New parameter and
9607 better handle reduction statement.
9608 (rdg_build_partitions): Revise comment.
9609 (distribute_loop): Compute statements in all partitions and pass it
9610 to classify_partition.
9612 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9614 * tree-loop-distribution.c (enum partition_type): New.
9615 (struct partition): New field type.
9616 (partition_merge_into): Add parameter. Update partition type.
9617 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9618 (build_rdg_partition_for_vertex): Compute partition type.
9619 (rdg_build_partitions): Dump partition type.
9620 (distribute_loop): Update calls to partition_merge_into.
9622 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9624 * tree-loop-distribution.c (struct ddr_hasher): New.
9625 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9627 (classify_partition): Call get_data_dependence.
9628 (pg_add_dependence_edges): Ditto.
9629 (distribute_loop): Release data dependence hash table.
9631 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9633 * tree-loop-distribution.c (ref_base_address): Delete.
9634 (similar_memory_accesses): Rename ...
9635 (share_memory_accesses): ... to this. Check if partitions access
9636 the same memory reference.
9637 (distribute_loop): Call share_memory_accesses.
9639 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9641 * tree-loop-distribution.c (struct partition): New field recording
9643 (partition_alloc, partition_free): Init and release data refs.
9644 (partition_merge_into): Merge data refs.
9645 (build_rdg_partition_for_vertex): Collect data refs for partition.
9646 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9648 (distribute_loop): Remve data refs from vertice data of partition
9651 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9653 * tree-loop-distribution.c (params.h): Include header file.
9654 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9655 (datarefs_vec): New global var.
9656 (create_rdg_vertices): Use datarefs_vec directly.
9657 (free_rdg): Don't free data references.
9658 (build_rdg): Update use. Don't free data references.
9659 (distribute_loop): Compute global variable for data references.
9660 Bail out if there are too many data references.
9662 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9664 * tree-loop-distribution.c (loop_nest): New global var.
9665 (build_rdg): Use loop directly, rather than loop nest.
9666 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9668 (distribute_loop): Compute global variable loop nest. Update use.
9670 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9672 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9673 (partition_merge_into): New parameter. Dump reason for fusion.
9674 (distribute_loop): Update use of partition_merge_into.
9676 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9678 * tree-loop-distribution.c (bb_top_order_index): New.
9679 (bb_top_order_index_size, bb_top_order_cmp): New.
9680 (stmts_from_loop): Use topological order.
9681 (pass_loop_distribution::execute): Compute and release topological
9682 order for basic blocks.
9684 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9686 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9689 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9691 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9692 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9693 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9694 * internal-fn.def (LOOP_DIST_ALIAS): New.
9695 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9696 (fold_loop_internal_call): ... this.
9697 (vect_loop_dist_alias_call): New function.
9698 (set_uid_loop_bbs): Call fold_loop_internal_call.
9699 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9702 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9705 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9706 Require dead FLAGS_REG at the beginning of a peephole.
9708 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9711 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9712 arguments in the call to __builtin_ia32_sbb_u32.
9713 (_subborrow_u64): Swap _X and _Y arguments in the call to
9714 __builtin_ia32_sbb_u64.
9716 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9719 * tree-vrp.c (compare_assert_loc): Turn into a function template
9720 with stable template parameter. Only test if a->e is NULL,
9721 !a->e == !b->e has been verified already. Use e == NULL or
9722 e != NULL instead of e or ! e tests. If stable is true, don't use
9723 iterative_hash_expr, on the other side allow a or b or both NULL
9724 and sort the NULLs last.
9725 (process_assert_insertions): Sort using compare_assert_loc<false>
9726 instead of compare_assert_loc, later sort using
9727 compare_assert_loc<true> before calling process_assert_insertions_for
9728 in a loop. Use break instead of continue once seen NULL pointer.
9730 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9732 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9733 Cortex-R7 and Cortex-R8 processors.
9735 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9737 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9738 uninitialized while src is not.
9740 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9742 * common/config/arm/arm-common.c: Adjust include path for
9744 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9745 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9746 (arm-cpu-data.h): Likewise.
9747 (arm-cpu-cdata.h): Likewise.
9748 * config/arm/arm-cpu.h: Delete.
9749 * config/arm/arm-cpu-cdata.h: Delete.
9750 * config/arm/arm-cpu-data.h: Delete.
9752 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9754 * config/arm/arm-cpus.in (cortex-a55): New.
9755 (cortex-a75): Likewise.
9756 (cortex-a75.cortex-a55): Likewise.
9757 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9759 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9760 * config/arm/arm-cpu-cdata.h: Regenerate.
9761 * config/arm/arm-cpu-data.h: Regenerate.
9762 * config/arm/arm-cpu.h: Regenerate.
9763 * config/arm/arm-tables.opt: Regenerate.
9764 * config/arm/arm-tune.md: Regenerate.
9766 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9768 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9770 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9772 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9775 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9777 PR tree-optimization/81292
9778 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9779 full_string_p, also call adjust_related_strinfos if the adjustment
9780 is simple, otherwise invalidate related strinfos.
9782 2017-07-04 Martin Liska <mliska@suse.cz>
9785 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9786 newly created variable as DECL_IGNORED_P.
9788 2017-07-04 Martin Liska <mliska@suse.cz>
9791 * ipa-inline.c (inline_small_functions):
9792 Use xstrdup_for_dump.
9794 2017-07-04 Tom de Vries <tom@codesourcery.com>
9796 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9798 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9801 * config/darwin.c (darwin_function_switched_text_sections):
9804 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9806 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9808 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9810 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9812 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9814 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9815 min_profitable_iters, and th as inclusive lower bounds.
9816 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9817 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9818 for min_profitable_iters and min_profitable_estimate.
9819 (vect_transform_loop): Treat th as an inclusive lower bound.
9820 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9822 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9825 * config/darwin.c (darwin_function_switched_text_sections):
9826 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9827 in two pieces, and suppress the use of buf.
9829 2017-07-03 Nathan Sidwell <nathan@acm.org>
9831 * hash-table.h (hash_table_mod1): Fix indentation.
9833 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9836 * predict.c (force_edge_cold): Be more careful about propagation
9838 * profile-count.h (profile_probability::guessed,
9839 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9841 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9843 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9845 * doc/invoke.texi (rcpc architecture extension): Document it.
9847 2017-07-03 Richard Biener <rguenther@suse.de>
9849 PR tree-optimization/60510
9850 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9851 the scalar reduction PHI and use it.
9852 (vectorizable_reduction): Properly guard the single_defuse_cycle
9853 path for non-SLP reduction chains where we cannot use it.
9854 Rework reduc_def/index and vector type deduction. Rework
9855 vector operand gathering during reduction op code-gen.
9856 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9857 chains dissolve the chain and leave it to non-SLP reduction
9860 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9862 * tree-data-ref.h (dr_alignment): Declare.
9863 * tree-data-ref.c (dr_alignment): New function.
9864 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9865 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9867 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9869 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9871 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9872 and base_misalignment fields.
9873 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9874 * tree-data-ref.c: Include builtins.h.
9875 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9876 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9877 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9878 * tree-vect-data-refs.c: Include tree-cfg.h.
9879 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9880 fields instead of calculating an alignment here.
9881 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9882 innermost_loop_behavior fields.
9884 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9886 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9888 (DR_STEP_ALIGNMENT): New macro.
9889 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9890 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9891 (create_data_ref): Print it.
9892 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9893 to tell whether the step preserves vector (mis)alignment.
9894 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9895 Move the check for an integer step and generalise to all INTEGER_CST.
9896 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9897 Print the outer step alignment.
9899 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9901 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9902 with offset_alignment.
9903 (DR_ALIGNED_TO): Delete.
9904 (DR_OFFSET_ALIGNMENT): New macro.
9905 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9906 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9907 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9908 (create_data_ref): Likewise.
9909 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9910 (vect_analyze_data_refs): Likewise.
9911 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9912 creating dummy innermost behavior.
9914 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9916 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9917 with a "innermost_loop_behavior *" and refeence tree.
9918 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9919 (create_data_ref): Update call accordingly.
9920 * tree-predcom.c (find_looparound_phi): Likewise.
9922 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9924 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9925 fields with dr_wrt_vec_loop.
9926 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9927 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9928 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9929 (vect_dr_behavior): New function.
9930 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9931 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9932 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9933 track whether the step preserves the misalignment.
9934 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9935 Use vect_dr_behavior.
9936 (vect_setup_realignment): Update call accordingly.
9937 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9938 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9939 call to vect_create_addr_base_for_vector_ref.
9940 (vect_create_cond_for_align_checks): Likewise.
9941 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9942 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9943 (vect_recog_mask_conversion_pattern): Likewise.
9944 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9945 (new_stmt_vec_info): Remove redundant zeroing.
9947 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9949 * common/config/arm/arm-common.c (arm_be8_option): New function.
9950 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9951 (ISA_ARMv6): Add isa_bit_be8.
9952 * config/arm/arm.h (arm_be8_option): Add prototype.
9953 (BE8_SPEC_FUNCTION): New define.
9954 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9955 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9956 (mlittle-endian): Similarly.
9957 (mbe8, mbe32): New options.
9958 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9959 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9963 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9967 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9968 (cleanup_tree_cfg_bb): Use it.
9969 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9971 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9973 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9976 * loop-doloop.c (add_test): Update profile.
9978 2017-07-03 Martin Liska <mliska@suse.cz>
9981 * sanopt.c (rewrite_usage_of_param): New function.
9982 (sanitize_rewrite_addressable_params): Likewise.
9983 (pass_sanopt::execute): Call rewrite_usage_of_param.
9985 2017-07-03 Richard Biener <rguenther@suse.de>
9987 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9988 back to using VIEW_CONVERT_EXPR.
9990 2017-07-03 Martin Liska <mliska@suse.cz>
9993 * doc/extend.texi: Document when a resolver function is
9994 generated for target_clones.
9996 2017-07-03 Martin Liska <mliska@suse.cz>
9998 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9999 and shadow memory of auto variables which are subject of
10000 use-after-scope sanitization.
10001 (asan_expand_mark_ifn): Add do set only when is_poison.
10003 2016-07-03 Richard Biener <rguenther@suse.de>
10005 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10007 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10008 (vectorizable_reduction): Perform reduction PHI creation when
10009 visiting a reduction PHI and adjust and simplify code generation
10010 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10011 (vect_transform_loop): Visit reduction PHIs.
10012 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10013 defs into the SLP tree.
10014 (vect_build_slp_tree): Reduction defs terminate the recursion.
10015 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10017 (vect_get_vec_defs_for_stmt_copy): Export.
10018 (vect_get_vec_defs): Likewise.
10019 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10021 (vect_get_vec_defs_for_stmt_copy): Declare.
10022 (vect_get_vec_defs): Likewise.
10024 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10026 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10027 parameter with a "loop" parameter and use it instead of the
10028 loop containing DR_STMT. Don't check simple_iv when doing
10029 BB analysis. Describe the two analysis modes in the comment.
10031 2017-07-03 Tom de Vries <tom@codesourcery.com>
10033 PR tree-optimization/69468
10034 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10035 (find_same_succ_bb): Handle ignore_edge_flags.
10037 2017-07-03 Tom de Vries <tom@codesourcery.com>
10039 PR tree-optimization/81192
10040 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10042 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10044 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10046 2017-07-03 Tom de Vries <tom@codesourcery.com>
10048 PR tree-optimization/81192
10049 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10050 BB_SAME_SUCC (bb) == NULL.
10052 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10054 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10057 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10059 * dumpfile.c: Include profile-count.h
10060 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10062 (insert_cond_bb): Update profile.
10063 * tree-cfg.h (insert_cond_bb): Update prototype.
10064 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10065 * tree-dump.c: Do not include tree-cfg.
10067 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10069 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10071 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10073 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10076 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10078 * tree-complex.c (expand_complex_div_wide): update profile.
10080 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10081 Alan Hayward <alan.hayward@arm.com>
10082 David Sherwood <david.sherwood@arm.com>
10084 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10085 (CORETYPES_H): New define.
10086 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10087 (insn-modes-inline.h, s-modes-inline-h): New rules.
10088 (generated_files): Add insn-modes-inline.h.
10089 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10090 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10091 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10092 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10093 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10094 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10095 (build/gencodes.o, build/genconditions.o): Likewise.
10096 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10097 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10098 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10099 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10100 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10101 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10102 * coretypes.h: Include everything up to real.h for generators.
10103 Include insn-modes.h first. Include wide-int-print.h after
10104 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10105 * machmode.h: Don't include insn-modes.h here.
10106 * function-tests.c: Remove includes of signop.h, machmode.h,
10107 double-int.h and wide-int.h.
10109 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10111 * optc-save-gen.awk: Likewise.
10112 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10113 * godump.c: Remove include of wide-int-print.h.
10114 * pretty-print.h: Likewise.
10115 * wide-int-print.cc: Likewise.
10116 * wide-int.cc: Likewise.
10117 * hash-map-tests.c: Remove include of signop.h.
10118 * hash-set-tests.c: Likewise.
10119 * rtl-tests.c: Likewise.
10120 * mkconfig.sh: Remove include of machmode.h.
10121 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10123 (emit_insn_modes_inline_h): ...this new function. Emit the code
10124 into an insn-modes-inline.h header file, adding appropriate
10125 include guards and end comments.
10126 (emit_insn_modes_c_header): Remove include of machmode.h.
10127 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10129 (main): Handle -i flag and call emit_insn_modes_inline_h when
10132 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10134 * tree-ssa-strlen.c (strinfo): Rename the length field to
10135 nonzero_chars. Add a full_string_p field.
10136 (compare_nonzero_chars, zero_length_string_p): New functions.
10137 (get_addr_stridx): Add an offset_out parameter.
10138 Use compare_nonzero_chars.
10139 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10140 (new_strinfo): Update after above changes to strinfo.
10141 (set_endptr_and_length): Set full_string_p.
10142 (get_string_length): Update after above changes to strinfo.
10143 (unshare_strinfo): Update call to new_strinfo.
10144 (maybe_invalidate): Likewise.
10145 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10146 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10147 as a uhwi instead of an shwi. Update after above changes to
10148 strinfo and new_strinfo.
10149 (zero_length_string): Assert that chainsi contains full strings.
10150 Use zero_length_string_p. Update call to new_strinfo.
10151 (adjust_related_strinfos): Update after above changes to strinfo.
10152 Copy full_string_p from origsi.
10153 (adjust_last_stmt): Use zero_length_string_p.
10154 (handle_builtin_strlen): Update after above changes to strinfo and
10155 new_strinfo. Install the lhs as the string length if the previous
10156 entry didn't describe a full string.
10157 (handle_builtin_strchr): Update after above changes to strinfo
10159 (handle_builtin_strcpy): Likewise.
10160 (handle_builtin_strcat): Likewise.
10161 (handle_builtin_malloc): Likewise.
10162 (handle_pointer_plus): Likewise.
10163 (handle_builtin_memcpy): Likewise. Track nonzero characters
10164 that aren't necessarily followed by a nul terminator.
10165 (handle_char_store): Likewise.
10167 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10169 PR tree-optimization/80769
10170 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10171 for malloc and calloc. Document the new invariant that all related
10172 strinfos have delayed lengths or none do.
10173 (verify_related_strinfos): Move earlier in file.
10174 (set_endptr_and_length): New function, split out from...
10175 (get_string_length): ...here. Also set the lengths of related
10177 (zero_length_string): Assert that chainsi has known (rather than
10179 (adjust_related_strinfos): Likewise.
10181 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10183 PR tree-optimization/81136
10184 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10185 assert that two references with the same misalignment have the same
10186 compile-time misalignment if those compile-time misalignments
10189 2017-07-01 Andi Kleen <ak@linux.intel.com>
10191 * print-tree.c (print_node): Print all attributes.
10193 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10195 * cfg.c (scale_bbs_frequencies): New function.
10196 * cfg.h (scale_bbs_frequencies): Declare it.
10197 * cfgloopanal.c (single_likely_exit): Cleanup.
10198 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10200 (scale_loop_profile): Likewise.
10201 (loop_version): Likewise.
10202 (create_empty_loop_on_edge): Update.
10203 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10204 scale_loop_frequencies, scale_loop_profile, loopify,
10205 loop_version): Update prototypes.
10206 * modulo-sched.c (sms_schedule): Update.
10207 * predict.c (unlikely_executed_edge_p): Also check probability.
10208 (probably_never_executed_edge_p): Fix typo.
10209 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10210 * tree-parloops.c (gen_parallel_loop): Update.
10211 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10212 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10213 * tree-ssa-loop-split.c (split_loop): Update.
10214 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10215 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10216 (vect_loop_versioning): Update.
10217 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10219 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10221 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10223 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10225 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10226 to keep profile consistent.
10228 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10230 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10231 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10232 * profile-count.h (max_safe_multiplier): Make unsigned.
10233 (profile_count::guessed_zero): New.
10235 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10237 * bb-reorder.c (fix_up_crossing_landing_pad,
10238 fix_crossing_conditional_branches): Use make_single_succ_edge
10239 to keep profile consistent.
10241 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10243 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10246 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10249 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10250 the right scopes, make sure cond_jump isn't preserved between multiple
10251 iterations. Search for fallthru edge whenever there are 3+ edges and
10252 use find_fallthru_edge for it.
10254 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10256 Patch by Alexander Monakov <amonakov@ispras.ru>
10257 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10258 probabilities consistently.
10260 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10262 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10263 emit_cmp_and_jump_insns.
10265 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10268 * tree-inline.c (expand_call_inline): Combine profile statuses.
10270 2017-06-30 Andrew Pinski <apinski@cavium.com>
10272 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10273 fold_stmt returned true.
10275 2017-06-30 Nathan Sidwell <nathan@acm.org>
10277 * ggc.h (empty_string): Delete.
10278 * cfgexpand.c (expand_asm_stmt): Use plain "".
10279 * optabs.c (expand_asm_memory_barrier): Likewise.
10280 * stringpool.c (empty_string): Delete.
10281 (digit_vector, digit_string): Delete.
10282 (ggc_alloc_string): Use plain "", don't optimize single digit
10283 strings. Use ggc_alloc_atomic.
10285 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10287 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10288 comparison set and one other set, use the cost of the non-comparison
10291 2017-06-30 Nathan Sidwell <nathan@acm.org>
10293 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10296 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10298 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10299 loops. Remove now unneeded calls to gimple_switch_set_label() that
10300 just set removed labels to NULL_TREE.
10302 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10304 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10305 (set_range_info): ...here. Only call set_range_info_raw if domain
10307 (set_nonzero_bits): Call set_range_info_raw.
10308 * tree-ssanames.h (set_range_info_raw): New.
10310 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10313 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10314 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10315 of nonimmediate_operand and <store_mask_constraint> instead of m
10316 for the input operand. For V8FI iterator, always split if input
10317 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10318 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10319 <store_mask_predicate> instead of register_operand and
10320 <store_mask_constraint> instead of v for the input operand. Make
10321 sure both operands aren't MEMs for if not <mask_applied>.
10323 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10325 * lto-wrapper.c (copy_file) Close both file descriptors before
10328 2017-06-30 Martin Liska <mliska@suse.cz>
10331 * multiple_target.c (create_dispatcher_calls): Make ifunc
10332 also for function that don't have calls or are not referenced.
10334 2017-06-30 Richard Biener <rguenther@suse.de>
10336 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10337 analyze the first scalar stmt. Move vector type computation
10338 for the BB case here from ...
10339 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10340 live operation processing in the SLP case properly.
10342 2017-06-30 Richard Biener <rguenther@suse.de>
10344 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10346 2017-06-30 Martin Liska <mliska@suse.cz>
10349 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10350 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10352 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10354 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10357 2017-06-30 Richard Biener <rguenther@suse.de>
10359 PR tree-optimization/81249
10360 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10361 condition reduction result to original scalar type.
10363 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10365 * profile-count.h (enum profile_quality): Fix typos and whitespace
10368 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10370 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10371 type for branch probabilities.
10373 2017-06-29 Julian Brown <julian@codesourcery.com>
10374 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10376 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10377 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10378 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10379 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10381 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10383 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10384 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10385 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10386 CC usage from generic code to here.
10387 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10388 CC usage into the target macros.
10390 2017-06-29 Maya Rashish <coypu@sdf.org>
10392 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10395 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10397 * arm/arm-builtins.c: Include profile-count.h
10398 * except.c (sjlj_emit_function_enter): Use
10399 profile_probability::unlikely.
10401 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10403 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10404 and tocrel_offset be pointer args rather than implicitly using
10406 (legitimate_constant_pool_address_p, rs6000_emit_move,
10407 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10408 tocrel_offset and use in toc_relative_expr_p call.
10409 (print_operand, print_operand_address): Use static tocrel_base_oac
10410 and tocrel_offset_oac.
10411 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10414 2017-06-29 Maya Rashish <coypu@sdf.org>
10416 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10418 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10420 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10421 objects, take into account only the alignment of 'op0' and 'mode1' if
10424 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10426 * ccmp.c (ccmp_tree_comparison_p): New function.
10427 (ccmp_candidate_p): Update to use above function.
10428 (get_compare_parts): New function.
10429 (expand_ccmp_next): Update to use new functions.
10430 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10432 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10433 take mode as argument.
10434 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10435 * expr.c (expand_expr_real_1): Pass mode as argument.
10437 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10439 * combine.c (combine_instructions): Print insns to dump_file, together
10442 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10444 * asan.c (asan_emit_stack_protection): Update.
10445 (create_cond_insert_point): Update.
10446 * auto-profile.c (afdo_propagate_circuit): Update.
10447 * basic-block.h (struct edge_def): Turn probability to
10448 profile_probability.
10449 (EDGE_FREQUENCY): Update.
10450 * bb-reorder.c (find_traces_1_round): Update.
10451 (better_edge_p): Update.
10452 (sanitize_hot_paths): Update.
10453 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10454 (make_single_succ_edge): Update.
10455 (check_bb_profile): Update.
10456 (dump_edge_info): Update.
10457 (update_bb_profile_for_threading): Update.
10458 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10459 probabilitycount to 0.
10460 * cfgbuild.c (compute_outgoing_frequencies): Update.
10461 * cfgcleanup.c (try_forward_edges): Update.
10462 (outgoing_edges_match): Update.
10463 (try_crossjump_to_edge): Update.
10464 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10465 (expand_gimple_tailcall): Update.
10466 (construct_init_block): Use make_single_succ_edge.
10467 (construct_exit_block): Use make_single_succ_edge.
10468 * cfghooks.c (verify_flow_info): Update.
10469 (redirect_edge_succ_nodup): Update.
10470 (split_edge): Update.
10471 (account_profile_record): Update.
10472 * cfgloopanal.c (single_likely_exit): Update.
10473 * cfgloopmanip.c (scale_loop_profile): Update.
10474 (set_zero_probability): Remove.
10475 (duplicate_loop_to_header_edge): Update.
10476 * cfgloopmanip.h (loop_version): Update prototype.
10477 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10478 (force_nonfallthru_and_redirect): Update.
10479 (update_br_prob_note): Update.
10480 (rtl_verify_edges): Update.
10481 (purge_dead_edges): Update.
10482 (rtl_lv_add_condition_to_bb): Update.
10483 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10484 * cgraphunit.c (init_lowered_empty_function): Update.
10485 (cgraph_node::expand_thunk): Update.
10486 * cilk-common.c: Include profile-count.h
10487 * dojump.c (inv): Remove.
10488 (jumpifnot): Update.
10489 (jumpifnot_1): Update.
10490 (do_jump_1): Update.
10492 (do_jump_by_parts_greater_rtx): Update.
10493 (do_compare_rtx_and_jump): Update.
10494 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10495 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10496 * dwarf2cfi.c: Include profile-count.h
10497 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10498 (sjlj_emit_dispatch_table): Likewise.
10499 * explow.c: Include profile-count.h
10500 * expmed.c (emit_store_flag_force): Update.
10501 (do_cmp_and_jump): Update.
10502 * expr.c (compare_by_pieces_d::generate): Update.
10503 (compare_by_pieces_d::finish_mode): Update.
10504 (emit_block_move_via_loop): Update.
10505 (store_expr_with_bounds): Update.
10506 (store_constructor): Update.
10507 (expand_expr_real_2): Update.
10508 (expand_expr_real_1): Update.
10509 * expr.h (try_casesi, try_tablejump): Update prototypes.
10510 * gimple-pretty-print.c (dump_probability): Update.
10511 (dump_profile): New.
10512 (dump_gimple_label): Update.
10513 (dump_gimple_bb_header): Update.
10514 * graph.c (draw_cfg_node_succ_edges): Update.
10515 * hsa-gen.c (convert_switch_statements): Update.
10516 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10517 (find_if_case_1): Update.
10518 (find_if_case_2): Update.
10519 * internal-fn.c (expand_arith_overflow_result_store): Update.
10520 (expand_addsub_overflow): Update.
10521 (expand_neg_overflow): Update.
10522 (expand_mul_overflow): Update.
10523 (expand_vector_ubsan_overflow): Update.
10524 * ipa-cp.c (good_cloning_opportunity_p): Update.
10525 * ipa-split.c (split_function): Use make_single_succ_edge.
10526 * ipa-utils.c (ipa_merge_profiles): Update.
10527 * loop-doloop.c (add_test): Update.
10528 (doloop_modify): Update.
10529 * loop-unroll.c (compare_and_jump_seq): Update.
10530 (unroll_loop_runtime_iterations): Update.
10531 * lra-constraints.c (lra_inheritance): Update.
10532 * lto-streamer-in.c (input_cfg): Update.
10533 * lto-streamer-out.c (output_cfg): Update.
10534 * mcf.c (adjust_cfg_counts): Update.
10535 * modulo-sched.c (sms_schedule): Update.
10536 * omp-expand.c (expand_omp_for_init_counts): Update.
10537 (extract_omp_for_update_vars): Update.
10538 (expand_omp_ordered_sink): Update.
10539 (expand_omp_for_ordered_loops): Update.
10540 (expand_omp_for_generic): Update.
10541 (expand_omp_for_static_nochunk): Update.
10542 (expand_omp_for_static_chunk): Update.
10543 (expand_cilk_for): Update.
10544 (expand_omp_simd): Update.
10545 (expand_omp_taskloop_for_outer): Update.
10546 (expand_omp_taskloop_for_inner): Update.
10547 * omp-simd-clone.c (simd_clone_adjust): Update.
10548 * optabs.c (expand_doubleword_shift): Update.
10549 (expand_abs): Update.
10550 (emit_cmp_and_jump_insn_1): Update.
10551 (expand_compare_and_swap_loop): Update.
10552 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10553 * predict.c (predictable_edge_p): Update.
10554 (edge_probability_reliable_p): Update.
10555 (set_even_probabilities): Update.
10556 (combine_predictions_for_insn): Update.
10557 (combine_predictions_for_bb): Update.
10558 (propagate_freq): Update.
10559 (estimate_bb_frequencies): Update.
10560 (force_edge_cold): Update.
10561 * profile-count.c (profile_count::dump): Add missing space into dump.
10562 (profile_count::debug): Add newline.
10563 (profile_count::differs_from_p): Explicitly convert to unsigned.
10564 (profile_count::stream_in): Update.
10565 (profile_probability::dump): New member function.
10566 (profile_probability::debug): New member function.
10567 (profile_probability::differs_from_p): New member function.
10568 (profile_probability::differs_lot_from_p): New member function.
10569 (profile_probability::stream_in): New member function.
10570 (profile_probability::stream_out): New member function.
10571 * profile-count.h (profile_count_quality): Rename to ...
10572 (profile_quality): ... this one.
10573 (profile_probability): New.
10574 (profile_count): Update.
10575 * profile.c (compute_branch_probabilities): Update.
10576 * recog.c (peep2_attempt): Update.
10577 * sched-ebb.c (schedule_ebbs): Update.
10578 * sched-rgn.c (find_single_block_region): Update.
10579 (compute_dom_prob_ps): Update.
10580 (schedule_region): Update.
10581 * sel-sched-ir.c (compute_succs_info): Update.
10582 * stmt.c (struct case_node): Update.
10583 (do_jump_if_equal): Update.
10584 (get_outgoing_edge_probs): Update.
10585 (conditional_probability): Update.
10586 (emit_case_dispatch_table): Update.
10587 (expand_case): Update.
10588 (expand_sjlj_dispatch_table): Update.
10589 (emit_case_nodes): Update.
10590 * targhooks.c: Update.
10591 * tracer.c (better_p): Update.
10592 (find_best_successor): Update.
10593 * trans-mem.c (expand_transaction): Update.
10594 * tree-call-cdce.c: Update.
10595 * tree-cfg.c (gimple_split_edge): Upate.
10596 (move_sese_region_to_fn): Upate.
10597 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10598 * tree-eh.c (lower_resx): Upate.
10599 (cleanup_empty_eh_move_lp): Upate.
10600 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10601 * tree-inline.c (copy_edges_for_bb): Update.
10602 (copy_cfg_body): Update.
10603 * tree-parloops.c (gen_parallel_loop): Update.
10604 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10605 (gimple_gen_time_profiler): Update.
10606 * tree-ssa-dce.c (remove_dead_stmt): Update.
10607 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10608 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10609 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10610 (unloop_loops): Update.
10611 (try_peel_loop): Update.
10612 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10613 * tree-ssa-loop-split.c (connect_loops): Update.
10614 (split_loop): Update.
10615 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10616 (hoist_guard): Update.
10617 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10618 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10619 (value_replacement): Update.
10620 * tree-ssa-reassoc.c (branch_fixup): Update.
10621 * tree-ssa-tail-merge.c (replace_block_by): Update.
10622 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10623 (create_edge_and_update_destination_phis): Update.
10624 (compute_path_counts): Update.
10625 (recompute_probabilities): Update.
10626 (update_joiner_offpath_counts): Update.
10627 (freqs_to_counts_path): Update.
10628 (duplicate_thread_path): Update.
10629 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10630 (struct switch_conv_info): Update.
10631 (gen_inbound_check): Update.
10632 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10633 (vect_do_peeling): Update.
10634 (vect_loop_versioning): Update.
10635 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10636 (optimize_mask_stores): Update.
10637 * ubsan.c (ubsan_expand_null_ifn): Update.
10638 * value-prof.c (gimple_divmod_fixed_value): Update.
10639 (gimple_divmod_fixed_value_transform): Update.
10640 (gimple_mod_pow2): Update.
10641 (gimple_mod_pow2_value_transform): Update.
10642 (gimple_mod_subtract): Update.
10643 (gimple_mod_subtract_transform): Update.
10644 (gimple_ic): Update.
10645 (gimple_stringop_fixed_value): Update.
10646 (gimple_stringops_transform): Update.
10647 * value-prof.h: Update.
10649 2017-06-29 Carl Love <cel@us.ibm.com>
10651 * config/rs6000/rs6000-c.c: Add support for built-in functions
10652 vector signed int vec_signed (vector float);
10653 vector signed long long vec_signed (vector double);
10654 vector signed int vec_signed2 (vector double, vector double);
10655 vector signed int vec_signede (vector double);
10656 vector signed int vec_signedo (vector double);
10657 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10658 instruction generator.
10659 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10660 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10661 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10663 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10664 vunsignede_v2df): Add define_expands.
10665 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10666 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10667 VEC_UNSIGNEDO): Add definitions.
10668 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10669 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10670 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10671 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10672 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10673 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10674 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10675 vec_unsignede, vec_unsignedo): Add builtin defines.
10676 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10678 * doc/extend.texi: Update the built-in documentation file for the
10679 new built-in functions.
10681 2017-06-29 Richard Biener <rguenther@suse.de>
10683 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10684 reduction chains to LOOP_VINFO_REDUCTIONS.
10685 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10686 SLP reductions after processing reduction chains.
10688 2017-06-29 Nathan Sidwell <nathan@acm.org>
10690 * builtins.c (fold_builtin_FUNCTION): Use
10691 lang_hooks.decl_printable_name.
10693 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10695 PR middle-end/81194
10696 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10697 with only one label.
10698 * stmt.c (expand_case): Assert NCASES is greater than one.
10700 2017-06-29 Richard Biener <rguenther@suse.de>
10702 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10704 (group_case_labels): Likewise.
10705 (find_taken_edge): Push sanity checking on val to workers...
10706 (find_taken_edge_cond_expr): ... here
10707 (find_taken_edge_switch_expr): ... and here, handle cases
10708 with just a default label.
10709 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10710 (group_case_labels): Likewise.
10711 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10712 group_case_labels does anything cleanup the CFG again.
10714 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10716 PR tree-optimization/81196
10717 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10718 exit condition comparing two IVs.
10720 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10722 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10723 profile to the dummy entry at the end of the list of architectures.
10724 * config/arm/arm-cpu-cdata.h: Regenerated.
10726 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10727 Michael Collison <michael.collison@arm.com>
10730 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10732 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10733 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10734 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10735 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10736 of shift when the shift amount is masked with constant equal to
10737 the size of the mode.
10738 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10741 2017-06-29 Martin Liska <mliska@suse.cz>
10743 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10745 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10748 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10751 (edge_growth_cache_entry::edge_growth_cache_entry): New
10753 (reset_edge_growth_cache): Update to use constructor.
10755 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10757 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10758 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10759 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10761 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10763 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10764 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10766 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10768 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10769 (*-linux-uclibc*): Add t-uclibc tmake_file.
10770 * config/t-musl: New.
10771 * config/t-uclibc: New.
10773 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10775 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10777 (gen_comm_data): Emit architectural setting of arch_prof.
10778 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10780 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10781 (armv8-m.base, armv8-m.main): Likewise.
10782 * arm-protos.h (arm_build_target): Add profile field.
10783 (arch_option): Likewise.
10784 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10786 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10787 arm_active_target.profile.
10789 2017-06-28 Richard Biener <rguenther@suse.de>
10791 PR middle-end/81227
10792 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10793 TYPE_OVERFLOW_WRAPS.
10794 * match.pd (negate_expr_p): Likewise.
10795 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10796 fold_build2, not fold_binary.
10798 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10800 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10801 Convert memory address to Pmode.
10802 (aarch64_print_operand): Assert MEM operands are always Pmode.
10804 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10807 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10808 Remove redundant if.
10809 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10810 * config/arm/aarch-common-protos.h
10811 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10812 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10814 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10817 * multiple_target.c (create_dispatcher_calls): Set the default
10818 clone to be static, not public.
10820 2017-06-28 Richard Biener <rguenther@suse.de>
10822 * tree-vect-loop.c (vectorizable_reduction): Move special
10823 cond reduction IV var creation ...
10824 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10825 parameter. Use STMT_VINFO_VECTYPE.
10826 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10829 2017-06-28 Martin Liska <mliska@suse.cz>
10832 * ipa-visibility.c (non_local_p): Handle visibility.
10834 2017-06-28 Martin Liska <mliska@suse.cz>
10837 * common.opt: Add IntegerRange to various options.
10838 * opt-functions.awk (integer_range_info): New function.
10839 * optc-gen.awk: Add integer_range_info to cl_options struct.
10840 * opts-common.c (decode_cmdline_option): Handle
10841 CL_ERR_INT_RANGE_ARG.
10842 (cmdline_handle_error): Likewise.
10843 * opts.c (print_filtered_help): Show valid interval in
10844 when --help is provided.
10845 * opts.h (struct cl_option): Add range_min and range_max fields.
10846 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10848 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10850 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10851 (x * C EQ/NE y * C): New transformation.
10853 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10855 * genmultilib (combination_space): Accept '+' in option names.
10857 2017-06-28 Martin Liska <mliska@suse.cz>
10860 * asan.c (instrument_derefs): Bail out inner references
10861 that are hard register variables.
10863 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10866 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10867 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10869 2017-06-28 Richard Biener <rguenther@suse.de>
10871 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10872 (vect_get_slp_defs): Adjust.
10873 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10875 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10877 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10878 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10879 (vectorizable_reduction): Adjust.
10880 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10882 (vect_get_slp_defs): Likewise.
10883 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10884 (vectorizable_bswap): Adjust.
10885 (vectorizable_call): Likewise.
10886 (vectorizable_conversion): Likewise.
10887 (vectorizable_assignment): Likewise.
10888 (vectorizable_shift): Likewise.
10889 (vectorizable_operation): Likewise.
10890 (vectorizable_store): Likewise.
10891 (vectorizable_condition): Likewise.
10892 (vectorizable_comparison): Likewise.
10894 2017-06-28 Michael Collison <michael.collison@arm.com>
10897 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10899 (arm_gen_movmemqi): Removed unused variable 'i'.
10900 Convert 'for' loop into 'while' loop.
10901 (arm_expand_prologue): Remove last unnecessary set of insn.
10902 (thumb_pop): Remove unused variable 'pushed_words'.
10903 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10905 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10907 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10908 * config/s390/s390-protos.h: Add prototype of
10909 s390_rel_address_ok_p.
10910 * config/s390/s390.c (s390_got_symbol): New function.
10911 (s390_rel_address_ok_p): New function.
10912 (legitimize_pic_address): Use s390_rel_address_ok_p.
10913 (s390_load_got): Use s390_got_symbol.
10914 (s390_option_override): Issue error if
10915 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10916 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10918 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10920 2017-06-27 Andrew Pinski <apinski@cavium.com>
10922 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10923 (X * copysign (1.0, X)): New pattern.
10924 (X * copysign (1.0, -X)): New pattern.
10925 (copysign (-1.0, CST)): New pattern.
10927 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10929 * genmultilib (combination_space): Remove variable.
10930 Validate reuse rules against regular expression for any sequence
10931 of multilib options in any order.
10933 2017-06-27 Michael Collison <michael.collison@arm.com>
10935 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10936 call aarch64_split_simd_combine.
10937 * (aarch64_combine_internal<mode>): Delete pattern.
10938 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10939 Allow register and subreg operands.
10941 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10943 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10944 specific need, just fallback on defaults.
10945 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10947 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10948 Olivier Hainque <hainque@adacore.com>
10950 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10952 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10953 targets. Pick a default if no particular attempt applied.
10954 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10957 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10959 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10960 (x86_64-wrs-vxworks7): Likewise.
10962 2017-06-27 Marek Polacek <polacek@redhat.com>
10965 * ubsan.c (instrument_null): Check get_base_address's result for null.
10967 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10969 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10971 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10973 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10974 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10975 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10976 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10977 New function types.
10978 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10979 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10980 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10981 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10982 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10983 BUILT_IN_FEUPDATEENV): New builtins.
10984 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10985 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10986 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10987 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10989 (builtin_structptr_types): Adjust size.
10990 * tree.c (builtin_structptr_types): Add four entries.
10992 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10993 Olivier Hainque <hainque@adacore.com>
10995 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10996 (TLS_SYM): New local macro, forcing reference to __tls__ on
10997 link command lines for VxWorks 7 RTPs, triggering initialization
10999 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11000 OS features TLS support, true for RTPs on VxWorks 7.
11001 * config/vxworks.c (vxworks_override_options): Setup emutls
11004 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11006 * predict.c (test_prediction_value_range): Use -1U instead of -1
11007 to avoid narrowing conversion warning.
11008 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11009 to avoid narrowing conversion warning.
11010 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11012 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11014 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11016 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11017 64bit configurations.
11018 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11019 (SIZE_TYPE): Likewise.
11020 * config/vxworks.c (vxworks_emutls_var_fields): Use
11021 long_unsigned_type_node instead of unsigned_type_node as the offset
11022 field type, which is "pointer" mode in emutls.c.
11024 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11027 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11029 PR middle-end/81207
11030 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11031 gimple_vuse copying separately from gimple_vdef copying.
11033 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11035 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11038 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11039 Olivier Hainque <hainque@adacore.com>
11041 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11042 for all vxworks7 targets.
11043 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11044 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11045 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11046 variations for VX6/VX7 and 32/64bits later on in ...
11047 (VXWORKS_LIB_SPEC): Leverage new macros.
11048 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11049 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11051 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11053 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11054 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11056 2017-06-26 Carl Love <cel@us.ibm.com>
11058 * config/rs6000/rs6000-c.c: Add support for built-in functions
11059 vector bool char vec_reve (vector bool char);
11060 vector signed char vec_reve (vector signed char);
11061 vector unsigned char vec_reve (vector unsigned char);
11062 vector bool int vec_reve (vector bool int);
11063 vector signed int vec_reve (vector signed int);
11064 vector unsigned int vec_reve (vector unsigned int);
11065 vector bool long long vec_reve (vector bool long long);
11066 vector signed long long vec_reve (vector signed long long);
11067 vector unsigned long long vec_reve (vector unsigned long long);
11068 vector bool short vec_reve (vector bool short);
11069 vector signed short vec_reve (vector signed short);
11070 vector double vec_reve (vector double);
11071 vector float vec_reve (vector float);
11072 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11073 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11074 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11075 (altivec_vreve): New pattern.
11076 * config/rs6000/altivec.h (vec_reve): New define.
11077 * doc/extend.texi (vec_rev): Update the built-in documentation file
11078 for the new built-in functions.
11080 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11082 PR tree-optimization/71815
11083 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11085 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11087 (slsr_process_phi): Likewise.
11088 (replace_uncond_cands_and_profitable_phis): Don't replace a
11089 multiply candidate with a stride of 1 (copy or cast).
11090 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11092 (lowest_cost_path): Likewise.
11093 (total_savings): Likewise.
11095 2017-06-26 Richard Biener <rguenther@suse.de>
11098 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11099 Use def_builtin_pure for all gather builtins.
11101 2017-06-26 Richard Biener <rguenther@suse.de>
11103 PR tree-optimization/81203
11104 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11105 non-dominating BBs.
11107 2017-06-26 Marek Polacek <polacek@redhat.com>
11110 * doc/invoke.texi: Document -Wmultistatement-macros.
11112 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11114 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11115 arm_neon_ok_no_float_abi effective target.
11117 2017-06-26 Richard Biener <rguenther@suse.de>
11119 PR tree-optimization/80928
11120 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11121 (copy_bbs): Set BB_DUPLICATED flag early.
11122 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11124 (execute_on_shrinking_pred): Likewise.
11125 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11126 BB_DUPLICATED blocks.
11127 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11128 iterate over all PHIs considering removal of *gsi.
11130 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11132 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11135 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11137 * config/rs6000/rs6000-string.c: (expand_block_clear,
11138 do_load_for_compare, select_block_compare_mode,
11139 compute_current_alignment, expand_block_compare,
11140 expand_strncmp_align_check, expand_strn_compare,
11141 expand_block_move, rs6000_output_load_multiple)
11142 Move functions related to string/block move/compare
11143 to a separate file.
11144 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11145 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11146 for this function which is now used in two files.
11147 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11148 * config.gcc: Add rs6000-string.o to extra_objs for
11149 targets powerpc*-*-* and rs6000*-*-*.
11151 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11154 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11155 32-bit, since indexed is not valid for DImode.
11156 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11157 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11158 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11159 (define_peephole2 for Altivec d-form store): Likewise.
11162 * multiple_target.c (create_dispatcher_calls): Only create the
11163 dispatcher call if the function is the default clone of a
11164 versioned function.
11166 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11168 PR middle-end/80902
11169 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11170 a call, force the call to not be a tail call.
11172 2017-06-23 Jeff Law <law@redhat.com>
11174 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11177 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11179 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11180 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11181 handling for early expansion of vector shifts (sl,sr,sra,rl).
11182 (builtin_function_type): Add vector shift right instructions
11183 to the unsigned argument list.
11185 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11187 rtl-optimizatoin/79286
11188 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11189 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11190 trap. PIC register plus a const unspec without offset can never trap.
11192 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11194 * tree.h (builtin_structptr_type): New type.
11195 (builtin_structptr_types): Declare new array.
11196 * tree.c (builtin_structptr_types): New array.
11197 (free_lang_data, build_common_tree_nodes): Use it.
11199 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11202 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11205 2017-06-22 Matt Turner <mattst88@gmail.com>
11207 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11208 Lake models to skylake case. Assume skylake for unknown
11209 models with clflushopt.
11211 2017-06-22 Jeff Law <law@redhat.com>
11213 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11214 frame sizes that do not satisfy aarch64_uimm12_shift.
11216 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11218 * profile-count.h (apply_probability,
11219 apply_scale, probability_in): Fix checks for zero.
11221 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11223 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11224 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11226 2016-06-22 Richard Biener <rguenther@suse.de>
11228 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11229 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11230 REDUC_MAX_EXPR support.
11231 (vectorizable_reduction): Likewise.
11232 (vect_create_epilog_for_reduction): Likewise.
11234 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11236 * match.pd (A / (1 << B) -> A >> B): New.
11237 * generic-match-head.c: Include optabs-tree.h.
11238 * gimple-match-head.c: Likewise.
11239 * optabs-tree.h (target_supports_op_p): New.
11240 * optabs-tree.c (target_supports_op_p): New.
11242 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11244 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11245 $gcc_cv_ld --help output.
11246 (gcc_cv_ld_demangle): Likewise.
11247 (gcc_cv_ld_eh_frame_hdr): Likewise.
11248 (gcc_cv_ld_pie): Likewise.
11249 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11250 (gcc_cv_ld_buildid): Likewise.
11251 (gcc_cv_ld_sysroot): Likewise.
11252 (ld_bndplt_support): Likewise.
11253 (ld_pushpopstate_support): Likewise.
11254 * configure: Regenerate.
11255 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11257 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11260 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11261 operands indexes to avoid gap between operands and match_dups.
11263 2017-06-21 Andrew Pinski <apinski@cavium.com>
11265 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11266 Increment Arith_shift and Arith_shift_reg by 1.
11267 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11269 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11270 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11271 (aarch64_strip_extend): Add new argument and test for it.
11272 (aarch64_cheap_mult_shift_p): New function.
11273 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11274 add a cost if it is true.
11275 Update calls to aarch64_strip_extend.
11276 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11278 2017-06-21 Andrew Pinski <apinski@cavium.com>
11280 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11282 (thunderxt88): Likewise.
11283 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11284 (thunderx_prefetch_tune): New variable.
11285 (thunderx2t99_prefetch_tune): Update for the correct values.
11286 (thunderxt88_tunings): New variable.
11287 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11288 generic_prefetch_tune.
11289 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11291 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11293 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11294 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11295 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11296 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11297 (aarch64_atomic_cas<mode>, GPI): Likewise.
11299 2017-06-21 Martin Liska <mliska@suse.cz>
11301 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11302 statements on cold and hot labels.
11303 * predict.c (tree_estimate_probability_bb): Remove the
11304 prediction from this place.
11306 2017-06-21 Martin Liska <mliska@suse.cz>
11308 PR tree-optimization/79489
11309 * gimplify.c (maybe_add_early_return_predict_stmt): New
11311 (gimplify_return_expr): Call the function.
11312 * predict.c (tree_estimate_probability_bb): Remove handling
11314 * predict.def: Update comment about early return predictor.
11315 * gimple-predict.h (is_gimple_predict): New function.
11316 * predict.def: Change default value of early return to 66.
11317 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11319 * passes.def: Put pass_strip_predict_hints to the beginning of
11322 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11324 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11325 FUNCTION_DECL declarations.
11326 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11328 (dwaf2out_decl): Likewise.
11329 * godump.c (go_early_global_decl): Skip call to the real debug hook
11330 for FUNCTION_DECL declarations.
11331 * passes.c (rest_of_decl_compilation): Skip call to the
11332 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11333 -fdump-go-spec is passed.
11335 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11337 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11338 Reorder for compactness.
11339 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11340 (def_builtin_pure, def_builtin_pure2): New functions.
11341 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11343 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11345 * match.pd (nop_convert): New predicate.
11346 ((A +- CST1) +- CST2): Allow some NOP conversions.
11348 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11351 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11352 with ctors/dtors if GOVD_SHARED is set.
11354 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11356 * config/aarch64/aarch64.md (movti_aarch64):
11357 Emit mov rather than orr.
11358 (movtf_aarch64): Likewise.
11359 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11360 Emit mov rather than orr.
11362 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11364 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11365 Swap alternatives, make integer dup more expensive.
11367 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11369 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11370 Return true for non-tls symbols.
11372 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11374 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11375 (cortex-a75): Likewise.
11376 (cortex-a75.cortex-a55): Likewise.
11377 * config/aarch64/aarch64-tune.md: Regenerate.
11378 * doc/invoke.texi (-mtune): Document new values for -mtune.
11380 2017-06-21 Tom de Vries <tom@codesourcery.com>
11382 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11383 stack_size feature.
11384 (Effective-Target Keywords, Other attributes): Suggest using
11385 dg-add-options stack_size feature to get stack limit in stack_size
11386 effective target documentation.
11388 2017-06-21 Julian Brown <julian@codesourcery.com>
11389 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11391 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11392 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11393 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11395 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11396 attribute type list for neon_multiply.
11397 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11398 attribute type list for neon_multiply.
11399 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11400 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11401 attribute type list for neon_multiply.
11402 * config/arm/types.md (crypto_pmull): Add.
11403 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11404 attribute type list.
11406 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11408 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11411 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11413 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11414 to make sure not to dereference a NULL cost_classes_ptr pointer.
11416 2017-06-20 Carl Love <cel@us.ibm.com>
11418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11419 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11420 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11421 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11422 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11423 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11424 VMULOSW): New enum "unspec" values.
11425 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11426 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11427 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11428 altivec_vmulosw): New patterns.
11429 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11430 VMULOSW): Add definitions.
11432 2017-06-20 Julia Koval <julia.koval@intel.com>
11434 * config/i386/i386.c: Fix rounding expand for new pattern.
11435 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11437 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11439 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11440 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11442 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11444 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11447 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11449 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11450 architecture, then by implementer ID.
11451 * config/aarch64/aarch64-tune.md: Regenerate.
11453 2017-06-20 Richard Biener <rguenther@suse.de>
11455 PR middle-end/81097
11456 * fold-const.c (split_tree): Fold to type before negating.
11458 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11460 * diagnostic-show-locus.c
11461 (selftest::test_fixit_deletion_affecting_newline): New function.
11462 (selftest::diagnostic_show_locus_c_tests): Call it.
11464 2017-06-20 Andreas Schwab <schwab@suse.de>
11467 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11470 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11472 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11473 __ARM_FEATURE_COPROC according to support.
11475 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11477 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11478 Rewritten to avoid overflow for > 32-bit pointers.
11481 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11482 by removing enum keyword.
11483 (ubsan_type_descriptor): Likewise. Formatting fix.
11486 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11487 splitter): Require TARGET_SSE2 in the condition.
11489 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11492 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11493 for doing vector set of SFmode on ISA 3.0.
11494 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11495 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11497 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11498 SFmode value into a V4SF variable that was extracted from another
11499 V4SF variable without converting the element to double precision
11500 and back to single precision vector format.
11501 (vsx_insert_extract_v4sf_p9_2): Likewise.
11503 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11505 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11506 in UWHI to avoid undefined overflow.
11509 * ubsan.h (enum ubsan_encode_value_phase): New.
11510 (ubsan_encode_value): Change second argument to
11511 enum ubsan_encode_value_phase with default value of
11512 UBSAN_ENCODE_VALUE_GENERIC.
11513 * ubsan.c (ubsan_encode_value): Change second argument to
11514 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11515 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11516 create_tmp_var_raw instead of create_tmp_var and use a
11518 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11519 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11520 ubsan_encode_value callers.
11523 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11524 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11525 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11527 2017-06-19 Richard Biener <rguenther@suse.de>
11529 PR middle-end/81118
11530 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11531 estimates if we changed anything.
11533 2017-06-19 Richard Biener <rguenther@suse.de>
11535 PR tree-optimization/80887
11536 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11537 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11538 simplified lookups, then reset mprts_hook.
11539 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11541 (try_to_simplify): Likewise.
11543 2017-06-19 Martin Liska <mliska@suse.cz>
11546 * gimplify.c (gimplify_switch_expr):
11547 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11549 2017-06-19 Martin Liska <mliska@suse.cz>
11551 * doc/install.texi: Document that PGO runs in 4 stages.
11553 2017-06-19 Martin Liska <mliska@suse.cz>
11556 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11557 to dispatcher function name.
11558 * multiple_target.c (replace_function_decl): New function.
11559 (create_dispatcher_calls): Redirect both edges and references.
11561 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11563 * profile-count.c (profile_count::dump): Dump quality.
11564 (profile_count::differs_from_p): Update for unsigned val.
11565 * profile-count.h (profile_count_quality): New enum.
11566 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11568 2017-06-19 Richard Biener <rguenther@suse.de>
11570 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11571 struct function as arg.
11572 (estimate_numbers_of_iterations): Export overload with loop arg.
11573 (free_numbers_of_iterations_estimates_loop): Use an overload of
11574 free_numbers_of_iterations_estimates instead.
11575 * tree-cfg.c (remove_bb): Adjust.
11576 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11577 * tree-parloops.c (gen_parallel_loop): Likewise.
11578 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11580 (tree_unroll_loops_completely): Likewise.
11581 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11582 Use an overload instead and export.
11583 (estimated_loop_iterations): Adjust.
11584 (max_loop_iterations): Likewise.
11585 (likely_max_loop_iterations): Likewise.
11586 (estimate_numbers_of_iterations): Take struct function as arg
11588 (loop_exits_before_overflow): Adjust.
11589 (free_numbers_of_iterations_estimates_loop): Use an overload.
11590 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11591 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11593 2017-06-19 Richard Biener <rguenther@suse.de>
11596 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11597 RANGE_EXPR conservatively.
11599 2017-06-16 Carl Love <cel@us.ibm.com>
11601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11602 definitions for vec_float, vec_float2, vec_floato,
11603 vec_floate built-ins.
11604 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11605 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11607 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11608 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11609 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11610 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11611 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11612 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11613 vec_floato): Add builtin defines.
11614 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11615 Update the built-in documentation file for the new built-in
11618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11620 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11621 (mthumb): Mark as the negative of -marm.
11623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11625 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11627 (ARM Options, -mtune): Document that this accepts the same
11628 extension options as -mcpu.
11629 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11631 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11633 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11634 permitted extensions.
11636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11638 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11639 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11640 (armv8-m.main): Add option +nodsp.
11641 * config/arm/arm-cpu-cdata.h: Regenerated.
11643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11645 * config/arm/t-fuchsia: New file.
11646 * config.gcc (arm*-*-fuchsia*): Use it.
11648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11650 * config/arm/t-symbian: Rewrite for new option infrastructure.
11652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11654 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11655 (MULTILIB_REQUIRED): Likewise.
11657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11659 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11660 (MULTILIB_RESUE): Likewise.
11661 (MULTILIB_MATCHES): Likewise.
11662 (MULTLIB_REQUIRED): Likewise.
11664 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11666 * config/arm/t-rtems: Rewrite for new option framework.
11668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11670 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11671 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11672 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11673 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11674 * config/arm/t-multilib: ... here.
11675 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11676 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11677 armv7-a and armv8*-a when A-profile libraries have not been built.
11678 * config/arm/t-rmprofile: Rewrite.
11680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11682 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11683 with a backslash. Remove the backslash after substituting unescaped
11685 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11689 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11690 the driver to pass through the default setting of -mfloat-abi.
11691 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11693 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11694 (all_feat_combs): New rule.
11695 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11697 * config/arm/t-aprofile: Rewrite.
11699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11701 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11702 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11703 fpu is not specified by the user/command-line.
11704 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11705 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11706 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11707 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11708 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11709 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11713 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11714 * config/arm/t-arm-elf: Rewritten.
11716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11718 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11719 have some floating-point instructions.
11720 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11721 (TARGET_MAYBE_HARD_FLOAT): New macro.
11722 * config/arm/arm-builtins.c (arm_init_builtins): Use
11723 TARGET_MAYBE_HARD_FLOAT.
11724 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11728 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11729 (configargs.h): Include it.
11730 (arm_print_hint_for_fpu_option): New function.
11731 (arm_parse_fpu_option): New function.
11732 (candidate_extension): New class.
11733 (arm_canon_for_multilib): New function.
11734 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11735 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11736 (ARCH_CANONICAL_SPECS): New macro.
11737 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11741 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11742 are set after handling multilib fragments. Set target_cpu_default2
11745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11747 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11749 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11750 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11752 (check_arch): Likewise.
11753 * config/arm/arm.c (arm_configure_build_target): Handle
11754 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11755 options in the default.
11757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11759 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11760 when an option is an alias of another.
11761 * config/arm/parsecpu.awk (optalias): New parser token.
11762 (gen_comm_data): Mark non-alias options as such. Emit entries
11763 for extension aliases.
11764 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11765 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11766 (armv6kz, armv6zk, armv6t2): Likewise.
11767 (armv7): Make vfpv3-d16 an alias.
11768 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11770 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11771 Sort in canonical order.
11772 (armv8-a): Sort in canonical order.
11773 (armv8.1-a, armv8.2-a): Likewise.
11774 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11776 (cortex-a9): Sort in canonical order.
11777 * config/arm/arm.c (selftests.h): Include it.
11778 (arm_test_cpu_arch_data): New function.
11779 (arm_run_self_tests): New function.
11780 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11781 (targetm): Move declaration to the end of the file.
11782 * arm-cpu-cdata.h: Regenerated.
11784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11786 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11787 call to target_mode_check describing the type of option passed.
11788 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11789 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11790 arm_parse_cpu_option_name to match parameters against list of
11792 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11793 arm_arch_core_flags data structure.
11794 * config/arm/arm-cpu_cdata.h: Regenerated.
11796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11798 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11800 (arm_print_hint_for_cpu_option): Likewise.
11801 (arm_print_hint_for_arch_option): Likewise.
11802 (arm_parse_cpu_option_name): Likewise.
11803 (arm_parse_arch_option_name): Likewise.
11804 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11805 of entries in the all_fpus list.
11806 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11807 (arm_parse_cpu_option_name): Declare.
11808 (arm_parse_arch_option_name): Declare.
11809 (arm_parse_option_features): Declare.
11810 (arm_intialize_isa): Declare.
11811 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11813 (gen_comm_data): ... here. Make definitions non-static.
11814 * config/arm/arm-cpu-data.h: Regenerated.
11815 * config/arm/arm-cpu-cdata.h: Regenerated.
11817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11819 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11820 (cpu_arch_extension): New structure.
11821 (cpu_arch_option, arch_option, cpu_option): New structures.
11822 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11823 architecture types.
11824 (gen_data): Generate new format data tables.
11825 * config/arm/arm.c (cpu_tune): New structure.
11826 (cpu_option, processors): Delete.
11827 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11828 (arm_print_hint_for_cpu_option): ... this and ...
11829 (arm_print_hint_for_arch_option): ... this.
11830 (arm_parse_arch_cpu_name): Delete. Replace with ...
11831 (arm_parse_cpu_option_name): ... this and ...
11832 (arm_parse_arch_option_name): ... this.
11833 (arm_unrecognized_feature): Change type of target parameter to
11835 (arm_parse_arch_cpu_features): Delete. Replace with ...
11836 (arm_parse_option_features): ... this.
11837 (arm_configure_build_target): Rework to use new configuration data
11839 (arm_print_tune_info): Rework for new configuration data tables.
11840 * config/arm/arm-cpu-data.h: Regenerated.
11841 * config/arm/arm-cpu.h: Regenerated.
11843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11845 * Makefile.in (OBJS): Move sbitmap.o from here ...
11846 (OBJS-libcommon): ... to here.
11848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11850 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11851 (ISA_ALL_CRYPTO): New macro.
11852 (ISA_ALL_SIMD): New macro
11853 (ISA_ALL_FP): New macro.
11854 * config/arm/arm.c (fpu_bitlist): Update initializer.
11855 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11857 (arm9e): Add fpu. Add option for nofp
11858 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11859 (arm926ej-s, arm1026ej-s): Likewise.
11860 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11861 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11862 neon-fp16, neon-vfpv4, nofp and nosimd.
11863 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11864 (cortex-a8): Add fpu. Add option for nofp.
11865 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11866 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11867 (cortex-r4f): Add fpu.
11868 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11869 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11871 (cortex-r8): Likewise.
11872 (cortex-m4): Add fpu. Add option for nofp.
11873 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11874 (cortex-a17.cortex-a7): Likewise.
11875 (cortex-a32): Add fpu. Add options for crypto and nofp.
11876 (cortex-a35, cortex-a53): Likewise.
11877 (cortex-a57): Add fpu. Add option for crypto.
11878 (cortex-a72, cortex-a73): Likewise.
11879 (exynos-m1): Likewise.
11880 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11881 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11882 (cortex-m33): Add fpu. Add option for nofp.
11883 * config/arm/arm-cpu-cdata.h: Regenerated
11884 * config/arm/arm-cpu-data.h: Regenerated.
11886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11888 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11889 (armv5te, armv5tej): Likewise.
11890 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11891 (armv7): Add options fp and vfpv3-d16.
11892 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11893 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11895 (armv7ve): Likewise.
11896 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11897 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11898 (armv8-a): Add nocrypto option.
11899 (armv8.1-a, armv8.2-a): Likewise.
11900 (armv8-m.main): add options fp, fp.dp and nofp.
11902 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11904 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11906 (armv8-a+crc): Delete.
11907 (armv8.1-a): Add options simd, crypto and nofp.
11908 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11909 (armv8.2-a+fp16): Delete.
11910 (armv8-m.main): Add option dsp.
11911 (armv8-m.main+dsp): Delete.
11912 (cortex-a8): Add fpu. Add nofp option.
11913 (cortex-a9): Add fpu. Add nofp and nosimd options.
11914 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11915 link to main cpu and architecture data structures.
11916 (gen_comm_data): Only put isa attributes from the main architecture
11918 (option): New statement for architecture and CPU entries.
11919 * arm.c (struct cpu_option): New structure.
11920 (struct processors): Add entry for options.
11921 (arm_unrecognized_feature): New function.
11922 (arm_parse_arch_cpu_name): Ignore any characters after the first
11924 (arm_parse_arch_cpu_feature): New function.
11925 (arm_configure_build_target): Separate out any CPU and architecture
11926 features and parse separately. Don't error out if -mfpu=auto is
11927 used with only an architecture string.
11928 (arm_print_asm_arch_directives): New function.
11929 (arm_file_start): Call it.
11930 * config/arm/arm-cpu-cdata.h: Regenerated.
11931 * config/arm/arm-cpu-data.h: Likewise.
11932 * config/arm/arm-tables.opt: Likewise.
11934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11936 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11937 assembler when it is not -mfpu=auto.
11939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11941 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11942 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11943 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11944 (ASM_CPU_SPEC): Rewrite.
11945 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11946 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11947 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11948 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11949 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11950 copied string is NUL-terminated. Also strip any characters prefixed
11952 (arm_rewrite_selected_arch): New function.
11953 (arm_rewrite_march): New function.
11955 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11957 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11958 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11959 (march, mcpu, mtune): Convert to string-based options.
11960 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11961 (arm_parse_arch_cpu_name): New function.
11962 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11963 identify selected architecture or CPU.
11964 (arm_option_save): New function.
11965 (TARGET_OPTION_SAVE): Redefine.
11966 (arm_option_restore): Restore string options.
11967 (arm_option_print): Print string options.
11969 2017-06-16 Martin Sebor <msebor@redhat.com>
11971 PR tree-optimization/80933
11972 PR tree-optimization/80934
11973 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11974 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11975 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11976 (gimple_fold_builtin): Call them.
11978 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11980 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11981 as unlikely; update profile.
11983 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11985 * predict.c (force_edge_cold): Handle declaring edges impossible
11988 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11990 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11992 (try_unroll_loop_completely): Fix reporting.
11994 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11996 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11998 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12001 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12002 if given a non-constant argument for an intrinsic which requires a
12005 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12007 * profile.c (compare_freqs): New function.
12008 (branch_prob): Sort edge list.
12009 (find_spanning_tree): Assume that the list is priority sorted.
12011 2017-06-16 Richard Biener <rguenther@suse.de>
12013 PR tree-optimization/81090
12014 * passes.def (pass_record_bounds): Remove.
12015 * tree-pass.h (make_pass_record_bounds): Likewise.
12016 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12017 make_pass_record_bounds): Likewise.
12018 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12019 not free niter estimates at the beginning but at the end.
12020 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12022 2017-06-16 Richard Biener <rguenther@suse.de>
12024 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12025 initializer to workaround ICE in host GCC 4.8.
12027 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12029 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12031 (clone_inlined_nodes): Update.
12033 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12035 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12036 prefetch settings, and enable prefetching by default at -O3.
12038 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12040 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12041 Set flag_prefetch_loop_arrays according to tuning data.
12043 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12045 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12046 New tune structure.
12047 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12048 [Unrelated to main purpose of the patch] Place the pointer field last
12049 to enable type checking errors when tune structure are wrongly merged.
12050 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12051 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12052 (thunderx2t99_prefetch_tune): New tune constants.
12053 (tune_params *_tunings): Update all tunings (no functional change).
12054 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12055 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12056 from tunings structures.
12058 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12061 * ubsan.c (instrument_null): Add T argument, use it instead
12062 of computing it based on IS_LHS.
12063 (instrument_object_size): Likewise.
12064 (pass_ubsan::execute): Adjust instrument_null and
12065 instrument_object_size callers to pass gimple_get_lhs or
12066 gimple_assign_rhs1 result to it. Use instrument_null instead of
12067 calling get_base_address and instrument_mem_ref. Handle
12068 aggregate call arguments for object-size sanitization.
12070 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12072 PR tree-optimization/81089
12073 * tree-vrp.c (is_masked_range_test): Validate operands of
12076 2017-06-15 Martin Sebor <msebor@redhat.com>
12079 * dumpfile.c (dump_register): Avoid calling memset to initialize
12080 a class with a default ctor.
12081 * gcc.c (struct compiler): Remove const qualification.
12082 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12083 * hash-table.h: Ditto.
12084 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12086 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12087 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12089 * params.h (struct param_info): Make struct members non-const.
12090 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12091 with default initialization.
12092 * vec.h (vec_copy_construct, vec_default_construct): New helper
12094 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12095 with vec_copy_construct.
12096 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12097 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12098 * doc/invoke.texi (-Wclass-memaccess): Document.
12100 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12102 * emit-rtl.h (is_leaf): Update comment about local
12103 register allocator.
12105 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12108 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12109 for a variable to have a section before checking if the section has a
12111 Set section to.persistent if persistent attribute is set.
12112 Warn if .persistent attribute is used on an automatic variable.
12114 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12116 PR rtl-optimization/80474
12117 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12119 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12121 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12124 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12126 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12127 (casesi): Emit bounds checking as RTL.
12128 (casesi_internal_mips16_<mode>): Remove bounds checking.
12130 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12132 * config/xtensa/xtensa.c (xtensa_option_override): Append
12133 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12134 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12135 xtensa_doloop_hooks): Define unconditionally.
12136 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12138 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12139 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12140 for it in xtensa_option_override.
12141 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12142 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12144 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12146 * doc/cppopts.texi: Document '-' special value to -MF.
12148 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12150 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12151 (cortex_a53_fconst): Likewise.
12152 (cortex_a53_fpmul): Likewise.
12153 (cortex_a53_f_load_64): Likewise.
12154 (cortex_a53_f_load_many): Likewise.
12155 (cortex_a53_advsimd_alu): Likewise.
12156 (cortex_a53_advsimd_alu_q): Likewise.
12157 (cortex_a53_advsimd_mul): Likewise.
12158 (cortex_a53_advsimd_mul_q): Likewise.
12159 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12160 Add missing fmul, r2f_cvt and fconst cases.
12162 2017-06-14 Richard Biener <rguenther@suse.de>
12164 PR middle-end/81088
12165 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12167 (fold_binary_loc): When associating do not treat pre-existing
12168 TREE_OVERFLOW on literal constants as a reason to allow
12169 TREE_OVERFLOW on associated literal constants.
12171 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12173 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12174 (MASK_FEATURES): New macro.
12175 * config/sparc/sparc.c (sparc_option_override): Remove the special
12176 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12178 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12180 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12181 a division of 0 if non-call exceptions are enabled.
12183 2017-06-14 Andrew Pinski <apinski@cavium.com>
12184 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12187 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12188 Improve vector initialization code gen for only variable case.
12190 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12192 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12194 2017-06-14 Richard Biener <rguenther@suse.de>
12196 PR tree-optimization/81083
12197 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12200 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12202 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12203 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12204 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12205 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12206 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12207 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12209 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12211 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12212 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12214 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12216 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12218 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12220 * config/rs6000/t-rtems: Don't handle SPE.
12222 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12224 * config/rs6000/t-linux: Don't handle SPE.
12226 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12228 * config/rs6000/eabispe.h: Delete file.
12230 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12232 * config/rs6000/t-spe: Delete file.
12234 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12236 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12237 (rs6000_legitimate_offset_address_p): Return false for anything in
12238 V2SImode or V2SFmode.
12240 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12242 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12243 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12244 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12246 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12247 (rs6000_legitimate_offset_address_p): Ditto.
12248 (rs6000_emit_move): Ditto.
12249 (rs6000_init_builtins): Remove V4HI_type_node.
12251 2017-06-13 Martin Liska <mliska@suse.cz>
12254 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12255 (gate_asan): Likewise.
12256 * asan.h (asan_no_sanitize_address_p): Remove the function.
12257 (sanitize_flags_p): New function.
12258 * builtins.def: Fix coding style.
12259 * common.opt: Use renamed enum value.
12260 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12261 * doc/extend.texi: Document no_sanitize attribute.
12262 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12263 to SANITIZE_UNDEFINED_NONDEFAULT.
12264 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12265 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12266 Use sanitize_flags_p.
12267 * gimplify.c (gimplify_function_tree): Likewise.
12268 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12269 * opts.c (parse_no_sanitize_attribute): New function.
12270 (common_handle_option): Use renamed enum value.
12271 * opts.h (parse_no_sanitize_attribute): Declare.
12272 * tree.c (sanitize_flags_p): New function.
12273 * tree.h: Declared here.
12274 * tsan.c: Use sanitize_flags_p.
12275 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12276 (instrument_mem_ref): Likewise.
12277 (instrument_bool_enum_load): Likewise.
12278 (do_ubsan_in_current_function): Remove the function.
12279 (pass_ubsan::execute): Use sanitize_flags_p.
12280 * ubsan.h: Remove do_ubsan_in_current_function
12281 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12282 (dump_function_to_file): Use it here.
12284 2017-06-13 Martin Jambor <mjambor@suse.cz>
12286 PR tree-optimization/80803
12287 PR tree-optimization/81063
12288 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12289 (propagate_subaccesses_across_link): Enqueue subtree whenever
12290 necessary instead of relying on the caller.
12292 2017-06-13 Martin Jambor <mjambor@suse.cz>
12294 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12295 that have a first_link.
12296 (sort_and_splice_var_accesses): Do not check first_link before
12298 (subtree_mark_written_and_enqueue): Likewise.
12299 (propagate_all_subaccesses): Likewise and do not stop at first
12300 parent with a first_link.
12302 2017-06-13 Martin Jambor <mjambor@suse.cz>
12304 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12307 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12309 * match.pd: New pattern.
12311 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12313 * tree-vrp.c (is_masked_range_test): New function.
12314 (register_edge_assert_for): Determine ranges for
12317 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12319 PR tree-optimization/67328
12320 * fold-const.c (maskable_range_p): New function.
12321 (build_range_check): Generate bittests if possible.
12323 2017-06-13 Martin Liska <mliska@suse.cz>
12325 * gimple-pretty-print.c (dump_probability): Add new argument.
12326 (dump_edge_probability): Dump both probability and count.
12327 (dump_gimple_label): Likewise.
12328 (dump_gimple_bb_header): Likewise.
12330 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12333 * config/avr/avr-devices.c: Fix indentation.
12334 * config/avr/gen-avr-mmcu-specs.c: Dito.
12336 2017-06-13 Richard Biener <rguenther@suse.de>
12338 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12339 instead get vector type from stmt_info.
12340 (vectorizable_reduction): Adjust. Remove dead code.
12342 2017-06-13 Richard Biener <rguenther@suse.de>
12344 PR middle-end/81065
12345 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12346 case of C * (x * C2 + C3).
12347 (fold_addr_of_array_ref_difference): Properly fold index difference.
12349 2017-06-12 David S. Miller <davem@davemloft.net>
12352 * config/sparc/sparc.md (return expander): Emit frame blockage if
12353 function uses alloca.
12355 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12357 * combine.c (make_field_assignment): Check len rather than the mode
12358 precision when calling force_to_mode.
12360 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12362 Support multilibs and devices that see flash in RAM address range.
12365 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12366 (avr_mcu_t) <flash_pm_offset>: New field.
12367 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12368 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12369 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12370 (AVR_TINY_PM_OFFSET): Remove macro.
12371 * config/avr/avr.opt (-mshort-calls): New option.
12372 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12373 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12374 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12375 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12376 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12377 instead of avr_arch->have_jmp_call.
12378 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12379 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12380 avr_arch->flash_pm_offset to define.
12381 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12382 new field flash_pm_offset. Add entry for avrxmega3.
12383 (avr_texinfo): Add entry for avrxmega3.
12384 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12385 attiny212, attiny214,
12386 attiny412, attiny414, attiny416, attiny417,
12387 attiny814, attiny816, attiny817,
12388 attiny1614, attiny1616, attiny1617,
12389 attiny3214, attiny3216, attiny3217.
12390 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12391 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12392 (avr_print_operand_address) [AVR_TINY]: Same.
12393 (avr_asm_init_sections) <readonly_data_section>: Only patch
12394 callback if avr_arch->flash_pm_offset = 0.
12395 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12396 for rodata if avr_arch->flash_pm_offset != 0.
12397 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12398 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12399 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12400 (m_options): Append opt_rcall.
12401 (m_dirnames): Append dir_rcall.
12402 * config/avr/t-multilib: Regenerate.
12404 * configure.ac [target=avr]: Check whether avrxmega3 default
12405 linker description file works as needed.
12406 * configure: Regenerate.
12407 * doc/avr-mmcu.texi: Regenerate.
12408 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12409 <__AVR_ARCH__>: Document avrxmega3 and 103.
12410 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12411 <__AVR_SHORT_CALLS__>: Document it.
12412 <__AVR_PM_BASE_ADDRESS__>: Document it.
12413 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12414 (AVR Variable Attributes) <progmem>: Document this is
12415 not needed for avrxmega3.
12416 (AVR Named Address Spaces) <__flash>: Dito.
12418 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12420 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12422 2017-06-12 Doug Rupp <rupp@adacore.com>
12424 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12425 Append vxworks-stdint.h to the tm_file list.
12426 * config/vxworks-stdint.h: New file.
12428 2017-06-12 Martin Liska <mliska@suse.cz>
12430 PR tree-optimization/81041
12431 * tree-profile.c (gimple_gen_ic_func_profiler):
12432 Create an extra BB in profile-generate
12433 (gimple_gen_time_profiler): Likewise.
12435 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12437 PR tree-optimization/81003
12438 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12439 (update_range_test): Use it instead of force_gimple_operand_gsi.
12441 2017-06-12 Richard Biener <rguenther@suse.de>
12443 PR tree-optimization/81053
12444 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12445 with backedge value not defined in loop. Simplify def stmt
12448 2017-06-11 Tom de Vries <tom@codesourcery.com>
12451 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12453 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12454 nvptx_cannot_force_const_mem.
12456 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12458 * opts.c (finish_options): Move test for flag_split_stack after
12459 it has been initialized.
12461 2017-06-11 Jason Merrill <jason@redhat.com>
12463 * tree.h (id_equal): New.
12464 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12465 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12466 instead of strcmp of IDENTIFIER_POINTER.
12468 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12470 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12471 (mark_all_inlined_calls_cdtor): Fix formating.
12472 (inline_transform): Rescale profile before inlining.
12474 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12476 * cgraph.h (cgraph_edge::clone): Update prototype.
12477 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12478 (cgraph_node::create_clone): Update.
12479 (cgraph_node::create_version_clone): Update.
12480 * tree-inline.c (copy_bb): Update.
12481 (expand_call_inline): Update.
12483 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12485 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12486 factored out from ...
12487 (rs6000_emit_prologue): ... here.
12489 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12491 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12492 factored out from ...
12493 (rs6000_emit_prologue): ... here.
12495 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12497 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12499 (handle_missing_profiles): Fix computation of tp_first_run.
12500 (counts_to_freqs): Do not touch freqs when count is 0.
12502 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12504 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12507 2017-06-10 Tom de Vries <tom@codesourcery.com>
12509 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12510 attributes): Document signal effective target.
12512 2017-06-10 Tom de Vries <tom@codesourcery.com>
12514 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12515 Document effective target stack_size.
12517 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12519 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12520 to the edit_context if they can be auto-applied.
12522 2017-06-9 Ian Lance Taylor <iant@golang.org>
12524 * opts.c (finish_options): If -fsplit-stack, disable implicit
12525 -forder-blocks-and-partition.
12526 * doc/invoke.texi (Optimize Options): Document that when using
12527 -fsplit-stack -forder-blocks-and-partition is not implicitly
12530 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12532 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12533 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12534 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12535 * builtins.def (abort, trap, unreachable): Declare cold.
12536 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12537 * tree-core.h (ECF_COLD): New.
12538 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12539 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12541 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12543 * predict.c (unlikely_executed_stmt_p): Cleanup.
12545 2017-06-09 Richard Biener <rguenther@suse.de>
12547 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12548 model if the ref is always written to.
12550 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12552 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12554 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12556 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12559 2017-06-09 Tom de Vries <tom@codesourcery.com>
12562 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12563 "target cannot support label values" when encountering LABEL_REF.
12565 2017-06-09 Martin Liska <mliska@suse.cz>
12567 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12568 (gimple_gen_ic_func_profiler): Emit direct comparison
12569 of __gcov_indirect_call_callee with NULL.
12570 (gimple_gen_time_profiler): Change probability from
12571 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12573 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12575 * profile.c (edge_gcov_counts): Turn to pointer.
12576 (compute_branch_probabilities, compute_branch_probabilities): Update.
12577 (branch_prob): Do not clear edge_gcov_count.
12578 * profile.h (edge_gcov_counts): Turn to pointer.
12579 (edge_gcov_count): Update.
12581 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12583 * gimple.h (gimple_check_failed): Mark cold.
12585 2017-06-09 Richard Biener <rguenther@suse.de>
12587 PR tree-optimization/66623
12588 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12589 refactor check_reduction into two parts, properly computing
12590 whether we have to check reduction validity for outer loop
12593 2017-06-09 Richard Biener <rguenther@suse.de>
12595 PR tree-optimization/79483
12596 * graphite-scop-detection.c (order): New global.
12597 (get_order): Compute bb to order mapping that satisfies code
12598 generation constraints.
12599 (cmp_pbbs): New helper.
12600 (build_scops): Start domwalk at entry block, sort generated
12603 2017-06-09 Richard Biener <rguenther@suse.de>
12605 PR middle-end/81007
12606 * ipa-polymorphic-call.c
12607 (ipa_polymorphic_call_context::restrict_to_inner_class):
12608 Skip FIELD_DECLs with error_mark_node type.
12609 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12612 2017-06-09 Martin Liska <mliska@suse.cz>
12614 * predict.c (struct branch_predictor): New struct.
12615 (test_prediction_value_range): New test.
12616 (predict_c_tests): New function.
12617 * selftest-run-tests.c (selftest::run_tests): Run the function.
12618 * selftest.h: Declare new tests.
12620 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12623 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12624 gen_add3_insn did not fail.
12625 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12626 r0, construct that number in a temporary reg and add that reg to r0.
12627 If asked to put the result in r0 as well, fail.
12629 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12632 for early expansion of vec_eqv.
12634 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12636 PR middle-end/81005
12637 * ubsan.c (instrument_null): Avoid pointless code temporary.
12638 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12640 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12644 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12647 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12650 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12652 * predict.c (unlikely_executed_edge_p): Move ahead.
12653 (probably_never_executed_edge_p): Use it.
12655 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12657 PR middle-end/79988
12658 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12659 gimple_call_builtin_p call.
12661 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12663 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12664 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12665 rtl_check_failed_type2, rtl_check_failed_code1,
12666 rtl_check_failed_code2, rtl_check_failed_code_mode,
12667 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12668 rtvec_check_failed_bounds, rtl_check_failed_flag,
12669 _fatal_insn_not_found, _fatal_insn): Likewise.
12670 * tree.h (tree_contains_struct_check_failed,
12671 tree_check_failed, tree_not_check_failed,
12672 tree_class_check_failed, tree_range_check_failed,
12673 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12674 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12675 tree_operand_check_failed, omp_clause_check_failed,
12676 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12679 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12681 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12682 flag_branch_probabilities.
12683 * ipa-inline.c (edge_badness): Likewise.
12684 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12685 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12686 * predict.c (maybe_hot_frequency_p): Likewise.
12687 (probably_never_executed): Likewise.
12688 * sched-ebb.c (schedule_ebbs): Likewise.
12689 * sched-rgn.c (find_single_block_region): Likewise.
12690 * tracer.c (tail_duplicate): Likewise.
12692 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12694 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12695 longer requires x_flag_profile_use.
12697 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12699 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12700 instead of flag_reorder_blocks_and_partition.
12701 * dbxout.c (dbxout_function_end): Likewise.
12702 * dwarf2out.c (gen_subprogram_die): Likewise.
12703 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12704 * hw-doloop.c (reorg_loops): Likewise.
12705 * varasm.c (assemble_start_function,
12706 assemble_end_function): Likewise.
12707 (decide_function_section): Do not check for
12708 flag_reorder_blocks_and_partition.
12710 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12712 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12714 (chkp_get_hard_register_fake_addr_expr): Ditto.
12715 (chkp_build_addr_expr): Add check for hard reg case.
12716 (chkp_parse_array_and_component_ref): Ditto.
12717 (chkp_find_bounds_1): Ditto.
12718 (chkp_process_stmt): Don't generate bounds store for
12721 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12723 * predict.c (maybe_hot_bb_p): Do not check profile status.
12724 (maybe_hot_edge_p): Likewise.
12725 (probably_never_executed): Check for zero counts even if profile
12727 (unlikely_executed_edge_p): New function.
12728 (unlikely_executed_stmt_p): New function.
12729 (unlikely_executed_bb_p): New function.
12730 (set_even_probabilities): Use unlikely predicates.
12731 (combine_predictions_for_bb): Likewise.
12732 (predict_paths_for_bb): Likewise.
12733 (predict_paths_leading_to_edge): Likewise.
12734 (determine_unlikely_bbs): New function.
12735 (estimate_bb_frequencies): Use it.
12736 (compute_function_frequency): Use zero counts even if profile is
12738 * profile-count.h: Fix typo.
12740 2017-08-08 Julia Koval <julia.koval@intel.com>
12742 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12743 _mm512_mask_cvtsepi16_storeu_epi8,
12744 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12745 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12746 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12747 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12748 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12749 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12750 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12751 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12752 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12753 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12754 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12755 __builtin_ia32_pmovuswb256mem_mask,
12756 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12757 __builtin_ia32_pmovwb512mem_mask): New builtins.
12759 2017-08-08 Julia Koval <julia.koval@intel.com>
12761 PR target/73350,80862
12762 * config/i386/subst.md (round): Fix round pattern.
12763 * config/i386/i386.c (ix86_erase_embedded_rounding):
12764 Fix erasing rounding for the fixed pattern.
12766 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12768 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12770 2017-06-08 Martin Liska <mliska@suse.cz>
12772 PR gcov-profile/80911
12773 * gcov.c (block_info::block_info): New constructor.
12775 2017-06-07 Carl Love <cel@us.ibm.com>
12777 * config/rs6000/rs6000-c: The return type of the following
12778 built-in functions was implemented as int not long long. Fix sign
12779 of return value for the unsigned version of vec_mulo and vec_mule.
12780 vector unsigned long long vec_bperm (vector unsigned long long,
12781 vector unsigned char)
12782 vector signed long long vec_mule (vector signed int,
12784 vector unsigned long long vec_mule (vector unsigned int,
12785 vector unsigned int)
12786 vector signed long long vec_mulo (vector signed int,
12788 vector unsigned long long vec_mulo (vector unsigned int,
12789 vector unsigned int)
12790 * doc/extend.texi: Fix the documentation for the built-in
12793 2017-06-07 Carl Love <cel@us.ibm.com>
12796 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12799 2017-06-07 Carl Love <cel@us.ibm.com>
12801 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12802 support, Generate doublehv for signed int/float for BE case only.
12804 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12806 * doc/invoke.texi (mcx16): Rewrite.
12808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12810 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12811 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12812 *mov<mode>_softfloat, and an anonymous splitter): Use
12813 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12815 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12817 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12819 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12820 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12821 (rs6000_debug_reg_global): Adjust.
12822 (rs6000_init_hard_regno_mode_ok): Adjust.
12823 (rs6000_dbx_register_number): Adjust.
12824 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12825 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12826 Remove SPE_ACC and SPEFSCR.
12827 (REG_ALLOC_ORDER): Ditto.
12828 (FRAME_POINTER_REGNUM): Change to 111.
12829 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12830 (REG_CLASS_NAMES): Ditto.
12831 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12832 (REGISTER_NAMES): Ditto.
12833 (ADDITIONAL_REG_NAMES): Ditto.
12834 (rs6000_reg_names): Ditto.
12835 * config/rs6000/rs6000.md: Renumber some register number
12838 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12840 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12842 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12843 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12845 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12846 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12847 Delete the SPE high registers.
12848 (REG_ALLOC_ORDER): Ditto.
12849 (enum reg_class): Remove SPE_HIGH_REGS.
12850 (REG_CLASS_NAMES): Ditto.
12851 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12852 (REGISTER_NAMES): Ditto.
12853 (rs6000_reg_names): Ditto.
12854 * doc/tm.texi.in: Remove SPE as example.
12855 * doc/tm.texi: Regenerate.
12857 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12859 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12860 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12861 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12862 * config/rs6000/rs6000.md (type): Remove "brinc".
12864 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12866 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12867 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12868 * config/rs6000/linuxspe.h: Delete file.
12869 * config/rs6000/rs6000.md: Don't include spe.md.
12870 * config/rs6000/spe.h: Delete file.
12871 * config/rs6000/spe.md: Delete file.
12872 * config/rs6000/t-rs6000: Remove spe.md.
12874 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12876 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12877 (reg_or_none500mem_operand): Delete.
12878 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12879 instead of reg_or_none500mem_operand.
12881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12884 handling of SPE flags.
12885 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12889 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12891 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12892 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12893 paired_divv2sf3): Similar.
12894 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12895 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12896 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12898 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12899 Rename the paired_* instruction patterns.
12900 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12902 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12903 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12904 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12905 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12906 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12907 PAIRED_VECTOR_MODE.
12908 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12909 (spe_func_has_64bit_regs_p): Delete.
12910 (spe_expand_predicate_builtin): Delete.
12911 (spe_expand_evsel_builtin): Delete.
12912 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12913 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12914 (invalid_e500_subreg): Delete.
12915 (rs6000_legitimize_address): Always force_reg op2 as well, for
12916 paired single memory accesses.
12917 (rs6000_member_type_forces_blk): Delete.
12918 (rs6000_spe_function_arg): Delete.
12919 (rs6000_expand_unop_builtin): Delete SPE handling.
12920 (rs6000_expand_binop_builtin): Ditto.
12921 (spe_expand_stv_builtin): Delete.
12922 (bdesc_2arg_spe): Delete.
12923 (spe_expand_builtin): Delete.
12924 (spe_expand_predicate_builtin): Delete.
12925 (spe_expand_evsel_builtin): Delete.
12926 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12927 (spe_init_builtins): Delete.
12928 (spe_func_has_64bit_regs_p): Delete.
12929 (savres_routine_name): Delete "info" parameter. Adjust callers.
12930 (rs6000_emit_stack_reset): Ditto.
12931 (rs6000_dwarf_register_span): Delete.
12932 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12933 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12934 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12936 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12938 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12939 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12940 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12941 mulv2sf3, divv2sf3): Delete expanders.
12943 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12945 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12947 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12949 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12950 * config/rs6000/rs6000.c: Ditto.
12952 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12954 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12955 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12956 comparison_operator.
12958 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12960 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12961 * config/rs6000/rs6000.opt: Ditto.
12962 * config/rs6000/t-rtems: Ditto.
12964 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12966 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12967 TARGET_E500_SINGLE by 0, simplify.
12968 * config/rs6000/rs6000.c: Ditto.
12969 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12970 (spe_build_register_parallel): Delete.
12971 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12972 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12973 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12974 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12975 (E500_CONVERT): Delete.
12976 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12980 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12981 * config/rs6000/dfp.md: Ditto.
12982 (negdd2, *negdd2_fpr): Merge.
12983 (absdd2, *absdd2_fpr): Merge.
12984 (negtd2, *negtd2_fpr): Merge.
12985 (abstd2, *abstd2_fpr): Merge.
12986 * config/rs6000/e500.h: Delete file.
12987 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12988 TARGET_FPRS by 1 and simplify.
12989 * config/rs6000/rs6000-c.c: Ditto.
12990 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12991 TARGET_DF_SPE by 0.
12992 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12994 * config/rs6000/rs6000.md: Ditto.
12995 (floatdidf2, *floatdidf2_fpr): Merge.
12996 (move_from_CR_gt_bit): Delete.
12997 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12998 (E500_CR_IOR_COMPARE): Delete.
12999 (All patterns that require !TARGET_FPRS): Delete.
13000 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13002 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13004 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13006 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13008 * graphds.c (add_edge): Intitialize edge's attached data.
13009 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13010 pointer parameter. Call pointed function on each edge during
13011 graph traversing. Skip traversing the edge when the function
13013 (graphds_dfs, graphds_scc): Ditto.
13014 (for_each_edge): New parameter. Pass the new parameter to callback
13016 * graphds.h (skip_edge_callback): New function pointer type.
13017 (graphds_dfs, graphds_scc): New function pointer parameter.
13018 (graphds_edge_callback, for_each_edge): New parameter.
13020 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13022 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13023 out code checking if runtime alias check is possible to below ...
13024 Call the new function.
13025 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13026 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13028 2017-06-07 Marek Polacek <polacek@redhat.com>
13031 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13032 TYPE_OVERFLOW_WRAPS check.
13034 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13036 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13037 if versioning is required.
13038 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13039 peeling with the check for versioning.
13041 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13043 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13044 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13045 Set true to new parameter if new ssa variable is defined.
13046 (vect_gen_vector_loop_niters): Refactor. Set range information
13047 for the new vector loop bound variable.
13048 (vect_do_peeling): Ditto.
13050 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13052 * tree-affine.c (ssa.h): Include header file.
13053 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13054 has wrapping overflow behavior.
13056 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13058 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13060 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13062 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13063 (tree_to_aff_combination): ... here.
13065 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13067 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13068 reg_pressure model function.
13069 (ivopts_global_cost_for_size): Delete.
13070 (determine_set_costs, iv_ca_recount_cost): Call new model function
13071 ivopts_estimate_reg_pressure.
13073 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13075 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13076 expensive than udiv. Remove floating point cases from mod.
13078 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13080 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13081 Increase idiv cost.
13083 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13085 * config/aarch64/aarch64.md
13086 (copysignsf3): Fix mask generation.
13088 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13090 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13092 (class dump_manager): Add register_dumps method.
13093 * dumpfile.c: Include langhooks.h.
13094 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13095 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13096 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13097 (dump_manager::dump_register): Start with 512 entries instead of 32.
13098 (dump_manager::register_dumps): New method.
13099 * toplev.c (general_init): Instead of invoking register_dumps
13100 langhook, invoke register_dumps method on the dump manager.
13101 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13104 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13106 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13107 Say that # requires an associated define_split to exist, and that
13108 the define_split must be suitable for use after register allocation.
13110 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13112 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13113 (compute_outgoing_frequencies): Also initialize zero counts.
13114 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13115 around loops; preserve more of profile when nothing changes.
13117 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13119 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13121 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13122 * config/arm/arm-cpu-cdata.h: Regenerate.
13123 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13124 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13125 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13126 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13127 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13129 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13131 * config/arm/t-rmprofile: Likewise.
13132 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13134 2017-06-06 David S. Miller <davem@davemloft.net>
13137 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13138 blockage if function uses alloca.
13140 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13142 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13143 New "uid" fields to hold pretty-print IDs of group and ref.
13144 Memory references are now identified as <group_id>:<ref_id>
13145 instead of using [random] addresses.
13146 (dump_mem_details): Simplify, no functional change.
13147 (dump_mem_ref): Simplify and make output more concise.
13148 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13149 (find_or_create_group): Initialize group uid.
13150 (record_ref): Initialize ref uid. Improve debug output.
13151 (prune_group_by_reuse, should_issue_prefetch_p,)
13152 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13153 (mark_nontemporal_store, determine_loop_nest_reuse):
13154 Improve debug output.
13156 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13158 * dbgcnt.def (prefetch): New debug counter.
13159 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13160 (schedule_prefetches): Stop issueing prefetches if debug counter
13163 2017-06-06 Tom de Vries <tom@codesourcery.com>
13165 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13166 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13168 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13170 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13171 Use aarch64_reg_or_zero predicate for operand 4.
13172 (aarch64_compare_and_swap<mode> define_insn_and_split):
13173 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13174 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13176 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13178 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13179 (arm_compute_save_core_reg_mask): This.
13180 (thumb1_compute_save_reg_mask): Rename into ...
13181 (thumb1_compute_save_core_reg_mask): This.
13182 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13183 (arm_compute_frame_layout): Likewise.
13185 2017-06-06 Richard Biener <rguenther@suse.de>
13187 PR tree-optimization/80974
13188 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13189 keep or clear leaders SSA info.
13191 2017-06-06 Tom de Vries <tom@codesourcery.com>
13193 * config/nvptx/nvptx.c (split_mode_p): New function.
13194 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13196 2017-06-06 Tom de Vries <tom@codesourcery.com>
13198 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13200 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13203 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13206 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13208 * shrink-wrap.c (handle_simple_exit): Update profile.
13209 (try_shrink_wrapping): Upate profile.
13211 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13213 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13214 (tree_guess_outgoing_edge_probabilities): New.
13215 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13216 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13218 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13220 * ipa-split.c (split_function): Initialize return bb profile.
13222 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13224 * profile.c (compute_branch_probabilities): Also initialize
13225 EXIT_BLOCK profile.
13227 2017-06-06 Richard Biener <rguenther@suse.de>
13229 PR tree-optimization/80928
13230 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13231 (vect_analyze_loop_operations): Properly guard analysis for
13233 (vect_transform_loop): Likewise.
13234 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13235 (vect_model_induction_cost): Do not cost for pure SLP.
13236 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13237 of induction in inner loop vectorization.
13238 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13239 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13240 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13242 (vect_analyze_slp_cost_1): Cost induction.
13243 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13244 (vect_get_slp_vect_defs): Likewise.
13245 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13246 (vect_transform_stmt): Handle SLP reductions.
13247 * tree-vectorizer.h (vectorizable_induction): Adjust.
13249 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13251 * config/rs6000/rs6000.c (make_resolver_func): Update
13252 init_lowered_empty_function call.
13254 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13256 * doc/invoke.texi: Document the -fprofile-abs-path option.
13257 * common.opt (fprofile-abs-path): New option.
13258 * gcov-io.h (gcov_write_filename): Declare.
13259 * gcov-io.c (gcov_write_filename): New function.
13260 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13261 * profile.c (output_location): Likewise.
13263 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13265 * shring-wrap.c: Revert accidental commit.
13267 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13269 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13271 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13273 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13275 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13276 profile in callgraph edge.
13277 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13278 (apply_scale): Likewise.
13279 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13280 Also scale profile when inlining function with zero profile.
13281 (initialize_cfun): Update exit block profile even when it is zero.
13282 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13283 when profile is read.
13285 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13287 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13288 (CLONE_*): New constants to define the processors we can generate
13289 code for with the target_clone attribute.
13290 (rs6000_clone_map): New array to identify which clone processors
13291 the current program is running on.
13292 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13293 target_clone attribute.
13294 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13295 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13296 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13297 (cpu_expand_builtin): Add support for target_clone attribute.
13298 (rs6000_valid_attribute_p): Allow "default" attribute.
13299 (get_decl_name): New debug function to simplify printing the
13300 current function name in debugging statements.
13301 (rs6000_clone_priority): New functions to support the target_clone
13302 attribute, and be able to generate code to switch between ISA 2.05
13303 through ISA 3.0 (power6 through power9).
13304 (rs6000_compare_version_priority): Likewise.
13305 (rs6000_get_function_versions_dispatcher): Likewise.
13306 (make_resolver_func): Likewise.
13307 (add_condition_to_bb): Likewise.
13308 (dispatch_function_versions): Likewise.
13309 (rs6000_generate_version_dispatcher_body): Likewise.
13310 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13311 (fusion_gpr_load_p): Fix a spacing issue.
13312 * doc/extend.texi (Common Function Attributes): Document that the
13313 PowerPC supports the target_clone attribute.
13315 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13317 * config/arm/arm.h: explain F symbol found in description of ARM
13318 register allocation in its legend.
13320 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13322 * config/mips/frame-header-opt.c: Include profile-count.h.
13323 * config/riscv/riscv.c: Include profile-count.h
13325 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13327 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13329 (sm_set_flag_if_changed): Add bbs field.
13330 (execute_sm_if_changed_flag_set): Pass BBS.
13331 (execute_sm): Update.
13333 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13335 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13338 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13340 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13341 (peephole2): New peephole2 to emit the above.
13342 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13344 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13346 * config/aarch64/aarch64.c (define_peephole2 above
13347 *sub_<shift>_<mode>): New peephole.
13349 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13351 * config/i386/i386.c (make_resolver_func): Update.
13352 * Makefile.in: Add profile-count.h and profile-count.o
13353 * auto-profile.c (afdo_indirect_call): Update to new API.
13354 (afdo_set_bb_count): Update.
13355 (afdo_propagate_edge): Update.
13356 (afdo_propagate_circuit): Update.
13357 (afdo_calculate_branch_prob): Update.
13358 (afdo_annotate_cfg): Update.
13359 * basic-block.h: Include profile-count.h
13360 (struct edge_def): Turn count to profile_count.
13361 (struct basic_block_def): Likewie.
13362 (REG_BR_PROB_BASE): Move to profile-count.h
13363 (RDIV): Move to profile-count.h
13364 * bb-reorder.c (max_entry_count): Turn to profile_count.
13365 (find_traces): Update.
13366 (rotate_loop):Update.
13367 (connect_traces):Update.
13368 (sanitize_hot_paths):Update.
13369 * bt-load.c (migrate_btr_defs): Update.
13370 * cfg.c (RDIV): Remove.
13371 (init_flow): Use alloc_block.
13372 (alloc_block): Uninitialize count.
13373 (unchecked_make_edge): Uninitialize count.
13374 (check_bb_profile): Update.
13375 (dump_edge_info): Update.
13376 (dump_bb_info): Update.
13377 (update_bb_profile_for_threading): Update.
13378 (scale_bbs_frequencies_int): Update.
13379 (scale_bbs_frequencies_gcov_type): Update.
13380 (scale_bbs_frequencies_profile_count): New.
13381 * cfg.h (update_bb_profile_for_threading): Update.
13382 (scale_bbs_frequencies_profile_count): Declare.
13383 * cfgbuild.c (compute_outgoing_frequencies): Update.
13384 (find_many_sub_basic_blocks): Update.
13385 * cfgcleanup.c (try_forward_edges): Update.
13386 (try_crossjump_to_edge): Update.
13387 * cfgexpand.c (expand_gimple_tailcall): Update.
13388 (construct_exit_block): Update.
13389 * cfghooks.c (verify_flow_info): Update.
13390 (dump_bb_for_graph): Update.
13391 (split_edge): Update.
13392 (make_forwarder_block): Update.
13393 (duplicate_block): Update.
13394 (account_profile_record): Update.
13395 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13396 (get_estimated_loop_iterations): Update.
13397 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13398 (single_likely_exit): Update.
13399 * cfgloopmanip.c (scale_loop_profile): Update.
13401 (set_zero_probability): Update.
13402 (lv_adjust_loop_entry_edge): Update.
13403 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13404 (purge_dead_edges): Update.
13405 (rtl_account_profile_record): Update.
13406 * cgraph.c (cgraph_node::create): Uninitialize count.
13407 (symbol_table::create_edge): Uninitialize count.
13408 (cgraph_update_edges_for_call_stmt_node): Update.
13409 (cgraph_edge::dump_edge_flags): Update.
13410 (cgraph_node::dump): Update.
13411 (cgraph_edge::maybe_hot_p): Update.
13412 * cgraph.h: Include profile-count.h
13413 (create_clone), create_edge, create_indirect_edge): Update.
13414 (cgraph_node): Turn count to profile_count.
13415 (cgraph_edge0: Likewise.
13416 (make_speculative, clone): Update.
13417 (create_edge): Update.
13418 (init_lowered_empty_function): Update.
13419 * cgraphclones.c (cgraph_edge::clone): Update.
13420 (duplicate_thunk_for_node): Update.
13421 (cgraph_node::create_clone): Update.
13422 * cgraphunit.c (cgraph_node::analyze): Update.
13423 (cgraph_node::expand_thunk): Update.
13424 * final.c (dump_basic_block_info): Update.
13425 * gimple-streamer-in.c (input_bb): Update.
13426 * gimple-streamer-out.c (output_bb): Update.
13427 * graphite.c (print_global_statistics): Update.
13428 (print_graphite_scop_statistics): Update.
13429 * hsa-brig.c: Include basic-block.h.
13430 * hsa-dump.c: Include basic-block.h.
13431 * hsa-gen.c (T sum_slice): Update.
13432 (convert_switch_statements):Update.
13433 * hsa-regalloc.c: Include basic-block.h.
13434 * ipa-chkp.c (chkp_produce_thunks): Update.
13435 * ipa-cp.c (struct caller_statistics): Update.
13436 (init_caller_stats): Update.
13437 (gather_caller_stats): Update.
13438 (ipcp_cloning_candidate_p): Update.
13439 (good_cloning_opportunity_p): Update.
13440 (get_info_about_necessary_edges): Update.
13441 (dump_profile_updates): Update.
13442 (update_profiling_info): Update.
13443 (update_specialized_profile): Update.
13444 (perhaps_add_new_callers): Update.
13445 (decide_about_value): Update.
13446 (ipa_cp_c_finalize): Update.
13447 * ipa-devirt.c (struct odr_type_warn_count): Update.
13448 (struct decl_warn_count): Update.
13449 (struct final_warning_record): Update.
13450 (possible_polymorphic_call_targets): Update.
13451 (ipa_devirt): Update.
13452 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13453 * ipa-icf.c (sem_function::merge): Update.
13454 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13455 * ipa-inline.c (compute_uninlined_call_time): Update.
13456 (compute_inlined_call_time): Update.
13457 (want_inline_small_function_p): Update.
13458 (want_inline_self_recursive_call_p): Update.
13459 (edge_badness): Update.
13460 (lookup_recursive_calls): Update.
13461 (recursive_inlining): Update.
13462 (inline_small_functions): Update.
13463 (dump_overall_stats): Update.
13464 (dump_inline_stats): Update.
13465 * ipa-profile.c (ipa_profile_generate_summary): Update.
13466 (ipa_propagate_frequency): Update.
13467 (ipa_profile): Update.
13468 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13469 * ipa-utils.c (ipa_merge_profiles): Update.
13470 * loop-doloop.c (doloop_modify): Update.
13471 * loop-unroll.c (report_unroll): Update.
13472 (unroll_loop_runtime_iterations): Update.
13473 * lto-cgraph.c (lto_output_edge): Update.
13474 (lto_output_node): Update.
13475 (input_node): Update.
13476 (input_edge): Update.
13477 (merge_profile_summaries): Update.
13478 * lto-streamer-in.c (input_cfg): Update.
13479 * lto-streamer-out.c (output_cfg): Update.
13480 * mcf.c (create_fixup_graph): Update.
13481 (adjust_cfg_counts): Update.
13482 (sum_edge_counts): Update.
13483 * modulo-sched.c (sms_schedule): Update.
13484 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13485 * predict.c (maybe_hot_count_p): Update.
13486 (probably_never_executed): Update.
13487 (dump_prediction): Update.
13488 (combine_predictions_for_bb): Update.
13489 (propagate_freq): Update.
13490 (handle_missing_profiles): Update.
13491 (counts_to_freqs): Update.
13492 (rebuild_frequencies): Update.
13493 (force_edge_cold): Update.
13494 * predict.h: Include profile-count.h
13495 (maybe_hot_count_p, counts_to_freqs): UPdate.
13496 * print-rtl-function.c: Do not include cfg.h
13497 * print-rtl.c: Include basic-block.h
13498 * profile-count.c: New file.
13499 * profile-count.h: New file.
13500 * profile.c (is_edge_inconsistent): Update.
13501 (correct_negative_edge_counts): Update.
13502 (is_inconsistent): Update.
13503 (set_bb_counts): Update.
13504 (read_profile_edge_counts): Update.
13505 (compute_frequency_overlap): Update.
13506 (compute_branch_probabilities): Update; Initialize and deinitialize
13508 (branch_prob): Update.
13509 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13510 (edge_gcov_count): New.
13511 (bb_gcov_count): New.
13512 * shrink-wrap.c (try_shrink_wrapping): Update.
13513 * tracer.c (better_p): Update.
13514 * trans-mem.c (expand_transaction): Update.
13515 (ipa_tm_insert_irr_call): Update.
13516 (ipa_tm_insert_gettmclone_call): Update.
13517 * tree-call-cdce.c: Update.
13518 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13519 (gimple_duplicate_sese_tail): Update.
13520 (gimple_account_profile_record): Update.
13521 (execute_fixup_cfg): Update.
13522 * tree-inline.c (copy_bb): Update.
13523 (copy_edges_for_bb): Update.
13524 (initialize_cfun): Update.
13525 (freqs_to_counts): Update.
13526 (copy_cfg_body): Update.
13527 (expand_call_inline): Update.
13528 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13529 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13530 (try_unroll_loop_completely): Update.
13531 (try_peel_loop): Update.
13532 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13533 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13534 * tree-ssa-loop-split.c (connect_loops): Update.
13535 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13536 * tree-ssa-reassoc.c (branch_fixup): Update.
13537 * tree-ssa-tail-merge.c (replace_block_by): Update.
13538 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13539 (compute_path_counts): Update.
13540 (update_profile): Update.
13541 (recompute_probabilities): Update.
13542 (update_joiner_offpath_counts): Update.
13543 (estimated_freqs_path): Update.
13544 (freqs_to_counts_path): Update.
13545 (clear_counts_path): Update.
13546 (ssa_fix_duplicate_block_edges): Update.
13547 (duplicate_thread_path): Update.
13548 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13549 (struct switch_conv_info): Update.
13550 * tree-tailcall.c (decrease_profile): Update.
13551 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13552 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13553 * value-prof.c (check_counter): Update.
13554 (gimple_divmod_fixed_value): Update.
13555 (gimple_mod_pow2): Update.
13556 (gimple_mod_subtract): Update.
13557 (gimple_ic_transform): Update.
13558 (gimple_stringop_fixed_value): Update.
13559 * value-prof.h (gimple_ic): Update.
13561 2017-06-02 Carl Love <cel@us.ibm.com>
13563 * config/rs6000/rs6000-c: Add support for built-in functions
13564 vector double vec_doublee (vector signed int);
13565 vector double vec_doublee (vector unsigned int);
13566 vector double vec_doublee (vector float);
13567 vector double vec_doubleh (vector signed int);
13568 vector double vec_doubleh (vector unsigned int);
13569 vector double vec_doubleh (vector float);
13570 vector double vec_doublel (vector signed int);
13571 vector double vec_doublel (vector unsigned int);
13572 vector double vec_doublel (vector float);
13573 vector double vec_doubleo (vector signed int);
13574 vector double vec_doubleo (vector unsigned int);
13575 vector double vec_doubleo (vector float);.
13576 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13577 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13579 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13580 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13581 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13583 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13584 vec_doublel, vec_doubleh.
13585 * doc/extend.texi: Update the built-in documentation file for the
13586 new built-in functions.
13588 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13591 * ipa-inline-analysis.c (free_growth_caches): Set
13592 edge_removal_hook_holder to NULL after removing it.
13594 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13596 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13597 comparision with zero.
13599 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13600 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13601 for early expansion of vec_min and vec_max builtins.
13602 (builtin_function_type): Add min/max unsigned variants to those
13603 identified as having unsigned arguments.
13605 2017-06-02 Olivier Hainque <hainque@adacore.com>
13607 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13609 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13611 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13612 Use VALL_F16 iterator rather than VALL.
13614 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13616 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13617 Emit CBNZ inside loop when doing a strong exchange and comparing
13618 against zero. Generate the CC flags after the loop.
13620 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13622 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13623 (dl_section_ref): New.
13624 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13625 On AIX, append an expression to subtract the size of the
13626 section length to dl_section_ref.
13628 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13631 for early expansion of vector absolute builtins.
13633 2017-06-02 Richard Biener <rguenther@suse.de>
13635 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13636 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13638 2017-06-02 Richard Biener <rguenther@suse.de>
13640 PR tree-optimization/80948
13641 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13642 stmt order as well.
13644 2017-06-02 Richard Biener <rguenther@suse.de>
13646 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13648 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13649 for inductions as relevant.
13651 2017-06-02 Richard Biener <rguenther@suse.de>
13653 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13654 (vectorizable_induction): ... this. Remove dead code.
13656 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13658 * builtins. (expand_builtin_alloca): Remove second parameter and
13659 infer its value from the first parameter instead.
13660 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13662 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13664 PR rtl-optimization/80903
13665 * loop-doloop.c (add_test): Unshare sequence.
13667 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13669 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13671 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13673 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13675 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13676 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13677 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13678 (xlogue_layout::get_stub_rtx): Make static.
13679 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13680 (xlogue_layout::compute_stub_managed_regs): Rename to...
13681 (xlogue_layout::count_stub_managed_regs): ...this.
13682 (xlogue_layout::is_stub_managed_reg): New function.
13683 (xlogue_layout::m_stub_names): Rename to...
13684 (xlogue_layout::s_stub_names): ...this, make static.
13685 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13686 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13687 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13688 xlogue_layout::s_stub_names): Instantiate statics.
13689 (stub_managed_regs): Remove.
13690 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13691 (disable_call_ms2sysv_xlogues): Rename to...
13692 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13693 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13695 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13696 change after reload_completed.
13697 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13699 (ix86_expand_prologue): Likewise.
13700 (ix86_expand_epilogue): Likewise.
13701 (ix86_expand_split_stack_prologue): Likewise.
13702 (ix86_compute_frame_layout): Remove frame parameter ...
13703 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13704 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13706 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13707 (ix86_frame): Move from here ...
13708 * config/i386/i386.h (ix86_frame): ... to here.
13709 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13710 complete ix86_frame data structure instead. Remove some_ld_name.
13712 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13714 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13715 symbols that hold a DECL_VALUE_EXPR.
13717 2017-06-01 Martin Jambor <mjambor@suse.cz>
13719 PR tree-optimization/80898
13720 * tree-sra.c (process_subtree_disqualification): Removed.
13721 (disqualify_candidate): Do not acll
13722 process_subtree_disqualification.
13723 (subtree_mark_written_and_enqueue): New function.
13724 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13725 RHS has been disqualified and re-queue LHS if necessary. Apart
13726 from that, ignore disqualified RHS.
13728 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13730 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13731 address fetch for z10 or later.
13733 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13735 * config/arc/arc.md (tst_movb): Add guard when splitting.
13737 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13739 * config/arc/arc.c (arc_can_eliminate): Test against
13740 arc_frame_pointer_needed.
13742 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13744 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13745 to prevent store reordering.
13746 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13747 (type): Add block type.
13748 (stack_tie): Define special instruction to be used in
13751 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13753 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13754 constraint. It is not valid for the pattern.
13755 (noncommutative_binary_comparison): Likewise.
13757 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13759 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13762 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13764 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13765 be used by the reg-alloc.
13767 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13769 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13770 reg-alloc when having mul64 or mul32x16 instructions.
13771 (mulsidi3): Likewise.
13772 (umulsidi3): Likewise.
13773 (mulsi32x16): New pattern.
13774 (mulsi64): Likewise.
13775 (mulsidi64): Likewise.
13776 (umulsidi64): Likewise.
13777 (MUL32x16_REG): Define.
13778 (mul64_600): Use MUL32x16_REG.
13779 (mac64_600): Likewise.
13780 (umul64_600): Likewise.
13781 (umac64_600): Likewise.
13783 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13785 * config/arc/arc.md (mulsi3_700): Make it commutative.
13787 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13789 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13791 (*sign_extendsidi2_insn): Likewise for movstosw.
13793 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13795 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13796 the type of the input discriminant value. Convert the
13797 discriminant value of signedness vary.
13799 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13801 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13802 Add to -Wall section.
13804 2017-06-01 Richard Biener <rguenther@suse.de>
13806 PR middle-end/66313
13807 * fold-const.c (fold_plusminus_mult_expr): If the factored
13808 factor may be zero use a wrapping type for the inner operation.
13809 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13810 and handle moved defs.
13811 (process_assignment): Properly guard the unary op case. Return a
13812 tri-state indicating that moving the stmt before the call may allow
13813 to continue. Pass through to_move.
13814 (find_tail_calls): Handle moving unrelated defs before
13817 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13820 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13821 splitter result in the canonical way.
13823 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13825 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13826 also for 32bit target. Update insn attributes.
13827 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13829 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13831 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13832 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13833 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13834 (_mm_maskz_min_ss): New intrinsics.
13836 2017-05-31 Martin Liska <mliska@suse.cz>
13838 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13839 Change comment style to one we normally use.
13840 (vectorizable_reduction): Likewise.
13841 (vectorizable_induction): Likewise.
13842 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13843 (vectorizable_call): Likewise.
13844 (vectorizable_simd_clone_call): Likewise.
13845 (vectorizable_conversion): Likewise.
13846 (vectorizable_assignment): Likewise.
13847 (vectorizable_shift): Likewise.
13848 (vectorizable_operation): Likewise.
13849 (vectorizable_store): Likewise.
13850 (vectorizable_load): Likewise.
13851 * tree-vectorizer.h: Likewise.
13853 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13855 * passes.c (emergency_dump_function): New.
13856 * tree-pass.h (emergency_dump_function): Declare.
13857 * plugin.c (plugins_internal_error_function): Remove.
13858 * plugin.h (plugins_internal_error_function): Remove declaration.
13859 * toplev.c (internal_error_function): New static function. Use it...
13860 (general_init): ...here.
13862 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13864 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13865 (arc_rtx_costs): Add costs for new patterns.
13866 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13867 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13869 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13871 * tree-ssa-strlen.c (get_next_strinfo): New function.
13872 (get_stridx_plus_constant): Use it.
13873 (zero_length_string): Likewise.
13874 (adjust_related_strinfos): Likewise.
13875 (adjust_last_stmt): Likewise.
13877 2017-05-31 Richard Biener <rguenther@suse.de>
13880 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13881 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13883 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13885 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13886 loop_vinfo argument and use of dependence distance vectors.
13887 Check instead whether the two references differ only in their
13888 initial value and assume that they have the same alignment if the
13889 difference is a multiple of the vector alignment.
13890 (vect_analyze_data_refs_alignment): Update call accordingly.
13892 2017-05-31 Martin Liska <mliska@suse.cz>
13895 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13897 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13899 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13900 (create_intersect_range_checks): Move from ...
13901 * tree-data-ref.c (create_intersect_range_checks_index)
13902 (create_intersect_range_checks): ... to here.
13903 (create_runtime_alias_checks): New function factored from ...
13904 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13905 here. Call above function.
13906 * tree-data-ref.h (create_runtime_alias_checks): New function.
13908 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13910 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13911 segment length for dr_b and compute it in wide_int.
13913 2017-05-31 Richard Biener <rguenther@suse.de>
13915 PR tree-optimization/80906
13916 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13917 and pass through iv_map.
13918 (copy_bb_and_scalar_dependences): Adjust.
13919 (translate_pending_phi_nodes): Likewise.
13920 (copy_loop_close_phi_args): Handle code-generating IVs instead
13923 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13925 * diagnostic-color.c (color_dict): Add "type-diff".
13926 (parse_gcc_colors): Update comment.
13927 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13928 -fdiagnostics-show-template-tree and -fno-elide-type.
13929 (GCC_COLORS): Add type-diff to example.
13931 (-fdiagnostics-show-template-tree): New.
13932 (-fno-elide-type): New.
13933 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13934 the pp_format_decoder callback. Call any m_format_postprocessor's
13936 (pretty_printer::pretty_printer): Initialize
13937 m_format_postprocessor.
13938 (pretty_printer::~pretty_printer): Delete any
13939 m_format_postprocessor.
13940 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13941 (class format_postprocessor): New class.
13942 (struct pretty_printer::format_decoder): Document the new parameters.
13943 (struct pretty_printer::m_format_postprocessor): New field.
13944 * tree-diagnostic.c (default_tree_printer): Update for new
13945 bool and const char ** params.
13946 * tree-diagnostic.h (default_tree_printer): Likewise.
13948 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13950 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13951 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13953 rs6000_gen_cell_microcode code.
13954 (rs6000_final_prescan_insn): Delete.
13955 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13956 "warn-cell-microcode" entries.
13957 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13958 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13959 throughout. Change cc_reg_not_micro_cr0_operand to
13960 cc_reg_not_cr0_operand throughout.
13961 (*extendhi<mode>2_noload): Delete.
13962 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13963 (mwarn-cell-microcode): Delete.
13964 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13965 -mgen-cell-microcode and -mwarn-cell-microcode.
13967 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13970 * config/i386/constraints.md (Yd): New constraint.
13972 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13973 and (?Yd, r) alternatives. Update insn attributes.
13974 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13975 and (?*Yd, r) alternatives. Update insn attributes.
13976 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13978 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13980 * gimplify.c (gimplify_modify_expr): Don't create a
13981 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13984 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13986 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13988 2017-05-30 Richard Biener <rguenther@suse.de>
13990 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13991 and reduc_def fields.
13992 (STMT_VINFO_REDUC_TYPE): New define.
13993 (STMT_VINFO_REDUC_DEF): Likewise.
13994 (vect_force_simple_reduction): Adjust prototype.
13995 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13996 (vect_is_simple_reduction): Remove check_reduction argument.
13997 (vect_force_simple_reduction): Adjust and set
13998 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13999 (vectorizable_reduction): Do not re-do reduction analysis
14000 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14001 * tree-parloops.c (gather_scalar_reductions): Adjust.
14003 2017-05-30 Richard Biener <rguenther@suse.de>
14005 PR middle-end/80901
14006 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14009 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14011 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14012 Introduce unknown_misalignment parameter and remove vf.
14013 (vect_peeling_hash_get_lowest_cost):
14014 Pass unknown_misalignment parameter.
14015 (vect_enhance_data_refs_alignment):
14016 Fix unsupportable data ref treatment.
14018 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14020 * tree-vect-data-refs.c (vect_get_data_access_cost):
14021 Workaround for SLP handling.
14022 (vect_enhance_data_refs_alignment):
14023 Compute costs for doing no peeling at all, compare to the best
14024 peeling costs so far and avoid peeling if cheaper.
14026 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14028 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14029 Return peeling info and set costs to zero for unlimited cost
14031 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14032 with unknown misalignment. Compute and costs for unknown
14033 misalignment, compare them to the costs for known misalignment
14034 and choose the cheapest for peeling.
14036 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14038 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14039 (vect_get_peeling_costs_all_drs): Create function.
14040 (vect_peeling_hash_get_lowest_cost):
14041 Use vect_get_peeling_costs_all_drs.
14042 (vect_peeling_supportable): Create function.
14043 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14045 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14047 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14048 DR_HAS_NEGATIVE_STEP.
14049 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14050 (vect_enhance_data_refs_alignment): Use.
14051 (vect_duplicate_ssa_name_ptr_info): Use.
14052 * tree-vectorizer.h (dr_misalignment): Use.
14053 (known_alignment_for_access_p): Use.
14055 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14058 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14060 (has_section_name): New function.
14062 2017-05-30 Martin Liska <mliska@suse.cz>
14065 * auto-profile.c (get_function_decl_from_block): Fix
14068 2017-05-30 Richard Biener <rguenther@suse.de>
14070 PR middle-end/80876
14071 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14073 2017-05-30 Martin Liska <mliska@suse.cz>
14075 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14076 * dumpfile.h (struct dump_file_info): Remove ctors.
14078 2017-05-30 Martin Liska <mliska@suse.cz>
14080 * predict.def: Fix GNU coding style.
14082 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14084 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14085 Mark 'to' argument with ATTRIBUTE_UNUSED.
14087 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14089 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14090 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14091 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14094 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14096 * doc/install.texi (Options specification): Restore entry of
14097 --enable-sjlj-exceptions.
14099 2017-05-27 Michael Eager <eager@eagercon.com>
14102 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14104 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14106 * config/microblaze/microblaze.h
14107 (FIXED_REGISTERS): Update in macro.
14108 (CALL_USED_REGISTERS): Update in macro.
14110 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14112 * doc/install.texi: Add links to macOS binary distributions.
14114 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14118 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14120 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14122 2017-05-26 Martin Liska <mliska@suse.cz>
14124 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14126 2017-05-26 Martin Liska <mliska@suse.cz>
14128 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14129 always leading ';; '.
14130 (dump_bb_info): Likewise.
14131 (brief_dump_cfg): Likewise.
14132 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14133 * dumpfile.c: Remove usage of TDF_VERBOSE.
14134 * dumpfile.h (enum dump_kind): Likewise.
14135 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14136 * print-tree.c (debug_verbose): Remove.
14137 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14138 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14139 * tree-diagnostic.c (default_tree_printer): Replace
14140 TDF_DIAGNOSTIC with TDF_SLIM.
14142 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14144 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14145 in parameter loop, rather than loop_vinfo.
14146 (create_intersect_range_checks): Ditto.
14147 (vect_create_cond_for_alias_checks): Update call to above functions.
14149 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14151 PR tree-optimization/80815
14152 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14153 for merging runtime alias checks. Handle negative DR_STEPs.
14155 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14157 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14159 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14160 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14161 out code pruning runtime alias checks.
14162 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14163 factored out from above.
14164 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14166 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14168 (prune_runtime_alias_test_list): New decalaration.
14170 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14172 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14173 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14174 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14175 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14176 (operator==, comp_dr_with_seg_len_pair): Ditto.
14177 (vect_prune_runtime_alias_test_list): Ditto.
14179 2017-05-26 Martin Liska <mliska@suse.cz>
14182 * params.def: Bound partial-inlining-entry-probability param.
14184 2017-05-26 Marek Polacek <polacek@redhat.com>
14187 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14190 2017-05-26 Richard Biener <rguenther@suse.de>
14192 PR tree-optimization/80842
14193 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14196 2017-05-26 Richard Biener <rguenther@suse.de>
14198 PR tree-optimization/80844
14199 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14201 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14203 * doc/md.texi (Machine Constraints): Update x86 family
14204 machine constraints section to match 'config/i386/constraints.md'.
14206 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14208 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14210 2017-05-25 Nathan Sidwell <nathan@acm.org>
14212 * doc/invoke.texi (--enable-languages): Update documentation.
14214 2017-05-25 Martin Liska <mliska@suse.cz>
14216 * dumpfile.c: Add TDF_FOLDING.
14217 * dumpfile.h (enum dump_kind): Likewise.
14218 * genmatch.c (dt_simplify::gen_1): Use it.
14220 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14222 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14224 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14226 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14227 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14229 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14231 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14232 * match.pd (X == C): Rewrite it here.
14233 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14234 with_certain_nonzero_bits2): New predicates.
14235 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14237 2017-05-24 Nathan Sidwell <nathan@acm.org>
14239 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14242 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14245 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14247 * config/powerpcspe: New port. Files are copied from the rs6000
14248 port, with "rs6000" in filenames replaced by "powerpcspe".
14250 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14252 PR rtl-optimization/80754
14253 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14255 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14257 * config/sparc/sparc.md (length): Return the correct value for -mflat
14258 sibcalls to match output_sibcall.
14260 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14264 * config/rs6000/rs6000.c (struct machine_function): Add new field
14266 (rs6000_get_separate_components): Init that field, use it.
14267 (rs6000_components_for_bb): Use the field.
14269 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14271 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14273 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14275 PR middle-end/80823
14276 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14278 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14281 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14282 address against address_operand predicate.
14283 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14285 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14287 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14288 parameters passed indirectly.
14290 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14292 * config/i386/i386.md (*movdi_internal): Remove SSE4
14293 alternative 18 (?r, *v). Update insn attributes.
14294 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14295 Update insn attributes.
14296 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14297 Update insn attributes.
14298 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14299 alternative 1 (r, v). Remove isa attribute.
14300 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14301 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14302 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14304 2017-05-23 Tom de Vries <tom@codesourcery.com>
14306 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14309 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14311 * cgraphunit.c (symbol_table::process_new_functions): Update.
14312 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14313 (inline_generate_summary): Rename to ...
14314 (ipa_fn_summary_generate): ... this one.
14315 (inline_read_summary): Rename to ...
14316 (ipa_fn_summary_read): ... this one.
14317 (inline_write_summary): Rename to ...
14318 (ipa_fn_summary_write): ... this one.
14319 (inline_free_summary): Rename to ...
14320 (ipa_free_fn_summary): ... this one.
14321 (pass_data_local_fn_summary, pass_local_fn_summary,
14322 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14323 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14324 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14325 make_pass_ipa_fn_summary): New.
14326 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14327 inline_write_summary, inline_free_summary): Remove.
14328 (ipa_free_fn_summary) : New.
14329 * ipa-inline.c (ipa_inline): Update.
14330 (pass_ipa_inline): Do not generate summaries.
14331 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14333 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14334 and add pass_ipa_fn_summary.
14335 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14337 (make_pass_inline_parameters): Remove.
14339 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14341 * omp-low.c (struct omp_context): Remove "default_kind" member.
14344 * omp-offload.c (execute_oacc_device_lower): Remove the
14345 parallelism dimensions function attributes for unparallelized
14346 OpenACC kernels constructs.
14348 2017-05-23 Martin Liska <mliska@suse.cz>
14350 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14352 (cgraph_edge::make_speculative): Likewise.
14353 (cgraph_edge::resolve_speculation): Likewise.
14354 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14355 (cgraph_node::dump): Likewise.
14356 * cgraph.h: Likewise.
14357 * cgraphunit.c (analyze_functions): Likewise.
14358 (symbol_table::compile): Likewise.
14359 * ipa-cp.c (print_all_lattices): Likewise.
14360 (determine_versionability): Likewise.
14361 (initialize_node_lattices): Likewise.
14362 (ipcp_verify_propagated_values): Likewise.
14363 (estimate_local_effects): Likewise.
14364 (update_profiling_info): Likewise.
14365 (create_specialized_node): Likewise.
14366 (perhaps_add_new_callers): Likewise.
14367 (decide_about_value): Likewise.
14368 (decide_whether_version_node): Likewise.
14369 (identify_dead_nodes): Likewise.
14370 (ipcp_store_bits_results): Likewise.
14371 * ipa-devirt.c (dump_targets): Likewise.
14372 (ipa_devirt): Likewise.
14373 * ipa-icf.c (sem_item::dump): Likewise.
14374 (sem_function::equals): Likewise.
14375 (sem_variable::equals): Likewise.
14376 (sem_item_optimizer::read_section): Likewise.
14377 (sem_item_optimizer::execute): Likewise.
14378 (congruence_class::dump): Likewise.
14379 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14380 (dump_inline_summary): Likewise.
14381 (estimate_node_size_and_time): Likewise.
14382 (inline_analyze_function): Likewise.
14383 * ipa-inline-transform.c (inline_call): Likewise.
14384 * ipa-inline.c (report_inline_failed_reason): Likewise.
14385 (want_early_inline_function_p): Likewise.
14386 (edge_badness): Likewise.
14387 (update_edge_key): Likewise.
14388 (inline_small_functions): Likewise.
14389 * ipa-profile.c (ipa_profile): Likewise.
14390 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14391 (ipa_make_edge_direct_to_target): Likewise.
14392 (remove_described_reference): Likewise.
14393 (ipa_impossible_devirt_target): Likewise.
14394 (propagate_controlled_uses): Likewise.
14395 (ipa_print_node_params): Likewise.
14396 (ipcp_transform_function): Likewise.
14397 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14398 (propagate_pure_const): Likewise.
14399 * ipa-reference.c (generate_summary): Likewise.
14400 (read_write_all_from_decl): Likewise.
14401 (propagate): Likewise.
14402 (ipa_reference_read_optimization_summary): Likewise.
14403 * ipa-utils.c (ipa_merge_profiles): Likewise.
14404 * ipa.c (walk_polymorphic_call_targets): Likewise.
14405 (symbol_table::remove_unreachable_nodes): Likewise.
14406 (ipa_single_use): Likewise.
14407 * passes.c (execute_todo): Likewise.
14408 * predict.c (drop_profile): Likewise.
14409 * symtab.c (symtab_node::get_dump_name): New function.
14410 (symtab_node::dump_name): Likewise.
14411 (symtab_node::dump_asm_name): Likewise.
14412 (symtab_node::dump_references): Likewise.
14413 (symtab_node::dump_referring): Likewise.
14414 (symtab_node::dump_base): Likewise.
14415 (symtab_node::debug_symtab): Likewise.
14416 * tree-sra.c (convert_callers_for_node): Likewise.
14417 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14418 * value-prof.c (init_node_map): Likewise.
14420 2017-05-23 Martin Liska <mliska@suse.cz>
14422 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14423 and symtab_node::debug_symtab to symbol_table::debug.
14424 * cgraphunit.c (analyze_functions): Use the renamed function.
14425 (symbol_table::compile): Likewise.
14426 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14427 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14428 * passes.c (execute_todo): Likewise.
14429 * symtab.c (symbol_table::dump): New function.
14430 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14432 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14434 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14435 that nonconst implies exec.
14437 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14439 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14440 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14441 (inline_edge_summary_vec): Turn into ...
14442 (ipa_call_summaries): ... this one.
14443 (redirect_to_unreachable, edge_set_predicate,
14444 evaluate_properties_for_edge, inline_summary_alloc,
14445 reset_ipa_call_summary, reset_inline_summary,
14446 inline_summary_t::duplicate): Update.
14447 (inline_edge_duplication_hook): Turn to ...
14448 (ipa_call_summary_t::duplicate): ... this one.
14449 (inline_edge_removal_hook): Turn to ...
14450 (ipa_call_summary_t::remove): ... this one.
14451 (dump_inline_edge_summary): Turn to ...
14452 (dump_ipa_call_summary): ... this one.
14453 (estimate_function_body_sizes): Update.
14454 (inline_update_callee_summaries): Update.
14455 (remap_edge_change_prob): Update.
14456 (remap_edge_summaries): Update.
14457 (inline_merge_summary): Update.
14458 (do_estimate_edge_time): Update.
14459 (inline_generate_summary): Update.
14460 (inline_read_section): Update.
14461 (inline_read_summary): Update.
14462 (inline_free_summary): Update.
14463 * ipa-inline.c (can_inline_edge_p): Update.
14464 (compute_inlined_call_time): Update.
14465 (want_inline_small_function_p): Update.
14466 (edge_badness): Update.
14467 (early_inliner): Update.
14468 * ipa-inline.h (inline_edge_summary): Turn to ...
14469 (ipa_call_summary): ... this one.
14470 (ipa_call_summary_t): New class.
14471 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14472 (ipa_call_summaries): New.
14473 (inline_edge_summary): Remove.
14474 (estimate_edge_growth): Update.
14475 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14476 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14477 * ipa-split.c (execute_split_functions): Update.
14478 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14480 2017-05-23 Tom de Vries <tom@codesourcery.com>
14482 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14483 attributes): Document rdrand effective target.
14485 2017-05-23 Tom de Vries <tom@codesourcery.com>
14487 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14488 attributes): Sort alphabetically.
14490 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14492 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14494 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14497 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14498 V2DF/V2DI splat into two separate patterns, one that handles
14499 registers, and the other that only handles memory. Drop support
14500 for splatting from a GPR on ISA 2.07 and then splitting the
14501 splat into direct move and splat.
14502 (vsx_splat_<mode>_reg): Likewise.
14503 (vsx_splat_<mode>_mem): Likewise.
14505 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14507 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14509 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14511 PR middle-end/80809
14512 * omp-low.c (finish_taskreg_remap): New function.
14513 (finish_taskreg_scan): If unit size of ctx->record_type
14514 is non-constant, unshare the size expression and replace
14515 decls in it with possible outer var refs.
14517 PR middle-end/80809
14518 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14519 GOVD_SHARED rather than GOVD_PRIVATE with it.
14520 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14521 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14523 PR middle-end/80853
14524 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14525 as last argument to build_outer_var_ref for pointer bases of array
14526 section reductions.
14528 2017-05-19 Martin Sebor <msebor@redhat.com>
14530 * print-tree.c (print_node): Print DECL_READ_P flag.
14532 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14534 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14535 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14536 * cgraph.c: Likewise.
14537 * cgraphunit.c: Likewise.
14538 * gengtype.c: Likewise.
14539 * ipa-cp.c: Likewise.
14540 * ipa-devirt.c: Likewise.
14541 * ipa-icf.c: Likewise.
14542 * ipa-predicate.c: Likewise.
14543 * ipa-profile.c: Likewise.
14544 * ipa-prop.c: Likewise.
14545 * ipa-split.c: Likewise.
14547 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14548 edge_predicate_pool, dump_inline_hints,
14549 inline_summary::account_size_time, redirect_to_unreachable,
14550 edge_set_predicate, set_hint_predicate,
14551 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14552 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14553 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14554 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14555 ipa_call_summary_t::remove, initialize_growth_caches,
14556 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14557 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14558 mark_modified, unmodified_parm_1, unmodified_parm,
14559 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14560 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14561 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14562 will_be_nonconstant_predicate, record_modified_bb_info,
14563 get_minimal_bb, record_modified, param_change_prob,
14564 phi_result_unknown_predicate, predicate_for_phi_result,
14565 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14566 estimate_function_body_sizes, compute_inline_parameters,
14567 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14568 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14569 inline_update_callee_summaries, remap_edge_change_prob,
14570 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14571 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14572 inline_analyze_function, inline_summary_t::insert,
14573 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14574 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14575 inline_free_summary): Move to ipa-fnsummary.h
14576 (predicate_t): Remove.
14577 * ipa-fnsummary.c: New file.
14578 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14579 (enum inline_hints_vals, inline_hints, agg_position_info,
14580 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14581 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14582 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14583 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14584 inline_read_summary, inline_write_summary, inline_free_summary,
14585 inline_analyze_function, initialize_inline_failed,
14586 inline_merge_summary, inline_update_overall_summary,
14587 compute_inline_parameters): Move to ipa-fnsummary.h
14588 * ipa-fnsummary.h: New file.
14589 * ipa-inline-transform.h: Include ipa-inline.h.
14590 * ipa-inline.c: LIkewise.
14592 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14594 * ipa-inline.c (edge_badness): Use inlined_time instead of
14595 inline_summaries->get.
14597 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14599 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14601 2017-05-22 Nathan Sidwell <nathan@acm.org>
14603 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14604 (fdump-lang): Document 'raw' option.
14605 * dumpfile.h (TDI_tu): Delete.
14606 * dumpfile.c (dump_files): Remove translation-unit.
14607 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14609 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14611 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14612 command option from $(AWK) call.
14613 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14614 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14615 [FORMAT]: Remove handling of variable.
14616 * config/avr/t-multilib: Regenerate.
14618 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14620 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14622 (dump_inline_summary): Do not print self_time.
14623 (estimate_function_body_sizes): Do not set self_time.
14624 (compute_inline_parameters): Likewise.
14625 (inline_read_section, inline_write_summary): Do not stream self_time.
14626 * ipa-inline.h (inline_summary): Drop self_time.
14628 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14630 * ipa-inline-analysis.c (account_size_time): Rename to ...
14631 (inline_summary::account_size_time): ... this one.
14632 (reset_ipa_call_summary): Turn to ...
14633 (ipa_call_summary::reset): ... this one.
14634 (reset_inline_summary): Turn to ...
14635 (inline_summary::reset): ... this one.
14636 (inline_summary_t::remove): Update.
14637 (inline_summary_t::duplicate): Update.
14638 (ipa_call_summary_t::remove): Update.
14639 (dump_inline_summary): Update.
14640 (estimate_function_body_sizes): Update.
14641 (compute_inline_parameters): Update.
14642 (estimate_node_size_and_time): Update.
14643 (inline_merge_summary): Update.
14644 (inline_update_overall_summary): Update.
14645 (inline_read_section): Update.
14646 (inline_write_summary): Update.
14647 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14648 add account_size_time and reset member functions.
14649 (ipa_call_summary): Add reset function.
14650 * ipa-predicate.h (predicate::operator &): Constify.
14652 2017-05-22 Richard Biener <rguenther@suse.de>
14654 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14656 2017-05-19 Jason Merrill <jason@redhat.com>
14658 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14660 2017-05-19 Marek Polacek <polacek@redhat.com>
14663 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14664 TYPE_OVERFLOW_WRAPS checks.
14666 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14668 * tree-core.h (enum omp_clause_default_kind): Add
14669 "OMP_CLAUSE_DEFAULT_PRESENT".
14670 * tree-pretty-print.c (dump_omp_clause): Handle it.
14671 * gimplify.c (enum gimplify_omp_var_data): Add
14672 "GOVD_MAP_FORCE_PRESENT".
14673 (gimplify_adjust_omp_clauses_1): Map it to
14674 "GOMP_MAP_FORCE_PRESENT".
14675 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14677 * gimplify.c (oacc_default_clause): Clarify.
14679 2017-05-19 Nathan Sidwell <nathan@acm.org>
14681 LANG_HOOK_REGISTER_DUMPS
14682 * toplev.c (general_init): Call register dump lang hook.
14683 * doc/invoke.texi: Document -fdump-lang option family.
14684 * dumpfile.c (dump_files): Remove class dump here.
14685 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14686 * dumpfile.h (tree_dump_index): Remove TDI_class.
14687 * langhooks-def.h (lhd_register_dumps): Declare.
14688 (LANG_HOOKS_REGISTER_DUMPS): Define.
14689 (LANG_HOOKS_INITIALIZER): Add it.
14690 * langhooks.c (lhd_register_dumps): Define.
14691 * langhooks.h (struct lang_hooks): Add register_dumps.
14693 2017-05-19 Nathan Sidwell <nathan@acm.org>
14695 * context.h (context::set_passes): New.
14696 * context.c (context::context): Do not create pass manager.
14697 * toplev.c (general_init): Create pass manager here.
14699 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14701 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14702 use this splitter if two add or or instructions would also work for
14703 the constant we want to generate.
14705 2017-05-19 Richard Biener <rguenther@suse.de>
14708 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14709 predicate evaluation.
14711 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14713 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14715 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14718 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14720 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14721 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14722 (inline_edge_summary_vec): Turn into ...
14723 (ipa_call_summaries): ... this one.
14724 (redirect_to_unreachable, edge_set_predicate,
14725 evaluate_properties_for_edge, inline_summary_alloc,
14726 reset_ipa_call_summary, reset_inline_summary,
14727 inline_summary_t::duplicate): Update.
14728 (inline_edge_duplication_hook): Turn to ...
14729 (ipa_call_summary_t::duplicate): ... this one.
14730 (inline_edge_removal_hook): Turn to ...
14731 (ipa_call_summary_t::remove): ... this one.
14732 (dump_inline_edge_summary): Turn to ...
14733 (dump_ipa_call_summary): ... this one.
14734 (estimate_function_body_sizes): Update.
14735 (inline_update_callee_summaries): Update.
14736 (remap_edge_change_prob): Update.
14737 (remap_edge_summaries): Update.
14738 (inline_merge_summary): Update.
14739 (do_estimate_edge_time): Update.
14740 (inline_generate_summary): Update.
14741 (inline_read_section): Update.
14742 (inline_read_summary): Update.
14743 (inline_free_summary): Update.
14744 * ipa-inline.c (can_inline_edge_p): Update.
14745 (compute_inlined_call_time): Update.
14746 (want_inline_small_function_p): Update.
14747 (edge_badness): Update.
14748 (early_inliner): Update.
14749 * ipa-inline.h (inline_edge_summary): Turn to ...
14750 (ipa_call_summary): ... this one.
14751 (ipa_call_summary_t): New class.
14752 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14753 (ipa_call_summaries): New.
14754 (inline_edge_summary): Remove.
14755 (estimate_edge_growth): Update.
14756 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14757 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14758 * ipa-split.c (execute_split_functions): Update.
14759 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14761 2017-05-19 Richard Biener <rguenther@suse.de>
14763 PR middle-end/80764
14764 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14766 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14768 * config/rs6000/rs6000.c (struct machine_function): Add field
14769 fpr_is_wrapped_separately.
14770 (rs6000_get_separate_components): Use 64 components. Handle the
14771 new FPR components.
14772 (rs6000_components_for_bb): Handle the FPR components.
14773 (rs6000_emit_prologue_components): Handle the FPR components.
14774 (rs6000_emit_epilogue_components): Handle the FPR components.
14775 (rs6000_set_handled_components): Handle the FPR components.
14776 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14777 that are already separately shrink-wrapped.
14778 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14779 that are already separately shrink-wrapped.
14781 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14784 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14787 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14788 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14789 cases where the register allocator uses a move and an offsettable
14790 memory operation to/from a FPR register on ISA 2.06/2.07.
14791 (define_peephole2 for Altivec d-form store): Likewise.
14793 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14796 * config/i386/mmx.md (*mov<mode>_internal): Enable
14797 alternatives 11, 12, 13 and 14 also for 32bit targets.
14798 Remove alternatives 15, 16, 17 and 18.
14799 * config/i386/sse.md (vec_concatv2di): Change
14800 alternative (!x, *y) to (x, ?!*Yn).
14802 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14804 * dumpfile.h (enum dump_kind): Remove stray comma.
14806 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14808 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14809 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14810 predicate::num_conditions
14811 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14812 (CHANGED): turn into predicate::changed.
14813 (agg_position_info): Move to ipa-predicate.h
14814 (add_condition, predicate::add_clause, predicate::operator &=,
14815 predicate::or_with, predicate::evaluate, predicate::probability,
14816 dump_condition, dump_clause, predicate::dump,
14817 predicate::remap_after_duplication, predicate::remap_after_inlining,
14818 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14819 (evaluate_conditions_for_known_args): Update.
14820 (set_cond_stmt_execution_predicate): Update.
14821 * ipa-inline.h: Include ipa-predicate.h
14822 (condition, inline_param_summary, conditions, agg_position_info,
14823 predicate): Move to ipa-predicate.h
14824 * ipa-predicate.c: New file.
14825 * ipa-predicate.h: New file.
14827 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14829 * final.c (leaf_function_p): Check we are not in a sequence.
14831 2017-05-18 Martin Liska <mliska@suse.cz>
14833 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14834 * dumpfile.c (dump_register): Use new enum dump_kind.
14835 (get_dump_file_name): Likewise.
14836 (dump_enable_all): Likewise.
14837 (dump_switch_p_1): Likewise.
14838 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14839 * dumpfile.h (enum dump_kind): New enum type.
14840 (struct dump_file_info): Create constructor and
14841 format fields and comments.
14842 * passes.c (pass_manager::register_one_dump_file):
14844 * statistics.c (statistics_early_init): Likewise.
14845 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14846 TDF_TREE with TDF_SLIM.
14847 (gather_memory_references_ref): Likewise.
14849 2017-05-18 Martin Liska <mliska@suse.cz>
14851 * vec.h (struct vnull): Use it.
14853 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14855 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14856 (true_predicate, false_predicate, true_predicate_p,
14857 false_predicate_p): Remove.
14858 (single_cond_predicate, not_inlined_predicate): Turn to member function
14860 (add_condition): Update.
14861 (add_clause): Turn to...
14862 (predicate::add_clause): ... this one; update; allow passing NULL
14864 (and_predicates): Turn to ...
14865 (predicate::operator &=): ... this one.
14866 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14867 (or_predicates): Turn to ...
14868 (predicate::or_with): ... this one.
14869 (evaluate_predicate): Turn to ...
14870 (predicate::evaluate): ... this one.
14871 (predicate_probability): Turn to ...
14872 (predicate::probability): ... this one.
14873 (dump_condition): Update.
14874 (dump_predicate): Turn to ...
14875 (predicate::dump): ... this one.
14876 (account_size_time): Update.
14877 (edge_set_predicate): Update.
14878 (set_hint_predicate): UPdate.
14879 (evaluate_conditions_for_known_args): Update.
14880 (evaluate_properties_for_edge): Update.
14881 (remap_predicate_after_duplication): Turn to...
14882 (predicate::remap_after_duplication): ... this one.
14883 (remap_hint_predicate_after_duplication): Update.
14884 (inline_summary_t::duplicate): UPdate.
14885 (dump_inline_edge_summary): Update.
14886 (dump_inline_summary): Update.
14887 (set_cond_stmt_execution_predicate): Update.
14888 (set_switch_stmt_execution_predicate): Update.
14889 (compute_bb_predicates): Update.
14890 (will_be_nonconstant_expr_predicate): Update.
14891 (will_be_nonconstant_predicate): Update.
14892 (phi_result_unknown_predicate): Update.
14893 (predicate_for_phi_result): Update.
14894 (array_index_predicate): Update.
14895 (estimate_function_body_sizes): Update.
14896 (estimate_node_size_and_time): Update.
14897 (estimate_ipcp_clone_size_and_time): Update.
14898 (remap_predicate): Rename to ...
14899 (predicate::remap_after_inlining): ... this one.
14900 (remap_hint_predicate): Update.
14901 (inline_merge_summary): Update.
14902 (inline_update_overall_summary): Update.
14903 (estimate_size_after_inlining): Update.
14904 (read_predicate): Rename to ...
14905 (predicate::stream_in): ... this one.
14906 (read_inline_edge_summary): Update.
14907 (write_predicate): Rename to ...
14908 (predicate::stream_out): ... this one.
14909 (write_inline_edge_summary): Update.
14910 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14911 (clause_t): Turn to uint32_t
14912 (predicate): Turn to class; implement constructor and operators
14914 (size_time_entry): Update.
14915 (inline_summary): Update.
14916 (inline_edge_summary): Update.
14918 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14920 * fold-const.c (fold_binary_loc): Move transformation...
14921 * match.pd (C - X CMP X): ... here.
14923 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14925 * config/sparc/sparc.c (sparc_option_override): Set function
14926 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14927 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14929 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14931 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14933 2017-05-18 Marek Polacek <polacek@redhat.com>
14936 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14937 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14938 gimple_assign_load_p.
14940 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14942 PR middle-end/80692
14943 * real.c (do_compare): Give decimal_do_compare preference over
14944 comparing just the signs.
14946 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14948 * doc/md.texi (Canonicalization of Instructions): Describe the
14949 canonical form of instructions that inherently set a condition
14952 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14954 PR middle-end/80775
14955 * tree-cfg.c: Move deletion of unreachable case statements to after
14956 the merging of consecutive case labels.
14958 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14960 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14961 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14962 restoring of callee-saved registers.
14964 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14966 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14967 * config/visium/visium.c (single_set_and_flags): Likewise.
14968 * config/visium/visium.md (Substitutions): Likewise.
14970 2017-05-17 Martin Liska <mliska@suse.cz>
14972 * cfg.c: Introduce dump_flags_t type and
14973 use it instead of int type.
14975 * cfghooks.c: Likewise.
14976 * cfghooks.h (struct cfg_hooks): Likewise.
14977 * cfgrtl.c: Likewise.
14978 * cfgrtl.h: Likewise.
14979 * cgraph.c (cgraph_node::get_body): Likewise.
14980 * coretypes.h: Likewise.
14981 * domwalk.c: Likewise.
14982 * domwalk.h: Likewise.
14983 * dumpfile.c (struct dump_option_value_info): Likewise.
14984 (dump_enable_all): Likewise.
14985 (dump_switch_p_1): Likewise.
14986 (opt_info_switch_p): Likewise.
14987 * dumpfile.h (enum tree_dump_index): Likewise.
14988 (struct dump_file_info): Likewise.
14989 * genemit.c: Likewise.
14990 * generic-match-head.c: Likewise.
14991 * gengtype.c (open_base_files): Likewise.
14992 * gimple-pretty-print.c: Likewise.
14993 * gimple-pretty-print.h: Likewise.
14994 * graph.c (print_graph_cfg): Likewise.
14995 * graphite-scop-detection.c (dot_all_sese): Likewise.
14996 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14997 * loop-unroll.c (report_unroll): Likewise.
14998 * passes.c (pass_manager::register_one_dump_file): Likewise.
14999 * print-tree.c: Likewise.
15000 * statistics.c: Likewise.
15001 * tree-cfg.c: Likewise.
15002 * tree-cfg.h: Likewise.
15003 * tree-dfa.c: Likewise.
15004 * tree-dfa.h: Likewise.
15005 * tree-dump.c (dump_function): Likewise.
15006 * tree-dump.h (struct dump_info): Likewise.
15007 * tree-pretty-print.c: Likewise.
15008 * tree-pretty-print.h: Likewise.
15009 * tree-ssa-live.c: Likewise.
15010 * tree-ssa-live.h: Likewise.
15011 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15012 * tree-vect-loop.c: Likewise.
15013 * tree-vect-slp.c: Likewise.
15015 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15016 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15018 PR tree-optimization/80457
15019 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15020 of all arguments to a statement as scalar_to_vec operations.
15021 (vectorizable_call): Adjust call to vect_model_simple_cost for
15023 (vectorizable_conversion): Likewise.
15024 (vectorizable_assignment): Likewise.
15025 (vectorizable_shift): Likewise.
15026 (vectorizable_operation): Likewise.
15027 (vectorizable_comparison): Likewise.
15028 (vect_is_simple_cond): Record the def types for operands.
15029 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15030 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15031 for statement argument count.
15033 2017-05-16 Carl Love <cel@us.ibm.com>
15035 * config/rs6000/rs6000-c: Add support for built-in functions
15036 vector unsigned long long vec_bperm (vector unsigned long long,
15037 vector unsigned char)
15038 vector signed long long vec_mule (vector signed int,
15040 vector unsigned long long vec_mule (vector unsigned int,
15041 vector unsigned int)
15042 vector signed long long vec_mulo (vector signed int,
15044 vector unsigned long long vec_mulo (vector unsigned int,
15045 vector unsigned int)
15046 vector signed char vec_sldw (vector signed char,
15047 vector signed char,
15049 vector unsigned char vec_sldw (vector unsigned char,
15050 vector unsigned char,
15052 vector signed short vec_sldw (vector signed short,
15053 vector signed short,
15055 vector unsigned short vec_sldw (vector unsigned short,
15056 vector unsigned short,
15058 vector signed int vec_sldw (vector signed int,
15061 vector unsigned int vec_sldw (vector unsigned int,
15062 vector unsigned int,
15064 vector signed long long vec_sldw (vector signed long long,
15065 vector signed long long,
15067 vector unsigned long long vec_sldw (vector unsigned long long,
15068 vector unsigned long long,
15070 * config/rs6000/rs6000-c: Add support for built-in functions
15071 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15072 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15073 * doc/extend.texi: Update the built-in documentation for the
15074 new built-in functions.
15076 2017-05-16 Marek Polacek <polacek@redhat.com>
15080 * tree.c (save_expr): Don't fold the expression.
15082 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15084 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15085 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15086 and (?*y,m). Update insn attributes.
15088 2017-05-16 Martin Liska <mliska@suse.cz>
15090 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15091 flags argument of print_gimple_stmt, print_gimple_expr,
15092 print_generic_stmt and print_generic_expr.
15093 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15094 * coretypes.h: Likewise.
15095 * except.c (dump_eh_tree): Likewise.
15096 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15097 * gimple-pretty-print.h: Likewise.
15098 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15099 (backprop::push_to_worklist): Likewise.
15100 (backprop::pop_from_worklist): Likewise.
15101 (backprop::process_use): Likewise.
15102 (backprop::intersect_uses): Likewise.
15103 (note_replacement): Likewise.
15104 * gimple-ssa-store-merging.c
15105 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15106 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15107 (pass_store_merging::execute): Likewise.
15108 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15109 (ssa_base_cand_dump_callback): Likewise.
15110 (dump_incr_vec): Likewise.
15111 (replace_refs): Likewise.
15112 (replace_mult_candidate): Likewise.
15113 (create_add_on_incoming_edge): Likewise.
15114 (create_phi_basis): Likewise.
15115 (insert_initializers): Likewise.
15116 (all_phi_incrs_profitable): Likewise.
15117 (introduce_cast_before_cand): Likewise.
15118 (replace_one_candidate): Likewise.
15119 * gimplify.c (gimplify_expr): Likewise.
15120 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15121 (set_rename): Likewise.
15122 (rename_uses): Likewise.
15123 (copy_loop_phi_nodes): Likewise.
15124 (add_close_phis_to_merge_points): Likewise.
15125 (copy_loop_close_phi_args): Likewise.
15126 (copy_cond_phi_args): Likewise.
15127 (graphite_copy_stmts_from_block): Likewise.
15128 (translate_pending_phi_nodes): Likewise.
15129 * graphite-poly.c (print_pdr): Likewise.
15130 (dump_gbb_cases): Likewise.
15131 (dump_gbb_conditions): Likewise.
15132 (print_scop_params): Likewise.
15133 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15134 (build_cross_bb_scalars_use): Likewise.
15135 (gather_bbs::before_dom_children): Likewise.
15136 * hsa-dump.c (dump_hsa_immed): Likewise.
15137 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15138 (get_replacement_map): Likewise.
15139 * ipa-inline-analysis.c (dump_condition): Likewise.
15140 (estimate_function_body_sizes): Likewise.
15141 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15142 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15143 * ipa-prop.c (ipa_dump_param): Likewise.
15144 (ipa_print_node_jump_functions_for_edge): Likewise.
15145 (ipa_modify_call_arguments): Likewise.
15146 (ipa_modify_expr): Likewise.
15147 (ipa_dump_param_adjustments): Likewise.
15148 (ipa_dump_agg_replacement_values): Likewise.
15149 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15150 * ipa-pure-const.c (check_stmt): Likewise.
15151 (pass_nothrow::execute): Likewise.
15152 * ipa-split.c (execute_split_functions): Likewise.
15153 * omp-offload.c (dump_oacc_loop_part): Likewise.
15154 (dump_oacc_loop): Likewise.
15155 * trans-mem.c (tm_log_emit): Likewise.
15156 (tm_memopt_accumulate_memops): Likewise.
15157 (dump_tm_memopt_set): Likewise.
15158 (dump_tm_memopt_transform): Likewise.
15159 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15160 (print_loop): Likewise.
15161 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15162 (chkp_gather_checks_info): Likewise.
15163 (chkp_get_check_result): Likewise.
15164 (chkp_remove_check_if_pass): Likewise.
15165 (chkp_use_outer_bounds_if_possible): Likewise.
15166 (chkp_reduce_bounds_lifetime): Likewise.
15167 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15168 (chkp_mark_completed_bounds): Likewise.
15169 (chkp_register_incomplete_bounds): Likewise.
15170 (chkp_mark_invalid_bounds): Likewise.
15171 (chkp_maybe_copy_and_register_bounds): Likewise.
15172 (chkp_build_returned_bound): Likewise.
15173 (chkp_get_bound_for_parm): Likewise.
15174 (chkp_build_bndldx): Likewise.
15175 (chkp_get_bounds_by_definition): Likewise.
15176 (chkp_generate_extern_var_bounds): Likewise.
15177 (chkp_get_bounds_for_decl_addr): Likewise.
15178 * tree-chrec.c (chrec_apply): Likewise.
15179 * tree-data-ref.c (dump_data_reference): Likewise.
15180 (dump_subscript): Likewise.
15181 (dump_data_dependence_relation): Likewise.
15182 (analyze_overlapping_iterations): Likewise.
15183 * tree-inline.c (expand_call_inline): Likewise.
15184 (tree_function_versioning): Likewise.
15185 * tree-into-ssa.c (dump_defs_stack): Likewise.
15186 (dump_currdefs): Likewise.
15187 (dump_names_replaced_by): Likewise.
15188 (dump_update_ssa): Likewise.
15189 (update_ssa): Likewise.
15190 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15191 * tree-parloops.c (build_new_reduction): Likewise.
15192 (try_create_reduction_list): Likewise.
15193 (ref_conflicts_with_region): Likewise.
15194 (oacc_entry_exit_ok_1): Likewise.
15195 (oacc_entry_exit_single_gang): Likewise.
15196 * tree-pretty-print.h: Likewise.
15197 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15198 (get_scalar_evolution): Likewise.
15199 (add_to_evolution): Likewise.
15200 (get_loop_exit_condition): Likewise.
15201 (analyze_evolution_in_loop): Likewise.
15202 (analyze_initial_condition): Likewise.
15203 (analyze_scalar_evolution): Likewise.
15204 (instantiate_scev): Likewise.
15205 (number_of_latch_executions): Likewise.
15206 (gather_chrec_stats): Likewise.
15207 (final_value_replacement_loop): Likewise.
15208 (scev_const_prop): Likewise.
15209 * tree-sra.c (dump_access): Likewise.
15210 (disqualify_candidate): Likewise.
15211 (create_access): Likewise.
15212 (reject): Likewise.
15213 (maybe_add_sra_candidate): Likewise.
15214 (create_access_replacement): Likewise.
15215 (analyze_access_subtree): Likewise.
15216 (analyze_all_variable_accesses): Likewise.
15217 (sra_modify_assign): Likewise.
15218 (initialize_constant_pool_replacements): Likewise.
15219 (find_param_candidates): Likewise.
15220 (decide_one_param_reduction): Likewise.
15221 (replace_removed_params_ssa_names): Likewise.
15222 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15223 * tree-ssa-copy.c (dump_copy_of): Likewise.
15224 (copy_prop_visit_cond_stmt): Likewise.
15225 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15226 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15227 (record_equivalences_from_stmt): Likewise.
15228 * tree-ssa-dse.c (compute_trims): Likewise.
15229 (delete_dead_call): Likewise.
15230 (delete_dead_assignment): Likewise.
15231 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15232 (forward_propagate_into_cond): Likewise.
15233 (pass_forwprop::execute): Likewise.
15234 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15235 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15237 (move_computations_worker): Likewise.
15238 (execute_sm): Likewise.
15239 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15240 (remove_exits_and_undefined_stmts): Likewise.
15241 (remove_redundant_iv_tests): Likewise.
15242 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15243 (adjust_iv_update_pos): Likewise.
15244 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15245 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15246 (value_replacement): Likewise.
15247 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15248 * tree-ssa-pre.c (print_pre_expr): Likewise.
15249 (get_representative_for): Likewise.
15250 (create_expression_by_pieces): Likewise.
15251 (insert_into_preds_of_block): Likewise.
15252 (eliminate_insert): Likewise.
15253 (eliminate_dom_walker::before_dom_children): Likewise.
15254 (eliminate): Likewise.
15255 (remove_dead_inserted_code): Likewise.
15256 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15257 * tree-ssa-reassoc.c (get_rank): Likewise.
15258 (eliminate_duplicate_pair): Likewise.
15259 (eliminate_plus_minus_pair): Likewise.
15260 (eliminate_not_pairs): Likewise.
15261 (undistribute_ops_list): Likewise.
15262 (eliminate_redundant_comparison): Likewise.
15263 (update_range_test): Likewise.
15264 (optimize_range_tests_var_bound): Likewise.
15265 (optimize_vec_cond_expr): Likewise.
15266 (rewrite_expr_tree): Likewise.
15267 (rewrite_expr_tree_parallel): Likewise.
15268 (linearize_expr): Likewise.
15269 (break_up_subtract): Likewise.
15270 (linearize_expr_tree): Likewise.
15271 (attempt_builtin_powi): Likewise.
15272 (attempt_builtin_copysign): Likewise.
15273 (transform_stmt_to_copy): Likewise.
15274 (transform_stmt_to_multiply): Likewise.
15275 (dump_ops_vector): Likewise.
15276 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15277 (print_scc): Likewise.
15278 (set_ssa_val_to): Likewise.
15279 (visit_reference_op_store): Likewise.
15280 (visit_use): Likewise.
15281 (sccvn_dom_walker::before_dom_children): Likewise.
15282 (run_scc_vn): Likewise.
15283 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15285 (expr_hash_elt::print): Likewise.
15286 (const_and_copies::pop_to_marker): Likewise.
15287 (const_and_copies::record_const_or_copy_raw): Likewise.
15288 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15289 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15290 (dump_predicates): Likewise.
15291 (find_uninit_use): Likewise.
15292 (warn_uninitialized_phi): Likewise.
15293 (pass_late_warn_uninitialized::execute): Likewise.
15294 * tree-ssa.c (verify_vssa): Likewise.
15295 (verify_ssa): Likewise.
15296 (maybe_optimize_var): Likewise.
15297 * tree-vrp.c (dump_value_range): Likewise.
15298 (dump_all_value_ranges): Likewise.
15299 (dump_asserts_for): Likewise.
15300 (register_edge_assert_for_2): Likewise.
15301 (vrp_visit_cond_stmt): Likewise.
15302 (vrp_visit_switch_stmt): Likewise.
15303 (vrp_visit_stmt): Likewise.
15304 (vrp_visit_phi_node): Likewise.
15305 (simplify_cond_using_ranges_1): Likewise.
15306 (fold_predicate_in): Likewise.
15307 (evrp_dom_walker::before_dom_children): Likewise.
15308 (evrp_dom_walker::push_value_range): Likewise.
15309 (evrp_dom_walker::pop_value_range): Likewise.
15310 (execute_early_vrp): Likewise.
15312 2017-05-16 Richard Biener <rguenther@suse.de>
15314 * dwarf2out.c (loc_list_from_tree_1): Do not create
15315 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15317 2017-05-16 Richard Biener <rguenther@suse.de>
15319 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15321 (note_variable_value_in_expr): If we resolved the decl ref
15322 do not push to the stack.
15324 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15326 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15327 operations in fast-math mode.
15328 (vaddq_f16): Likewise.
15329 (vmul_f16): Likewise.
15330 (vmulq_f16): Likewise.
15331 (vsub_f16): Likewise.
15332 (vsubq_f16): Likewise.
15333 * config/arm/neon.md (add<mode>3): New.
15335 (fma:<VH:mode>3): New. Also remove outdated comment.
15338 2017-05-16 Martin Liska <mliska@suse.cz>
15342 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15343 (odr_types_equivalent_p): Likewise.
15345 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15347 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15349 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15352 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15353 non-interunit SSE move alternatives with '?'.
15354 (zero-extendsidi peephole2): New peephole to skip intermediate
15355 general register in SSE zero-extend sequence.
15357 2017-05-15 Jeff Law <law@redhat.com>
15359 * reorg.c (relax_delay_slots): Create a new variable to hold
15360 the temporary target rather than clobbering TARGET_LABEL.
15362 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15363 missing argument to extract_bit_field call.
15364 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15366 2017-05-15 Martin Liska <mliska@suse.cz>
15369 * gcc.c (process_command): Do not allow empty argument of -o option.
15371 2017-05-15 Renlin Li <renlin.li@arm.com>
15373 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15374 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15375 * config/aarch64/constraints.md (Usf): Add long call check.
15376 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15377 (call_value): Likewise.
15378 (sibcall): Likewise.
15379 (sibcall_value): Likewise.
15381 (call_value_insn): New.
15382 (sibcall_insn): Update rtx pattern.
15383 (sibcall_value_insn): Likewise.
15384 (call_internal): Remove.
15385 (call_value_internal): Likewise.
15386 (sibcall_internal): Likewise.
15387 (sibcall_value_internal): Likewise.
15388 (call_reg): Likewise.
15389 (call_symbol): Likewise.
15390 (call_value_reg): Likewise.
15391 (call_value_symbol): Likewise.
15393 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15396 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15398 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15400 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15401 compatible with CCGOCmode and with CCZmode.
15403 2017-05-14 Martin Sebor <msebor@redhat.com>
15405 PR middle-end/77671
15406 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15407 (gimple_fold_builtin_snprintf): Same.
15408 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15409 (gimple_fold_builtin_snprintf): Same.
15410 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15411 of character types.
15412 (is_call_safe): New function.
15413 (try_substitute_return_value): Call it.
15414 (try_simplify_call): New function.
15415 (pass_sprintf_length::handle_gimple_call): Call it.
15417 2017-05-14 Martin Sebor <msebor@redhat.com>
15419 PR middle-end/80669
15420 * builtins.c (expand_builtin_stpncpy): Simplify.
15422 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15424 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15425 * config/i386/i386.h
15426 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15427 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15428 (struct machine_function): Add new members call_ms2sysv,
15429 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15430 (struct machine_frame_state): New fields sp_realigned and
15431 sp_realigned_offset.
15432 * config/i386/i386.c
15433 (enum xlogue_stub): New enum.
15434 (enum xlogue_stub_sets): New enum.
15435 (class xlogue_layout): New class.
15436 (struct ix86_frame): New fields stack_realign_allocate_offset,
15437 stack_realign_offset and outlined_save_offset. Modify comments to
15438 detail stack layout when using out-of-line stubs.
15439 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15440 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15441 -mcall-ms2sysv-xlogues.
15442 (stub_managed_regs): New static variable.
15443 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15444 registers managed by out-of-line stub.
15445 (disable_call_ms2sysv_xlogues): New function.
15446 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15447 m->call_ms2sysv when appropriate and compute frame layout for
15449 (sp_valid_at, fp_valid_at): New inline functions.
15450 (choose_basereg): New function.
15451 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15453 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15454 Use align parameter of choose_baseaddr to generated aligned SSE movs
15456 (pro_epilogue_adjust_stack): Modify to track
15457 machine_frame_state::sp_realigned.
15458 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15459 (ix86_nsaved_sseregs): Likewise.
15460 (ix86_emit_save_regs): Likewise.
15461 (ix86_emit_save_regs_using_mov): Likewise.
15462 (ix86_emit_save_sse_regs_using_mov): Likewise.
15463 (get_scratch_register_on_entry): Likewise.
15464 (gen_frame_set): New function.
15465 (gen_frame_load): Likewise.
15466 (gen_frame_store): Likewise.
15467 (emit_outlined_ms2sysv_save): Likewise.
15468 (emit_outlined_ms2sysv_restore): Likewise.
15469 (ix86_expand_prologue): Modify stack re-alignment code and call
15470 emit_outlined_ms2sysv_save when appropriate.
15471 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15472 parameter rtx_insn *insn, which allows the function to be used to only
15473 generate the notes.
15474 (ix86_expand_epilogue): Modify validity checks of frame and stack
15475 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15476 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15477 * config/i386/predicates.md
15478 (save_multiple): New predicate.
15479 (restore_multiple): Likewise.
15480 * config/i386/sse.md
15481 (save_multiple<mode>): New pattern.
15482 (save_multiple_realign<mode>): Likewise.
15483 (restore_multiple<mode>): Likewise.
15484 (restore_multiple_and_return<mode>): Likewise.
15485 (restore_multiple_leave_return<mode>): Likewise.
15486 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15488 2017-05-14 Julia Koval <julia.koval@intel.com>
15490 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15491 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15492 (__builtin_ia32_xsetbv): New builtins.
15493 * config/i386/i386.c (ix86_expand_special_args_builtin):
15495 (ix86_expand_builtin): Special expand for new intrinsics.
15496 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15497 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15498 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15500 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15502 * cfganal.c (inverted_post_order_compute): Change argument type
15504 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15505 * df-core.c (rest_of_handle_df_initialize): Adjust.
15506 (rest_of_handle_df_finish): Likewise.
15507 (df_analyze_1): Likewise.
15508 (df_analyze): Likewise.
15509 (loop_inverted_post_order_compute): Change argument to be a vec *.
15510 (df_analyze_loop): Adjust.
15511 (df_get_n_blocks): Likewise.
15512 (df_get_postorder): Likewise.
15513 * df.h (struct df_d): Change field to be a vec.
15514 * lcm.c (compute_laterin): Adjust.
15515 (compute_available): Likewise.
15516 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15517 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15518 * tree-ssa-pre.c (compute_antic): Likewise.
15520 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15522 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15523 (depth_first_search::depth_first_search): Change structure init
15524 function to this constructor.
15525 (depth_first_search::add_bb): Rename function to this member.
15526 (depth_first_search::execute): Likewise.
15527 (flow_dfs_compute_reverse_finish): Adjust.
15529 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15531 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15532 (longest_simple_path): Likewise.
15533 * shrink-wrap.c (spread_components): Likewise.
15534 (disqualify_problematic_components): Likewise.
15535 (emit_common_heads_for_components): Likewise.
15536 (emit_common_tails_for_components): Likewise.
15537 (insert_prologue_epilogue_for_components): Likewise.
15539 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15541 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15544 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15546 * df-core.c (df_set_blocks): Start using auto_bitmap.
15547 (df_compact_blocks): Likewise.
15548 * df-problems.c (df_rd_confluence_n): Likewise.
15549 * df-scan.c (df_insn_rescan_all): Likewise.
15550 (df_process_deferred_rescans): Likewise.
15551 (df_update_entry_block_defs): Likewise.
15552 (df_update_exit_block_uses): Likewise.
15553 (df_entry_block_bitmap_verify): Likewise.
15554 (df_exit_block_bitmap_verify): Likewise.
15555 (df_scan_verify): Likewise.
15556 * lra-constraints.c (lra_constraints): Likewise.
15557 (undo_optional_reloads): Likewise.
15558 (lra_undo_inheritance): Likewise.
15559 * lra-remat.c (calculate_gen_cands): Likewise.
15560 (do_remat): Likewise.
15561 * lra-spills.c (assign_spill_hard_regs): Likewise.
15562 (spill_pseudos): Likewise.
15563 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15564 (bitmap_set_subtract_values): Likewise.
15566 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15568 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15569 management with auto_bitmap.
15570 (fix_inter_tick): Likewise.
15571 (fix_recovery_deps): Likewise.
15572 * ira.c (add_store_equivs): Likewise.
15573 (find_moveable_pseudos): Likewise.
15574 (split_live_ranges_for_shrink_wrap): Likewise.
15575 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15576 (rtx_reuse_manager::seen_def_p): Likewise.
15577 (rtx_reuse_manager::set_seen_def): Likewise.
15578 * print-rtl.h (class rtx_reuse_manager): Likewise.
15580 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15582 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15584 (migrate_btr_def): Likewise.
15585 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15586 * df-core.c (loop_post_order_compute): Likewise.
15587 (loop_inverted_post_order_compute): Likewise.
15588 * hsa-common.h: Likewise.
15589 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15590 * init-regs.c (initialize_uninitialized_regs): Likewise.
15591 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15592 (inline_small_functions): Likewise.
15593 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15594 * ira.c (combine_and_move_insns): Likewise.
15595 (build_insn_chain): Likewise.
15596 * loop-invariant.c (find_invariants): Likewise.
15597 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15598 * predict.c (tree_predict_by_opcode): Likewise.
15599 (predict_paths_leading_to): Likewise.
15600 (predict_paths_leading_to_edge): Likewise.
15601 (estimate_loops_at_level): Likewise.
15602 (estimate_loops): Likewise.
15603 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15604 (spread_components): Likewise.
15605 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15606 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15607 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15608 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15609 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15610 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15611 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15612 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15613 (mark_threaded_blocks): Likewise.
15614 (thread_through_all_blocks): Likewise.
15615 * tree-ssa.c (verify_ssa): Likewise.
15616 (execute_update_addresses_taken): Likewise.
15617 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15619 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15621 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15623 (post_order_compute): Likewise.
15624 (inverted_post_order_compute): Likewise.
15625 (pre_and_rev_post_order_compute_fn): Likewise.
15627 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15629 * genrecog.c (int_set::int_set): Explicitly construct our
15630 auto_vec base class.
15631 * vec.h (auto_vec::auto_vec): New constructor.
15633 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15635 * bitmap.h (class auto_bitmap): New constructor taking
15636 bitmap_obstack * argument.
15638 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15640 * bitmap.h (class auto_bitmap): Change type of m_bits to
15641 bitmap_head, and adjust ctor / dtor and member operators.
15643 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15645 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15646 when returned register mode doesn't match original mode.
15648 2017-05-12 Jeff Law <law@redhat.com>
15649 Jakub Jelinek <jakub@redhat.com>
15651 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15652 we look for cc setter after the compare-elim changes.
15653 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15654 within the vector to match what compare-elim now expects.
15655 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15656 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15658 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15659 after the compare-elim changes.
15660 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15661 the vector to match what compare-elim now expects.
15662 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15663 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15664 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15665 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15666 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15668 * config/visium/visium.c (single_set_and_flags): Fix where
15669 we look for cc setter after the compare-elim changes.
15670 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15671 with the vector to match what compare-elim now expects.
15672 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15673 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15674 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15675 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15676 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15677 (neg<mode>2_insn_set_overflow): Likewise.
15679 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15681 PR middle-end/79794
15682 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15683 maybe_expand_insn call, set ops[0].target. If still set after call,
15684 set alt_rtl. Add extra arg to recursive calls.
15685 (extract_bit_field): Add alt_rtl argument. Pass to
15687 * expmed.h (extract_bit_field): Fix prototype.
15688 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15689 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15690 to extract_bit_field_calls.
15691 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15692 Pass alt_rtl to extract_bit_field calls.
15693 * calls.c (store_unaligned_arguments_into_psuedos)
15694 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15695 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15697 * optabs.h (struct expand_operand): Add target bitfield.
15699 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15701 * compare-elim.c (try_eliminate_compare): Canonicalize
15702 operation with embedded compare to
15703 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15704 (set (reg) (operation)].
15706 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15708 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15711 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15712 cost of adding a carry flag for ADC instruction.
15713 [case MINUS]: Ignore the cost of subtracting a carry flag
15714 for SBB instruction.
15716 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15718 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15720 * config/rs6000/bmiintrin.h: New file.
15721 * config/rs6000/bmi2intrin.h: New file.
15722 * config/rs6000/x86intrin.h: New file.
15724 2017-05-12 Jeff Law <law@redhat.com>
15726 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15729 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15731 PR middle-end/80707
15732 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15734 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15736 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15737 early expansion of vector divide builtins.
15738 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15739 builtins identified as having unsigned arguments.
15741 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15743 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15744 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15745 expansion of vector logical operations (and, andc, or, xor,
15748 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15750 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15751 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15753 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15756 early GIMPLE expansion of vector multiplies.
15758 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15760 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15761 TARGET_HAVE_MOVT conditional.
15762 (movt splitter): Likewise.
15764 2017-05-12 Richard Biener <rguenther@suse.de>
15766 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15767 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15768 Fold all stmts not inplace.
15770 2017-05-12 Richard Biener <rguenther@suse.de>
15772 PR tree-optimization/80713
15773 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15774 inserted_exprs bit for not removed stmts.
15776 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15778 PR middle-end/69921
15779 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15780 parallelized" attribute for parallelized OpenACC kernels.
15781 * omp-offload.c (execute_oacc_device_lower): Use it.
15783 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15784 Set "oacc kernels" attribute.
15785 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15786 parameter. Adjust all users.
15787 (oacc_fn_attrib_kernels_p): Remove function.
15788 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15789 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15790 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15791 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15792 assert "oacc kernels" attribute is set.
15794 2017-05-11 Carl Love <cel@us.ibm.com>
15796 * config/rs6000/rs6000-c: Add support for built-in functions
15797 vector unsigned char vec_popcnt (vector signed char)
15798 vector unsigned char vec_popcnt (vector unsigned char)
15799 vector unsigned short vec_popcnt (vector signed short)
15800 vector unsigned short vec_popcnt (vector unsigned short)
15801 vector unsigned int vec_popcnt (vector signed int)
15802 vector unsigned int vec_popcnt (vector unsigned int)
15803 vector unsigned long long vec_popcnt (vector signed long long)
15804 vector unsigned long long vec_popcnt (vector unsigned long long)
15805 vector signed long long vec_slo (vector signed long long,
15806 vector signed char)
15807 vector signed long long vec_slo (vector signed long long,
15808 vector unsigned char)
15809 vector unsigned long long vec_slo (vector unsigned long long,
15810 vector signed char)
15811 vector unsigned long long vec_slo (vector unsigned long long,
15812 vector unsigned char)
15813 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15814 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15815 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15816 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15817 * doc/extend.texi: Update the built-in documentation file for the
15818 new built-in functions.
15820 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15822 * attribs.h (sorted_attr_string): Move machine independent
15823 functions for target clone support from the i386 port to common
15824 code. Rename ix86_function_versions to common_function_versions.
15825 Rename make_name to make_unique_name.
15826 (common_function_versions): Likewise.
15827 (make_unique_name): Likewise.
15828 (make_dispatcher_decl): Likewise.
15829 (is_function_default_version): Likewise.
15830 * attribs.c (attr_strcmp): Likewise.
15831 (sorted_attr_string): Likewise.
15832 (common_function_versions): Likewise.
15833 (make_unique_name): Likewise.
15834 (make_dispatcher_decl): Likewise.
15835 (is_function_default_version): Likewise.
15836 * config/i386/i386.c (attr_strcmp): Likewise.
15837 (sorted_attr_string): Likewise.
15838 (ix86_function_versions): Likewise.
15839 (make_name): Likewise.
15840 (make_dispatcher_decl): Likewise.
15841 (is_function_default_version): Likewise.
15842 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15844 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15847 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15848 Account for direct move costs for vec_construct of integer
15851 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15854 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15855 (UNSPEC_STX_ATOMIC): Ditto.
15856 (loaddi_via_sse): New insn.
15857 (storedi_via_sse): Ditto.
15858 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15859 Update corresponding peephole2 patterns.
15860 (atomic_storedi_fpu): Ditto.
15862 2017-05-11 Julia Koval <julia.koval@intel.com>
15864 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15865 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15867 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15868 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15869 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15871 2017-05-11 Nathan Sidwell <nathan@acm.org>
15873 * graphite-poly.c: Include dumpfile.h.
15875 * dumpfle.h (dump_function): Declare here ...
15876 * tree-dump.h (dump_function): ... not here.
15877 * dumpfile.c: #include tree-cfg.h.
15878 (dump_function): Move here from ...
15879 * tree-dump.c (dump_function): ... here.
15880 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15881 * graphite-poly.c: Don't include tree-dump.h.
15882 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15883 * print-tree.c: Likewise.
15884 * stor-layout.c: Likewise.
15885 * tree-nested.c: Likewise.
15887 * dumpfile.c (dump_start): Use TDF_FLAGS.
15888 (dump_enable_all): Fix TDF_KIND check thinko.
15890 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15892 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15893 array entries to represent two legal parameterizations of the
15894 overloaded __builtin_cmpb function, as represented by the
15895 P6_OV_BUILTIN_CMPB constant.
15896 (altivec_resolve_overloaded_builtin): Add special case handling
15897 for the __builtin_cmpb function, as represented by the
15898 P6_OV_BUILTIN_CMPB constant.
15899 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15900 (BU_P6_64BIT_2): New macro.
15901 (BU_P6_OVERLOAD_2): New macro
15902 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15903 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15904 (CMPB): Add overload support to represent both 32-bit and 64-bit
15905 compare-bytes function.
15906 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15907 support for TARGET_CMPB.
15908 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15909 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15910 documentation of the __builtin_cmpb overloaded built-in function.
15912 2017-05-11 Richard Biener <rguenther@suse.de>
15914 PR tree-optimization/80705
15915 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15916 bases are not vectorizable.
15918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15920 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15921 when counting register pressure.
15923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15925 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15926 (iv_ca_more_deps): Renamed to ...
15927 (iv_ca_compare_deps): ... this.
15928 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15932 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15934 (determine_group_iv_costs): ... here.
15935 (find_inv_vars_cb): Record inv var if it's not recorded before.
15937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15939 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15940 (get_shiftadd_cost): Ditto.
15942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15944 * tree-ssa-address.c: Include header file.
15945 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15947 (add_to_parts): Refactor.
15948 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15949 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15954 PR tree-optimization/53090
15955 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15957 (extract_cond_operands): Detect condition with IV on both sides
15958 and return COMP_IV_EXPR_2.
15959 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15960 (rewrite_use_compare): Simplify by removing call to function
15961 extract_cond_operands.
15963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15965 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15966 (extract_cond_operands): Detect condition comparing against non-
15967 invariant bound and return appropriate enum value.
15968 (find_interesting_uses_cond): Update use of extract_cond_operands.
15969 Handle its return value accordingly.
15970 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15974 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15975 nonlinear iv_use computation in loop invariant sensitive way.
15977 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15979 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15980 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15982 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15984 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15985 (dump_cand): Support iv_cand.inv_exprs.
15986 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15988 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15991 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15993 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15995 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15996 as local function. Include necessary header files.
15997 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15999 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16001 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16003 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16005 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16006 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16007 RSHIFT_EXPR and BIT_NOT_EXPR.
16009 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16011 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16012 (adjust_setup_cost): New parameter supporting round up adjustment.
16013 (struct address_cost_data): Delete.
16014 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16015 (split_address_cost, ptr_difference_cost): Delete.
16016 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16017 (struct ainc_cost_data): New struct.
16018 (get_address_cost_ainc): New function.
16019 (get_address_cost, get_computation_cost): Reimplement.
16020 (determine_group_iv_cost_address): Record inv_expr for all uses of
16022 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16023 (iv_ca_has_deps): Reimplemented to ...
16024 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16026 (iv_ca_extend): Call iv_ca_more_deps.
16028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16030 * tree-ssa-address.c (struct mem_address): Move to header file.
16031 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16032 * tree-ssa-address.h (struct mem_address): Move from C file.
16033 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16035 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16037 * tree-affine.h (aff_combination_type): New interface.
16038 (aff_combination_zero_p): Remove static.
16039 (aff_combination_const_p): New interface.
16040 (aff_combination_singleton_var_p): New interfaces.
16042 2017-05-11 Richard Biener <rguenther@suse.de>
16044 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16045 Skip unreachable blocks and destinations.
16046 (eliminate): Move stmt removal and fixup ...
16047 (fini_eliminate): ... here. Skip inserted exprs.
16048 (pass_pre::execute): Move fini_pre after fini_eliminate.
16049 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16050 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16051 PRE to get rid of dead code that has invalid SSA form and
16052 split critical edges again.
16054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16056 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16058 2017-05-11 Richard Biener <rguenther@suse.de>
16060 * passes.c (execute_function_todo): Verify loops if they are
16061 said to be up-to-date.
16062 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16063 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16065 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16068 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16069 handle calling assemble_external ourself.
16072 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16073 modes with zero size. Enhance comment.
16075 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16077 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16078 built-ins for vec_xl and vec_xst with short and char pointer
16081 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16083 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16084 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16085 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16086 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16087 (_mm_maskz_min_round_ss): New intrinsics.
16088 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16089 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16090 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16091 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16092 (__builtin_ia32_minss_mask_round): New builtins.
16093 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16094 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16095 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16097 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16098 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16100 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16103 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16105 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16106 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16107 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16108 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16109 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16110 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16111 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16112 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16113 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16114 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16115 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16116 (__builtin_ia32_mulss_mask_round): New builtins.
16117 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16118 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16119 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16121 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16122 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16124 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16127 2017-05-10 Julia Koval <julia.koval@intel.com>
16129 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16130 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16131 (_mm256_setr_m128i): New intrinsics.
16133 2017-05-10 Julia Koval <julia.koval@intel.com>
16135 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16136 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16137 (_mm_maskz_rcp14_ss): New intrinsics.
16138 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16139 (__builtin_ia32_rcp14ss_mask): New builtins.
16140 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16142 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16144 PR tree-optimization/51513
16145 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16146 (assert_unreachable_fallthru_edge_p): Use it.
16147 (group_case_labels_stmt): Likewise.
16148 * tree-cfg.h: Prototype it.
16149 * stmt.c: Include cfghooks.h and tree-cfg.h.
16150 (emit_case_dispatch_table) <gap_label>: New local variable.
16151 Use it to fill dispatch table gaps.
16152 Test for default_label before updating probabilities.
16153 (expand_case) <default_label>: Remove unneeded initialization.
16154 Test for unreachable default case statement and remove its edge.
16155 Set default_label accordingly.
16156 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16158 2017-05-10 Carl Love <cel@us.ibm.com>
16160 * config/rs6000/rs6000-c: Add support for built-in functions
16161 vector signed char vec_neg (vector signed char)
16162 vector signed short int vec_neg (vector short int)
16163 vector signed int vec_neg (vector signed int)
16164 vector signed long long vec_neg (vector signed long long)
16165 vector float vec_neg (vector float)
16166 vector double vec_neg (vector double)
16167 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16169 * config/rs6000/altivec.h: Add define for vec_neg
16170 * doc/extend.texi: Update the built-in documentation for the
16171 new built-in functions.
16173 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16175 PR tree-optimization/77644
16176 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16178 2017-05-10 Nathan Sidwell <nathan@acm.org>
16180 * dumpfile.h (TDI_lang_all): New.
16181 (TDF_KIND): New. Renumber others
16182 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16184 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16186 (get_dump_file_name): Adjust suffix generation.
16187 (dump_enable_all): Use TDF_KIND.
16188 * doc/invoke.texi (-fdump-lang-all): Document.
16190 * dumpfile.h: Tabify.
16192 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16195 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16196 Move member access before delete.
16198 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16200 * tree-inline.c (expand_call_inline): Split block at stmt
16203 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16206 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16207 are now unused after splitting mov{sf,sd}_hardfloat.
16208 (f32_lr2): Likewise.
16209 (f32_lm): Likewise.
16210 (f32_lm2): Likewise.
16211 (f32_li): Likewise.
16212 (f32_li2): Likewise.
16213 (f32_lv): Likewise.
16214 (f32_sr): Likewise.
16215 (f32_sr2): Likewise.
16216 (f32_sm): Likewise.
16217 (f32_sm2): Likewise.
16218 (f32_si): Likewise.
16219 (f32_si2): Likewise.
16220 (f32_sv): Likewise.
16221 (f32_dm): Likewise.
16222 (f32_vsx): Likewise.
16223 (f32_av): Likewise.
16224 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16225 For movsf, order stores so the VSX stores occur before the GPR
16226 store which encourages the register allocator to use a traditional
16227 FPR instead of a GPR. For movsd, order the stores so that the GPR
16228 store comes before the VSX stores to allow the power6 to work.
16229 This is due to the power6 not having a 32-bit integer store
16230 instruction from a FPR.
16231 (movsf_hardfloat): Likewise.
16232 (movsd_hardfloat): Likewise.
16234 2017-05-09 Martin Sebor <msebor@redhat.com>
16236 PR translation/80280
16237 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16240 PR translation/80280
16241 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16242 data member added in r247778.
16243 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16245 2017-05-09 Nathan Sidwell <nathan@acm.org>
16247 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16249 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16252 2017-05-09 Marek Polacek <polacek@redhat.com>
16254 * doc/invoke.texi: Fix typo.
16256 2017-05-09 Richard Biener <rguenther@suse.de>
16258 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16259 (vrp_val_is_min): Likewise.
16260 (set_value_range_to_value): Likewise.
16261 (set_value_range_to_nonnegative): Likewise.
16262 (gimple_assign_nonzero_p): Likewise.
16263 (gimple_stmt_nonzero_p): Likewise.
16264 (vrp_int_const_binop): Likewise. Remove unreachable case.
16265 (adjust_range_with_scev): Adjust comments.
16266 (compare_range_with_value): Likewise.
16267 (extract_range_from_phi_node): Likewise.
16268 (test_for_singularity): Likewise.
16270 2017-05-09 Richard Biener <rguenther@suse.de>
16272 * tree-vrp.c (get_single_symbol): Add assert that we don't
16273 get overflowed constants as invariant part.
16274 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16275 checks. Use wi::cmp instead of recursing for integer constants.
16276 (compare_values): Just ignore whether we assumed undefined
16277 overflow instead of failing the compare.
16278 (extract_range_for_var_from_comparison_expr): Add comment before the
16279 TREE_NO_WARNING sets.
16280 (test_for_singularity): Likewise.
16281 (extract_range_from_comparison): Do not disable optimization
16282 when we assumed undefined overflow.
16283 (extract_range_basic): Remove init of unused var.
16285 2017-05-09 Richard Biener <rguenther@suse.de>
16287 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16288 (extract_range_from_multiplicative_op_1): Adjust.
16289 (extract_range_from_binary_expr_1): Use int_const_binop.
16291 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16294 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16295 rs6000_store_data_bypass_p in seven define_bypass directives and
16296 in several comments.
16297 * config/rs6000/rs6000-protos.h: Add prototype for
16298 rs6000_store_data_bypass_p function.
16299 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16300 function implements slightly different (rs6000-specific) semantics
16301 than store_data_bypass_p, returning false rather than aborting
16302 with assertion error when arguments do not satisfy the
16303 requirements of store data bypass.
16304 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16305 rs6000_store_data_bypass_p.
16307 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16309 * config/xtensa/xtensa-protos.h
16310 (xtensa_initial_elimination_offset): New declaration.
16311 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16312 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16313 macro definition, add case for FRAME_POINTER_REGNUM when
16314 FRAME_GROWS_DOWNWARD.
16315 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16316 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16317 xtensa_initial_elimination_offset.
16319 2017-05-08 Nathan Sidwell <nathan@acm.org>
16321 * doc/invoke.texi: Alphabetize -fdump options.
16323 2017-05-08 Martin Sebor <msebor@redhat.com>
16325 PR translation/80280
16326 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16328 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16330 * target.def (compute_frame_layout): New optional target hook.
16331 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16332 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16333 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16335 * reload1.c (verify_initial_elim_offsets): Likewise.
16336 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16337 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16338 (arm_get_frame_offsets): Split up into this ...
16339 (arm_compute_frame_layout): ... and this function.
16341 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16343 * config/aarch64/constraints.md (Usa): New constraint.
16344 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16346 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16348 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16349 with_multilib_list after it has been checked.
16351 2017-05-08 Richard Biener <rguenther@suse.de>
16353 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16354 (bitmap_set_subtract_values): Likewise.
16356 2017-05-08 Richard Biener <rguenther@suse.de>
16358 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16359 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16361 (gimple_stmt_nonzero_warnv_p): Rename to ...
16362 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16364 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16365 (extract_range_basic): Adjust, do not disable propagation on
16366 strict overflow sensitive simplification.
16367 (vrp_visit_cond_stmt): Likewise.
16369 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16371 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16372 body size unconditionally.
16374 2017-05-07 Jeff Law <law@redhat.com>
16377 2017-05-06 Jeff Law <law@redhat.com>
16378 PR tree-optimization/78496
16379 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16382 PR tree-optimization/78496
16383 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16384 (simplify_stmt_using_ranges): Call it.
16385 (vrp_dom_walker::before_dom_children): Extract equivalences
16386 from an ASSERT_EXPR with an equality comparison against a
16389 2017-05-06 Jeff Law <law@redhat.com>
16391 PR tree-optimization/78496
16392 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16395 PR tree-optimization/78496
16396 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16397 (simplify_stmt_using_ranges): Call it.
16398 (vrp_dom_walker::before_dom_children): Extract equivalences
16399 from an ASSERT_EXPR with an equality comparison against a
16402 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16404 * lra-constraints.c (lra_copy_reg_equiv): New function.
16405 (split_reg): Use it to copy equivalence information from the
16406 original register to the spill register.
16408 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16410 PR rtl-optimization/75964
16411 * simplify-rtx.c (simplify_const_relational_operation): Remove
16412 invalid handling of comparisons of integer ABS.
16414 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16416 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16417 initialize to zero.
16418 (init_regs): Remove declaration.
16419 (function_arg_advance_32): Initialize error_p as boolean variable.
16421 2017-05-05 Nathan Sidwell <nathan@acm.org>
16423 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16424 lines. Use for (;;).
16426 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16428 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16429 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16430 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16431 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16432 VF=2 that require versioning.
16434 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16436 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16439 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16441 * diagnostic.h (diagnostic_override_option_index): Convert from
16442 macro to inline function.
16444 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16446 * diagnostic.c (last_module_changed_p): New function.
16447 (set_last_module): New function.
16448 (diagnostic_report_current_module): Convert macro usage to
16449 the above functions.
16450 * diagnostic.h (diagnostic_context::last_module): Strengthen
16451 from const line_map * to const line_map_ordinary *.
16452 (diagnostic_last_module_changed): Delete macro.
16453 (diagnostic_set_last_module): Delete macro.
16455 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16457 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16458 with diagnostic_report_diagnostic.
16459 (diagnostic_n_impl_richloc): Likewise.
16460 * diagnostic.h (report_diagnostic): Delete macro.
16461 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16462 with diagnostic_report_diagnostic.
16463 * substring-locations.c (format_warning_va): Likewise.
16465 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16467 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16468 save/restor of format_spec. Move option-printing code to...
16469 (print_option_information): ...this new function, and
16470 reimplement by simply printing to the pretty_printer,
16471 rather than appending to the format string.
16473 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16475 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16476 handling logic into...
16477 (update_effective_level_from_pragmas): ...this new function.
16479 2017-05-04 Andrew Waterman <andrew@sifive.com>
16481 * config/riscv/riscv.opt (mstrict-align): New option.
16482 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16483 (SLOW_UNALIGNED_ACCESS): Define.
16484 (riscv_slow_unaligned_access): Declare.
16485 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16487 (riscv_slow_unaligned_access): New variable.
16488 (rocket_tune_info): Set slow_unaligned_access to true.
16489 (optimize_size_tune_info): Set slow_unaligned_access to false.
16490 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16491 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16492 (riscv_option_override): Set riscv_slow_unaligned_access.
16493 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16495 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16497 * config/riscv/riscv.md: Unify indentation.
16499 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16504 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16506 (extendsi<mode>2): Add support for doing sign extension via
16507 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16508 don't have ISA 3.0 instructions.
16509 (extendsi<mode>2 splitter): Likewise.
16510 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16511 generate the normal insns since SImode can now go in vector
16512 registers. Disallow the special UNSPECs needed for previous
16513 machines to hide SImode being used. Add new insns
16514 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16515 (fix_trunc<mode>si2_stfiwx): Likewise.
16516 (fix_trunc<mode>si2_internal): Likewise.
16517 (fixuns_trunc<mode>si2): Likewise.
16518 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16519 (fctiw<u>z_<mode>_smallint): Likewise.
16520 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16521 of floating point to 32-bit integer from doing a direct move to
16522 the GPR registers to do a store.
16523 (fctiwz_<mode>): Break long line.
16525 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16527 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16528 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16529 (addr_list, addr_offset_valid_p): New.
16530 (split_address_groups): Check offset validity with above function.
16531 (gt-tree-ssa-loop-ivopts.h): Include header file.
16533 2017-05-05 Nathan Sidwell <nathan@acm.org>
16535 * config.gcc (arm*-*-*): Add missing 'fi'.
16537 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16539 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16540 included in -fopt-info does not matter.
16541 * doc/optinfo.texi (-fopt-info): Fix description of default
16542 behavour. Explicitly say order of options included in -fopt-info
16545 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16547 * config.gcc: Allow combinations of aprofile and rmprofile values for
16548 --with-multilib-list.
16549 * config/arm/t-multilib: New file.
16550 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16551 variables. Remove setting of ISA and floating-point ABI in
16552 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16553 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16554 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16555 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16557 * config/arm/t-rmprofile: Likewise except for the matches changes.
16558 * doc/install.texi (--with-multilib-list): Document the combination of
16559 aprofile and rmprofile values and warn about pitfalls in doing that.
16561 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16563 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16564 (movdi_aarch64): Likewise.
16566 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16568 PR tree-optimization/80632
16569 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16571 (build_arrays): Initialize it for virtual phis.
16572 (fix_phi_nodes): Use it for virtual phis.
16574 PR tree-optimization/80558
16575 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16576 [x, y] op z into [x op, y op z] for op & or | if conditions
16579 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16580 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16583 * config/arm/arm.md (use_literal_pool): Remove.
16584 (64-bit immediate split): No longer takes cost into consideration
16585 if arm_disable_literal_pool is enabled.
16586 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16587 used when arm_disable_literal_pool is enabled.
16588 (arm_max_const_double_inline_cost): Remove use of
16589 arm_disable_literal_pool.
16590 (push_minipool_fix): Add assert.
16591 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16592 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16593 (no_literal_pool_sf_immediate): New.
16595 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16597 PR tree-optimization/80613
16598 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16599 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16601 2017-05-05 Richard Biener <rguenther@suse.de>
16603 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16605 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16607 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16608 of this flag from insn conditions due to removal from r247495.
16610 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16612 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16614 (arm_early_store_addr_dep_ptr): Likewise.
16615 * config/arm/aarch-common-protos.h
16616 (arm_early_load_addr_dep_ptr): Add prototype.
16617 (arm_early_store_addr_dep_ptr): Likewise.
16618 * config/arm/cortex-a53.md: Add new bypasses.
16620 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16622 * tree.c (next_type_uid): Change type to unsigned.
16623 (type_hash_canon): Decrement back next_type_uid if
16624 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16625 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16628 2017-05-04 Martin Sebor <msebor@redhat.com>
16630 * builtins.c: Fix a trivial typo in a comment.
16632 PR middle-end/79234
16633 * builtins.c (check_sizes): Adjust to handle reading past the end.
16634 Avoid printing excessive upper bound of ranges. Use %E to print
16635 tree nodes instead of converting them to %wu.
16636 (expand_builtin_memchr): New function.
16637 (compute_dest_size): Rename...
16638 (compute_objsize): ...to this.
16639 (expand_builtin_memcpy): Adjust.
16640 (expand_builtin_mempcpy): Adjust.
16641 (expand_builtin_strcat): Adjust.
16642 (expand_builtin_strcpy): Adjust.
16643 (check_strncat_sizes): Adjust.
16644 (expand_builtin_strncat): Adjust.
16645 (expand_builtin_strncpy): Adjust and simplify.
16646 (expand_builtin_memset): Adjust.
16647 (expand_builtin_bzero): Adjust.
16648 (expand_builtin_memcmp): Adjust.
16649 (expand_builtin): Handle memcmp.
16650 (maybe_emit_chk_warning): Check strncat just once.
16652 2017-05-04 Martin Sebor <msebor@redhat.com>
16654 PR preprocessor/79214
16655 PR middle-end/79222
16656 PR middle-end/79223
16657 * builtins.c (check_sizes): Add inlining context and issue
16658 warnings even when -Wno-system-headers is set.
16659 (check_strncat_sizes): Same.
16660 (expand_builtin_strncat): Same.
16661 (expand_builtin_memmove): New function.
16662 (expand_builtin_stpncpy): Same.
16663 (expand_builtin): Handle memmove and stpncpy.
16665 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16667 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16668 which is not used any more.
16670 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16672 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16674 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16676 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16677 (cortexa53_tunings): Likewise.
16678 (cortexa57_tunings): Likewise.
16679 (cortexa72_tunings): Likewise.
16680 (cortexa73_tunings): Likewise.
16682 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16684 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16685 Set loop alignment to 8.
16687 2017-05-04 Martin Sebor <msebor@redhat.com>
16689 PR translation/80280
16690 * builtins.c (expand_builtin_object_size): Add missing quoting to
16691 %D and like directives.
16692 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16693 (hsa_type_for_tree_type): Same.
16694 (verify_function_arguments): Same.
16695 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16696 * varasm.c (get_section): Same.
16699 2017-05-04 Martin Sebor <msebor@redhat.com>
16701 PR translation/80280
16702 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16704 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16706 * config/aarch64/aarch64.c (generic_addrcost_table):
16707 Change HI/TI mode setting.
16709 2017-05-04 Martin Jambor <mjambor@suse.cz>
16711 PR tree-optimization/80622
16712 * tree-sra.c (comes_initialized_p): New function.
16713 (build_accesses_from_assign): Only set write lazily when
16714 comes_initialized_p is false.
16715 (analyze_access_subtree): Use comes_initialized_p.
16716 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16717 instead of testing for PARM_DECL.
16719 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16721 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16722 constraint on operand 0 to allow more general addressing modes.
16723 Adjust output template.
16724 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16726 * config/aarch64/aarch64-protos.h
16727 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16728 * config/aarch64/constraints.md (Dp): New address constraint.
16729 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16732 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16734 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16735 update use of estimate_ipcp_clone_size_and_time.
16736 (estimate_local_effects): Update use of
16737 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16738 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16739 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16740 Return nonspecialized time.
16742 2017-05-04 Richard Biener <rguenther@suse.de>
16744 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16745 for the last VUSE which def dominates the PHI. Directly call
16747 (get_continuation_for_phi_1): Remove.
16749 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16751 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16752 to explain the use of truncating division. Cap the number of
16753 iterations to the maximum given by nb_iterations_upper_bound,
16756 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16758 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16759 * configure: Regenerate.
16760 * config.in: Regenerate.
16761 * config/i386/driver-mingw32.c: new file.
16762 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16763 * config.host: Link driver-mingw32.o on MinGW host.
16764 * doc/install.texi: Document new --enable-mingw-wildcard configure
16767 2017-05-04 Marek Polacek <polacek@redhat.com>
16769 PR tree-optimization/80612
16770 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16772 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16773 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16775 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16776 (movt splitter): Likewise.
16777 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16778 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16779 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16780 block for Thumb-1 with MOVT.
16781 (thumb2_legitimate_address_p): Move code block ...
16782 (can_avoid_literal_pool_for_label_p): ... into this new function.
16783 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16785 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16786 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16787 "M-profile targets with the MOVT instruction".
16789 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16791 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16792 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16793 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16795 2017-05-04 Martin Liska <mliska@suse.cz>
16797 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16798 variable cond_code.
16800 2017-05-04 Richard Biener <rguenther@suse.de>
16802 * tree.c (array_at_struct_end_p): Handle arrays at struct
16803 end with flexarrays more conservatively. Refactor and treat
16804 arrays of arrays or aggregates more strict. Fix
16805 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16806 * tree.c (array_at_struct_end_p): Adjust prototype.
16807 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16808 * gimple-fold.c (get_range_strlen): Likewise.
16809 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16811 2017-05-04 Richard Biener <rguenther@suse.de>
16813 PR tree-optimization/31130
16814 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16816 (supports_overflow_infinity): Likewise.
16817 (is_negative_overflow_infinity): Likewise.
16818 (is_positive_overflow_infinity): Likewise.
16819 (is_overflow_infinity): Likewise.
16820 (stmt_overflow_infinity): Likewise.
16821 (overflow_infinity_range_p): Likewise.
16822 (usable_range_p): Remove as always returning true.
16823 (make_overflow_infinity): Remove.
16824 (negative_overflow_infinity): Likewise.
16825 (positive_overflow_infinity): Likewise.
16826 (avoid_overflow_infinity): Likewise.
16827 (set_value_range): Adjust accordingly.
16828 (set_value_range_to_nonnegative): Likewise, remove now unused
16829 overflow_infinity arg.
16830 (vrp_operand_equal_p): Adjust.
16831 (update_value_range): Likewise.
16832 (range_int_cst_singleton_p): Likewise.
16833 (operand_less_p): Likewise.
16834 (compare_values_warnv): Likewise.
16835 (extract_range_for_var_from_comparison_expr): Likewise.
16836 (vrp_int_const_binop): Likewise.
16837 (zero_nonzero_bits_from_vr): Likewise.
16838 (extract_range_from_multiplicative_op_1): Likewise.
16839 (extract_range_from_binary_expr_1): Likewise.
16840 (extract_range_from_unary_expr): Likewise.
16841 (extract_range_from_comparison): Likewise.
16842 (extract_range_basic): Likewise.
16843 (adjust_range_with_scev): Likewise.
16844 (compare_ranges): Likewise.
16845 (compare_range_with_value): Likewise.
16846 (dump_value_range): Likewise.
16847 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16849 (simplify_cond_using_ranges): Adjust.
16851 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16853 * brig-builtins.def: Added a builtin for class_f64.
16854 * builtin-types.def: Added a builtin type needed by class_f64.
16856 2017-05-03 Jason Merrill <jason@redhat.com>
16858 * timevar.def: Add TV_CONSTEXPR.
16860 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16862 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16864 2017-05-03 Martin Jambor <mjambor@suse.cz>
16866 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16867 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16868 * ipa-cp.c (ipcp_propagate_stage): Do not call
16869 ipa_update_after_lto_read.
16870 * ipa-inline.c (ipa_inline): Likewise.
16872 2017-05-03 Martin Jambor <mjambor@suse.cz>
16874 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16875 tag. Added a default constructor and a destructor.
16876 (ipa_edge_args_sum_t): New class;
16877 (ipa_edge_args_sum): Declare.
16878 (ipa_edge_args_vector): Remove declaration.
16879 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16880 (ipa_free_edge_args_substructures): Remove declaration.
16881 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16882 (ipa_edge_args_info_available_for_edge_p): Likewise.
16883 * ipa-prop.c (ipa_edge_args_vector): Removed.
16884 (edge_removal_hook_holder): Likewise.
16885 (edge_duplication_hook_holder): Likewise.
16886 (ipa_edge_args_sum): New variable.
16887 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16888 ipa_edge_args_vector.
16889 (ipa_free_edge_args_substructures): Likewise.
16890 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16891 ipa_edge_args_vector.
16892 (ipa_edge_removal_hook): Turned into method
16893 ipa_edge_args_sum_t::remove.
16894 (ipa_edge_duplication_hook): Turned into method
16895 ipa_edge_args_sum_t::duplicate.
16896 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16897 registering edge hooks.
16898 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16899 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16900 ipa_edge_args_sum instead of ipa_edge_args_vector.
16901 * ipa-profile.c (ipa_profile): Likewise.
16903 2017-05-03 Martin Jambor <mjambor@suse.cz>
16905 * symbol-summary.h (function_summary): New method exists.
16906 (function_summary::symtab_removal): Deallocate through release.
16907 (call_summary): New class.
16908 (gt_ggc_mx): New overload.
16909 (gt_pch_nx): Likewise.
16910 (gt_pch_nx): Likewise.
16912 2017-05-03 Jeff Law <law@redhat.com>
16914 PR tree-optimization/78496
16915 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16916 from simplify_cond_using_ranges. Split off code to walk
16917 backwards through casts into ...
16918 (simplify_cond_using_ranges_2): New function.
16919 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16920 (execute_vrp): After identifying jump threads, call
16921 simplify_cond_using_ranges_2.
16923 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16926 * ipa-inline.h (inline_summary): Add ctor.
16927 (create_ggc): Do not use ggc_cleared_alloc.
16929 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16930 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16932 * gcc.c (handle_braces): Support escaping in switch matching
16934 * doc/invoke.texi (Spec Files): Document it.
16935 Remove superfluous @code markup in items.
16937 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16939 * diagnostic-show-locus.c (struct column_range): New struct.
16940 (get_affected_columns): New function.
16941 (get_printed_columns): New function.
16942 (struct correction): New struct.
16943 (correction::ensure_capacity): New function.
16944 (correction::ensure_terminated): New function.
16945 (struct line_corrections): New struct.
16946 (line_corrections::~line_corrections): New dtor.
16947 (line_corrections::add_hint): New function.
16948 (layout::print_trailing_fixits): Reimplement in terms of the new
16950 (selftest::test_overlapped_fixit_printing): New function.
16951 (selftest::diagnostic_show_locus_c_tests): Call it.
16953 2017-05-03 Nathan Sidwell <nathan@acm.org>
16955 Canonicalize canonical type hashing
16956 * tree.h (type_hash_canon_hash): Declare.
16957 * tree.c (type_hash_list, attribute_hash_list): Move into
16958 type_hash_canon_hash.
16959 (build_type_attribute_qual_variant): Break out hash code calc into
16960 type_hash_canon_hash.
16961 (type_hash_canon_hash): New. Generic type hash computation.
16962 (build_range_type_1, build_array_type_1, build_function_type,
16963 build_method_type_directly, build_offset_type, build_complex_type,
16964 make_vector_type): Call it.
16966 2017-05-03 Richard Biener <rguenther@suse.de>
16968 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16969 When all DRs have unknown misaligned do not always peel
16970 when there is a store but apply the same costing model as if
16971 there were only loads.
16973 2017-05-03 Richard Biener <rguenther@suse.de>
16976 PR tree-optimization/80492
16977 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16978 compare_base_decls returning dont-know properly.
16980 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16982 * config/arm/iterators.md (CCSI): New mode iterator.
16983 (arch): New mode attribute.
16984 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16985 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16986 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16987 code iterator for success result mode.
16988 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16989 the corresponding new insn generators.
16991 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16994 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16995 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16997 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16999 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17000 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17001 (DDR_A): Wrap DDR argument in brackets.
17002 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17003 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17004 (DDR_REVERSED_P): Likewise.
17006 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17008 PR tree-optimization/79472
17009 * tree-switch-conversion.c (struct switch_conv_info): Add
17010 contiguous_range and default_case_nonstandard fields.
17011 (collect_switch_conv_info): Compute contiguous_range and
17012 default_case_nonstandard fields, don't clear final_bb if
17013 contiguous_range and only the default case doesn't have the required
17015 (check_all_empty_except_final): Set default_case_nonstandard instead
17016 of failing if contiguous_range and the default case doesn't have empty
17018 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17019 and only the default case doesn't have the required constants. Skip
17021 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17022 if default_case_nonstandard.
17023 (build_constructors): Build constant 1 just once. Assert that default
17024 values aren't inserted in between cases if contiguous_range. Skip
17026 (build_arrays): Skip virtual phis.
17027 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17028 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17029 Handle virtual phis.
17030 (gen_inbound_check): Handle default_case_nonstandard case.
17031 (process_switch): Adjust check_final_bb caller. Call
17032 gather_default_values with the first non-default case instead of
17033 default case if default_case_nonstandard.
17035 2017-05-02 Nathan Sidwell <nathan@acm.org>
17037 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17038 check. Fix formatting.
17040 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17042 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17043 errors when comparing specialized and unspecialized times.
17045 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17047 * diagnostic-show-locus.c
17048 (layout::should_print_annotation_line_p): Make private.
17049 (layout::print_annotation_line): Make private.
17050 (layout::annotation_line_showed_range_p): Make private.
17051 (layout::show_ruler): Make private.
17052 (layout::print_source_line): Make private. Pass in line and
17053 line_width, rather than calling location_get_source_line. Drop
17055 (layout::print_leading_fixits): New method.
17056 (layout::print_any_fixits): Rename to...
17057 (layout::print_trailing_fixits): ...this, and make private.
17058 Don't print newline fixits.
17059 (diagnostic_show_locus): Move logic for printing one row into...
17060 (layout::print_line): ...this new function. Move the
17061 location_get_source_line call and error-handling from
17062 print_source_line to here. Call print_leading_fixits, and rename
17063 print_any_fixits to print_trailing_fixits.
17064 (selftest::test_fixit_insert_containing_newline): Update now that
17065 newlines are partially supported.
17066 (selftest::test_fixit_insert_containing_newline_2): New test.
17067 (selftest::test_fixit_replace_containing_newline): Update comments.
17068 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17069 * edit-context.c (class added_line): New class.
17070 (class edited_line): Describe newline handling in comment.
17071 (edited_line::actually_edited_p): New method.
17072 (edited_line::print_content): Delete redundant decl.
17073 (edited_line::m_predecessors): New field.
17074 (edited_file::print_content): Call edited_line::print_content.
17075 (edited_file::print_diff): Update to support newlines.
17076 (edited_file::print_diff_hunk): Likewise.
17077 (edited_file::print_run_of_changed_lines): New function.
17078 (edited_file::print_diff_line): Convert to...
17079 (print_diff_line): ...this.
17080 (edited_file::get_effective_line_count): New function.
17081 (edited_line::edited_line): Initialize new field m_predecessors.
17082 (edited_line::~edited_line): Clean up m_predecessors.
17083 (edited_line::apply_fixit): Handle newlines.
17084 (edited_line::get_effective_line_count): New function.
17085 (edited_line::print_content): New function.
17086 (edited_line::print_diff_lines): New function.
17087 (selftest::test_applying_fixits_insert_containing_newline): New
17089 (selftest::test_applying_fixits_replace_containing_newline): New
17091 (selftest::insert_line): New function.
17092 (selftest::test_applying_fixits_multiple_lines): Add example of
17094 (selftest::edit_context_c_tests): Call the new tests.
17096 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17098 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17099 parameter cand. Update dump information.
17100 (get_computation_cost): Update uses.
17102 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17104 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17105 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17106 (get_computation_at, rewrite_use_address): Update use of
17107 get_computation_aff.
17109 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17111 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17112 (get_computation): Delete.
17113 (get_computation_cost): Implement like get_computation_cost_at.
17114 Use get_computation_at.
17115 (get_computation_cost_at): Delete.
17116 (rewrite_use_nonlinear_expr): Use get_computation_at.
17117 (rewrite_use_compare, remove_unused_ivs): Ditto.
17119 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17121 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17123 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17125 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17126 (ivopts_global_cost_for_size): Rename parameter and update uses.
17127 (iv_ca_recount_cost): Update uses.
17128 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17129 candidates seperately in n_invs and n_cands.
17130 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17132 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17134 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17135 (find_inv_vars_cb): New.
17136 (find_depends): Renamed to ...
17137 (find_inv_vars): ... this.
17138 (add_candidate_1, force_var_cost): Call find_inv_vars.
17139 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17141 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17143 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17144 inv_vars. Add inv_exprs.
17145 (struct iv_cand): Rename depends_on to inv_vars.
17146 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17147 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17148 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17149 (dump_cand): Dump inv_vars.
17150 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17151 (record_invariant, find_depends, add_candidate_1): Ditto.
17152 (set_group_iv_cost, force_var_cost): Ditto.
17153 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17154 (get_computation_cost_at, get_computation_cost): Ditto.
17155 (determine_group_iv_cost_generic): Ditto.
17156 (determine_group_iv_cost_address): Ditto.
17157 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17158 (determine_group_iv_costs): Ditto.
17159 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17160 (iv_ca_set_remove_invariants): Renamed to ...
17161 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17162 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17163 (iv_ca_set_add_invariants): Renamed to ...
17164 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17165 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17166 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17167 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17168 (create_new_ivs): Remove useless dump.
17170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17172 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17174 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17175 (iv_ca_set_no_cp, create_new_iv): Ditto.
17177 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17179 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17181 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17183 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17184 function tree_check2.
17186 2017-05-02 Martin Liska <mliska@suse.cz>
17188 * doc/gcov.texi: Add missing preposition.
17189 * gcov.c (function_info::function_info): Properly fill up
17190 all member variables.
17192 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17194 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17196 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17198 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17200 2017-05-02 Martin Liska <mliska@suse.cz>
17203 * lto-streamer-in.c (lto_read_tree_1): Remove
17204 LTO_STREAMER_DEBUG.
17205 * lto-streamer.c (struct tree_hash_entry): Likewise.
17206 (struct tree_entry_hasher): Likewise.
17207 (tree_entry_hasher::hash): Likewise.
17208 (tree_entry_hasher::equal): Likewise.
17209 (lto_streamer_init): Likewise.
17210 (lto_orig_address_map): Likewise.
17211 (lto_orig_address_get): Likewise.
17212 (lto_orig_address_remove): Likewise.
17213 * lto-streamer.h: Likewise.
17214 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17215 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17217 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17219 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17220 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17221 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17222 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17223 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17224 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17225 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17226 (mm_maskz_sub_ss): New intrinsics.
17227 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17228 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17229 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17230 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17231 (__builtin_ia32_subss_mask_round): New builtins.
17232 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17233 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17234 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17236 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17237 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17239 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17242 2017-05-02 Martin Jambor <mjambor@suse.cz>
17244 PR tree-optimization/78687
17245 * tree-sra.c (access): New field parent.
17246 (process_subtree_disqualification): New function.
17247 (disqualify_candidate): Call it.
17248 (build_accesses_from_assign): Reset write flag if creating an
17250 (build_access_subtree): Fill in parent field and also prpagate
17251 down grp_write flag.
17252 (create_artificial_child_access): New parameter set_grp_write, set
17253 grp_write to its value.
17254 (propagate_subaccesses_across_link): Also propagate grp_write flag
17256 (propagate_all_subaccesses): Push the closest parent back to work
17257 queue if add_access_to_work_queue returned true.
17259 2017-05-02 Richard Biener <rguenther@suse.de>
17261 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17262 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17263 -fstrict-overflow documentation.
17264 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17265 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17266 flag_strict_overflow.
17267 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17268 * lto-opts.c (lto_write_options): Do not stream it.
17269 * lto-wrapper.c (merge_and_complain): Do not handle it.
17270 * opts.c (default_options_table): Do not set -fstrict-overflow.
17271 (finish_options): Likewise do not clear it when sanitizing.
17272 * simplify-rtx.c (simplify_const_relational_operation): Do not
17273 test flag_strict_overflow.
17275 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17277 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17278 using enabled attribute.
17279 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17280 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17281 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17282 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17283 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17284 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17285 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17286 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17287 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17288 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17290 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17292 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17294 2017-05-02 Richard Biener <rguenther@suse.de>
17296 PR tree-optimization/80591
17298 2017-04-10 Richard Biener <rguenther@suse.de>
17300 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17303 2017-05-02 Richard Biener <rguenther@suse.de>
17305 PR tree-optimization/80549
17306 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17307 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17308 headers if they do not have a preheader.
17310 2017-05-02 Martin Liska <mliska@suse.cz>
17313 * common.opt: Fix typo.
17314 * doc/invoke.texi: Likewise.
17316 2017-05-01 Jan Beulich <jbeulich@suse.com>
17318 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17319 swapping, add (x,x,m,x,n) alternative.
17321 2017-05-01 Nathan Sidwell <nathan@acm.org>
17323 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17324 unnecessary unadjusted_alignment check.
17326 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17329 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17331 * gimplify.c (gimplify_cilk_detach): New function.
17332 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17333 * tree-core.h: Document EXPR_CILK_SPAWN.
17334 * tree.h (EXPR_CILK_SPAWN): Define.
17336 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17338 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17339 to use new fixit_hint representation, using the "replace" logic.
17340 (get_line_span_for_fixit_hint): Likewise.
17341 (layout::print_any_fixits): Likewise.
17342 (selftest::test_one_liner_many_fixits): Rename to...
17343 (selftest::test_one_liner_many_fixits_1): ...this, and update
17344 comment and expected output to reflect that the multiple fix-it
17345 hints are now consolidated into one insertion.
17346 (selftest::test_one_liner_many_fixits_2): New test.
17347 (selftest::test_diagnostic_show_locus_one_liner): Update for
17349 (selftest::test_fixit_consolidation): Update for fix-it API
17351 * diagnostic.c (print_parseable_fixits): Likewise.
17352 * edit-context.c (edited_line::m_line_events): Convert from
17353 auto_vec <line_event *> to auto_vec <line_event>.
17354 (class line_event): Convert from abstract base class to a concrete
17355 class, taking over the role of replace_event.
17356 (class insert_event): Delete.
17357 (class replace_event): Rename to class line_event. Convert to
17359 (edit_context::add_fixits): Reimplement.
17360 (edit_context::apply_insert): Delete.
17361 (edit_context::apply_replace): Rename to...
17362 (edit_context::apply_fixit): ...this. Convert to half-open range.
17363 (edited_file::apply_insert): Delete.
17364 (edited_file::apply_replace): Rename to...
17365 (edited_file::apply_fixit): ...this.
17366 (edited_line::~edited_line): Drop deletion of events.
17367 (edited_line::apply_insert): Delete.
17368 (edited_line::apply_replace): Rename to...
17369 (edited_line::apply_fixit): ...this. Convert to half-open range.
17370 Update for change to type of m_line_events.
17371 * edit-context.h (edit_context::apply_insert): Delete.
17372 (edit_context::apply_replace): Rename to...
17373 (edit_context::apply_fixit): ...this.
17375 2017-05-01 Martin Sebor <msebor@redhat.com>
17377 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17380 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17383 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17384 __get_cpuid_max returns 0.
17385 (__get_cpuid_count): Ditto.
17387 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17389 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17390 replacement expression is another instance of one of its arguments.
17392 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17395 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17396 check for stack push/pop autoinc.
17397 * config/i386/i386.c (ix86_agi_dependent): Return false
17398 if the only reason why modified_in_p returned true is that
17399 addr is SP based and set_insn is a push or pop.
17401 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17403 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17406 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17409 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17410 (account_size_time): Use two predicates - exec_pred and
17412 (evaluate_conditions_for_known_args): Compute both clause and
17414 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17415 (inline_summary_t::duplicate): Update.
17416 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17418 (compute_inline_parameters): Likewise.
17419 (estimate_edge_size_and_time): Update caluclation of time.
17420 (estimate_node_size_and_time): Compute both time and nonspecialized
17422 (estimate_ipcp_clone_size_and_time): Update.
17423 (inline_merge_summary): Update.
17424 (do_estimate_edge_time): Update.
17425 (do_estimate_edge_size): Update.
17426 (do_estimate_edge_hints): Update.
17427 (inline_read_section, inline_write_summary): Stream both new predicates.
17428 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17430 (compute_inlined_call_time): Cleanup.
17431 (big_speedup_p): Update.
17432 (edge_badness): Update.
17433 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17434 (size_time_entry): Replace predicate by exec_predicate and
17435 nonconst_predicate.
17436 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17437 (estimate_edge_time): Return also nonspec_time.
17438 (reset_edge_growth_cache): Update.
17440 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17442 PR rtl-optimization/80491
17443 * ifcvt.c (noce_process_if_block): When looking for x setter
17444 with missing else_bb, don't check only the insn right before
17445 cond_earliest, but look for the last insn that x is modified in
17446 within the same bb.
17448 PR rtl-optimization/80491
17449 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17451 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17453 PR tree-optimization/80487
17454 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17456 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17458 PR tree-optimization/79697
17459 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17460 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17461 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17463 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17464 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17466 2017-04-28 Martin Sebor <msebor@redhat.com>
17468 PR tree-optimization/80523
17469 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17470 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17472 (maybe_warn, format_directive, parse_directive): Use new functions.
17473 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17475 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17477 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17479 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17481 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17482 target_header_dir): Set correctly.
17483 * configure: Regenerated.
17484 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17485 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17486 instead of SYSTEM_HEADER_DIR.
17488 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17490 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17491 (estimate_local_effects): Likewise.
17492 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17493 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17494 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17495 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17496 do_estimate_edge_time, estimate_edge_time): Likewise.
17497 * ipa-inline-analysis.c (estimate_node_size_and_time,
17498 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17499 (estimate_time_after_inlining): Remove.
17501 2017-04-28 Martin Liska <mliska@suse.cz>
17503 * doc/gcov.texi: Enhance documentation of gcov.
17505 2017-04-28 Martin Liska <mliska@suse.cz>
17507 * doc/gcov.texi: Sort options in alphabetic order.
17508 * doc/gcov-dump.texi: Likewise.
17509 * doc/gcov-tool.texi: Likewise.
17510 * gcov.c (print_usage): Likewise.
17511 * gcov-dump.c (print_usage): Likewise.
17512 * gcov-tool.c (print_merge_usage_message): Likewise.
17513 (print_rewrite_usage_message): Likewise.
17514 (print_overlap_usage_message): Likewise.
17516 2017-04-28 Martin Liska <mliska@suse.cz>
17518 PR gcov-profile/53915
17519 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17521 2017-04-28 Martin Liska <mliska@suse.cz>
17523 PR gcov-profile/79891
17524 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17525 is marked by compiler as living on a line.
17526 (get_cycles_count): Remove usage of the union.
17527 (output_intermediate_file): Likewise.
17528 (find_source): Fix GNU coding style.
17529 (accumulate_line_counts): Remove old non-all block mode.
17530 (output_lines): Remove usage of the union.
17531 * profile.c (output_location): Include all BBs, even if
17532 belonging to a same line (and file) as a previous BB.
17534 2017-04-28 Martin Liska <mliska@suse.cz>
17536 * gcov.c (process_args): Handle new argument 'w'.
17537 (read_graph_file): Assign ID to BBs.
17538 (output_branch_count): Display BB # if verbose flag is set.
17539 (output_lines): Likewise for arcs.
17540 (print_usage): Add '--verbose' option help.
17541 * doc/gcov.texi: Document --verbose (-w) option.
17543 2017-04-28 Martin Liska <mliska@suse.cz>
17545 * gcov.c (struct block_location_info): New struct.
17546 (process_file): Fill up the new structure.
17547 (read_graph_file): Replace usage of encoding by the newly added
17549 (add_line_counts): Likewise.
17550 (accumulate_line_counts): Remove usage of the union.
17551 (function_info::function_info): New function.
17552 (function_info::~function_info): Likewise.
17553 (process_file): Call delete instead of release_function.
17554 (release_function): Release the function.
17555 (release_structures): Call delete instead of release_function.
17556 (solve_flow_graph): Replace usage of num_blocks.
17557 (find_exception_blocks): Likewise.
17558 (output_lines): Fix GNU coding style.
17560 2017-04-28 Martin Liska <mliska@suse.cz>
17563 * coverage.c (coverage_remove_note_file): New function.
17564 * coverage.h: Declare the function.
17565 * toplev.c (finalize): Clean if an error has been seen.
17567 2017-04-28 Martin Liska <mliska@suse.cz>
17569 PR gcov-profile/80031
17570 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17571 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17572 * gcov.c (read_graph_file): Read just number of blocks.
17573 * profile.c (branch_prob): Do not stream 0 flags per a basic
17576 2017-04-28 Martin Liska <mliska@suse.cz>
17578 * gcov-dump.c (tag_*): Add new argument to declarations.
17579 (dump_gcov_file): Likewise.
17580 (tag_blocks): Add and use new argument depth.
17581 (tag_arcs): Likewise.
17582 (tag_lines): Likewise.
17583 (tag_counters): Likewise.
17584 (tag_summary): Likewise.
17585 (dump_working_sets): Use depth to do a proper indentation.
17587 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17590 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17591 * symtab.c (symtab_node::debug_symtab): Move definition here.
17593 2017-04-28 Richard Biener <rguenther@suse.de>
17595 * lto-streamer.h (LTO_major_version): Bump to 7.
17597 2017-04-28 Richard Biener <rguenther@suse.de>
17599 * tree-vrp.c (assert_info): New struct.
17600 (add_assert_info): New helper.
17601 (register_edge_assert_for_2): Refactor to add asserts to a vector
17603 (register_edge_assert_for_1): Likewise.
17604 (register_edge_assert_for): Likewise.
17605 (finish_register_edge_assert_for): New helper actually registering
17606 asserts where live on edge.
17607 (find_conditional_asserts): Adjust.
17608 (find_switch_asserts): Likewise.
17609 (evrp_dom_walker::try_find_new_range): Generalize.
17610 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17612 2017-04-27 Marek Polacek <polacek@redhat.com>
17615 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17616 arg10 and arg11 to itype.
17618 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17620 * doc/extend.texi (Object Size Checking): Improve grammar.
17622 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17625 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17626 that the logic for permitting reciprocal estimates matches that
17629 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17632 * tree.c (type_cache_hasher::equal): Only compare
17633 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17634 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17635 non-aggregate element types.
17636 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17637 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17639 2017-04-27 Richard Biener <rguenther@suse.de>
17641 PR middle-end/80533
17642 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17643 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17644 keeping a reference to a trailing array.
17646 2017-04-27 Richard Biener <rguenther@suse.de>
17648 PR middle-end/80539
17649 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17650 being in loop-closed SSA form conservatively.
17651 (chrec_fold_multiply_poly_poly): Likewise.
17653 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17655 PR middle-end/79665
17656 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17657 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17659 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17662 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17663 (aarch64_function_arg_alignment): Return unsigned int again, but still
17664 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17665 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17666 Don't emit -Wpsabi note.
17667 (aarch64_function_arg_boundary): Likewise.
17668 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17671 2017-04-26 Nathan Sidwell <nathan@acm.org>
17673 * tree.h (crc32_unsigned_n): Declare.
17674 (crc32_unsigned, crc32_unsigned): Make inline.
17675 * tree.c (crc32_unsigned_bits): Replace with ...
17676 (crc32_unsigned_n): ... this.
17677 (crc32_unsigned, crc32_byte): Remove.
17678 (crc32_string): Remove unnecessary braces.
17680 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17682 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17683 * ipa-inline-analysis.c (MAX_TIME): Remove.
17684 (account_size_time): Use sreal for time.
17685 (dump_inline_summary): Update.
17686 (estimate_function_body_sizes): Update.
17687 (estimate_edge_size_and_time): Update.
17688 (estimate_calls_size_and_time): Update.
17689 (estimate_node_size_and_time): Update.
17690 (inline_merge_summary): Update.
17691 (inline_update_overall_summary): Update.
17692 (estimate_time_after_inlining): Update.
17693 (inline_read_section): Update.
17694 (inline_write_summary): Update.
17695 * ipa-inline.c (compute_uninlined_call_time): Update.
17696 (compute_inlined_call_time): Update.
17697 (recursive_inlining): Update.
17698 (inline_small_functions): Update.
17699 (dump_overall_stats): Update.
17700 * ipa-inline.h: Include sreal.h.
17701 (size_time_entry): Turn time to sreal.
17702 (inline_summary): Turn self_time nad time to sreal.
17704 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17706 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17708 (sreal::stream_out, sreal::stream_in): New.
17709 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17711 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17713 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17716 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17719 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17721 (dimode_scalar_chain::compute_convert_gain): Ditto.
17722 (dimode_scalar_chain::make_vector_copies): Ditto.
17723 (dimode_scalar_chain::convert_reg): Ditto.
17724 (dimode_scalar_chain::convert_insn): Ditto.
17725 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17726 (VI248_AVX512BW_1): New mode iterator.
17727 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17728 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17731 2017-04-25 Martin Sebor <msebor@redhat.com>
17733 PR tree-optimization/80497
17734 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17735 constants are representable in HOST_WIDE_INT.
17736 (parse_directive): Ditto.
17738 2017-04-25 Martin Sebor <msebor@redhat.com>
17741 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17742 (new_zero_array): Adjust signature.
17743 (dom_info::dom_init): Used unsigned rather that size_t.
17744 (dom_info::dom_info): Same.
17746 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17747 Jakub Jelinek <jakub@redhat.com>
17750 * config/arm/arm.c: Include gimple.h.
17751 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17752 returns negative, increment ncrn only if it returned positive.
17753 (arm_needs_doubleword_align): Return int instead of bool,
17754 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17755 members, but if there is any such non-FIELD_DECL
17756 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17757 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17758 returns negative, increment nregs only if it returned positive.
17759 (arm_setup_incoming_varargs): Likewise.
17760 (arm_function_arg_boundary): Emit -Wpsabi note if
17761 arm_needs_doubleword_align returns negative, return
17762 DOUBLEWORD_ALIGNMENT only if it returned positive.
17764 2017-04-25 Marek Polacek <polacek@redhat.com>
17767 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17768 first argument to type.
17770 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17773 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17774 type checks to test for compatibility instead of equality.
17776 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17777 Jakub Jelinek <jakub@redhat.com>
17780 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17782 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17783 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17784 the alignment computation, but return their maximum in warn_alignment.
17785 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17786 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17788 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17789 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17792 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17794 * config/arc/simdext.md (dmpyh): Fix typo.
17796 2017-04-25 Richard Biener <rguenther@suse.de>
17798 PR tree-optimization/80492
17799 * alias.c (compare_base_decls): Handle registers with asm
17800 specification conservatively.
17801 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17802 compare_base_decls returning dont-know properly.
17804 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17806 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17807 (legitimate_offset_address_p): New function.
17808 (arc_legitimate_address_p): Use above function.
17810 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17812 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17814 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17816 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17817 ACCH registers whenever they are available.
17819 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17821 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17822 double regs fix when not used.
17824 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17826 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17828 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17829 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17831 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17833 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17834 when emitting short ADD instructions.
17836 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17838 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17840 (cmpsi_cc_c_insn): Likewise.
17841 (cbranchsi4_scratch): Compute proper instruction length using
17842 compact_hreg_operand.
17843 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17845 2017-04-25 Richard Biener <rguenther@suse.de>
17847 PR middle-end/80509
17848 * passes.c (pass_manager::pass_manager): Initialize
17849 m_name_to_pass_map.
17851 2017-04-25 Richard Biener <rguenther@suse.de>
17853 PR tree-optimization/79201
17854 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17856 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17859 * config/s390/vector.md: Split MEM->GPR vector moves for
17860 non-s_operand addresses.
17862 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17865 * config/s390/predicates.md (reload_const_wide_int_operand): New
17867 * config/s390/s390.md ("movti"): Remove d/P alternative.
17868 ("movti_bigconst"): New pattern definition.
17870 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17873 * s390-protos.h (s390_expand_cs_hqi): Removed.
17874 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17875 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17876 modes as well as CCZ1mode and CCZmode.
17877 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17878 signature of s390_emit_compare_and_swap.
17879 (s390_expand_cs_hqi): Likewise, make static.
17880 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17881 compare-and-swap, in some cases.
17882 (s390_expand_cs): Wrapper function.
17883 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17885 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17886 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17887 patterns for small and large integers. Forbid symref memory operands.
17888 Move expander to s390.c. Require cc register.
17889 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17890 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17891 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17892 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17893 symref memory operands. Remove CC mode and call s390_match_ccmode
17895 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17897 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17899 * config/s390/s390.md (define_peephole2): New peephole to help
17900 combining the load-and-test pattern with volatile memory.
17902 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17904 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17905 with CCZmode for TARGET_Z196.
17907 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17909 PR rtl-optimization/80501
17910 * combine.c (make_compound_operation_int): Set subreg_code to SET
17911 even for AND with mask of the sign bit of mode.
17913 PR rtl-optimization/80500
17914 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17915 sum's initial value.
17917 2017-04-25 Julian Brown <julian@codesourcery.com>
17918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17920 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17922 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17924 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17926 2017-04-25 Julian Brown <julian@codesourcery.com>
17927 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17929 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17930 (thunderx2t99_sha): New Reservation.
17932 2017-04-25 Julian Brown <julian@codesourcery.com>
17933 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17935 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17936 type for 1-element load.
17938 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17940 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17942 2017-04-24 Martin Jambor <mjambor@suse.cz>
17944 PR tree-optimization/80293
17945 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17946 char arrays not totally scalarizable if it is false.
17947 (analyze_all_variable_accesses): Pass correct value in the new
17948 parameter. Add a statistics counter.
17950 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17952 PR middle-end/79931
17953 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17955 2017-04-24 Richard Biener <rguenther@suse.de>
17957 PR tree-optimization/80494
17958 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17959 out for complex types.
17961 2017-04-24 Richard Biener <rguenther@suse.de>
17963 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17964 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17965 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17966 (DFS): Adjust and never fail.
17967 (sccvn_dom_walker::fail): Remove.
17968 (sccvn_dom_walker::before_dom_children): Adjust.
17969 (run_scc_vn): Likewise and never fail.
17970 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17971 (pass_fre::execute): Likewise.
17973 2017-04-24 Richard Biener <rguenther@suse.de>
17975 PR tree-optimization/79725
17976 * tree-ssa-sink.c (statement_sink_location): Return whether
17977 failure reason was zero uses. Move that check later.
17978 (sink_code_in_bb): Deal with zero uses by removing the stmt
17981 2017-04-24 Richard Biener <rguenther@suse.de>
17984 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17985 pointer-based references.
17987 2017-04-24 Richard Biener <rguenther@suse.de>
17990 * pass_manager.h (pass_manager::operator new): Remove.
17991 (pass_manager::operator delete): Likewise.
17992 * passes.c (pass_manager::operator new): Remove.
17993 (pass_manager::operator delete): Likewise.
17994 (pass_manager::pass_manager): Zero individual pass members.
17996 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17999 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18000 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18001 Check "XEXP (src, 1)" operand here.
18002 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18003 Check "XEXP (src, 1)" operand here.
18004 (dimode_scalar_chain::make_vector_copies): Detect count register
18005 of a shift instruction. Zero extend count register from QImode
18006 to DImode to satisfy vector shift pattern count operand predicate.
18007 Substitute vector shift count operand with a DImode copy.
18008 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18011 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18013 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18014 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18015 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18016 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18017 (UNSPEC_NOREX_MEM): Remove definition.
18019 2017-04-21 Richard Biener <rguenther@suse.de>
18021 PR tree-optimization/79547
18022 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18023 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18024 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18025 without any constraints.
18027 2017-04-21 Richard Biener <rguenther@suse.de>
18029 PR tree-optimization/78847
18030 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18032 2017-04-21 Richard Biener <rguenther@suse.de>
18034 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18035 (build_distinct_type_copy): Likewise.
18036 (build_variant_type_copy): Likewise.
18037 * tree.c (build_qualified_type): Pass down mem-stat info.
18038 (build_distinct_type_copy): Likewise.
18039 (build_variant_type_copy): Likewise.
18041 2017-04-21 Richard Biener <rguenther@suse.de>
18043 PR tree-optimization/80237
18044 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18046 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18047 for a simplified result.
18049 2016-04-21 Richard Biener <rguenther@suse.de>
18051 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18052 sth as strict as a simple_iv but a chrec without symbols and an
18053 operand defined in the loop we are peeling (and not some subloop).
18054 (propagate_constants_for_unrolling): Propagate all constants.
18056 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18059 * config/i386/i386.c (print_reg): Remove assert for disalowed
18060 regno values, call output_operand_lossage instead.
18062 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18065 * config/i386/constraints.md (Yc): New register constraint.
18066 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18067 Use Yc constraint for alternative 2 of operand 0. Remove
18068 preferred_for_speed attribute.
18070 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18072 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18073 lastprivate clauses in SIMT case.
18075 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18077 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18079 2017-04-20 Richard Biener <rguenther@suse.de>
18081 PR tree-optimization/57796
18082 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18083 as N scalar stores.
18084 (vect_model_load_cost): Cost gathers as N scalar loads.
18086 2017-04-20 Richard Biener <rguenther@suse.de>
18088 * ggc-page.c (ggc_allocated_p): Rename to ...
18089 (safe_lookup_page_table_entry): ... this and return the lookup
18091 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18093 2017-04-20 Richard Biener <rguenther@suse.de>
18095 PR tree-optimization/80453
18096 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18097 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18098 from the conditions.
18099 (vn_phi_eq): Pass them down.
18100 (vn_phi_lookup): Record them.
18101 (vn_phi_insert): Likewise.
18103 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18105 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18106 uninitialized variable warning to avoid buffer overrun.
18108 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18111 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18112 is suppressed for '{ 0 }' in C.
18114 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18116 * BASE-VER: Set to 8.0.0.
18118 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18120 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18121 priority .init_array and .fini_array section with SECTION_NOTYPE
18124 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18126 PR middle-end/80423
18127 * tree.h (build_array_type): Add typeless_storage default argument.
18128 * tree.c (type_cache_hasher::equal): Also compare
18129 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18130 (build_array_type): Add typeless_storage argument, set
18131 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18133 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18134 (build_array_type): Likewise. Add typeless_storage argument.
18136 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18137 Jakub Jelinek <jakub@redhat.com>
18139 PR tree-optimization/80426
18140 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18141 operation on symbolic operands, also compute the overflow for the
18142 invariant part when the operation degenerates into a negation.
18144 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18147 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18148 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18151 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18153 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18156 * config/avr/avr.c (tree.h): Include it.
18157 (cgraph.h): Include it.
18158 (avr_encode_section_info): Don't warn for uninitialized progmem
18159 variable if it's just an alias.
18161 2017-04-19 Richard Biener <rguenther@suse.de>
18164 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18165 when needed by AutoPGO.
18167 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18170 * doc/lto.texi: Remove an extra 'that'.
18172 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18174 PR rtl-optimization/80429
18175 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18176 are only used in debug insns.
18178 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18179 Vladimir Makarov <vmakarov@redhat.com>
18181 * config/sparc/predicates.md (input_operand): Add comment. Return
18182 true for any memory operand when LRA is in progress.
18183 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18185 2017-04-18 Jeff Law <law@redhat.com>
18188 * mips.md ({return,simple_return}_internal): Do not overwrite
18191 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18193 PR tree-optimization/80443
18194 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18195 instead of adding 1, subtract -1 and similarly instead of subtracting
18198 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18200 PR rtl-optimization/80357
18201 * haifa-sched.c (tmp_bitmap): New variable.
18202 (model_recompute): Handle duplicate use records.
18203 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18204 (free_global_sched_pressure_data): Free it.
18206 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18209 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18210 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18211 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18212 instead of SYSTEM_HEADER_DIR.
18214 2017-04-18 Jeff Law <law@redhat.com>
18216 PR middle-end/80422
18217 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18218 predecessors after walking up the insn chain.
18220 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18223 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18224 sizetype type into debug info.
18226 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18229 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18230 unneeded test for TARGET_UPPER_REGS_SF.
18231 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18233 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18236 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18237 instead of gsi_after_labels.
18239 2017-04-18 Jeff Law <law@redhat.com>
18241 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18245 2017-04-13 Jeff Law <law@redhat.com>
18246 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18247 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18249 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18252 * config/avr/avr.c (intl.h): Include it.
18253 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18255 2017-04-18 Martin Liska <mliska@suse.cz>
18257 PR gcov-profile/78783
18258 * gcov-tool.c (gcov_output_files): Validate that destination
18259 file is either removed by the tool or by a user.
18261 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18262 Guy Benyei <guybe@mellanox.com>
18264 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18265 block, and do not negate it, the stored id is already negative.
18267 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18269 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18271 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18274 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18275 masks of options that should be turned off if the VSX vector
18276 options are turned off.
18277 (OTHER_P8_VECTOR_MASKS): Likewise.
18278 (OTHER_VSX_VECTOR_MASKS): Likewise.
18279 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18280 rs6000_disable_incompatible_switches to validate no type switches
18282 (rs6000_incompatible_switch): New function to disallow turning on
18283 other vector options if -mno-vsx, -mno-power8-vector, or
18284 -mno-power9-vector are specified.
18286 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18288 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18290 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18292 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18293 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18294 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18295 (ARG_POINTER_CFA_OFFSET): Likewise.
18297 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18299 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18300 conditions to take advantage of various optimizations.
18302 2017-04-13 Jeff Law <law@redhat.com>
18304 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18305 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18306 (zero_extendsidi2_dext): Likewise.
18308 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18311 * fold-const.c (fold_ternary_loc): Revert
18312 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18315 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18317 PR rtl-optimization/80343
18318 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18319 new scratch pseudo.
18321 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18324 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18325 to ubsan_encode_value.
18327 2017-04-13 Jeff Law <law@redhat.com>
18329 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18330 appearing in DEBUG_INSNs.
18332 2017-04-13 Martin Liska <mliska@suse.cz>
18334 PR gcov-profile/80413
18335 * gcov-io.c (gcov_write_string): Copy to buffer just when
18336 allocated size is greater than zero.
18338 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18341 * dwarf2out.c (decls_for_scope): Ignore declarations of
18342 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18344 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18347 * ipa-visibility.c (non_local_p): Fix typos.
18348 (localize_node): When localizing symbol in same comdat group,
18349 dissolve the group only when we know external symbols are going
18351 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18353 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18355 PR tree-optimization/79390
18356 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18357 order does not result in usable sequence, retry with reversed operand
18363 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18364 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18365 op0 instead of fold_convert_loc (loc, type, arg0).
18367 2017-04-12 Jeff Law <law@redhat.com>
18369 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18370 has a delay slot in the generated code.
18372 * config/cris/cris.md (cris_preferred_reload_class): Return
18373 GENNONACR_REGS rather than GENERAL_REGS.
18375 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18378 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18379 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18380 signedness of the result type.
18382 2017-04-12 Richard Biener <rguenther@suse.de>
18383 Jeff Law <law@redhat.com>
18385 PR tree-optimization/80359
18386 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18387 trim stores to TARGET_MEM_REFs.
18389 2017-04-12 Richard Biener <rguenther@suse.de>
18391 PR tree-optimization/79390
18392 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18393 threading case even more.
18395 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18398 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18399 for quad_address_p for TImode, instead of just not indexed_address.
18401 2017-04-12 Richard Biener <rguenther@suse.de>
18402 Bernd Edlinger <bernd.edlinger@hotmail.de>
18404 PR middle-end/79671
18405 * alias.c (component_uses_parent_alias_set_from): Handle
18406 TYPE_TYPELESS_STORAGE.
18407 (get_alias_set): Likewise.
18408 * tree-core.h (tree_type_common): Add typeless_storage flag.
18409 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18410 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18411 for types containing members with TYPE_TYPELESS_STORAGE.
18412 (place_field): Likewise.
18413 (layout_type): Likewise for ARRAY_TYPE.
18414 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18415 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18416 TYPE_TYPELESS_STORAGE.
18417 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18419 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18422 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18423 first argument to type.
18425 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18429 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18430 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18431 (rs6000_expand_binop_builtin): Likewise.
18432 (rs6000_expand_ternop_builtin): Likewise; also add missing
18433 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18434 vshasigma built-ins.
18435 * doc/extend.texi: Document that vec_xxpermdi's third argument
18436 must be a constant.
18438 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18440 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18441 Use shift_const cost parameter when calculating gain of STV shifts.
18443 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18445 PR rtl-optimization/70478
18446 * lra-constraints.c (process_alt_operands): Check memory for
18447 disfavoring memory insn operand.
18449 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18451 PR middle-end/80100
18452 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18453 left shift in unsigned HOST_WIDE_INT type.
18455 PR rtl-optimization/80385
18456 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18457 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18460 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18461 if they have any depend clauses.
18463 2017-04-11 Martin Liska <mliska@suse.cz>
18466 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18467 * ipa-split.c (split_function): Create a local comdat symbol
18468 if caller is in a comdat group.
18470 2017-04-11 Martin Liska <mliska@suse.cz>
18473 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18476 2017-04-11 Martin Sebor <msebor@redhat.com>
18478 PR middle-end/80364
18479 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18480 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18482 (directive::set_width, directive::set_precision, format_character):
18484 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18487 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18490 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18491 conflict, set target->arch_name instead of target->cpu_name.
18493 2017-04-11 Richard Biener <rguenther@suse.de>
18495 PR tree-optimization/80374
18496 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18497 build_zero_cst, remove fold_convertible_p check again.
18499 2017-04-11 Martin Liska <mliska@suse.cz>
18502 * ubsan.c (instrument_object_size): Do not instrument register
18505 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18508 * config/i386/i386-builtin-types.def
18509 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18510 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18511 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18512 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18513 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18514 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18515 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18516 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18517 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18518 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18519 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18520 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18521 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18522 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18523 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18524 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18525 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18526 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18527 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18528 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18529 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18530 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18531 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18532 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18533 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18534 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18535 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18536 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18537 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18538 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18539 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18540 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18541 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18542 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18543 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18544 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18545 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18546 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18547 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18548 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18549 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18550 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18551 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18552 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18553 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18554 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18556 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18557 flag to second_arg_count, handle 4 argument function type _COUNT
18558 aliases, handle second_arg_count on second argument rather than last.
18560 2017-04-10 Jeff Law <law@redhat.com>
18562 PR tree-optimization/80374
18563 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18564 record anything if we can not convert integer_zero_node to the
18567 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18570 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18571 Enhance special handling given to the TARGET_P9_MINMAX option in
18572 relation to certain other options.
18574 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18576 PR tree-optimization/80153
18577 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18578 remove POINTER_PLUS_EXPR's base part directly, rather than through
18581 2017-04-10 Richard Biener <rguenther@suse.de>
18582 Bin Cheng <bin.cheng@arm.com>
18584 PR tree-optimization/80153
18585 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18586 the first element of pointer type aff_tree. Build result expr in
18588 (add_elt_to_tree): Convert to type unconditionally. Remove other
18589 fold_convert calls.
18590 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18591 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18593 2017-04-10 Richard Biener <rguenther@suse.de>
18595 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18598 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18600 PR rtl-optimization/70478
18601 * lra-constraints.c (curr_small_class_check): New.
18602 (update_and_check_small_class_inputs): New.
18603 (process_alt_operands): Update curr_small_class_check. Disfavor
18604 alternative insn memory operands. Check available regs for small
18607 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18610 * config/mips/mips.opt (-mvirt): Update description.
18611 * doc/invoke.texi (-mvirt): Likewise.
18613 2017-04-10 Richard Biener <rguenther@suse.de>
18615 PR middle-end/80362
18616 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18617 looking for NEGATE_EXPR in -A / -B to A / B folding.
18619 2017-04-10 Martin Liska <mliska@suse.cz>
18621 PR gcov-profile/80224
18622 * gcov.c (print_usage): Fix usage string.
18623 (get_gcov_intermediate_filename): Remove.
18624 (output_gcov_file): Use both for normal and intermediate format.
18625 (generate_results): Do not initialize special file for
18626 intermediate format.
18628 2017-04-10 Richard Biener <rguenther@suse.de>
18630 PR tree-optimization/80304
18631 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18634 2017-04-10 Nathan Sidwell <nathan@acm.org>
18637 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18638 (rs6000_init_builtins): Use it.
18640 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18642 * config/arm/arm.md (<mrc>): Add mode to SET source.
18643 (<mrrc>): Likewise.
18645 2017-04-10 Richard Biener <rguenther@suse.de>
18647 PR middle-end/80344
18648 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18650 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18653 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18654 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18655 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18656 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18657 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18658 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18659 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18660 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18661 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18662 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18663 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18664 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18665 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18666 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18667 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18668 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18669 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18670 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18671 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18672 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18673 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18674 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18675 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18677 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18679 PR rtl-optimization/70478
18680 * lra-constraints.c: Reverse the last patch.
18682 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18684 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18685 Add comment for WCHAR_T.
18687 2017-04-08 Martin Liska <mliska@suse.cz>
18690 2017-04-07 Martin Liska <mliska@suse.cz>
18693 * ipa-split.c (split_function): Add function part to a same comdat
18696 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18699 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18701 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18703 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18704 (vec_pairing): ...to this.
18705 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18706 (rs6000_sched_init): Adjust for name change.
18707 (struct rs6000_sched_context): Likewise.
18708 (rs6000_init_sched_context): Likewise.
18709 (rs6000_set_sched_context): Likewise.
18711 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18717 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18719 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18720 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18721 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18723 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18725 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18727 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18729 PR rtl-optimization/70703
18730 * ira-color.c (update_conflict_hard_regno_costs): Use
18731 int64_t instead of HOST_WIDE_INT.
18733 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18735 PR rtl-optimization/70478
18736 * lra-constraints.c (process_alt_operands): Disfavor alternative
18737 insn memory operands.
18739 2017-04-07 Jeff Law <law@redhat.com>
18741 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18742 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18744 2017-04-07 Martin Liska <mliska@suse.cz>
18747 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18748 Show error message instead of an ICE.
18750 2017-04-07 Martin Liska <mliska@suse.cz>
18753 * ipa-split.c (split_function): Add function part to a same comdat
18756 2017-04-07 Richard Biener <rguenther@suse.de>
18758 PR middle-end/80341
18759 * tree.c (get_unwidened): Also handle ! for_type case for
18761 * convert.c (do_narrow): Split out from ...
18762 (convert_to_integer_1): ... here. Do not pass final truncation
18763 type to get_unwidened for TRUNC_DIV_EXPR.
18765 2017-04-07 Richard Biener <rguenther@suse.de>
18767 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18769 (aff_combination_const): Adjust.
18770 (aff_combination_scale): Likewise.
18771 (aff_combination_add_elt): Likewise.
18772 (aff_combination_add_cst): Likewise.
18773 (aff_combination_convert): Likewise.
18774 (add_elt_to_tree): Likewise. Remove unused argument.
18775 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18777 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18779 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18781 * config/arm/arm.c (arm_default_short_enums): Use
18782 ARM_DEFAULT_SHORT_ENUMS.
18783 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18785 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18788 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18789 members with redundant out-of-class redeclaration.
18791 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18794 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18795 * config/i386/i386.md (*zero_extendsidi2):
18796 Add (?*x,*x) and (?*v,*v) alternatives.
18798 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18801 * config/i386/i386.c (ix86_expand_builtin)
18802 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18803 mode from insn data. Convert operands to insn operand mode.
18804 Copy operands that don't satisfy insn predicate to a register.
18806 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18808 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18811 2017-04-06 Richard Biener <rguenther@suse.de>
18813 PR tree-optimization/80334
18814 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18815 preserve alignment of accesses.
18817 2017-04-06 Richard Biener <rguenther@suse.de>
18819 PR tree-optimization/80262
18820 * tree-sra.c (build_ref_for_offset): Preserve address-space
18822 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18823 Drop useless address-space information on MEM_REF offsets.
18825 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18827 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18829 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18831 PR rtl-optimization/70703
18832 * ira-color.c (update_conflict_hard_regno_costs): Use
18833 HOST_WIDE_INT instead of long.
18835 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18838 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18839 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18841 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18842 for x86_64 target. Handle -m3dnowa option.
18844 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18846 PR rtl-optimization/70703
18847 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18848 (update_conflict_hard_regno_costs): Use long instead of unsigned
18849 arithmetic for cost calculation.
18851 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18852 Bernd Edlinger <bernd.edlinger@hotmail.de>
18855 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18858 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18861 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18862 ptr_mode with Pmode throughout.
18863 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18864 into probe_stack_range and use DImode.
18866 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18869 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18870 call_eh_return is true.
18872 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18874 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18875 Initialize last_match_fntype_index.
18877 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18880 * tree-nvr.c: Include internal-fn.h.
18881 (pass_return_slot::execute): Ignore internal calls without
18884 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18885 Richard Biener <rguenther@suse.de>
18888 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18889 captures used multiple times, except for the last use.
18890 * generic-match-head.c: Include gimplify.h.
18892 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18894 PR tree-optimization/79390
18895 * target.h (struct noce_if_info): Declare.
18896 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18897 * target.def (noce_conversion_profitable_p): New target hook.
18898 * ifcvt.h (struct noce_if_info): New type, moved from ...
18899 * ifcvt.c (struct noce_if_info): ... here.
18900 (noce_conversion_profitable_p): Renamed to ...
18901 (default_noce_conversion_profitable_p): ... this. No longer
18903 (noce_try_store_flag_constants, noce_try_addcc,
18904 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18905 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18906 instead of noce_conversion_profitable_p.
18907 * config/i386/i386.c: Include ifcvt.h.
18908 (ix86_option_override_internal): Don't override
18909 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18910 (ix86_noce_conversion_profitable_p): New function.
18911 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18912 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18913 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18914 * doc/tm.texi: Regenerated.
18916 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18918 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18921 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18924 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18925 instructions for small multiply cores.
18927 2017-04-04 Jeff Law <law@redhat.com>
18929 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18931 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18932 that are not set by the loop over the elements.
18934 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18937 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18938 int mode, convert_modes it to mode as unsigned, otherwise use
18939 lowpart_subreg to mode rather than SImode.
18940 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18941 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18942 Use DImode instead of SImode for the shift count operand.
18943 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18946 2017-04-04 Richard Biener <rguenther@suse.de>
18948 PR middle-end/80281
18949 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18950 arithmetic done for the negate or the plus. Simplify.
18951 (A - (-B) -> A + B): Likewise.
18952 * fold-const.c (split_tree): Make sure to not negate pointers.
18954 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18956 PR rtl-optimization/60818
18957 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18958 a compare of comparisons with the thing compared if this results
18959 in a different machine mode.
18961 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18963 * alias.c (base_alias_check): Fix typo in comment.
18964 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18965 * cgraphunit.c (symbol_table::compile): Likewise.
18966 * collect2.c (maybe_run_lto_and_relink): Likewise.
18967 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18968 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18969 * config/avr/avr.c (avr_map_op_t): Likewise.
18970 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18971 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18972 * config/epiphany/epiphany.md (movcc): Likewise.
18973 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18974 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18976 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18977 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18978 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18979 * config/sh/sh.c (sh_rtx_costs): Likewise.
18980 * fold-const.c (fold_truth_andor): Likewise.
18981 * genautomata.c (collapse_flag): Likewise.
18982 * gengtype.h (struct type::u::s): Likewise.
18983 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18984 * input.c (FORMAT_AMOUNT): Likewise.
18985 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18986 (known_aggs_to_agg_replacement_list): Likewise.
18987 * ipa-inline-analysis.c: Likewise.
18988 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18989 * ipa-polymorphic-call.c
18990 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18991 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18992 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18994 * modulo-sched.c (apply_reg_moves): Likewise.
18995 * omp-expand.c (build_omp_regions_1): Likewise.
18996 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18997 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18998 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18999 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19000 * value-prof.c: Likewise.
19001 * var-tracking.c (val_reset): Likewise.
19003 2017-04-03 Richard Biener <rguenther@suse.de>
19005 PR tree-optimization/80275
19006 * fold-const.c (split_address_to_core_and_offset): Handle
19009 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19011 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19012 descriptors is at least equal to that of functions.
19014 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19016 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19018 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19021 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19022 (mov<IMOD4:mode>): New expander.
19023 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19025 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19027 PR rtl-optimization/79405
19028 * fwprop.c (propagations_left): New variable.
19029 (forward_propagate_into): Decrement it.
19030 (fwprop_init): Initialize it.
19031 (fw_prop): If the variable has reached zero, stop propagating.
19032 (fwprop_addr): Ditto.
19034 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19037 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19038 a FUNCTION_DECL, pass it as decl instead of origin to
19041 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19043 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19046 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19049 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19050 TARGET_VSX_SMALL_INTEGER.
19052 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19054 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19055 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19057 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19059 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19060 extraction from odd-numbered MSA register.
19062 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19064 PR middle-end/80173
19065 * expmed.c (store_bit_field_1): Don't attempt to create
19066 a word subreg out of hard registers wider than word if they
19067 have HARD_REGNO_NREGS of 1 for their mode.
19069 PR middle-end/80163
19070 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19071 conversions to integer types wider than word and pointer.
19074 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19075 (rtx_equal_for_cselib_p): Pass 0 to it.
19076 * cselib.c (cselib_hasher::equal): Likewise.
19077 (rtx_equal_for_cselib_1): Add depth argument. If depth
19078 is 128, don't look up VALUE locs and punt. Increment
19079 depth in recursive calls when walking VALUE locs.
19081 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19083 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19084 (make_gcov_file_name): Use the canonical path name for generating
19086 (read_line): Fix handling of files with ascii null bytes.
19088 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19090 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19091 to initialise a vector register instead
19092 of using a const_int.
19094 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19096 PR translation/80189
19097 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19098 diagnostic messages.
19100 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19103 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19104 (dfp_diex_<mode>): Update mode of operand 1.
19105 * doc/extend.texi (dxex, dxexq): Document change to return type.
19106 (diex, diexq): Document change to argument type.
19108 2017-03-30 Martin Jambor <mjambor@suse.cz>
19111 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19112 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19113 it reflects the signature changes performed at the callee side.
19114 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19115 to cgraph_build_function_type_skip_args.
19116 (build_function_decl_skip_args): Adjust call to the above function.
19118 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19121 * config/i386/sse.md
19122 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19123 register as dest whenever it is a MEM not rtx_equal_p to the
19124 corresponding dup operand, and when forcing into reg move the
19125 reg into the memory afterwards.
19126 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19127 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19128 for the force_reg mode.
19129 (avx512vl_vextractf128<mode>): Use register as dest either
19130 always when a MEM, or when it is a MEM not rtx_equal_p to the
19131 corresponding dup operand, or even not when it is a CONST_VECTOR
19132 depending on the mode and lo vs. hi.
19133 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19135 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19136 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19137 Likewise. Require that operands[2] is even.
19138 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19139 Remove extraneous parens. Require that operands[2] is a multiple
19141 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19142 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19143 disallow memory then.
19145 2017-03-30 Richard Biener <rguenther@suse.de>
19147 PR tree-optimization/77498
19148 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19149 to non-constants over backedges.
19151 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19153 PR rtl-optimization/80233
19154 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19155 as last_combined_insn. Do not test for BARRIER_P separately.
19157 2017-03-29 Andreas Schwab <schwab@suse.de>
19160 * calls.c (prepare_call_address): Convert funexp to Pmode before
19161 copying to temp reg.
19163 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19165 PR tree-optimization/80158
19166 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19167 Handle possible future case of more than one alternate
19169 (replace_rhs_if_not_dup): Likewise.
19170 (replace_one_candidate): Likewise.
19172 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19174 PR rtl-optimization/80193
19175 * ira.c (ira): Do not check allocation for LRA.
19177 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19179 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19180 (nvptx_output_simt_exit): Declare.
19181 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19182 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19183 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19184 (nvptx_declare_function_name): ...here. Emit declaration of local
19185 memory space buffer for omp_simt_enter insn.
19186 (nvptx_output_unisimt_switch): New.
19187 (nvptx_output_softstack_switch): New.
19188 (nvptx_output_simt_enter): New.
19189 (nvptx_output_simt_exit): New.
19190 * config/nvptx/nvptx.h (struct machine_function): New fields
19191 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19192 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19193 (UNSPECV_SIMT_EXIT): Ditto.
19194 (omp_simt_enter_insn): New insn.
19195 (omp_simt_enter): New expansion.
19196 (omp_simt_exit): New insn.
19197 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19199 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19200 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19201 (expand_GOMP_SIMT_EXIT): New.
19202 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19203 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19204 (GOMP_SIMT_EXIT): Ditto.
19205 * target-insns.def (omp_simt_enter): New insn.
19206 (omp_simt_exit): Ditto.
19207 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19209 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19210 (lower_rec_input_clauses): Likewise.
19211 (lower_lastprivate_clauses): Handle SIMT privatization.
19213 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19214 (ompdevlow_adjust_simt_enter): New.
19215 (find_simtpriv_var_op): New.
19216 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19217 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19219 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19220 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19221 (copy_decl_for_dup_finish): Ditto.
19223 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19225 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19228 * config/i386/i386.c (ix86_option_override_internal): Always
19229 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19231 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19233 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19235 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19237 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19238 mark new edge's irreducible flag accordign to it.
19239 (vect_do_peeling): Check loop preheader edge's irreducible flag
19240 and pass it to function slpeel_add_loop_guard.
19242 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19244 PR tree-optimization/80218
19245 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19246 Update block frequencies and counts.
19248 2017-03-28 Richard Biener <rguenther@suse.de>
19250 PR tree-optimization/78644
19251 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19252 of a simplification result we may not use it at all.
19254 2017-03-28 Richard Biener <rguenther@suse.de>
19257 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19258 without arguments, generate default definition of a SSA name.
19260 2017-03-28 Richard Biener <rguenther@suse.de>
19262 PR middle-end/80222
19263 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19264 TYPE_REF_CAN_ALIAS_ALL references.
19265 * fold-const.c (fold_indirect_ref_1): Likewise.
19267 2017-03-28 Martin Liska <mliska@suse.cz>
19270 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19271 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19273 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19274 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19276 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19277 (EXTRA_SPECS): Define.
19278 (SUBTARGET_EXTRA_SPECS): Likewise.
19279 (SUBTARGET_CPP_SPEC): Likewise.
19280 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19281 SUBTARGET_EXTRA_SPECS.
19282 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19284 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19286 * config/arc/simdext.md (vst64_insn): Update pattern.
19287 (vld32wh_insn): Likewise.
19288 (vld32wl_insn): Likewise.
19289 (vld64_insn): Likewise.
19290 (vld32_insn): Likewise.
19292 2017-03-28 Marek Polacek <polacek@redhat.com>
19295 * fold-const.c (fold_comparison): Use protected_set_expr_location
19296 instead of SET_EXPR_LOCATION.
19298 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19300 * tree.c (add_expr): Avoid name lookup warning.
19302 2017-03-27 Jeff Law <law@redhat.com>
19304 PR tree-optimization/80216
19305 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19306 function name. Limit recursion depth.
19307 (record_temporary_equivalences): Corresponding changes.
19309 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19311 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19314 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19317 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19319 * cfgcleanup.c (reg_note_cfa_p): New array.
19320 (insns_have_identical_cfa_notes): New function.
19321 (old_insns_match_p): Don't cross-jump in between /f
19322 and non-/f instructions. If both i1 and i2 are frame related,
19323 verify all CFA notes, their order and content.
19325 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19328 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19329 HImode and SImode with zero extend to DImode to one insn.
19330 (bswap<mode>2_extenddi): Likewise.
19331 (bswapsi2_extenddi): Likewise.
19332 (bswaphi2_extendsi): Likewise.
19333 (bswaphi2): Combine bswap HImode and SImode into one insn.
19334 Separate memory insns from swapping register.
19335 (bswapsi2): Likewise.
19336 (bswap<mode>2): Likewise.
19337 (bswaphi2_internal): Delete, no longer used.
19338 (bswapsi2_internal): Likewise.
19339 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19340 store, and gpr<-gpr swap insns.
19341 (bswap<mode>2_store): Likewise.
19342 (bswaphi2_reg): Register only splitter, combine with the splitter.
19343 (bswaphi2 splitter): Likewise.
19344 (bswapsi2_reg): Likewise.
19345 (bswapsi2 splitter): Likewise.
19346 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19347 the insns into load, store, and register/register insns.
19348 (bswapdi2_ldbrx): Likewise.
19349 (bswapdi2_load): Likewise.
19350 (bswapdi2_store): Likewise.
19351 (bswapdi2_reg): Likewise.
19353 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19355 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19356 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19358 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19361 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19363 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19364 special handling for target option conflicts between dform
19365 options (-mpower9-dform, -mpower9-dform-vector,
19366 -mpower9-dform-scalar) and -mno-direct-move.
19368 2017-03-27 Richard Biener <rguenther@suse.de>
19370 PR tree-optimization/80181
19371 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19373 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19375 * config/arc/predicates.md (move_double_src_operand): Replace the
19376 call to move_double_src_operand with a call to address_operand.
19378 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19380 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19381 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19382 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19384 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19386 * config/arc/predicates.md (long_immediate_loadstore_operand):
19387 Consider scaled addresses cases.
19389 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19391 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19392 restored when in interrupt.
19393 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19394 doesn't have delay slot.
19396 2017-03-27 Richard Biener <rguenther@suse.de>
19399 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19400 inlined thunk clones.
19402 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19405 * asan.c (instrument_derefs): Copy over last operand from
19406 original COMPONENT_REF to the new COMPONENT_REF with
19407 DECL_BIT_FIELD_REPRESENTATIVE.
19408 * ubsan.c (instrument_object_size): Likewise.
19410 2017-03-27 Richard Biener <rguenther@suse.de>
19412 PR tree-optimization/80170
19413 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19414 sure DR/SCEV didnt fold in constants we do not see when looking
19415 at the reference base alignment.
19417 2017-03-27 Richard Biener <rguenther@suse.de>
19419 PR middle-end/80171
19420 * gimple-fold.c (fold_ctor_reference): Properly guard against
19421 NULL return value from canonicalize_constructor_val.
19423 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19426 * config/i386/i386.c (ix86_expand_builtin)
19427 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19428 flags reg setting and flags reg using instructions.
19429 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19430 clobbering instructions to zero extend op2.
19432 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19434 * doc/install.texi (Configuration) <--with-aix-soname>:
19435 Update link to AIX ld.
19437 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19439 PR rtl-optimization/80160
19440 PR rtl-optimization/80159
19441 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19442 reg_alternate_class into account.
19444 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19447 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19448 to consider in curr_insn_transform.
19450 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19452 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19453 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19454 and emit_mode_inner.
19456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19458 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19459 argument to the overloaded builtin variants. Use the new flag to
19460 deprecate certain builtin variants.
19461 * config/s390/s390-builtin-types.def: Add new builtin types.
19462 * config/s390/s390-builtins.h: Support new flags field for
19463 overloaded builtins.
19464 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19465 (s390_macro_to_expand): Enable vector float data type.
19466 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19467 builtins by incrementing the __VEC__ version number.
19468 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19470 (s390_resolve_overloaded_builtin): Emit error messages depending
19471 on the builtin flags.
19472 * config/s390/s390.c (s390_expand_builtin): Support additional
19473 flags argument. Change error message to match the messages
19474 emitted in s390-c.c.
19475 * config/s390/s390.md: New UNSPEC_* constants.
19476 (op_type): Add new instruction types.
19477 * config/s390/vecintrin.h: Add new builtins and test data class
19479 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19480 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19481 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19482 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19483 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19484 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19486 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19487 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19488 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19489 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19491 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19492 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19493 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19494 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19496 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19497 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19498 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19499 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19500 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19501 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19502 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19504 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19505 ("vec_scatter_element<V_HW_4:mode>_DI")
19506 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19507 ("vec_fpint<mode>", "vflls")
19508 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19509 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19510 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19511 ("*vec_cmphe<mode>_cc"): ... these.
19513 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19514 mode constant instead of magic value.
19516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19518 * config/s390/s390.c (s390_expand_vec_compare): Support other
19519 vector floating point modes than just V2DF.
19520 (s390_expand_vcond): Likewise.
19521 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19522 (s390_cannot_change_mode_class): Prevent mode changes between TF
19523 and V1TF in vector registers.
19524 * config/s390/s390.md (DF, SF): New mode attributes.
19525 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19526 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19527 SFmode support for VRs.
19528 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19530 (VFT, VF_HW): New mode iterators.
19531 (vw, sdx): New mode attributes.
19532 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19533 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19534 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19535 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19536 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19537 also the new vector floating point modes. Renaming to ...
19539 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19540 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19541 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19542 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19543 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19544 ("vec_unordered<mode>"): ... these.
19546 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19547 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19548 ("*vec_extendv2df"): New insn definitions.
19550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19552 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19553 ("mulditi3_2", "*muldi3_sign"): New patterns.
19554 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19555 rename the pattern definition.
19557 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19559 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19561 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19565 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19566 instruction if possible.
19567 * config/s390/vector.md (vec_halfnumelts): New mode
19569 ("*vec_vllezlf<mode>"): New pattern.
19571 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19573 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19574 ("popcountv4si2", "popcountv2di2"): Rename to ...
19575 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19576 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19578 ("popcount<mode>2_vxe"): New pattern.
19580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19582 * common/config/s390/s390-common.c (processor_flags_table): Add
19584 * config.gcc: Add arch12.
19585 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19586 Default to arch12 for unknown CPU model numbers.
19587 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19588 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19589 PROCESSOR_max sanity check.
19590 * config/s390/s390-opts.h (enum processor_type): Add
19592 * config/s390/s390.c (processor_table): Add arch12.
19593 (s390_expand_builtin): Add check for B_VXE flag.
19594 (s390_issue_rate): Add PROCESSOR_ARCH12.
19595 (s390_get_sched_attrmask): Likewise.
19596 (s390_get_unit_mask): Likewise.
19597 (s390_sched_score): Enable z13 scheduling for arch12.
19598 (s390_sched_reorder): Likewise.
19599 (s390_sched_variable_issue): Likewise.
19600 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19602 (s390_tune_attr): Use z13 scheduling also for arch12.
19603 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19604 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19605 (TARGET_VXE_P): New macros.
19606 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19607 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19608 * config/s390/s390.opt: Add arch12 as processor_type.
19610 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19612 * config/s390/s390.md
19613 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19614 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19615 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19617 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19618 Rename expanders to ...
19620 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19621 ("fixuns_truncdddi2_emu"): ... these.
19623 ("fixuns_trunc<mode>si2_emu"): New expander.
19625 ("*fixuns_truncdfdi2_z13"): Rename to ...
19626 ("*fixuns_truncdfdi2_vx"): ... this.
19628 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19630 * config/s390/2964.md: Remove the single element vector compare
19631 instructions which are no longer used.
19632 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19634 (s390_canonicalize_comparison): Remove handling of DFmode
19636 (s390_expand_vec_compare_scalar): Remove function.
19637 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19638 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19640 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19642 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19644 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19646 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19647 will anyway by matched by mov<mode>_64dfp.
19649 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19651 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19652 vlef/vstef. Add missing operand to vleif.
19654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19656 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19657 pair for all vector types with 64 bit elements.
19658 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19659 * config/s390/vector.md (V_HW_64): ... here.
19660 (V_128_NOSINGLE): New mode iterator.
19661 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19662 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19663 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19664 ("*vec_load_pairv2di"): Change to ...
19665 ("*vec_load_pair<mode>"): ... this one.
19667 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19669 * config/s390/constraints.md: Add comments.
19670 (jKK): Reject element sizes > 8 bytes.
19671 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19673 * config/s390/s390.md: Add the s_operand checks formerly in
19674 s390_split_ok_p to various splitters where they are still
19676 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19677 for 128 bit vectors. Plus two splitters.
19679 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19681 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19684 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19687 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19688 error if the boundary argument is not constant.
19690 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19692 PR rtl-optimization/80112
19693 * loop-doloop.c (doloop_condition_get): Don't check condition
19694 if cmp isn't SET with IF_THEN_ELSE src.
19696 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19698 PR tree-optimization/80158
19699 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19700 replacing a candidate statement, also replace it for the
19701 candidate's alternate interpretation.
19702 (replace_rhs_if_not_dup): Likewise.
19703 (replace_one_candidate): Likewise.
19705 2017-03-24 Richard Biener <rguenther@suse.de>
19707 PR tree-optimization/80167
19708 * graphite-isl-ast-to-gimple.c
19709 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19711 (translate_isl_ast_to_gimple::get_rename): Likewise.
19713 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19715 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19716 handling of certain combinations of target options, including the
19717 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19718 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19720 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19723 * config/arm/arm.md (*load_multiple): Add reload_completed to
19724 matching condition.
19726 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19727 Richard Biener <rguenth@suse.de>
19729 PR tree-optimization/79908
19730 PR tree-optimization/80136
19731 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19732 been cast away, gimplify_and_add suffices.
19734 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19736 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19738 2017-03-23 Richard Biener <rguenther@suse.de>
19740 PR tree-optimization/80032
19741 * gimplify.c (gimple_push_cleanup): Forced unconditional
19742 cleanups still have to go to the conditional_cleanups
19745 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19747 PR tree-optimization/80072
19748 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19750 (next_operand_entry_id): Change type to unsigned int.
19751 (sort_by_operand_rank): Make sure to return the right return value
19752 even if unsigned fields are bigger than INT_MAX.
19753 (struct oecount): Change cnt and id type to unsigned int.
19754 (oecount_hasher::equal): Formatting fix.
19755 (oecount_cmp): Make sure to return the right return value
19756 even if unsigned fields are bigger than INT_MAX.
19757 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19760 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19761 TREE_READONLY on result if writing it more than once.
19764 * doc/invoke.texi (-fsanitize=thread): Document that with
19765 -fnon-call-exceptions atomics are not able to throw
19769 * tsan.c: Include tree-eh.h.
19770 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19771 maybe_clean_or_replace_eh_stmt where needed.
19772 (instrument_memory_accesses): Add cfg_changed argument.
19773 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19774 if it returned true.
19775 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19777 PR rtl-optimization/63191
19778 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19779 wrapper function, moved the whole old content into ...
19780 (ix86_delegitimize_address_1): ... this. New inline function.
19781 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19782 true as last argument instead of ix86_delegitimize_address.
19784 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19786 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19787 cortexa57_branch_cost.
19789 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19791 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19793 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19796 * doc/md.texi (Constraints): Document wA constraint.
19797 * config/rs6000/constraints.md (wA): New.
19798 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19799 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19800 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19801 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19803 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19806 * gimplify.c (is_oacc_declared): New function.
19807 (oacc_default_clause): Use it to set default flags for acc declared
19808 variables inside parallel regions.
19809 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19810 declared variables.
19811 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19812 declare attribute to any decl as necessary.
19814 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19817 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19818 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19819 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19820 (arm_arch_lpae): This.
19821 * config/arm/arm.c (arm_arch7ve): Rename into ...
19822 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19823 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19826 2017-03-22 Martin Liska <mliska@suse.cz>
19829 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19830 error message instead of an ICE.
19832 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19834 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19836 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19838 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19843 2017-03-21 Martin Sebor <msebor@redhat.com>
19845 * doc/extend.texi: Use "cannot" instead of "can't."
19846 * doc/hostconfig.texi: Same.
19847 * doc/install.texi: Same.
19848 * doc/invoke.texi: Same.
19849 * doc/loop.texi: Same.
19850 * doc/md.texi: Same.
19851 * doc/objc.texi: Same.
19852 * doc/rtl.texi: Same.
19853 * doc/tm.texi: Same.
19854 * doc/tm.texi.in: Same.
19855 * doc/trouble.texi: Same.
19857 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19860 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19861 (collect_checksum_attributes): Set it.
19862 (die_checksum_ordered): Use it.
19864 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19866 PR tree-optimization/79908
19867 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19868 change: For a VA_ARG whose LHS has been cast away, use
19869 force_gimple_operand to construct the side effects.
19871 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19873 PR translation/80001
19874 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19875 more amenable to translation.
19876 (oacc_loop_auto_partitions): Likewise.
19878 2017-03-21 Marek Polacek <polacek@redhat.com>
19879 Martin Sebor <msebor@redhat.com>
19881 PR tree-optimization/80109
19882 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19883 on INTEGRAL_TYPE_P.
19885 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19886 Segher Boessenkool <segher@kernel.crashing.org>
19889 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19890 check reg_used_between_p between insn and one of succ or succ2
19891 depending on if succ is artificial insn not inserted into insn
19894 2017-03-21 Martin Liska <mliska@suse.cz>
19896 PR gcov-profile/80081
19897 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19898 * doc/gcc.texi: Include gcov-dump stuff.
19899 * doc/gcov-dump.texi: New file.
19901 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19903 PR rtl-optimization/79150
19904 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19905 conditional jump, if the jump is the last insn of the loop.
19907 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19908 Richard Biener <rguenth@suse.de>
19910 PR tree-optimization/79908
19911 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19912 been cast away, use force_gimple_operand to construct the side
19915 2017-03-21 Martin Liska <mliska@suse.cz>
19917 PR libfortran/79956
19918 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19921 2017-03-21 Brad Spengler <spender@grsecurity.net>
19924 * plugin.c (htab_hash_plugin): New function.
19925 (add_new_plugin): Use it and adjust.
19926 (parse_plugin_arg_opt): Adjust.
19927 (init_one_plugin): Likewise.
19929 2017-03-21 Richard Biener <rguenther@suse.de>
19931 PR tree-optimization/80032
19932 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19933 if set force the cleanup to happen unconditionally.
19934 (gimplify_target_expr): Push inserted clobbers with force_uncond
19935 to avoid them being removed by control-dependent DCE.
19937 2017-03-21 Richard Biener <rguenther@suse.de>
19939 PR tree-optimization/80122
19940 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19941 va_arg_pack_len when the inlined call stmt requires pack
19943 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19945 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19948 * tsan.c (instrument_builtin_call): If the memory model argument
19949 is not a constant, assume it is valid.
19952 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19955 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19957 PR rtl-optimization/79910
19958 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19959 if its dest is used by an insn before I2 (other than the combined
19960 insns themselves, which are properly handled already).
19962 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19965 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19967 * combine.c (record_used_regs): New static function.
19968 (try_combine): Handle situations where there is an additional
19969 instruction between I2 and I3 which needs to have a LOG_LINK
19973 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19975 * combine.c (try_combine): Delete redundant i1 test. Call
19976 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19978 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19981 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19982 alternatives 13/14.
19984 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19986 PR tree-optimization/80054
19987 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19988 the optimization if a PHI or any of its arguments is not dominated
19989 by the candidate's basis. Use gphi* rather than gimple* as
19991 (replace_profitable_candidates): Clean up a gimple* variable that
19992 should be a gphi* variable.
19994 2017-03-20 Martin Sebor <msebor@redhat.com>
19997 * doc/extend.texi (attribute constructor): Document present limitation.
19999 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20002 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20003 __POWER9_VECTOR__ #ifdef control, change template definition to
20004 use Power9-specific built-in function.
20005 (vec_any_eq): Likewise.
20006 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20007 to control outcomes from this test.
20008 (vector_ae_<mode>p): For VEC_F modes, likewise.
20010 2017-03-20 Ian Lance Taylor <iant@google.com>
20012 * config/i386/i386.c (ix86_function_regparm): Save an extra
20013 register for -fsplit-stack with DECL_STATIC_CHAIN.
20015 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20018 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20019 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20021 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20023 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20025 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20028 2017-03-20 Marek Polacek <polacek@redhat.com>
20031 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20033 2017-03-20 Richard Biener <rguenther@suse.de>
20035 PR tree-optimization/80113
20036 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20037 allocate extra SSA name for PHI def.
20038 (add_close_phis_to_outer_loops): Likewise.
20039 (add_close_phis_to_merge_points): Likewise.
20040 (copy_loop_close_phi_args): Likewise.
20041 (copy_cond_phi_nodes): Likewise.
20043 2017-03-20 Martin Liska <mliska@suse.cz>
20045 PR middle-end/79753
20046 * tree-chkp.c (chkp_build_returned_bound): Do not build
20047 returned bounds for a LHS that's not a BOUNDED_P type.
20049 2017-03-20 Martin Liska <mliska@suse.cz>
20053 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20054 COMPLEX_CST and VECTOR_CST.
20056 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20059 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20060 target operand. A new splitter adds the clobber statement in case
20061 the target operand is dead anyway.
20063 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20065 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20066 to age-old versions of binutils and glibc.
20068 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20070 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20072 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20074 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20076 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20078 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20079 requirement for binutils 2.13.
20081 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20083 * combine.c (try_combine): Delete redundant i1 test. Call
20084 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20086 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20088 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20089 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20091 <riscv64-*-elf>: Re-arrange section
20092 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20093 <riscv32-*-linux>: Likewise.
20094 <riscv64-*-elf>: Likewise
20095 <riscv64-*-linux>: Likewise.
20097 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20100 * aarch64.opt(verbose-cost-dump): Fix typo.
20102 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20105 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20106 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20108 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20110 * reload.c (find_reloads): When reloading a nonoffsettable address,
20111 use RELOAD_OTHER for it and its address reloads.
20113 PR rtl-optimization/79910
20114 * combine.c (record_used_regs): New static function.
20115 (try_combine): Handle situations where there is an additional
20116 instruction between I2 and I3 which needs to have a LOG_LINK
20119 2017-03-17 Jeff Law <law@redhat.com>
20121 PR tree-optimization/71437
20122 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20123 conditional in the hash table first.
20124 (vrp_dom_walker::before_dom_children): Extract condition from
20125 ASSERT_EXPR. Record condition, its inverion and any implied
20126 conditions as well.
20128 2017-03-17 Marek Polacek <polacek@redhat.com>
20129 Markus Trippelsdorf <markus@trippelsdorf.de>
20131 PR tree-optimization/80079
20132 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20135 2017-03-17 Richard Biener <rguenther@suse.de>
20137 PR middle-end/80075
20138 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20139 Properly verify the LHS before the RHS possibly claims to be
20141 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20144 2017-03-17 Martin Jambor <mjambor@suse.cz>
20146 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20147 (List of -O2 options): Likewise.
20148 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20151 2017-03-17 Tom de Vries <tom@codesourcery.com>
20153 * gcov-dump.c (print_usage): Print bug_report_url.
20155 2017-03-17 Richard Biener <rguenther@suse.de>
20157 PR middle-end/80050
20158 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20159 (parser::peek): Likewise.
20161 2017-03-17 Richard Biener <rguenther@suse.de>
20163 PR tree-optimization/80048
20164 * sese.c (free_sese_info): Properly release rename_map and
20165 copied_bb_map elements.
20167 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20169 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20170 Add linked-list forward and backlinks. Insert on
20171 construction, remove on destruction.
20172 (class pass_store_merging): Add m_stores_head field.
20173 (pass_store_merging::terminate_and_process_all_chains):
20174 Iterate over m_stores_head list.
20175 (pass_store_merging::terminate_all_aliasing_chains):
20177 (pass_store_merging::execute): Check for debug stmts first.
20178 Push new chains onto the m_stores_head stack.
20180 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20183 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20184 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20185 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20187 2017-03-16 Jeff Law <law@redhat.com>
20189 PR tree-optimization/71437
20190 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20191 member function. Implementation moved into after_dom_children
20192 member function and into the threader's thread_outgoing_edges
20194 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20195 some code into new thread_outgoing_edges.
20196 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20197 definition. Simplify marker handling (do it here). Assume we always
20198 have the available expression and the const/copies tables.
20199 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20201 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20202 * tree-vrp.c (equiv_stack): No longer file scoped.
20203 (vrp_dom_walker): New class.
20204 (vrp_dom_walker::before_dom_children): New member function.
20205 (vrp_dom_walker::after_dom_children): Likewise.
20206 (identify_jump_threads): Setup domwalker. Use it rather than
20207 walking edges in a random order by hand. Simplify setup/finalization.
20208 (finalize_jump_threads): Remove.
20209 (vrp_finalize): Do not call identify_jump_threads here.
20210 (execute_vrp): Do it here instead and call thread_through_all_blocks
20213 PR tree-optimization/71437
20214 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20216 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20218 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20219 (dom_opt_dom_walker::thread_across_edge): Remove
20220 handle_dominating_asserts argument. All callers changed.
20221 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20222 changes. Remove calls to lhs_of_dominating_assert. Other
20223 uses of handle_dominating_asserts turn into unconditional code
20224 (simplify_control_stmt_condition_1): Likewise.
20225 (simplify_control_stmt_condition): Likewise.
20226 (thread_through_normal_block, thread_across_edge): Likewise.
20227 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20228 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20229 object if it is not an SSA_NAME.
20230 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20231 before calling into the VRP specific simplifiers.
20232 (identify_jump_threads): Remove handle_dominating_asserts
20235 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20238 * tree-diagnostic.c (default_tree_printer): No longer static.
20239 * tree-diagnostic.h (default_tree_printer): New prototype.
20241 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20243 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20244 Change ins into fmov.
20246 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20248 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20249 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20250 Use h_con constraint for operand 1.
20251 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20252 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20254 2017-03-15 Jeff Law <law@redhat.com>
20256 PR tree-optimization/71437
20257 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20258 (record_temporary_equivalences): Use it.
20260 PR tree-optimization/71437
20261 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20262 tree-ssa-scopedtables.
20263 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20264 (record_conditions, record_cond, vuse_eq): Likewise.
20265 (record_edge_info): Adjust to API tweak of record_conditions.
20266 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20267 (record_temporary_equivalences, optimize_stmt): Likewise.
20268 (eliminate_redundant_computations): Likewise.
20269 (record_equivalences_from_stmt): Likewise.
20270 * tree-ssa-scopedtables.c: Include options.h and params.h.
20271 (vuse_eq): New function, moved from tree-ssa-dom.c
20272 (build_and_record_new_cond): Likewise.
20273 (record_conditions): Likewise. Accept vector of conditions rather
20274 than edge_equivalence structure for first argument.
20275 for the first argument.
20276 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20277 from tree-ssa-dom.c.
20278 (avail_exprs_stack::record_cond): Likewise.
20279 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20280 from tree-ssa-dom.c.
20281 (avail_exprs_stack): Add new member functions lookup_avail_expr
20283 (record_conditions): Declare.
20285 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20288 * lra-constraints.c (process_alt_operands): Increase reject for
20289 reloading an input/output operand.
20291 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20294 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20295 insns to convert from signed/unsigned char/short to IEEE 128-bit
20297 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20299 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20302 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20303 subreg of inner mode for values already in registers.
20305 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20307 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20308 iteration reg is used after the loop.
20310 2017-03-14 Martin Sebor <msebor@redhat.com>
20312 PR tree-optimization/79800
20313 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20314 precision in negative-positive range.
20315 (format_floating): Call non-const overload with adjusted precision.
20317 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20320 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20323 2017-03-14 Martin Sebor <msebor@redhat.com>
20325 PR middle-end/80020
20326 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20327 * builtins.def (aligned_alloc): Use it.
20330 * Makefile.in (GTFILES): Add calls.c.
20331 * calls.c: Include "gt-calls.h".
20333 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20335 PR rtl-optimization/79728
20336 * regs.h (struct target_regs): New field
20337 x_contains_allocatable_regs_of_mode.
20338 (contains_allocatable_regs_of_mode): New macro.
20339 * reginfo.c (init_reg_sets_1): Initialize it, and change
20340 contains_reg_of_mode so it includes global regs as well.
20341 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20342 rather than contains_regs_of_mode.
20344 2017-03-14 Martin Liska <mliska@suse.cz>
20346 * doc/invoke.texi: Document options that can't be combined with
20347 -fcheck-pointer-bounds.
20349 2017-03-14 Martin Liska <mliska@suse.cz>
20351 PR middle-end/79831
20352 * doc/invoke.texi (-Wchkp): Document the option.
20354 2017-03-14 Martin Liska <mliska@suse.cz>
20356 * Makefile.in: Install gcov-dump.
20358 2017-03-14 Martin Liska <mliska@suse.cz>
20360 * multiple_target.c (expand_target_clones): Bail out for
20361 an invalid attribute.
20363 2017-03-14 Richard Biener <rguenther@suse.de>
20365 * alias.c (struct alias_set_entry): Pack properly.
20366 * cfgloop.h (struct loop): Likewise.
20367 * cse.c (struct set): Likewise.
20368 * ipa-utils.c (struct searchc_env): Likewise.
20369 * loop-invariant.c (struct invariant): Likewise.
20370 * lra-remat.c (struct cand): Likewise.
20371 * recog.c (struct change_t): Likewise.
20372 * rtl.h (struct address_info): Likewise.
20373 * symbol-summary.h (function_summary): Likewise.
20374 * tree-loop-distribution.c (struct partition): Likewise.
20375 * tree-object-size.c (struct object_size_info): Likewise.
20376 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20377 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20378 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20379 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20380 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20381 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20382 (struct _stmt_vec_info): Likewise.
20384 2017-03-14 Martin Liska <mliska@suse.cz>
20387 * multiple_target.c (create_dispatcher_calls): Check that
20388 a target can create a function dispatcher.
20390 2017-03-14 Martin Liska <mliska@suse.cz>
20393 * multiple_target.c (expand_target_clones): Drop local.local
20394 flag for default implementation.
20396 2017-03-14 Richard Biener <rguenther@suse.de>
20398 PR tree-optimization/80030
20399 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20401 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20403 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20404 gcc_fallthrough() instead of __attribute__((fallthrough));
20406 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20408 * doc/gcc.texi: Remove "up" link to (DIR).
20409 * doc/gccint.texi: Ditto.
20411 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20413 * doc/install.texi (Specific) <avr>: Remove reference to
20416 2017-03-13 Jeff Law <law@redhat.com>
20418 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20419 attribute rather than comments.
20421 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20422 match_scratch operand is highest.
20424 2017-03-13 Martin Liska <mliska@suse.cz>
20426 PR middle-end/78339
20427 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20428 is a CHKP clone, use original declaration.
20430 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20432 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20433 (arc_conditional_register_usage): Use a different allocation order
20434 when optimizing for size.
20435 * common/config/arc/arc-common.c (arc_option_optimization_table):
20436 Section anchors default on when optimizing for size.
20438 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20440 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20442 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20444 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20445 * config/arc/arc.md (cpu_facility): Add cd variant.
20446 (*movqi_insn): Add code density variant.
20447 (*movhi_insn): Likewise.
20448 (*movqi_insn): Likewise.
20449 (*addsi3_mixed): Likewise.
20450 (subsi3_insn): Likewise.
20452 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20454 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20456 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20458 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20459 expressions with MINUS and UNARY ops.
20461 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20464 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20466 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20467 between vec_select and vector argument.
20468 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20469 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20470 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20471 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20472 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20473 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20475 2017-03-13 Richard Biener <rguenther@suse.de>
20478 * params.def (vect-max-peeling-for-alignment): Fix typo.
20480 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20482 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20483 issue that only occurred with binutils below 2.18.
20485 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20487 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20488 refer to binutils 2.11/2.12 minimum.
20490 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20492 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20493 ftp.kernel.org and simplify binutils requirement.
20495 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20497 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20499 (Optimize Options): Ditto. Also remove redundancy.
20501 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20503 PR translation/79848
20504 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20506 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20507 to G_ to avoid double translation.
20509 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20511 PR translation/79923
20512 * auto-profile.c (get_combined_location): Convert leading
20513 character of diagnostics to lower case and remove trailing period.
20514 (read_profile): Likewise for various diagnostics.
20515 * config/arm/arm.c (arm_option_override): Remove trailing period
20516 from various diagnostics.
20517 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20518 (msp430_expand_delay_cycles): Likewise.
20520 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20523 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20524 full command-line argument, rather than just "str".
20525 (aarch64_validate_march): Likewise.
20526 (aarch64_validate_mtune): Likewise.
20528 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20530 PR rtl-optimization/78911
20531 * lra-assigns.c (must_not_spill_p): New function.
20532 (spill_for): Use it.
20534 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20536 PR tree-optimization/79981
20537 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20538 ATOMIC_COMPARE_EXCHANGE ifn result.
20539 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20540 IFN_ATOMIC_COMPARE_EXCHANGE.
20542 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20545 * opts.c (parse_sanitizer_options): Add missing question mark to
20546 "did you mean" message.
20548 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20550 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20552 (VMULEUH_UNS): Likewise.
20553 (VMULOUB_UNS): Likewise.
20554 (VMULOUH_UNS): Likewise.
20555 * config/rs6000/rs6000.c (builtin_function_type): Remove
20556 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20558 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20561 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20562 x with result of extra_parsing_for_operand_code_0.
20563 (function_reader::extra_parsing_for_operand_code_0): Convert
20564 return type from void to rtx, returning x. When reading
20565 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20566 larger size containing struct block_symbol.
20568 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20570 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20571 -mfloat128-hardware without -m64.
20573 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20576 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20577 entries to the case statement that marks unsigned arguments to
20578 overloaded functions.
20580 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20582 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20583 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20585 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20588 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20589 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20591 2017-03-10 Martin Liska <mliska@suse.cz>
20595 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20596 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20599 2017-03-10 Olivier Hainque <hainque@adacore.com>
20601 * tree-switch-conversion (array_value_type): Start by resetting
20602 candidate type to it's main variant.
20604 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20606 PR rtl-optimization/79909
20607 * combine.c (try_combine): Use simplify_replace_rtx on individual
20608 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20609 of the whole CALL_INSN_FUNCTION_USAGE.
20611 PR tree-optimization/79972
20612 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20613 get_range_info on SSA_NAMEs. Formatting fixes.
20615 2017-03-10 Richard Biener <rguenther@suse.de>
20616 Jakub Jelinek <jakub@redhat.com>
20618 PR tree-optimization/77975
20619 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20620 edge to be constant.
20621 (get_val_for): For constant x return it. Formatting fix.
20622 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20623 would use the same bases as the current one.
20625 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20627 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20628 instead of vec_select for V1TImode.
20629 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20631 (VSX_LE_128): Add V1TI to this mode iterator.
20632 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20633 (*vsx_le_perm_store_<mode>): Likewise.
20634 (pre-reload splitter for VSX stores): Likewise.
20635 (post-reload splitter for VSX stores): Likewise.
20636 (*vsx_xxpermdi2_le_<mode>): Likewise.
20637 (*vsx_lxvd2x2_le_<mode>): Likewise.
20638 (*vsx_stxvd2x2_le_<mode>): Likewise.
20640 2017-03-09 Michael Eager <eager@eagercon.com>
20642 Correct failures with --enable-checking=yes,rtl.
20644 * config/microblaze/microblaze.c (microblaze_expand_shift):
20645 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20646 test for const0_rtx.
20647 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20648 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20650 2017-03-09 Richard Biener <rguenther@suse.de>
20652 PR tree-optimization/79977
20653 * graphite-scop-detection.c (scop_detection::merge_sese):
20654 Handle the case of extra exits to blocks dominating the entry.
20656 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20658 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20661 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20663 PR rtl-optimization/79949
20664 * lra-constraints.c (process_alt_operands): Check memory when
20665 trying to predict a cycle. Print about the overall increase.
20667 2017-03-09 Richard Biener <rguenther@suse.de>
20669 PR middle-end/79971
20670 * gimple-expr.c (useless_type_conversion_p): Preserve
20671 TYPE_SATURATING for fixed-point types.
20673 2017-03-09 Richard Biener <rguenther@suse.de>
20676 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20677 alignment of BLKmode params.
20679 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20682 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20683 (VALL_NO_V2Q): Likewise.
20685 * config/aarch64/aarch64-simd.md
20686 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20688 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20689 VALL_NO_V2Q mode iterator.
20690 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20692 2017-03-09 Martin Liska <mliska@suse.cz>
20694 PR tree-optimization/79631
20695 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20696 tree_int_cst_sign_bit just for INTEGER constants.
20698 2017-03-09 Martin Liska <mliska@suse.cz>
20702 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20705 2017-03-09 Marek Polacek <polacek@redhat.com>
20708 * tree.c (inchash::add_expr): Handle TREE_VEC.
20710 2017-03-09 Martin Liska <mliska@suse.cz>
20713 (chkp_narrow_size_and_offset): New function.
20714 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20715 (void chkp_parse_bit_field_ref): New function.
20716 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20717 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20719 2017-03-09 Martin Liska <mliska@suse.cz>
20722 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20723 (chkp_find_bounds_1): Remove gcc_unreachable.
20725 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20728 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20729 BUILT_IN_SYNC*, determine the access type from the size suffix and
20730 always build a MEM_REF with that type. Handle forgotten
20731 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20734 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20735 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20736 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20737 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20738 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20739 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20740 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20741 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20742 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20743 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20744 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20745 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20746 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20747 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20748 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20749 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20750 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20751 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20752 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20753 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20754 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20755 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20756 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20757 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20758 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20759 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20760 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20761 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20762 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20763 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20764 definitions outside of __OPTIMIZE__ guarded section.
20767 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20768 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20769 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20770 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20773 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20775 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20776 ("vfenez<mode>"): Add missing constraints.
20778 2017-03-08 Martin Sebor <msebor@redhat.com>
20781 * config/nds32/nds32.c (nds32_option_override):
20782 Fix misspelled diagnostic.
20784 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20787 * gimplify.c (gimplify_omp_for): Replace index var in outer
20788 taskloop statement with an artificial variable and add
20789 OMP_CLAUSE_PRIVATE clause for it.
20791 2017-03-08 Richard Biener <rguenther@suse.de>
20793 PR tree-optimization/79955
20794 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20795 for accesses that are completely outside of the variable.
20797 2017-03-08 Andrew Haley <aph@redhat.com>
20799 PR tree-optimization/79943
20800 * tree-ssa-loop-split.c (compute_new_first_bound): When
20801 calculating the new upper bound, (END-BEG) should be added, not
20804 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20806 * config/avr/avr.md (setmemhi): Make sure match_dup
20807 operand number comes before match_scratch.
20809 2017-03-08 Richard Biener <rguenther@suse.de>
20811 PR tree-optimization/79920
20812 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20813 with ncopies == 1 to ...
20814 (vect_transform_slp_perm_load): ... here. Properly compute
20815 all element loads by iterating VF times over the group. Do
20816 not handle ncopies (computed in a broken way) in
20817 vect_create_mask_and_perm.
20819 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20822 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20823 is a uniform vector, use uniform_vector_p return value instead of
20824 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20826 2017-03-07 Marek Polacek <polacek@redhat.com>
20828 PR middle-end/79809
20829 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20830 (alloca_call_type): Likewise.
20832 2017-03-07 Martin Liska <mliska@suse.cz>
20834 * gcov.c (process_args): Put comment to correct location.
20836 2017-03-07 Martin Liska <mliska@suse.cz>
20838 PR middle-end/68270
20839 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20840 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20841 (chkp_narrow_bounds_for_field): Likewise.
20842 (chkp_parse_array_and_component_ref): Pass one more argument to
20845 2017-03-07 Richard Biener <rguenther@suse.de>
20847 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20850 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20852 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20853 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20855 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20858 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20859 to end of description.
20860 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20862 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20864 PR rtl-optimization/79901
20865 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20867 (*avx512f_<code><mode>3<mask_name>): ... this.
20868 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20869 iterator instead of VI8_AVX2_AVX512BW.
20871 PR rtl-optimization/79901
20872 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20873 min/max expander, expand it using expand_vec_cond_expr.
20876 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20879 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20882 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20883 to void * for PCH reasons.
20884 * dwarf2out.c (output_loc_operands, output_die): Cast
20885 v.val_vec.array to unsigned char *.
20887 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20890 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20893 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20895 PR rtl-optimization/79571
20896 * lra-constraints.c (process_alt_operands): Calculate static
20897 reject and subtract it from overall when only addresses will be
20900 2017-03-06 Julia Koval <julia.koval@intel.com>
20903 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20904 incoming stack boundary to 128 for 64-bit targets.
20906 2017-03-06 Richard Biener <rguenther@suse.de>
20908 PR tree-optimization/79894
20909 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20910 to NULL after folding it.
20912 2017-03-06 Richard Biener <rguenther@suse.de>
20914 PR tree-optimization/79824
20915 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20916 check disabling peeling for gaps.
20918 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20920 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20921 attributes): Document gettimeofday.
20923 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20925 * config/s390/s390.c (s390_option_override_internal): Set
20926 PARAM_MIN_VECT_LOOP_BOUND
20928 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20930 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20931 * config/s390/s390.md: Likewise.
20933 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20936 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20937 (<avx2_avx512>_perm<mode>): Rename to ...
20938 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20940 (<avx512>_perm<mode>_mask): Rename to ...
20941 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20943 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20944 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20945 instead of VI8F_256_512.
20946 (avx512f_perm<mode>): New define_expand.
20947 (avx512f_perm<mode>_mask): Likewise.
20948 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20949 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20951 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20953 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20954 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20956 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20958 2017-03-06 Martin Liska <mliska@suse.cz>
20961 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20962 when having a SSA NAME w/o VAR_DECL assigned to it.
20964 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20966 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20967 msa_dpsub_<su>_d): Fix MODE for vec_select.
20969 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20971 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20973 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20975 2017-03-06 Richard Biener <rguenther@suse.de>
20977 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20978 * plugin.c (register_plugin_info): Likewise.
20979 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20981 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20983 * config/i386/sse.md (sse_storehps, sse_storelps,
20984 avx_<castmode><avxsizesuffix>_<castmode>,
20985 avx512f_<castmode><avxsizesuffix>_<castmode>,
20986 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20987 in condition that at least one operand is not a MEM.
20989 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20991 PR middle-end/79805
20992 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20993 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20995 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20996 gimple_call_nothrow_p flag based on whether original builtin can throw.
20997 If it can, emit following stmts on the fallthrough edge.
20998 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20999 don't create new bb if inserting just debug stmts on the edge, try to
21000 insert them on the fallthru bb or just reset debug stmts.
21002 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21005 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21006 restore recog_data (including the operand rtxes inside it) around
21007 the call to get_insn_template.
21009 2017-03-03 Martin Sebor <msebor@redhat.com>
21011 PR tree-optimization/79699
21012 * context.c (context::~context): Free MPFR caches to avoid
21013 a memory leak on program exit.
21015 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21017 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21018 Use wide_int::ulow () instead of .elt (0).
21020 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21022 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21023 (*pushxf): Limit oF constraint to 32bit targets and add oC
21024 constraint for 64bit targets.
21025 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21026 (*pushdf): Change rmF constraint to rmC.
21028 2017-03-03 Martin Liska <mliska@suse.cz>
21030 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21031 Remove unused variable.
21033 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21036 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21037 is a memory operand, increase num_memory.
21038 (ix86_expand_args_builtin): Likewise.
21040 2017-03-03 Jan Hubicka <jh@suse.cz>
21043 * ipa-devirt.c (maybe_record_node): Properly handle
21044 __cxa_pure_virtual visibility.
21046 2017-03-03 Martin Liska <mliska@suse.cz>
21048 PR tree-optimization/79803
21049 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21051 (pass_loop_prefetch::execute): Disabled optimization if an
21052 assumption about L1 cache size is not met.
21054 2017-03-03 Martin Liska <mliska@suse.cz>
21056 PR rtl-optimization/79574
21057 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21058 (hash_scan_set): Likewise.
21059 (dump_hash_table): Likewise.
21060 (hoist_code): Likewise.
21062 2017-03-03 Richard Biener <rguenther@suse.de>
21064 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21065 in place of uhigh ().
21066 (fixed_convert_from_real): Likewise.
21068 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21071 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21073 2017-03-03 Richard Biener <rguenther@suse.de>
21075 PR middle-end/79818
21076 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21077 TYPE_OVERFLOW_UNDEFINED check.
21079 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21081 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21083 (vector_ae_<mode>_p): Likewise.
21084 (vector_nez_<mode>_p): Likewise.
21085 (vector_ne_v2di_p): Likewise.
21086 (vector_ae_v2di_p): Likewise.
21087 (vector_ne_<mode>_p): Likewise.
21088 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21090 (vsx_tsqrt<mode>2_fe): Likewise.
21092 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21095 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21097 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21099 PR rtl-optimization/79780
21100 * cprop.c (one_cprop_pass): When second and further conditional trap
21101 in a single basic block is turned into an unconditional trap, turn it
21102 into a deleted note to avoid RTL verification failures.
21104 2017-03-02 Richard Biener <rguenther@suse.de>
21106 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21108 2017-03-02 Richard Biener <rguenther@suse.de>
21110 PR tree-optimization/79345
21112 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21113 param and abort the walk, returning -1 if it is hit.
21114 (walk_aliased_vdefs): Take a limit param and pass it on.
21115 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21116 defaulting to 0 and return a signed int.
21117 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21118 (check_defs): New helper.
21119 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21120 about uninitialized memory.
21121 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21122 bogus uninitialized warning.
21123 (fixed_convert_from_real): Likewise.
21125 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21127 PR tree-optimization/66768
21128 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21129 iv_use if base object can't be determined.
21131 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21133 PR tree-optimization/79345
21134 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21135 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21136 (get_pattern_stats): Initialize it.
21137 * genemit.c (gen_expand): Verify match_scratch numbers come after
21138 match_operand/match_dup numbers.
21139 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21140 match_scratch numbers.
21141 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21143 * config/s390/s390.md (trunctdsd2): Likewise.
21145 2017-03-02 Richard Biener <rguenther@suse.de>
21147 * wide-int.h (wide_int_storage::operator=): Implement in terms
21150 2017-03-02 Richard Biener <rguenther@suse.de>
21152 PR tree-optimization/79777
21153 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21154 the to insert expression to sth existing.
21156 2017-03-01 Martin Sebor <msebor@redhat.com>
21158 PR middle-end/79692
21159 * gimple-ssa-sprintf.c
21160 (directive::known_width_and_precision): New function.
21161 (format_integer): Use it.
21162 (get_mpfr_format_length): Consider the full range of precision
21163 when computing %g output with the # flag. Set the likely byte
21164 count to 3 rather than 1 when precision is indeterminate.
21165 (format_floating): Correct the lower bound of precision.
21167 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21169 * doc/invoke.texi: Document default code model for 64-bit Linux.
21171 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21174 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21175 udiv rather than div since input pattern is unsigned.
21177 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21179 * config/i386/i386.c (print_reg): Warn for values of
21180 unsupported size in integer register.
21182 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21185 * config/rs6000/predicates.md (current_file_function_operand): Do
21186 not allow self calls to be local if the function is replaceable.
21188 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21191 * config/rs6000/altivec.h (vec_ctz and others): Change the
21192 preprocessor macro that controls conditional compilation from
21193 _ARCH_PWR9 to __POWER9_VECTOR__.
21194 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21195 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21196 control (instead of _ARCH_PWR9 control) so that template
21197 definition uses power9-specific function.
21198 (vec_any_eq): Likewise.
21199 (vec_all_ne): Change macro definition to use a power9-specific
21200 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21201 _ARCH_PWR9 control).
21202 (vec_any_eq) Likewise.
21203 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21204 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21205 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21206 support for xvcmpnedp instruction.
21207 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21208 macro expansion so that Power9 implementation of vec_all_ne does
21209 not use the AltiVec predicate framework.
21210 (VCMPNEH_P): Likewise.
21211 (VCMPNEW_P): Likewise.
21212 (VCMPNED_P): Likewise.
21213 (VCMPNEFP_P): Likewise.
21214 (VCMPNEDP_P): Likewise.
21215 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21216 implementation of vec_any_eq to not use AltiVec predicate
21218 (VCMPAEH_P): Likewise.
21219 (VCMPAEW_P): Likewise.
21220 (VCMPAED_P): Likewise.
21221 (VCMPAEFP_P): Likewise.
21222 (VCMPAEDP_P): Likewise.
21223 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21224 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21225 not use the AltiVec predicate framework.
21226 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21227 of vec_any_eq to not use AltiVec predicate framework.
21228 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21229 support for predefined __POWER9_VECTOR__ macro to indicate that
21230 Power9 instruction selection is enabled.
21231 (altivec_overloaded_builtins): Remove extraneous
21232 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21233 function argument types RS6000_BTI_bool_V16QI and
21234 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21235 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21236 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21237 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21238 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21239 Power9 for implementations of vec_cmpne. Change the signature for
21240 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21241 (representing vec_all_ne) to remove the previously described first
21242 argument of type RS6000_BTI_INTSI, as this was an artifact of
21243 reliance on the AltiVec predicate framework, which is no longer
21244 used in the implementation of these functions. Add
21245 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21246 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21247 since, unlike the AltiVec predicate framework implementation, we
21248 do not share function descriptors between vec_alle and vec_anyeq.
21249 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21250 set of modes that receive special treatment even when
21251 TARGET_P9_VECTOR is true. The special treatment emits code that
21252 does not depend on Power9 instructions.
21253 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21254 define_expand to not rely on AltiVec predicate framework.
21255 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21257 (vector_ne_v2di_p): Change this define_expand to not rely on
21258 AltiVec predicate framework.
21259 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21261 (vector_ne_<mode>_p): Change this define_expand to not rely on
21262 AltiVec predicate framework.
21263 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21265 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21266 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21267 define_insn pattern.
21268 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21269 define_insn pattern because the xvcmpne<VSs>. instruction is not
21271 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21272 instruction is not supported.
21274 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21276 * config/nvptx/nvptx.c: Include intl.h.
21278 2017-03-01 Martin Jambor <mjambor@suse.cz>
21281 * ipa-prop.h (ipa_bits): Removed field known.
21282 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21283 to pointers. Adjusted their comments to warn about their sharing.
21284 (ipcp_transformation_summary): Change bits to a vector of pointers.
21285 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21286 (ipa_get_ipa_bits_for_value): Declare.
21287 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21288 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21289 (ipa_bits_hash_table): Likewise.
21290 (ipa_vr_ggc_hash_traits): Likewise.
21291 (ipa_vr_hash_table): Likewise.
21292 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21293 being pointers and vr_known being removed.
21294 (ipa_set_jf_unknown): Likewise.
21295 (ipa_get_ipa_bits_for_value): New function.
21296 (ipa_set_jfunc_bits): Likewise.
21297 (ipa_get_value_range): New overloaded functions.
21298 (ipa_set_jfunc_vr): Likewise.
21299 (ipa_compute_jump_functions_for_edge): Use the above functions to
21300 construct bits and vr parts of jump functions.
21301 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21302 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21304 (ipcp_grow_transformations_if_necessary): Also allocate
21305 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21307 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21308 them. Fix too long lines.
21309 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21310 vr_known being removed.
21311 (ipa_read_jump_function): Use new setter functions to construct bits
21312 and vr parts of jump functions or set them to NULL.
21313 (write_ipcp_transformation_info): Adjust for bits being pointers.
21314 (read_ipcp_transformation_info): Likewise.
21315 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21317 Include gt-ipa-prop.h.
21318 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21320 (ipcp_store_bits_results): Likewise.
21321 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21322 Do not write to existing jump functions but use a temporary instead.
21324 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21327 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21328 attempt to use its first operand as BIT_FIELD_REF base.
21330 2017-03-01 Richard Biener <rguenther@suse.de>
21332 PR middle-end/79721
21333 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21334 interpolating formula in wrapping arithmetic.
21335 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21337 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21339 PR tree-optimization/79734
21340 * tree-vect-generic.c (expand_vector_condition): Optimize
21341 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21342 Handle VEC_COND_EXPR where comparison has different inner width from
21343 type's inner width.
21345 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21347 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21348 markup, and similar issues. Remove @opindex entries for things
21349 that aren't options. Add missing -mmpy-option entries.
21351 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21353 PR tree-optimization/79737
21354 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21355 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21356 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21357 instead of byte_size. Formatting fix.
21358 (shift_bytes_in_array_right): Formatting fix.
21360 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21363 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21364 condition on optimize for the leaf function test.
21366 2017-02-28 Martin Liska <mliska@suse.cz>
21369 * read-rtl-function.c (function_reader::handle_unknown_directive):
21370 Bail out when one uses -flto.
21372 2017-02-28 Martin Liska <mliska@suse.cz>
21374 * common.opt: Replace space with tabular for options of <number>
21376 * config/i386/i386.opt: Show <number> value for
21377 -mlarge-data-threshold.
21378 * opts.c (print_filtered_help): Do not display number in hexadecimal
21381 2017-02-28 Martin Liska <mliska@suse.cz>
21383 * common.opt: Fix --help=option -Q for options which are of
21386 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21388 * config/i386/i386.c (print_reg): Error out for values
21389 of 8-bit size in invalid integer register.
21391 2017-02-28 Martin Sebor <msebor@redhat.com>
21393 PR tree-optimization/79691
21394 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21396 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21399 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21400 gcc_unreachable with output_operand_lossage.
21402 2017-02-28 Richard Biener <rguenther@suse.de>
21404 PR tree-optimization/79740
21405 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21407 (visit_nary_op): Insert the nary into the hashtable if we
21408 pattern-matched sth.
21409 * tree-ssa-pre.c (eliminate_insert): Robustify.
21411 2017-02-28 Richard Biener <rguenther@suse.de>
21413 PR middle-end/79731
21414 * fold-const.c (decode_field_reference): Reject out-of-bound
21417 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21419 * config/i386/i386.c: Include intl.h.
21420 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21421 instead of just cond ? "..." : "...".
21422 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21423 * coverage.c (read_counts_file): Likewise.
21424 * omp-offload.c: Include intl.h.
21425 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21426 of just cond ? "..." : "...".
21427 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21428 of just cond ? "..." : "...".
21430 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21433 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21435 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21436 'tune for' CPU name.
21437 * config/arm/arm-cpu-data.h: Regenerated.
21439 2017-02-28 Richard Biener <rguenther@suse.de>
21441 PR tree-optimization/79732
21442 * tree-inline.c (expand_call_inline): Do not shadow var.
21444 2017-02-28 Richard Biener <rguenther@suse.de>
21446 PR tree-optimization/79723
21447 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21448 address-space properly.
21450 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21452 * doc/optinfo.texi (Optimization groups): Fix option used for
21454 * doc/invoke.texi (-fopt-info): Document "omp".
21455 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21456 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21457 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21458 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21459 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21461 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21463 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21464 OPTGROUP_OMP with "omp".
21466 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21469 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21470 for arithmetic shift of unsigned V2DI.
21472 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21474 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21475 arc/linux.h headers.
21476 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21477 (LINK_SPEC): Likewise.
21478 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21479 (EXTRA_SPECS): Likewise.
21480 (STARTFILE_SPEC): Likewise.
21481 (ENDFILE_SPEC): Likewise.
21482 (LIB_SPEC): Likewise.
21483 (TARGET_SDATA_DEFAULT): Likewise.
21484 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21485 (MULTILIB_DEFAULTS): Likewise.
21486 (DWARF2_UNWIND_INFO): Likewise.
21487 * config/arc/big.h: New file.
21488 * config/arc/elf.h: Likewise.
21489 * config/arc/linux.h: Likewise.
21490 * config/arc/t-uClibc: Remove.
21492 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21494 PR tree-optimization/77536
21495 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21496 (tree_transform_and_unroll_loop): Use above function to compute the
21497 estimated niter of unrolled loop and use it when scaling profile.
21498 Also use count info rather than frequency if it's non-zero.
21499 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21500 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21501 (vect_transform_loop): Call above function.
21503 2017-02-27 Richard Biener <rguenther@suse.de>
21505 PR tree-optimization/45397
21506 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21507 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21508 (visit_nary_op): Add pattern matching for CSEing sign-changed
21509 or truncated operations with wider ones.
21511 2017-02-27 Richard Biener <rguenther@suse.de>
21513 PR tree-optimization/79690
21514 * tree-vect-stmts.c (vectorizable_store): Use vector type
21515 built from the DR with address-space.
21517 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21519 * doc/invoke.texi (Optimize Options): Refine the description
21520 of asan-use-after-return.
21522 2017-02-25 Alan Modra <amodra@gmail.com>
21524 PR rtl-optimization/79584
21525 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21526 base, not ad->base_term, the reg within base. Remove assertion
21527 that ad->base == ad->base_term. Replace gen_int_mode using
21528 bogus mode with const0_rtx.
21530 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21532 PR middle-end/79396
21533 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21534 FMA_EXPR like tcc_binary or tcc_unary.
21536 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21539 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21541 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21542 (output_loc_operands): Handle DW_OP_call_ref and
21543 DW_OP_GNU_variable_value.
21544 (struct variable_value_struct): New type.
21545 (struct variable_value_hasher): Likewise.
21546 (variable_value_hash): New variable.
21547 (string_types): Remove.
21548 (copy_loc_descr): New function.
21549 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21550 (prepend_loc_descr_to_each): New function.
21551 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21552 instead of add_loc_descr_to_each if the first argument is single
21553 location list and the second has multiple.
21554 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21555 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21556 when looking for variable value which doesn't have other location info.
21557 (loc_list_from_tree): Formatting fix.
21558 (gen_array_type_die): Simplify DW_AT_string_length handling.
21559 (adjust_string_types): Remove.
21560 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21561 string_types. Call resolve_variable_values.
21562 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21563 (resolve_addr_in_expr): Likewise. Add A argument.
21564 (copy_deref_exprloc): Remove deref argument. Adjust for the
21565 original expression being DW_OP_GNU_variable_value with optionally
21566 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21567 optionally after it.
21568 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21569 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21570 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21571 (variable_value_hasher::hash, variable_value_hasher::equal): New
21573 (resolve_variable_value_in_expr, resolve_variable_value,
21574 resolve_variable_values, note_variable_value_in_expr,
21575 note_variable_value): New functions.
21576 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21579 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21582 * opts.h (handle_generated_option): Add GENERATED_P argument.
21583 * opts-common.c (handle_option): Adjust function comment.
21584 (handle_generated_option): Add GENERATED_P argument, pass it to
21586 (control_warning_option): Pass false to handle_generated_option
21588 * opts.c (maybe_default_option): Pass true to handle_generated_option
21590 * optc-gen.awk: Likewise.
21592 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21594 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21595 a REG, look at the REG it is a SUBREG of.
21596 (splitter for cmpeqsi_t): Ditto.
21598 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21600 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21601 the special USEs with the pattern of the insn, not the insn itself.
21603 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21606 * doc/invoke.texi: Document -mload-store-pairs.
21608 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21609 Sandra Loosemore <sandra@codesourcery.com>
21611 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21612 argument isn't a CONST_INT.
21613 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21614 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21615 (nios2_validate_compare): Bypass alternate compare logic if *op2
21616 is not a CONST_INT.
21617 (ldstwm_operation_p): Return false if first_base is not a REG or
21618 if first_offset is not a CONST_INT.
21620 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21622 * config/cris/cris.md: Use correct operand in a define_peephole2.
21624 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21626 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21628 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21630 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21631 this_insn if it is an INSN or JUMP_INSN.
21632 (force_offsettable): Look at base, not at addr.
21633 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21634 on things that aren't necessarily CONST_INTs.
21636 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21638 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21639 -mfpmath=sse is the default also for x86-32 targets with SSE2
21640 instruction set when @option{-ffast-math} is enabled
21642 2017-02-24 Jeff Law <law@redhat.com>
21644 PR rtl-optimizatoin/79286
21645 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21648 2017-02-24 Richard Biener <rguenther@suse.de>
21650 PR tree-optimization/79389
21651 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21654 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21656 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21657 function comment to reflect reality.
21658 (loop_exits_before_overflow): Fix typo in function description.
21660 2017-02-24 Richard Biener <rguenther@suse.de>
21662 PR tree-optimization/79389
21663 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21664 properly that a threading opportunity exists. Detect conditional
21665 copy/constant propagation opportunities.
21667 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21669 * config/visium/visium.md (type): Add trap.
21670 (b): New mode attribute.
21671 (*btst): Rename into...
21672 (*btst<mode>): ...this and adjust.
21673 (*cbranchsi4_btst_insn): Rename into...
21674 (*cbranch<mode>4_btst_insn): ...this and adjust.
21675 (trap): New define_insn.
21677 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21679 PR tree-optimization/79389
21680 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21681 (noce_reversed_cond_code): New function.
21682 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21683 reversed_comparison_code. Formatting fix.
21684 (noce_try_store_flag): Test rev_cond != NULL in addition to
21685 reversed_comparison_code.
21686 (noce_try_store_flag_constants): Likewise.
21687 (noce_try_store_flag_mask): Likewise.
21688 (noce_try_addcc): Use rev_cond if non-NULL instead of
21689 reversed_comparison_code.
21690 (noce_try_cmove_arith): Likewise. Formatting fixes.
21691 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21692 (noce_find_if_block): Initialize rev_cond.
21693 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21694 instead of false as last argument never attempt to reverse it
21697 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21699 PR tree-optimization/79663
21700 * tree-predcom.c (combine_chains): Process refs in reverse order
21701 only for ZERO length chains, and add explaining comment.
21703 2017-02-23 Jeff Law <law@redhat.com>
21705 PR tree-optimization/79578
21706 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21707 in call to operand_equal_p.
21709 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21712 * config/i386/cpuid.h: Fix another undefined behavior.
21714 2017-02-23 Richard Biener <rguenther@suse.de>
21716 PR tree-optimization/79683
21717 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21718 vector types for data-refs.
21720 2017-02-23 Martin Liska <mliska@suse.cz>
21722 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21724 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21726 PR middle-end/79665
21727 * internal-fn.c (get_range_pos_neg): Moved to ...
21728 * tree.c (get_range_pos_neg): ... here. No longer static.
21729 * tree.h (get_range_pos_neg): New prototype.
21730 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21731 are known to be in between 0 and signed maximum inclusive, try to
21732 expand both unsigned and signed divmod and use the cheaper one from
21735 2017-02-22 Jeff Law <law@redhat.com>
21737 PR tree-optimization/79578
21738 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21739 to compare base operands.
21741 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21744 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21745 gpc_reg_operand instead of fpr_reg_operand.
21747 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21749 * config/mips/mips.c (mips_return_in_memory): Force FP
21750 vector types to be returned in memory for o32 ABI.
21752 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21754 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21755 instead of DW_TAG_member for static data member declarations and don't
21756 set no_linkage_name for static inline data members.
21757 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21760 2017-02-22 Martin Liska <mliska@suse.cz>
21762 * doc/invoke.texi: Replace inequality signs with square brackets
21765 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21767 PR tree-optimization/68644
21768 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21770 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21773 * lra-constraints.c (simplify_operand_subreg): Handle
21774 WORD_REGISTER_OPERATIONS targets.
21776 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21779 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21780 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21781 elimination by swapping fld*.
21783 2017-02-22 Richard Biener <rguenther@suse.de>
21785 PR tree-optimization/79673
21786 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21787 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21788 irrelevant address-space qualifiers and avoiding a
21789 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21791 2017-02-22 Richard Biener <rguenther@suse.de>
21793 PR tree-optimization/79666
21794 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21795 to not symbolically negate if that may introduce undefined
21798 2017-02-22 Martin Liska <mliska@suse.cz>
21801 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21802 * data-streamer-out.c (streamer_write_gcov_count_stream):
21804 * value-prof.c (stream_out_histogram_value): Make assert more
21805 precise based on type of counter.
21807 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21810 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21811 Use nonimmediate_operand instead of memory_operand for operand 1.
21812 (float-extend standard_x87sse_constant_load splitter): Ditto.
21814 2017-02-21 Jeff Law <law@redhat.com>
21816 PR tree-optimization/79621
21817 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21818 blocks with edges to themselves.
21820 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21823 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21824 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21825 Use gimple_call_builtin_p.
21828 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21829 on temporarily removed DEBUG_INSNs.
21831 PR tree-optimization/79649
21832 * tree-loop-distribution.c (classify_partition): Give up on
21833 non-generic address space loads/stores.
21835 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21837 * doc/loop.texi (Loop manipulation): Remove nonexistent
21838 tree_ssa_loop_version from the documentation.
21839 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21841 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21844 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21845 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21846 * config/rs6000/rs6000.c: Include except.h.
21847 (rs6000_expand_split_stack_prologue): Call
21848 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21850 2017-02-21 Martin Jambor <mjambor@suse.cz>
21853 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21854 have been analyzed.
21856 2017-02-21 Martin Jambor <mjambor@suse.cz>
21858 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21859 for backward compatibility only.
21860 * doc/invoke.texi (Option Summary): Remove all references to
21861 -fipa-cp-alignment.
21863 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21867 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21869 * lra-constraints.c (curr_insn_transform): Handle
21870 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21872 2017-02-21 Martin Liska <mliska@suse.cz>
21874 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21876 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21878 PR translation/79638
21879 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21881 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21884 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21885 (arm_function_ok_for_sibcall): Return false for an indirect call by
21886 descriptor if all the argument registers are used.
21887 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21888 alignment of the function.
21890 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21892 PR tree-optimization/61441
21893 * simplify-rtx.c (simplify_const_unary_operation): For
21894 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21895 the sNaN unmodified.
21897 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21899 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21900 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21901 instead of SYSTEM_HEADER_DIR.
21903 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21904 Martin Liška <mliska@suse.cz>
21906 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21907 Fix typos and grammar, use active voice, and clarify.
21909 2017-02-20 Marek Polacek <polacek@redhat.com>
21911 PR middle-end/79537
21912 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21915 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21917 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21920 * config/i386/i386.c (ix86_expand_builtin): Handle
21921 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21922 ix86_builtins_isa[fcode].isa as a requirement of those
21923 flags and any other flag in the bitmask.
21924 (ix86_init_mmx_sse_builtins): Use 0 instead of
21925 ~OPTION_MASK_ISA_64BIT as mask.
21926 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21927 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21928 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21929 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21931 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21934 * lra-constraints.c (split_reg): Check requested split mode
21935 is supported by the register.
21937 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21939 * lra-constraints.c (simplify_operand_subreg): Remove early
21942 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21945 * lra-constraints.c (curr_insn_transform): Tighten condition
21946 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21948 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21951 * lra-constraints.c (curr_insn_transform): Handle
21952 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21954 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21957 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21959 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21961 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21964 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21967 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21969 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21972 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21974 * final.c (last_columnnum, override_columnnum): New variables.
21975 (final_start_function): Set last_columnnum, pass it to begin_prologue
21976 hook and pass 0 to dwarf2out_begin_prologue.
21977 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21978 to source_line debug hook.
21979 (notice_source_line): Compute last_columnnum and for debug_column_info
21980 return true on column changes.
21981 * debug.h (struct gcc_debug_hooks): Add column argument to
21982 source_line and begin_prologue hooks.
21983 (debug_nothing_int_charstar_int_bool): Remove prototype.
21984 (debug_nothing_int_int_charstar,
21985 debug_nothing_int_int_charstar_int_bool): New prototypes.
21986 (dwarf2out_begin_prologue): Add column argument.
21987 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21988 begin_prologue hooks.
21989 (debug_nothing_int_charstar_int_bool): Remove.
21990 (debug_nothing_int_int_charstar,
21991 debug_nothing_int_int_charstar_int_bool): New functions.
21992 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21993 through to dwarf2out_source_line.
21994 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21995 (dwarf2out_source_line): Add column argument, emit it if requested.
21996 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21998 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21999 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22000 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22001 through to dwarf2out_begin_prologue.
22002 (vmsdbgout_source_line): Add column argument, pass it through to
22003 dwarf2out_source_line.
22004 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22005 dbxout_source_line caller.
22006 (dbxout_source_line): Add column argument.
22008 * common.opt (gno-column-info, gcolumn-info): New options.
22009 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22010 (check_die): Also test for multiple DW_AT_decl_column attributes.
22011 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22012 DW_AT_decl_column if requested.
22013 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22015 (gen_variable_die): Likewise.
22016 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22017 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22020 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22021 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22022 (ix86_handle_option): Handle OPT_m3dnowa.
22023 * doc/invoke.texi (-m3dnowa): Document.
22024 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22025 -m3dnowa instead of -m3dnow -march=athlon.
22028 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22029 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22031 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22034 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22035 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22036 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22037 generator for vsx_xxpermdi_<mode>_be.
22038 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22039 force big-endian semantics.
22040 (vsx_xxpermdi_<mode>_be): New define_expand with same
22041 implementation as previous version of vsx_xxpermdi_<mode>.
22043 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22045 PR tree-optimization/79327
22046 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22047 variable, its initialization and use.
22049 2017-02-17 Julia Koval <julia.koval@intel.com>
22051 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22052 (OPTION_MASK_ISA_PKU_UNSET): New.
22053 (ix86_handle_option): Handle -mrdpid.
22054 * config/i386/cpuid.h (bit_RDPID): New.
22055 * config/i386/driver-i386.c (host_detect_local_cpu):
22056 Detect RDPID feature.
22057 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22058 * config/i386/i386-c.c (ix86_target_macros_internal):
22060 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22061 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22062 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22063 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22064 * config/i386/i386.md (define_insn "rdpid"): New.
22065 * config/i386/i386.opt Add -mrdpid.
22066 * config/i386/immintrin.h (_rdpid_u32): New.
22068 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22070 PR rtl-optimization/79541
22071 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22072 instead of transforming it into USE.
22074 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22076 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22077 If HONOR_SNANS (SFmode) force the input to a register.
22078 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22079 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22080 an frsp or similar insn.
22082 2017-02-17 Martin Liska <mliska@suse.cz>
22084 PR rtl-optimization/79577
22085 * params.def (selsched-max-sched-times): Increase minimum to 1.
22087 2017-02-17 Martin Liska <mliska@suse.cz>
22089 PR rtl-optimization/79574
22090 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22092 2017-02-17 Martin Liska <mliska@suse.cz>
22094 PR tree-optimization/79529
22095 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22096 ssa_defined_default_def_p to handle cases which are implicitly
22098 * tree-ssa.c (ssa_defined_default_def_p): New function.
22099 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22100 which are implicitly defined.
22101 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22103 2017-02-17 Richard Biener <rguenther@suse.de>
22105 PR middle-end/79576
22106 * params.def (max-ssa-name-query-depth): Limit to 10.
22108 2017-02-17 Richard Biener <rguenther@suse.de>
22110 PR tree-optimization/79552
22111 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22114 2017-02-17 Richard Biener <rguenther@suse.de>
22117 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22119 2017-02-17 Marek Polacek <polacek@redhat.com>
22121 PR middle-end/79536
22122 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22123 (fold_negate_expr): New wrapper.
22125 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22127 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22128 Correct terminology and de-emphasize pre-standard behavior.
22130 2017-02-16 Alan Modra <amodra@gmail.com>
22132 PR rtl-optimization/79286
22133 * ira.c (def_dominates_uses): New function.
22134 (update_equiv_regs): Don't create an equivalence for insns that
22135 may trap where the register def does not dominate the use.
22137 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22139 PR rtl-optimization/78127
22140 * lra.c (lra): Call lra_eliminate before finish the loop after
22143 2017-02-16 Richard Biener <rguenther@suse.de>
22145 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22147 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22148 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22149 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22150 (isl_val_int_from_wi): New function.
22151 (extract_affine_gmp): Rename to ...
22152 (extract_affine_wi): ... this, take a widest_int.
22153 (extract_affine_int): Just wrap extract_affine_wi.
22154 (add_param_constraints): Use isl_val_int_from_wi.
22155 (add_loop_constraints): Likewise, and extract_affine_wi.
22157 2017-02-15 Jeff Law <law@redhat.com>
22159 PR middle-end/79521
22160 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22161 ira_init_register_move_cost_if_necessary.
22163 2017-02-15 Martin Sebor <msebor@redhat.com>
22165 PR middle-end/32003
22166 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22167 removed in a prior commit.
22169 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22171 PR tree-optimization/79347
22172 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22173 counters during peeling.
22175 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22177 * Makefile.in (site.exp): Remove "set ISLVER".
22179 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22182 * real.c (real_from_integer): Call real_convert even for decimal.
22184 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22187 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22189 2017-02-14 Andrew Pinski <apinski@cavium.com>
22191 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22192 cores and change the partno/implementer to be correct.
22193 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22194 the 'B" as the implementer.
22195 * config/aarch64/aarch64-tune.md: Regenerate.
22197 2017-02-14 Carl Love <cel@us.ibm.com>
22199 * config/rs6000/rs6000.c: Add case statement entry to make the
22200 xvcvuxdsp built-in argument unsigned.
22201 * config/rs6000/vsx.md: Fix the source and return operand types so they
22202 match the instruction definitions from the ISA document. Fix typo
22203 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22206 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22209 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22210 member early_clobber_alts.
22211 * lra-lives.c (reg_early_clobber_p): New.
22212 (process_bb_lives): Use it.
22213 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22214 (debug_operand_data): Initialize early_clobber_alts.
22215 (setup_operand_alternative): Set up early_clobber_alts.
22216 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22217 alternatives to new_insn_reg.
22218 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22220 (lra_update_insn_regno_info): Pass the new arg.
22222 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22224 PR middle-end/79505
22225 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22226 (new_oacc_loop_raw): Don't clear already cleared fields.
22229 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22230 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22231 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22233 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22236 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22238 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22241 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22242 the extra instruction to the right place to store 128-bit constant
22245 2017-02-14 Martin Sebor <msebor@redhat.com>
22247 PR middle-end/79448
22248 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22249 warning for strings of unknown length.
22251 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22253 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22255 2017-02-14 Jeff Law <law@redhat.com>
22258 * ira-costs.c (scan_one_insn): Initialize register move costs
22259 for pseudos seen in USE/CLOBBER insns.
22261 PR tree-optimization/79095
22262 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22263 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22264 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22265 if the operands are known to be not equal, then the resulting range
22267 (intersect_ranges): If the new range is ~[0,0] and the old range is
22268 wide, then prefer ~[0,0].
22269 * tree-vrp.c (overflow_comparison_p_1): New function.
22270 (overflow_comparison_p): New function.
22271 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22272 if NAME is used in an overflow test.
22273 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22274 overflow check that can be expressed as an equality test, then adjust
22275 ops to be that equality test.
22277 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22279 * config/s390/s390-builtin-types.def: Remove flags argument.
22280 * config/s390/s390.c (s390_init_builtins): Likewise.
22282 2017-02-14 Martin Liska <mliska@suse.cz>
22284 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22285 vector. Fix trailing white spaces.
22287 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22289 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22292 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22294 PR rtl-optimization/68664
22295 * config/arm/arm.c (arm_sched_can_speculate_insn):
22296 New function. Declare prototype.
22297 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22299 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22301 PR rtl-optimization/68664
22302 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22304 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22306 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22308 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22309 max skip bytes for function, loop and jump.
22311 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22313 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22314 ABS_EXPR for gimple dump.
22316 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22319 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22321 PR tree-optimization/79408
22322 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22323 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22324 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22325 also if rhs1 is INTEGER_CST.
22327 2017-02-14 Richard Biener <rguenther@suse.de>
22329 PR middle-end/79432
22330 * tree-into-ssa.c (insert_phi_nodes): When the function can
22331 have abnormal edges rewrite SSA names with broken use-def
22332 dominance out of SSA and register them for PHI insertion.
22334 2017-02-13 Martin Sebor <msebor@redhat.com>
22336 PR middle-end/79496
22337 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22338 clearing info.nowrite flag when snprintf size argument is a range.
22340 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22342 * cprop.c (cprop_jump): Add missing space in string literal.
22343 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22344 (get_constraint_for_component_ref): Likewise.
22345 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22346 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22347 * lra-constraints.c (process_alt_operands): Likewise.
22348 * ipa-inline.c (inline_small_functions): Likewise.
22349 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22350 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22351 * trans-mem.c (diagnose_tm_1_op): Likewise.
22352 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22353 (grid_parallel_clauses_gridifiable): Likewise.
22355 * config/nvptx/mkoffload.c (process): Add space in between
22358 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22359 "MOD4_SSE_REGS" and "ALL_REGS".
22361 * spellcheck.c (test_data): Add , in between "foo" and "food".
22363 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22366 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22367 boundary crossing check and subsequent code generation agree.
22369 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22371 * config/aarch64/aarch64.c (has_memory_op): Delete.
22372 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22375 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22377 PR rtl-optimization/79388
22378 PR rtl-optimization/79450
22379 * combine.c (distribute_notes): When removing TEM_INSN for which
22380 corresponding dest has last value recorded, invalidate that last
22383 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22385 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22386 of explicit '@'. Add missing assembly comment marker on branch costs
22389 2017-02-13 Nathan Sidwell <nathan@acm.org>
22391 * gengtype-lex.l (<in_struct>): Add '/'.
22393 2017-02-13 Martin Liska <mliska@suse.cz>
22396 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22398 2017-02-13 Richard Biener <rguenther@suse.de>
22400 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22402 * configure: Re-generate.
22403 * config.in: Likewise.
22404 * graphite-dependences.c: Simplify as if
22405 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22406 * graphite-isl-ast-to-gimple.c: Likewise.
22407 * graphite-optimize-isl.c: Likewise.
22408 * graphite-poly.c: Likewise.
22409 * graphite-sese-to-poly.c: Likewise.
22410 * graphite.h: Likewise.
22411 * toplev.c: Include isl/version.h and use isl_version () for
22412 printing the ISL version.
22413 * doc/install.texi: Update ISL requirement.
22415 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22417 * doc/standards.texi (Standards): Update reference to
22420 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22422 * doc/extend.texi (Named Address Spaces): sourceware.org now
22424 * doc/install.texi (Binaries): Ditto.
22427 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22429 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22430 standard terminology "stringize"/"stringizing" throughout.
22431 * doc/cppinternals.texi: Likewise.
22433 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22435 * doc/extend.texi: Fix some spelling mistakes and typos.
22436 * doc/invoke.texi: Likewise.
22438 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22441 * params.def (inline-min-speedup) Change from 10 to 8.
22443 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22445 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22448 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22451 * ipa-inline-analysis.c (get_minimal_bb): New function.
22452 (record_modified): Use it.
22453 (remap_edge_change_prob): Handle also ancestor functions.
22455 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22457 * doc/contrib.texi (Contributors): Remove broken link into
22458 the Mauve CVS repository.
22460 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22462 PR middle-end/79454
22463 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22464 result computation whenever lhs doesn't have vector mode, not
22465 just when it has BLKmode.
22467 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22469 * doc/makefile.texi (profiledbootstrap): Refer to the
22470 installation instructions only in textual form.
22472 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22475 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22477 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22479 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22480 (Specific): Update mingw-w64 reference.
22482 (Specific): Remove broken link to Renesas RX processor.
22484 2017-02-10 Richard Biener <rguenther@suse.de>
22486 * toplev.c (process_options): Do not mention obsolete graphite
22487 options when printing sorry message about missing graphite support.
22488 Mention -floop-nest-optimize.
22490 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22492 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22493 (vtst_p16): Likewise.
22494 (vtstq_p8): Likewise.
22495 (vtstq_p16): Likewise.
22497 (vtstq_p64): Likewise.
22498 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22499 (vset_lane_p64): New.
22500 (vsetq_lane_p64): New.
22502 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22504 PR tree-optimization/79411
22505 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22506 stmt operands are SSA_NAMEs used in abnormal phis.
22507 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22510 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22513 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22516 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22518 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22520 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22522 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22526 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22527 not arbitrary TREE_CONSTANT.
22530 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22531 "omp declare target link" attribute unless is_global_var.
22532 * omp-offload.c (find_link_var_op): Likewise.
22534 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22535 Chung-Lin Tang <cltang@codesourcery.com>
22537 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22539 (gimplify_adjust_omp_clauses): Don't delete TILE.
22540 (gimplify_omp_for): Deal with TILE.
22541 * internal-fn.c (expand_GOACC_TILE): New function.
22542 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22544 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22545 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22547 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22548 avoid DIV for outermost collapse var.
22549 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22550 Remove out of date comments, fix whitespace.
22551 * omp-general.c (omp_extract_for_data): Deal with tiling.
22552 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22553 adjust OLF_DIM_BASE value.
22554 (struct omp_for_data): Add tiling field.
22555 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22556 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22557 for auto loops. Remove default auto determining, moved to
22558 oacc_loop_fixed_partitions.
22559 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22560 stmts, add e_mask field.
22561 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22562 (oacc_thread_numbers): Use oacc_dim_call.
22563 (oacc_xform_tile): New.
22564 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22565 (finish_oacc_loop): Adjust for ifns vector.
22566 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22567 add case for GOACC_TILE fns.
22568 (oacc_loop_xform_loop): Delete.
22569 (oacc_loop_process): Iterate over call list directly, and add
22570 handling for GOACC_TILE fns.
22571 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22573 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22574 vector partitioning to outer loops. Assign 2 partitions to loops
22575 when available. Add TILE handling.
22576 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22577 (execite_oacc_device_lower): Process GOACC_TILE fns,
22578 ignore unknown specs.
22579 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22580 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22581 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22583 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22585 * configure.ac (ACX_BUGURL): Update.
22586 * configure: Regenerate.
22588 2017-02-09 Richard Biener <rguenther@suse.de>
22590 PR tree-optimization/69823
22591 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22592 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22594 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22596 * config/arc/arc-c.def: Add __NPS400__ definition.
22597 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22598 (TARGET_NPS400): Define.
22600 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22602 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22604 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22605 pointer, arch_info.
22606 (arc_cpu_types): Fill the arch_info field with a pointer into the
22607 arc_arch_types table.
22608 (arc_selected_cpu): Declare.
22609 * config/arc/arc.c (arc_selected_cpu): Make global.
22610 (arc_selected_arch): Delete.
22611 (arc_base_cpu): Delete.
22612 (arc_override_options): Remove references to deleted variables,
22613 update access to arch information.
22614 (ARC_OPT): Update access to arch information.
22615 (ARC_OPTX): Likewise.
22616 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22617 (TARGET_ARC600): Update access to arch information.
22618 (TARGET_ARC601): Likewise.
22619 (TARGET_ARC700): Likewise.
22620 (TARGET_EM): Likewise.
22621 (TARGET_HS): Likewise.
22622 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22625 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22628 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22629 condition/operands for integer GE/LE/GEU/LEU operations.
22631 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22633 PR translation/79397
22634 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22637 2017-02-08 Martin Jambor <mjambor@suse.cz>
22640 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22641 whether allocation happened.
22642 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22643 nothing was allocated.
22645 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22647 PR tree-optimization/79408
22648 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22649 constant, but SSA_NAME with a known integer range, use the minimum
22650 of that range instead of op1 to determine if modulo can be replaced
22651 with its first operand.
22653 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22655 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22657 2017-02-08 Richard Biener <rguenther@suse.de>
22659 PR tree-optimization/71824
22660 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22661 Check all loops contained in the merged region.
22663 2017-02-07 Andrew Pinski <apinski@cavium.com>
22665 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22667 2017-02-07 Andrew Pinski <apinski@cavium.com>
22669 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22670 (thunderxt88): Likewise.
22671 (thunderxt81): Disable LSE and change v8.1 to v8.
22672 (thunderxt83): Likewise.
22674 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22675 Richard Biener <rguenther@suse.de>
22677 PR middle-end/79399
22678 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22679 type from int to size_t.
22680 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22682 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22684 PR rtl-optimization/79386
22685 * cprop.c (bypass_conditional_jumps): Initialize
22686 bypass_last_basic_block already before splitting bbs after
22687 unconditional traps...
22688 (bypass_conditional_jumps): ... rather than here.
22691 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22692 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22693 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22694 fix -masm=intel patterns.
22696 2017-02-07 Richard Biener <rguenther@suse.de>
22698 PR tree-optimization/79256
22699 PR middle-end/79278
22700 * builtins.c (get_object_alignment_2): Use min_align_of_type
22701 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22702 and ADJUST_FIELD_ALIGN.
22704 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22706 * doc/tm.texi: Regenerate.
22707 * stor-layout.c (layout_decl): Adjust.
22708 (update_alignment_for_field): Likewise.
22709 (place_field): Likewise.
22710 (min_align_of_type): Likewise.
22711 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22712 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22713 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22714 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22715 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22716 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22717 * config/i386/i386.c (x86_field_alignment): Likewise.
22718 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22719 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22720 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22721 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22722 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22723 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22727 2017-01-30 Richard Biener <rguenther@suse.de>
22729 PR tree-optimization/79256
22730 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22731 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22734 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22736 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22737 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22738 builtins to SImode and emit a zero-extend, if necessary.
22740 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22742 * docs/invoke.texi (RISC-V Options): Alphabetize.
22744 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22746 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22749 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22751 * config/riscv/riscv.c: New file.
22752 * common/config/riscv/riscv-common.c: Likewise.
22753 * config.gcc: Likewise.
22754 * config/riscv/constraints.md: Likewise.
22755 * config/riscv/elf.h: Likewise.
22756 * config/riscv/generic.md: Likewise.
22757 * config/riscv/linux.h: Likewise.
22758 * config/riscv/multilib-generator: Likewise.
22759 * config/riscv/peephole.md: Likewise.
22760 * config/riscv/pic.md: Likewise.
22761 * config/riscv/predicates.md: Likewise.
22762 * config/riscv/riscv-builtins.c: Likewise.
22763 * config/riscv/riscv-c.c: Likewise.
22764 * config/riscv/riscv-ftypes.def: Likewise.
22765 * config/riscv/riscv-modes.def: Likewise.
22766 * config/riscv/riscv-opts.h: Likewise.
22767 * config/riscv/riscv-protos.h: Likewise.
22768 * config/riscv/riscv.h: Likewise.
22769 * config/riscv/riscv.md: Likewise.
22770 * config/riscv/riscv.opt: Likewise.
22771 * config/riscv/sync.md: Likewise.
22772 * config/riscv/t-elf-multilib: Likewise.
22773 * config/riscv/t-linux: Likewise.
22774 * config/riscv/t-linux-multilib: Likewise.
22775 * config/riscv/t-riscv: Likewise.
22776 * configure.ac: Likewise.
22777 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22778 Waterman as RISC-V maintainers.
22779 * doc/install.texi: Add RISC-V entries.
22780 * doc/invoke.texi: Add RISC-V options section.
22781 * doc/md.texi: Add RISC-V constraints section.
22782 * configure: Regenerated.
22784 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22787 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22788 false values to be constant vectors with all 0 or all 1 bits set.
22789 (vcondu<mode><mode>): Likewise.
22790 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22792 (fpmask_comparison_operator): Update comment.
22793 (vecint_comparison_operator): New predicate.
22794 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22795 vector conditionals when the true and false values are constant
22796 vectors with all 0 bits or all 1 bits set.
22798 2017-02-06 Martin Sebor <msebor@redhat.com>
22800 PR tree-optimization/79376
22801 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22803 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22805 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22806 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22807 to simplify split condition.
22809 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22811 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22812 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22815 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22817 PR rtl-optimization/68664
22818 * target.def (can_speculate_insn): New hook.
22819 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22820 * doc/tm.texi: Regenerate.
22821 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22822 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22823 (rs6000_sched_can_speculate_insn): New function.
22825 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22827 PR tree-optimization/79284
22828 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22829 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22830 vectorizable_mask_load_store, vectorizable_operation,
22831 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22832 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22833 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22834 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22835 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22836 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22837 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22838 is_gimple_assign (stmt). Replace another such test with
22839 is_gimple_assign (stmt).
22841 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22844 * config/avr/avr.c (rtl-iter.h): Include it.
22845 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22846 (avr_legitimate_combined_insn): ...and implementation.
22848 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22850 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22851 * config/s390/s390.c (s390_const_operand_ok)
22852 (s390_canonicalize_comparison, s390_extract_part)
22853 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22854 (s390_contiguous_bitmask_p, s390_rtx_costs)
22855 (legitimize_pic_address): Likewise.
22856 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22857 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22858 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22859 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22860 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22862 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22864 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22865 REGNO($0) == REGNO($1).
22867 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22869 * config/s390/linux.h(SIZE_TYPE): Add comment.
22871 2017-02-06 Julian Brown <julian@codesourcery.com>
22872 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22873 Virendra Pathak <virendra.pathak@broadcom.com>
22875 * config/aarch64/aarch64-cores.def: Change the scheduler
22877 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22878 * config/aarch64/thunderx2t99.md: New file.
22880 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22882 * doc/standards.texi (Go Language): Update link to language
22885 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22887 * tree-eh.c (lower_resx): Sanitize profile.
22888 (cleanup_empty_eh_move_lp): Likewise.
22890 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22893 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22895 * cfgloopmanip.h (loop_version): Update prototype.
22896 * modulo-sched.c (sms_schedule): Update call of loop_version.
22897 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22898 * tree-parloops.c (gen_parallel_loop): Likewise.
22899 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22900 * tree-ssa-loop-split.c (split_loop): Likewise.
22901 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22902 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22904 2017-02-05 Martin Liska <mliska@suse.cz>
22907 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22909 (print_operand_address): Initialize a struct to zero.
22911 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22913 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22914 garbage collector only in textual form.
22916 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22918 * doc/extend.texi (x86 specific memory model extensions for
22919 transactional memory): Simplify a phrase.
22921 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22924 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22925 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22926 (atomic_storedi_1): Likewise.
22928 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22930 PR tree-optimization/79338
22931 * tree-parloops.c (gather_scalar_reductions): Don't call
22932 vect_analyze_loop_form for loop->inner before destroying loop's
22935 2017-02-03 Martin Sebor <msebor@redhat.com>
22937 PR tree-optimization/79327
22938 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22939 when precision has resulted in leading zeros.
22940 (format_integer): Adjust the likely counter to assume an unknown
22941 argument that may be zero is non-zero.
22943 2017-02-03 Jason Merrill <jason@redhat.com>
22946 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22947 avoid copying non-taken branch.
22949 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22951 PR tree-optimization/79340
22952 * tree-vect-loop.c (vectorizable_reduction): Release
22953 vec_defs elements after safe_splicing them into other vectors.
22956 PR tree-optimization/79327
22957 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22958 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22960 (format_integer): Use wide_int_to_tree instead of build_int_cst
22961 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22962 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22963 of shortest and longest sequence.
22965 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22967 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22968 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22970 2017-02-03 Walter Lee <walt@tilera.com>
22973 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22974 after initial stackframe link reg save.
22975 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22977 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22980 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22981 wu for stxssp alternative.
22983 2017-02-03 Martin Sebor <msebor@redhat.com>
22985 PR tree-optimization/79352
22986 * gimple-fold.c (get_range_strlen): Add argument.
22987 (get_range_strlen): Change return type to bool.
22988 (get_maxval_strlen): Pass in a dummy argument.
22989 * gimple-fold.h (get_range_strlen): Change return type to bool.
22990 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22991 * tree.h (array_at_struct_end_p): Add argument.
22992 * tree.c (array_at_struct_end_p): Handle it.
22994 2017-02-03 Martin Liska <mliska@suse.cz>
22997 * multiple_target.c (create_dispatcher_calls): Redirect edge
22998 from a caller of a dispatcher.
22999 (expand_target_clones): Make the clones local.
23000 (ipa_target_clone): Do both target clones and resolvers.
23001 (ipa_dispatcher_calls): Remove the pass.
23002 (pass_dispatcher_calls::gate): Likewise.
23003 (make_pass_dispatcher_calls): Likewise.
23004 * passes.def (pass_target_clone): Put as very first IPA early
23007 2017-02-03 Martin Liska <mliska@suse.cz>
23009 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23010 in case of a function with ifunc attribute.
23012 2017-02-03 Martin Liska <mliska@suse.cz>
23014 * cgraph.c (cgraph_node::dump): Dump function version info.
23015 * symtab.c (symtab_node::dump_base): Add missing new line.
23017 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23019 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23020 (ifcombine_ifandif): Use it.
23022 2017-02-03 Martin Liska <mliska@suse.cz>
23024 * doc/invoke.texi: Document default value for
23025 use-after-scope-direct-emission-threshold.
23027 2017-02-03 Martin Liska <mliska@suse.cz>
23029 PR tree-optimization/79339
23030 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23031 (format_floating): Likewise.
23033 2017-02-03 Martin Liska <mliska@suse.cz>
23036 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23038 (ipa_node_params_t::remove): Likewise.
23039 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23040 initialization from removed ipa_node_params_t::insert.
23041 (ipa_node_params::~ipa_node_params): Move from removed
23042 ipa_node_params_t::release.
23043 * symbol-summary.h (symbol_summary::m_released): New member.
23044 Do not release a summary twice. Do not allow to call finalizer
23045 for types of a summary that live in GGC memory.
23047 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23049 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23052 2017-02-02 Martin Sebor <msebor@redhat.com>
23054 PR middle-end/79275
23055 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23056 (format_string): Tighten up the range of output for non-constant
23057 strings and correct the expected range for wide non-constant strings.
23059 2017-02-02 Martin Sebor <msebor@redhat.com>
23061 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23063 PR middle-end/32003
23064 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23066 (-fdump-tree-@var): Add to index and document how to come up
23067 with pass-specific option and dump file names.
23068 (-fdump-passes): Clarify where to look for output.
23070 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23072 PR middle-end/77445
23073 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23074 statistics of the analyzed path; allow threading for speed when
23075 any of BBs along the path are optimized for speed.
23077 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23079 PR middle-end/78468
23080 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23081 settings of the virtual registers.
23084 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23086 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23087 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23090 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23092 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23093 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23095 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23097 * config/s390/s390.md: Add missing comments with the expanded
23099 * config/s390/vector.md: Likewise.
23100 * config/s390/vx-builtins.md: Likewise.
23102 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23105 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23106 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23107 conditions on a single line.
23109 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23111 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23112 __S390_VX__ to __VX__.
23114 2017-02-01 Andrew Pinski <apinski@cavium.com>
23116 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23117 stmt_info to record_stmt_cost.
23118 (vect_get_known_peeling_cost): Pass stmt_info if known to
23120 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23121 cpu_vector_cost field into
23122 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23123 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23124 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23125 splitting of scalar_stmt_cost and vec_stmt_cost.
23126 (thunderx_vector_cost): Likewise.
23127 (cortexa57_vector_cost): LIkewise.
23128 (exynosm1_vector_cost): Likewise.
23129 (xgene1_vector_cost): Likewise.
23130 (thunderx2t99_vector_cost): Improve after the splitting of the two
23132 (aarch64_builtin_vectorization_cost): Update for the splitting of
23133 scalar_stmt_cost and vec_stmt_cost.
23135 2017-02-01 Torvald Riegel <triegel@redhat.com>
23136 Richard Henderson <rth@redhat.com>
23138 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23139 conditional on existance of a fast atomic load.
23140 * optabs-query.c (can_atomic_load_p): New function.
23141 * optabs-query.h (can_atomic_load_p): Declare it.
23142 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23143 no fast atomic load is available for the particular size of access.
23144 (expand_atomic_compare_and_swap): Likewise.
23145 (expand_atomic_load): Likewise.
23146 (expand_atomic_store): Likewise.
23147 (expand_atomic_fetch_op): Likewise.
23148 * testsuite/lib/target-supports.exp
23149 (check_effective_target_sync_int_128): Remove x86 because it provides
23150 no fast atomic load.
23151 (check_effective_target_sync_int_128_runtime): Likewise.
23153 2017-02-01 Richard Biener <rguenther@suse.de>
23155 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23156 (graphite_transform_loops): Provide opt-info for optimized nests.
23157 * tree-parloop.c (parallelize_loops): Provide opt-info for
23158 parallelized loops.
23160 2017-02-01 Richard Biener <rguenther@suse.de>
23162 PR middle-end/79315
23163 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23164 was not set before.
23166 2017-02-01 Richard Biener <rguenther@suse.de>
23168 PR tree-optimization/71824
23169 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23170 Verify the loops are valid in the merged SESE region.
23171 (scop_detection::can_represent_loop_1): Check analyzing the
23172 evolution of the number of iterations in the region succeeds.
23174 2017-01-31 Ian Lance Taylor <iant@golang.org>
23176 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23177 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23179 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23181 PR preprocessor/79210
23182 * input.c (get_substring_ranges_for_loc): Replace line_width
23183 assertion with error-handling.
23185 2017-01-31 Richard Biener <rguenther@suse.de>
23187 PR tree-optimization/77318
23188 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23189 (create_pw_aff_from_tree): Take loop parameter.
23190 (add_condition_to_pbb): Pass loop of the condition to
23191 create_pw_aff_from_tree.
23193 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23195 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23196 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23198 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23202 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23204 (convert_int_to_float128): Likewise.
23205 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23206 (convert_int_to_float128): Likewise.
23207 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23208 (UNSPEC_IEEE128_CONVERT): Likewise.
23209 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23210 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23211 Use local variables for IBM extended format.
23212 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23213 (fix_trunc<mode>si2_fprs): Likewise.
23214 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23215 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23216 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23217 to know that we can now have integers of all sizes in vector
23219 (fix<uns>_<mode>di2_hw): Likewise.
23220 (float<uns>_<mode>si2_hw): Likewise.
23221 (fix_<mode>si2_hw): Likewise.
23222 (fixuns_<mode>si2_hw): Likewise.
23223 (float<uns>_<mode>di2_hw): Likewise.
23224 (float_<mode>di2_hw): Likewise.
23225 (float_<mode>si2_hw): Likewise.
23226 (floatuns_<mode>di2_hw): Likewise.
23227 (floatuns_<mode>si2_hw): Likewise.
23228 (xscvqp<su>wz_<mode>): Delete, no longer used.
23229 (xscvqp<su>dz_<mode>): Likewise.
23230 (xscv<su>dqp_<mode>): Likewise.
23231 (ieee128_mfvsrd_64bit): Likewise.
23232 (ieee128_mfvsrd_32bit): Likewise.
23233 (ieee128_mfvsrwz): Likewise.
23234 (ieee128_mtvsrw): Likewise.
23235 (ieee128_mtvsrd_64bit): Likewise.
23236 (ieee128_mtvsrd_32bit): Likewise.
23238 2017-01-31 Martin Liska <mliska@suse.cz>
23241 * ipa-prop.c (ipa_free_all_node_params): Call release method
23242 instead of ~sumbol_summary to not to trigger double times
23245 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23247 PR tree-optimization/71691
23248 * bitmap.h (class auto_bitmap): New.
23249 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23250 is_maybe_undefined instead of ssa_undefined_value_p.
23252 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23254 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23255 __S390_ARCH_LEVEL__ to __ARCH__.
23257 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23259 PR tree-optimization/79267
23260 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23261 if should_remove_lhs_p is true.
23263 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23266 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23267 (add_alignment_attribute): New.
23268 (base_type_die): Add alignment attribute.
23269 (subrange_type_die): Likewise.
23270 (modified_type_die): Likewise.
23271 (gen_array_type_die): Likewise.
23272 (gen_descr_array_type_die: Likewise.
23273 (gen_enumeration_type_die): Likewise.
23274 (gen_subprogram_die): Likewise.
23275 (gen_variable_die): Likewise.
23276 (gen_field_die): Likewise.
23277 (gen_ptr_to_mbr_type_die): Likewise.
23278 (gen_struct_or_union_type_die): Likewise.
23279 (gen_subroutine_type_die): Likewise.
23280 (gen_typedef_die): Likewise.
23281 (base_type_cmp): Compare alignment attribute.
23283 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23286 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23287 (setb_unsigned) New pattern for setb with CCUNS.
23288 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23289 subfc./subfe sequence to avoid overflow problems. Generate a
23290 shorter sequence with cmpld/setb for power9.
23291 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23292 for generating subfc. instruction.
23293 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23294 now uses this instruction.
23296 2017-01-30 Ian Lance Taylor <iant@google.com>
23299 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23300 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23302 2017-01-30 Martin Sebor <msebor@redhat.com>
23304 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23305 Move constant to the right of a relational operator.
23306 (get_mpfr_format_length, format_character, format_string): Ditto.
23307 (should_warn_p, maybe_warn): Same.
23309 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23311 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23314 * asan.c (get_translation_unit_decl): Remove function.
23315 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23317 2017-01-30 Martin Liska <mliska@suse.cz>
23319 PR gcov-profile/79259
23320 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23321 -fprofile-generate.
23323 2017-01-30 Martin Liska <mliska@suse.cz>
23326 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23327 Initialize variables with NULL value.
23329 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23332 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23334 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23336 2017-01-30 Richard Biener <rguenther@suse.de>
23338 PR tree-optimization/79276
23339 * tree-vrp.c (process_assert_insertions): Properly adjust common
23340 when removing a duplicate.
23342 * gcc.dg/torture/pr79276.c: New testcase.
23344 2017-01-30 Richard Biener <rguenther@suse.de>
23346 PR tree-optimization/79256
23347 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23348 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23350 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23352 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23355 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23356 ("*r<noxa>sbg_<mode>_sll_bitmask")
23357 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23358 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23359 Use contiguous_bitmask_nowrap_operand.
23361 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23364 * config/rs6000/altivec.h (vec_xl): Revise #define.
23365 (vec_xst): Likewise.
23367 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23369 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23371 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23373 PR rtl-optimization/79194
23374 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23375 traps before call to bypass_conditional_jumps.
23377 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23379 PR tree-optimization/71374
23380 * lra-constraints.c (check_conflict_input_operands): New.
23381 (match_reload): Use it.
23383 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23386 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23387 account to calculate conflict_set.
23389 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23391 PR rtl-optimization/78559
23392 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23393 other_insn in combine.
23395 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23397 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23398 uint16_type_node for BT_UINT16.
23400 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23402 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23403 "RTL Tests" to menu.
23404 (GIMPLE Tests): New node.
23405 (RTL Tests): New node.
23407 2017-01-27 Richard Biener <rguenther@suse.de>
23409 PR tree-optimization/79245
23410 * tree-loop-distribution.c (distribute_loop): Apply cost
23411 modeling also to detected patterns.
23413 2017-01-27 Richard Biener <rguenther@suse.de>
23415 PR tree-optimization/71433
23416 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23417 (compare_assert_loc): New function.
23418 (process_assert_insertions): Sort and optimize assert locations
23419 to remove duplicates and push down identical assertions on
23420 edges to their destination block.
23422 2017-01-27 Richard Biener <rguenther@suse.de>
23424 PR tree-optimization/79244
23425 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23426 out SSA names even if abnormal.
23428 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23430 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23431 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23432 instead of MPFR_RNDN.
23434 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23437 * arm.c (arm_option_override): Don't call build_target_option_node
23438 until after doing all option overrides.
23439 (arm_valid_target_attribute_tree): Likewise.
23441 2017-01-27 Martin Liska <mliska@suse.cz>
23443 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23444 for {cd}tors and C++ {cd}tors.
23446 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23448 * config/s390/s390.md ("*setmem_long_and")
23449 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23451 2017-01-26 Martin Sebor <msebor@redhat.com>
23453 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23456 2017-01-26 Martin Sebor <msebor@redhat.com>
23458 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23459 HAVE_DFmode before using XFmode or DFmode.
23460 (parse_directive): Avoid using the z length modifier to avoid
23461 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23463 PR middle-end/78703
23464 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23465 to accept adjustment as an array.
23466 (get_int_range): New function.
23467 (struct directive): Make width and prec arrays.
23468 (directive::set_width, directive::set_precision): Call get_int_range.
23469 (format_integer, format_floating): Handle width and precision ranges.
23470 (format_string, parse_directive): Same.
23472 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23475 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23476 comdat_type_p set, just clone them, but keep the children in the
23480 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23481 which have direct callers with -fvar-tracking-assignments enabled
23483 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23484 inside of type units.
23486 2017-01-26 Martin Sebor <msebor@redhat.com>
23488 PR middle-end/78703
23489 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23491 (struct format_result): Replace number_chars, number_chars_min,
23492 and number_chars_max with a single member of struct result_range.
23494 (format_result::operator+=): Adjust.
23495 (struct fmtresult): Remove bounded. Handle likely and unlikely
23497 (fmtresult::adjust_for_width_or_precision): New function.
23498 (fmtresult:type_max_digits): New function.
23499 (bytes_remaining): Handle likely and unlikely counters.
23500 (min_bytes_remaining): Remove.
23501 (format_percent): Simplify.
23502 (format_integer, format_floating): Set likely and unlikely counters.
23503 (get_string_length, format_character, format_string): Same.
23504 (format_plain, should_warn_p): New function.
23505 (maybe_warn): Call should_warn_p. Update diagnostic messages
23506 and handle those for all directives, including plain strings.
23507 (format_directive): Handle likely and unlikely counters.
23508 Remove unnecessary quoting from diagnostics. Add an informational
23510 (add_bytes): Remove.
23511 (pass_sprintf_length::compute_format_length): Simplify.
23512 (try_substitute_return_value): Handle likely and unlikely counters.
23514 2017-01-26 Carl Love <cel@us.ibm.com>
23516 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23517 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23519 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23522 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23523 endianess for subregs into account.
23524 * lra-constraints.c (lra_constraints): Do risky transformations
23525 always on the first iteration.
23526 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23527 last_call_used_reg_set.
23528 (process_bb_lives): Define and use last_call_used_reg_set.
23529 * lra.c (lra): Always continue after lra_constraints on the first
23532 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23534 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23536 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23538 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23540 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23541 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23542 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23543 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23544 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23545 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23546 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23547 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23548 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23550 2017-01-26 Marek Polacek <polacek@redhat.com>
23553 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23554 for the third operand.
23556 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23558 PR middle-end/79236
23559 * omp-low.c (struct omp_context): Add simt_stmt field.
23560 (scan_omp_for): Return omp_context *.
23561 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23562 context to the _simt_ SIMD stmt.
23563 (lower_omp_for): For combined SIMD with sibling _simt_
23564 SIMD, make sure to use the same decls in _looptemp_
23565 clauses as in the sibling.
23567 2017-01-26 David Sherwood <david.sherwood@arm.com>
23569 PR middle-end/79212
23570 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23573 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23576 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23577 emit fld b; fld a; if possible.
23579 * brig-builtins.def: Update copyright years.
23580 * config/arm/arm_acle_builtins.def: Update copyright years.
23582 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23585 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23586 constraint instead of o for the stxsd instruction.
23588 2017-01-25 Carl Love <cel@us.ibm.com>
23590 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23591 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23593 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23595 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23597 2017-01-25 Richard Biener <rguenther@suse.de>
23599 PR tree-optimization/69264
23600 * target.def (vector_alignment_reachable): Improve documentation.
23601 * doc/tm.texi: Regenerate.
23602 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23604 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23605 earlier changes with respect to TYPE_USER_ALIGN.
23606 (vector_alignment_reachable_p): Likewise. Improve dumping.
23608 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23611 * config/arm/arm.md (xordi3): Force constant operand into a register
23614 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23616 * doc/invoke.texi (-fstore-merging): Correct default optimization
23617 levels at which it is enabled.
23618 (-O): Move -fstore-merging from list to...
23621 2017-01-25 Richard Biener <rguenther@suse.de>
23624 * omp-expand.c: Include debug.h.
23625 (expand_omp_taskreg): Make sure to generate early debug before
23626 outlining anything from a function.
23627 (expand_omp_target): Likewise.
23628 (grid_expand_target_grid_body): Likewise.
23630 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23633 * asan.c (get_translation_unit_decl): New function.
23634 (asan_add_global): Extract modules file name from globals
23635 TRANSLATION_UNIT_DECL name.
23637 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23640 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23641 for long calls with APCS frame and VFP.
23643 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23645 * cfg.c (original_copy_tables_initialized_p): New function.
23646 * cfg.h (original_copy_tables_initialized_p): New decl.
23647 * cfgrtl.c (relink_block_chain): Guard the call to
23648 free_original_copy_tables with a call to
23649 original_copy_tables_initialized_p.
23650 * cgraph.h (symtab_node::native_rtl_p): New decl.
23651 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23652 (symtab_node::needed_p): Don't assert for early assembly output
23653 for __RTL functions.
23654 (cgraph_node::finalize_function): Set "force_output" for __RTL
23656 (cgraph_node::analyze): Bail out early for __RTL functions.
23657 (analyze_functions): Update assertion to support __RTL functions.
23658 (cgraph_node::expand): Bail out early for __RTL functions.
23659 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23661 * function.h (struct function): Update comment for field
23663 * gimple-expr.c: Include "tree-pass.h".
23664 (gimple_has_body_p): Return false for __RTL functions.
23665 * Makefile.in (OBJS): Add run-rtl-passes.o.
23666 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23668 (gcc::pass_manager::get_clean_slate): New accessor.
23669 * passes.c: Include "insn-addr.h".
23670 (should_skip_pass_p): Add logging. Update logic for running
23671 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23672 property-provider override so it is only done for gimple passes.
23674 (skip_pass): New function.
23675 (execute_one_pass): Call skip_pass when skipping passes.
23676 * read-md.c (md_reader::read_char): Support filtering
23677 the input to a subset of line numbers.
23678 (md_reader::md_reader): Initialize fields
23679 m_first_line and m_last_line.
23680 (md_reader::read_file_fragment): New function.
23681 * read-md.h (md_reader::read_file_fragment): New decl.
23682 (md_reader::m_first_line): New field.
23683 (md_reader::m_last_line): New field.
23684 * read-rtl-function.c (function_reader::create_function): Only
23685 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23686 curr_properties. Set DECL_INITIAL to a dummy block.
23687 (read_rtl_function_body_from_file_range): New function.
23688 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23690 * run-rtl-passes.c: New file.
23691 * run-rtl-passes.h: New file.
23693 2017-01-24 Jeff Law <law@redhat.com>
23695 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23698 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23700 PR tree-optimization/79159
23701 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23702 (record_nonwrapping_iv): Improve boundary using above function if no
23703 value range information.
23705 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23706 Martin Jambor <mjambor@suse.cz>
23708 * brig-builtins.def: New file.
23709 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23710 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23711 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23712 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23713 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23714 * builtin-types.def (BT_INT8): New.
23715 (BT_INT16): Likewise.
23716 (BT_UINT8): Likewise.
23717 (BT_UINT16): Likewise.
23718 (BT_FN_ULONG): Likewise.
23719 (BT_FN_UINT_INT): Likewise.
23720 (BT_FN_UINT_ULONG): Likewise.
23721 (BT_FN_UINT_LONG): Likewise.
23722 (BT_FN_UINT_PTR): Likewise.
23723 (BT_FN_ULONG_PTR): Likewise.
23724 (BT_FN_INT8_FLOAT): Likewise.
23725 (BT_FN_INT16_FLOAT): Likewise.
23726 (BT_FN_UINT32_FLOAT): Likewise.
23727 (BT_FN_UINT16_FLOAT): Likewise.
23728 (BT_FN_UINT8_FLOAT): Likewise.
23729 (BT_FN_UINT64_FLOAT): Likewise.
23730 (BT_FN_UINT16_UINT32): Likewise.
23731 (BT_FN_UINT32_UINT16): Likewise.
23732 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23733 (BT_FN_INT_PTR_INT): Likewise.
23734 (BT_FN_UINT_PTR_UINT): Likewise.
23735 (BT_FN_LONG_PTR_LONG): Likewise.
23736 (BT_FN_ULONG_PTR_ULONG): Likewise.
23737 (BT_FN_VOID_UINT64_UINT64): Likewise.
23738 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23739 (BT_FN_INT8_INT8_INT8): Likewise.
23740 (BT_FN_INT16_INT16_INT16): Likewise.
23741 (BT_FN_INT_INT_INT): Likewise.
23742 (BT_FN_UINT_FLOAT_UINT): Likewise.
23743 (BT_FN_FLOAT_UINT_UINT): Likewise.
23744 (BT_FN_ULONG_UINT_UINT): Likewise.
23745 (BT_FN_ULONG_UINT_PTR): Likewise.
23746 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23747 (BT_FN_UINT_UINT_UINT): Likewise.
23748 (BT_FN_VOID_UINT_PTR): Likewise.
23749 (BT_FN_UINT_UINT_PTR: Likewise.
23750 (BT_FN_UINT32_UINT64_PTR): Likewise.
23751 (BT_FN_INT_INT_UINT_UINT): Likewise.
23752 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23753 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23754 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23755 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23756 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23757 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23758 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23759 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23760 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23761 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23762 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23763 * doc/frontends.texi: List BRIG FE.
23764 * doc/install.texi (Testing): Add BRIG tesring requirements.
23765 * doc/invoke.texi (Overall Options): Mention BRIG.
23766 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23768 2017-01-24 Richard Biener <rguenther@suse.de>
23770 PR translation/79208
23771 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23773 2017-01-24 Martin Jambor <mjambor@suse.cz>
23776 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23777 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23778 and known_contexts.
23780 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23782 PR middle-end/79123
23783 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23784 casts from signed to unsigned really don't have a range.
23786 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23788 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23789 GMP_RNDx for compatiblity.
23791 2017-01-24 Martin Liska <mliska@suse.cz>
23794 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23795 that would prevent us to call alloca with -1 as argument.
23797 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23799 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23800 Avoid -Wformat-security warning.
23802 2017-01-23 Andrew Pinski <apinski@cavium.com>
23804 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23807 2017-01-23 Martin Sebor <msebor@redhat.com>
23809 PR middle-end/78703
23810 * gimple-ssa-sprintf.c (warn_level): New global.
23811 (format_integer): Use it here and throughout the rest of the file.
23812 Use the same switch to compute sign as base.
23813 (maybe_warn): New function.
23814 (format_directive): Factor out warnings into maybe_warn.
23815 Add debugging output. Use warn_level.
23816 (add_bytes): Use warn_level.
23817 (pass_sprintf_length::compute_format_length): Add debugging output.
23818 (try_substitute_return_value): Same.
23819 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23821 PR middle-end/78703
23822 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23823 (struct fmtresult, format_integer, format_floating): Adjust.
23824 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23825 (get_string_length, format_string,format_directive): Same.
23826 (pass_sprintf_length::compute_format_length): Same.
23827 (try_substitute_return_value): Simplify slightly.
23829 PR middle-end/78703
23830 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23831 (fmtresult::operator+=): Outlined.
23832 (struct fmtresult): Add ctors.
23833 (struct conversion_spec): Rename...
23834 (struct directive): ...to this. Add and remove data members.
23835 (directive::set_width, directive::set_precision): New functions.
23836 (format_percent): Use fmtresult ctor.
23837 (get_width_and_precision): Remove.
23838 (format_integer): Make naming changes. Avoid computing width and
23840 (format_floating): Same. Adjust indentation.
23841 (format_character, format_none): New functions.
23842 (format_string): Moved character handling to format_character.
23843 (format_directive): Remove arguments, change return type.
23844 (parse_directive): New function.
23845 (pass_sprintf_length::compute_format_length): Move directive
23846 parsing to parse_directive.
23848 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23850 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23851 (assign_assembler_name_if_needed): ... this.
23852 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23853 (assign_assembler_name_if_needed): ... this.
23854 (free_lang_data_in_cgraph): Adjust callers.
23855 * cgraphunit.c (cgraph_node::analyze): Likewise.
23856 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23859 2017-01-23 Richard Biener <rguenther@suse.de>
23861 PR tree-optimization/79088
23862 PR tree-optimization/79188
23863 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23864 resetting loop bounds after last path deletion. Reset loop
23865 bounds of the target loop, make code match the comments.
23866 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23867 Make sure loops need no fixups.
23869 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23871 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23872 exponent support with double type for first argument.
23873 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23874 type returned by __builtin_vec_extract_sig,
23875 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23876 functions from "vector int" to "vector unsigned int" or from
23877 "vector long long int" to "vector unsigned long long int".
23878 Changed type returned by __builtin_vec_extract_exp,
23879 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23880 functions from "vector int" to "vector unsigned int" or from
23881 "vector long long int" to "vector unsigned long long int".
23882 Changed return type of __builtin_vec_test_data_class,
23883 __builtin_vec_test_data_class_sp, and
23884 __builtin_vec_test_data_class_dp from "vector int" to
23885 "vector bool int" or from "vector long long int" to "vector bool
23886 long long int" and changed second argument type from "unsigned
23887 int" to "int". Added new overloaded function forms "vector float
23888 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23889 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23890 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23891 double, vector unsigned long long int)" and "vector double
23892 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23893 long int)". Changed return type of
23894 __builtin_scalar_test_data_class and
23895 __builtin_scalar_test_data_class_sp and
23896 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23897 int" and changed second argument from "unsigned int" to "int".
23898 Changed type returned by __builtin_scalar_test_neg,
23899 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23900 from "int" to "bool int". Added new overloaded function form
23901 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23902 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23903 exponent double-precision with floating point first argument.
23904 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23905 documentation of scalar_test_data_class, scalar_test_neg,
23906 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23907 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23908 vec_test_data_class built-in functions to reflect refinements in
23909 their type signatures.
23911 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23913 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23915 (aarch64_elf_asm_destructor): Likewise.
23917 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23919 PR rtl-optimization/78634
23920 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23921 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23922 * ifcvt.c (noce_try_cmove): Add missing cost check.
23924 PR rtl-optimization/71724
23925 * combine.c (if_then_else_cond): Look for situations where it is
23926 beneficial to undo the work of one of the recursive calls.
23928 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23930 PR tree-optimization/70754
23931 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23932 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23933 combined stmt before it if not NULL.
23934 (combine_chains): Process refs reversely and compute dominance point
23937 2017-01-23 Martin Liska <mliska@suse.cz>
23939 PR tree-optimization/79196
23940 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23941 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23943 (strlen_optimize_stmt): Call the renamed function.
23945 2017-01-23 Michael Matz <matz@suse.de>
23947 PR tree-optimization/78384
23948 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23950 2017-01-23 Richard Biener <rguenther@suse.de>
23952 PR tree-optimization/79186
23953 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23954 both incoming edges before moving an assert.
23956 2017-01-23 Martin Jambor <mjambor@suse.cz>
23958 * ipa-prop.c (load_from_param_1): Removed.
23959 (load_from_unmodified_param): Bits from load_from_param_1 put back
23961 (load_from_param): Removed.
23962 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23963 with stmt. Reverted back to use of load_from_unmodified_param.
23965 2017-01-23 Martin Jambor <mjambor@suse.cz>
23968 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23969 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23970 field a pointer to garbage collected vector, mark lattices and
23971 ipcp_orig_node with GTY((skip)).
23972 (ipa_get_param_count): Adjust to descriptors being a pointer.
23973 (ipa_get_param): Likewise.
23974 (ipa_get_type): Likewise.
23975 (ipa_get_param_move_cost): Likewise.
23976 (ipa_set_param_used): Likewise.
23977 (ipa_get_controlled_uses): Likewise.
23978 (ipa_set_controlled_uses): Likewise.
23979 (ipa_is_param_used): Likewise.
23980 (ipa_node_params_t): Move into garbage collector. New methods insert
23982 (ipa_node_params_sum): Annotate wth GTY(()).
23983 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23985 (ipa_load_from_parm_agg): Adjust declaration.
23986 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23987 * ipa-profile.c (ipa_profile): Likewise.
23988 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23989 (ipa_populate_param_decls): Make descriptors parameter garbage
23991 (ipa_dump_param): Adjust to descriptors being a pointer.
23992 (ipa_alloc_node_params): Likewise.
23993 (ipa_initialize_node_params): Likewise.
23994 (load_from_param_1): Make descriptors parameter garbage collected.
23995 (load_from_unmodified_param): Likewise.
23996 (load_from_param): Likewise.
23997 (ipa_load_from_parm_agg): Likewise.
23998 (ipa_node_params::~ipa_node_params): Removed.
23999 (ipa_free_all_node_params): Remove call to delete operator.
24000 (ipa_node_params_t::insert): New.
24001 (ipa_node_params_t::remove): Likewise.
24002 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24003 copy known_csts and known_contexts vectors.
24004 (ipa_read_node_info): Adjust to descriptors being a pointer.
24005 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24007 (ipcp_transform_function): Make descriptors variable garbage
24010 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24012 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24013 * config/i386/avx512dqintrin.h: Ditto.
24014 * config/i386/avx512fintrin.h: Ditto.
24015 * config/i386/i386.c: Handle new builtins.
24016 * config/i386/i386-builtin.def: Add new builtins.
24017 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24018 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24020 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24021 Martin Liska <mliska@suse.cz>
24023 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24024 * asan.c (asan_expand_poison_ifn): Support stores and use
24025 appropriate ASAN report function.
24026 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24027 * internal-fn.def (ASAN_POISON_USE): Declare.
24028 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24029 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24030 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24031 ASAN_POISON calls w/o LHS.
24032 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24033 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24034 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24035 * gimplify.c (asan_poison_variables): Add attribute
24036 use_after_scope_memory to variables that really needs to live
24038 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24039 having the attribute.
24041 2017-01-23 Martin Liska <mliska@suse.cz>
24043 * asan.c (create_asan_shadow_var): New function.
24044 (asan_expand_poison_ifn): Likewise.
24045 * asan.h (asan_expand_poison_ifn): New declaration.
24046 * internal-fn.c (expand_ASAN_POISON): Likewise.
24047 * internal-fn.def (ASAN_POISON): New builtin.
24048 * sanopt.c (pass_sanopt::execute): Expand
24049 asan_expand_poison_ifn.
24050 * tree-inline.c (copy_decl_for_dup_finish): Make function
24052 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24053 * tree-ssa.c (is_asan_mark_p): New function.
24054 (execute_update_addresses_taken): Rewrite local variables
24055 (identified just by use-after-scope as addressable) into SSA.
24057 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24059 * doc/install.texi (Specific): opensource.apple.com uses https
24060 now. Remove trailing slash.
24062 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24064 * README.Portability: Remove note on an Irix compatibility issue.
24066 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24068 * gcov.c (INCLUDE_ALGORITHM): Define.
24069 (INCLUDE_VECTOR): Define.
24070 No longer include <vector> and <algorithm> directly.
24072 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24074 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24076 * doc/invoke.texi (Code Gen Options): Ditto.
24078 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24081 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24083 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24085 rtl-optimization/79125
24086 * cprop.c (local_cprop_pass): Handle cases where we make an
24087 unconditional trap.
24089 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24093 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24094 read from, for big endian.
24096 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24098 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24099 register pauth builtins for LP64 only.
24101 2017-01-20 Marek Polacek <polacek@redhat.com>
24104 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24107 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24109 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24111 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24112 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24113 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24115 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24118 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24119 in big-endian mode when they are not a single duplicated value.
24121 2017-01-20 Richard Biener <rguenther@suse.de>
24123 * BASE-VER: Bump to 7.0.1.
24125 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24127 * omp-low.c (omplow_simd_context): New struct. Use it...
24128 (lower_rec_simd_input_clauses): ...here and...
24129 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24130 references to idx, lane, max_vf, is_simt.
24132 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24134 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24137 2017-01-20 Martin Jambor <mjambor@suse.cz>
24139 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24140 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24142 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24143 (GTFILES): Rename hsa.c to hsa-common.c.
24144 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24145 * hsa-dump.c: Likewise.
24146 * hsa-gen.c: Likewise.
24147 * hsa-regalloc.c: Likewise.
24148 * ipa-hsa.c: Likewise.
24149 * omp-expand.c: Likewise.
24150 * omp-low.c: Likewise.
24151 * toplev.c: Likewise.
24153 2017-01-20 Marek Polacek <polacek@redhat.com>
24156 * doc/invoke.texi: Document -Wduplicated-branches.
24157 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24158 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24159 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24160 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24161 return 0 only when not OEP_LEXICOGRAPHIC.
24162 (fold_build_cleanup_point_expr): Use the expression
24163 location when building CLEANUP_POINT_EXPR.
24164 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24165 * tree.c (add_expr): Handle error_mark_node.
24167 2017-01-20 Martin Liska <mliska@suse.cz>
24170 * tree-profile.c (init_ic_make_global_vars): Do not call
24172 (gimple_init_gcov_profiler): Likewise.
24174 2017-01-20 Martin Liska <mliska@suse.cz>
24177 * cgraph.h (maybe_create_reference): Remove argument and
24179 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24181 * ipa-cp.c (create_specialized_node): Likewise.
24182 * symtab.c (symtab_node::maybe_create_reference): Handle
24183 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24185 2017-01-20 Martin Liska <mliska@suse.cz>
24187 * read-rtl-function.c (function_reader::create_function): Use
24188 build_decl instread of build_decl_stat.
24190 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24192 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24193 * config/i386/avx512dqintrin.h: Ditto.
24194 * config/i386/avx512fintrin.h: Ditto.
24195 * config/i386/i386-builtin-types.def: Add new types.
24196 * config/i386/i386.c: Handle new types.
24197 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24198 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24199 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24200 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24201 (__builtin_ia32_kshiftridi): New.
24202 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24204 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24208 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24209 define to rs6000_init_stack_protect_guard.
24210 (rs6000_init_stack_protect_guard): New function.
24212 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24213 Yunqiang Su <yunqiang.su@imgtec.com>
24215 * config.gcc (supported_defaults): Add madd4.
24216 (with_madd4): Add validation.
24217 (all_defaults): Add madd4.
24218 * config/mips/mips.opt (mmadd4): New option.
24219 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24221 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24223 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24224 (ISA_HAS_FUSED_MADD4): Likewise.
24225 * doc/invoke.texi (-mmadd4): Document the new option.
24226 * doc/install.texi (--with-madd4): Document the new option.
24228 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24230 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24231 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24232 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24233 (aarch64_init_pauth_hint_builtins): New.
24234 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24235 (aarch64_expand_builtin): Expand new builtins.
24237 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24239 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24240 * combine-stack-adj.c (no_unhandled_cfa): Handle
24241 REG_CFA_TOGGLE_RA_MANGLE.
24242 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24243 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24244 info for return address signing.
24245 (aarch64_expand_epilogue): Likewise.
24247 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24249 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24250 * config/aarch64/aarch64-protos.h
24251 (aarch64_return_address_signing_enabled): New declaration.
24252 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24254 (aarch64_expand_prologue): Sign return address before it's pushed onto
24256 (aarch64_expand_epilogue): Authenticate return address fetched from
24258 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24259 (aarch64_attributes): New function attributes for "sign-return-address".
24260 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24261 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24262 ("*do_return"): Generate combined instructions according to key index.
24263 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24264 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24266 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24267 * config/aarch64/aarch64.opt (msign-return-address=): New.
24268 * doc/extend.texi (AArch64 Function Attributes): Documents
24269 "sign-return-address=".
24270 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24272 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24274 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24275 overall option summary.
24277 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24279 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24280 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24281 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24282 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24284 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24286 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24287 -mpower9-minmax by default for -mcpu=power9.
24288 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24289 128-bit floating point.
24291 2017-01-20 Alan Modra <amodra@gmail.com>
24293 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24294 optimizing for size.
24296 2017-01-20 Alan Modra <amodra@gmail.com>
24299 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24300 for strcmp and strncmp from corresponding builtin decl.
24302 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24304 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24305 instead of i386/rtems-64.h.
24306 * config/i386/rtems-64.h: Remove.
24308 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24312 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24314 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24316 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24318 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24319 Change int to HOST_WIDE_INT.
24320 * config/aarch64/aarch64-protos.h
24321 (aarch64_simd_gen_const_vector_dup): Likewise.
24322 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24324 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24326 * langhooks-def.h (lhd_type_for_size): New decl.
24327 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24328 * langhooks.c (lhd_type_for_size): New function, taken from
24331 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24333 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24334 define_bypass for CR latency.
24335 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24336 (power9-alu2): Add define_bypass for CR latency.
24338 (power9-mul): Update insn latency.
24339 (power9-mul-compare): Update insn latency, bypass latency and remove
24342 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24344 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24346 * config/aarch64/aarch64.md
24347 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24348 aarch64_nopcrelative_literal_loads.
24349 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24351 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24353 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24354 TARGET_LOONGSON_3A.
24355 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24357 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24360 * config.gcc (supported_defaults): Add lxc1-sxc1.
24361 (with_lxc1_sxc1): Add validation.
24362 (all_defaults): Add lxc1-sxc1.
24363 * config/mips/mips.opt (mlxc1-sxc1): New option.
24364 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24366 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24367 __mips_no_lxc1_sxc1.
24368 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24369 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24370 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24372 2017-01-19 Richard Biener <rguenther@suse.de>
24374 PR tree-optimization/72488
24375 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24376 sure to restore SSA info.
24377 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24379 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24381 PR rtl-optimization/79121
24382 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24383 of the inner type when shifting an extended value.
24385 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24388 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24389 interposable aliases.
24391 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24394 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24395 Use the evmergelohi instruction.
24396 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24397 (mov_sitf_e500_subreg8_2_be): Likewise.
24398 (mov_sitf_e500_subreg12_2_le): Likewise.
24399 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24400 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24401 (mov_sitf_e500_subreg8_2_le): Likewise.
24402 (mov_sitf_e500_subreg12_2_be): Likewise.
24404 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24406 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24407 attribute from vecsimple to vecperm.
24408 (altivec_vbpermq2): Likewise.
24410 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24413 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24415 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24416 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24417 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24418 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24419 case where N arg is SIZE_MAX.
24420 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24421 (cmpstrsi): Add pattern.
24423 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24425 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24426 __builtin_vec_revb builtins.
24427 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24428 built-in functions to support generation of the ISA 3.0 XXBR<x>
24429 vector byte reverse instructions.
24430 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24431 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24432 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24433 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24434 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24435 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24436 (P9V_BUILTIN_VEC_REVB): Likewise.
24437 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24438 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24439 (p9_xxbrq_v16qi): Likewise.
24440 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24441 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24442 (p9_xxbrh_v8hi): Likewise.
24443 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24444 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24445 vec_revb built-in functions.
24447 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24449 PR rtl-optimization/78952
24450 * config/i386/i386.md (any_extract): New code iterator.
24451 (*insvqi_2): Use any_extract for source operand.
24452 (*insvqi_3): Use any_shiftrt for source operand.
24454 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24456 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24458 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24460 2017-01-18 Matthias Klose <doko@ubuntu.com>
24462 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24464 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24466 * config/rs6000/altivec.h (vec_bperm): Change #define.
24467 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24468 (altivec_vbpermq2): New define_insn.
24469 (altivec_vbpermd): Likewise.
24470 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24471 function interface.
24472 (VBPERMD): Likewise.
24473 (VBPERM): New polymorphic function interface.
24474 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24475 Add entries for P9V_BUILTIN_VEC_VBPERM.
24476 * doc/extend.texi: Add interfaces for vec_bperm.
24478 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24480 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24481 first letter of error messages.
24482 (s390_resolve_overloaded_builtin): Likewise.
24483 * config/s390/s390.c (s390_expand_builtin): Likewise.
24484 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24485 (s390_valid_target_attribute_inner_p): Likewise.
24486 * config/s390/s390.md ("tabort"): Likewise.
24488 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24490 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24491 (ISA_AVOID_DIV_HILO): New macro.
24492 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24493 (ISA_HAS_DDIV): Likewise.
24495 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24497 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24499 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24501 * doc/invoke.texi (fabi-version): Spelling fix.
24503 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24506 * doc/invoke.texi (fabi-version): Mention mangling fix for
24509 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24512 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24514 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24517 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24518 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24520 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24522 (stack_protect_test): Ditto.
24523 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24524 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24526 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24527 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24528 -mstack-protector-guard-offset=.
24529 (RS/6000 and PowerPC Options): Ditto.
24531 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24533 * config/i386/i386.h (MASK_CLASS_P): New define.
24534 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24535 there are no registers from different register sets also when
24536 mask registers are used. Update function comment.
24537 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24538 to (*k/*r) and (*k/*km) alternatives.
24540 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24542 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24543 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24544 (EH_RETURN_HANDLER_RTX): New define.
24545 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24546 Force frame pointer in EH return functions.
24547 (aarch64_expand_epilogue): Add barrier for eh_return.
24548 (aarch64_final_eh_return_addr): Remove.
24549 (aarch64_eh_return_handler_rtx): New function.
24550 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24552 (aarch64_eh_return_handler_rtx): New prototype.
24554 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24556 * config/rs6000/altivec.h (vec_rlmi): New #define.
24557 (vec_vrlnm): Likewise.
24558 (vec_rlnm): Likewise.
24559 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24560 (UNSPEC_VRLNM): Likewise.
24561 (VIlong): New mode iterator.
24562 (altivec_vrl<VI_char>mi): New define_insn.
24563 (altivec_vrl<VI_char>nm): Likewise.
24564 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24566 (VRLDNM): Likewise.
24567 (RLNM): New polymorphic function entry.
24568 (VRLWMI): New monomorphic function entry.
24569 (VRLDMI): Likewise.
24570 (RLMI): New polymorphic function entry.
24571 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24572 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24573 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24576 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24579 * dwarf2out.c (field_byte_offset): Restore the
24580 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24581 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24582 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24585 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24588 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24590 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24593 * dwarf2out.c (add_data_member_location_attribute): For constant
24594 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24595 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24596 DW_AT_byte_size attributes.
24598 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24600 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24601 after forcing to constant memory when the code model is medium.
24603 2017-01-17 Julia Koval <julia.koval@intel.com>
24606 * config/i386/avx512fintrin.h
24607 (_mm512_i32gather_ps): Change __addr type to void const*.
24608 (_mm512_mask_i32gather_ps): Ditto.
24609 (_mm512_i32gather_pd): Ditto.
24610 (_mm512_mask_i32gather_pd): Ditto.
24611 (_mm512_i64gather_ps): Ditto.
24612 (_mm512_mask_i64gather_ps): Ditto.
24613 (_mm512_i64gather_pd): Ditto.
24614 (_mm512_mask_i64gather_pd): Ditto.
24615 (_mm512_i32gather_epi32): Ditto.
24616 (_mm512_mask_i32gather_epi32): Ditto.
24617 (_mm512_i32gather_epi64): Ditto.
24618 (_mm512_mask_i32gather_epi64): Ditto.
24619 (_mm512_i64gather_epi32): Ditto.
24620 (_mm512_mask_i64gather_epi32): Ditto.
24621 (_mm512_i64gather_epi64): Ditto.
24622 (_mm512_mask_i64gather_epi64): Ditto.
24623 (_mm512_i32scatter_ps): Change __addr type to void*.
24624 (_mm512_mask_i32scatter_ps): Ditto.
24625 (_mm512_i32scatter_pd): Ditto.
24626 (_mm512_mask_i32scatter_pd): Ditto.
24627 (_mm512_i64scatter_ps): Ditto.
24628 (_mm512_mask_i64scatter_ps): Ditto.
24629 (_mm512_i64scatter_pd): Ditto.
24630 (_mm512_mask_i64scatter_pd): Ditto.
24631 (_mm512_i32scatter_epi32): Ditto.
24632 (_mm512_mask_i32scatter_epi32): Ditto.
24633 (_mm512_i32scatter_epi64): Ditto.
24634 (_mm512_mask_i32scatter_epi64): Ditto.
24635 (_mm512_i64scatter_epi32): Ditto.
24636 (_mm512_mask_i64scatter_epi32): Ditto.
24637 (_mm512_i64scatter_epi64): Ditto.
24638 (_mm512_mask_i64scatter_epi64): Ditto.
24639 * config/i386/avx512pfintrin.h
24640 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24641 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24642 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24643 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24644 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24645 (_mm512_prefetch_i32scatter_ps): Ditto.
24646 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24647 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24648 (_mm512_prefetch_i64scatter_pd): Ditto.
24649 (_mm512_prefetch_i64scatter_ps): Ditto.
24650 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24651 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24652 * config/i386/avx512vlintrin.h
24653 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24654 (_mm_mmask_i32gather_ps): Ditto.
24655 (_mm256_mmask_i32gather_pd): Ditto.
24656 (_mm_mmask_i32gather_pd): Ditto.
24657 (_mm256_mmask_i64gather_ps): Ditto.
24658 (_mm_mmask_i64gather_ps): Ditto.
24659 (_mm256_mmask_i64gather_pd): Ditto.
24660 (_mm_mmask_i64gather_pd): Ditto.
24661 (_mm256_mmask_i32gather_epi32): Ditto.
24662 (_mm_mmask_i32gather_epi32): Ditto.
24663 (_mm256_mmask_i32gather_epi64): Ditto.
24664 (_mm_mmask_i32gather_epi64): Ditto.
24665 (_mm256_mmask_i64gather_epi32): Ditto.
24666 (_mm_mmask_i64gather_epi32): Ditto.
24667 (_mm256_mmask_i64gather_epi64): Ditto.
24668 (_mm_mmask_i64gather_epi64): Ditto.
24669 (_mm256_i32scatter_ps): Change __addr type to void*.
24670 (_mm256_mask_i32scatter_ps): Ditto.
24671 (_mm_i32scatter_ps): Ditto.
24672 (_mm_mask_i32scatter_ps): Ditto.
24673 (_mm256_i32scatter_pd): Ditto.
24674 (_mm256_mask_i32scatter_pd): Ditto.
24675 (_mm_i32scatter_pd): Ditto.
24676 (_mm_mask_i32scatter_pd): Ditto.
24677 (_mm256_i64scatter_ps): Ditto.
24678 (_mm256_mask_i64scatter_ps): Ditto.
24679 (_mm_i64scatter_ps): Ditto.
24680 (_mm_mask_i64scatter_ps): Ditto.
24681 (_mm256_i64scatter_pd): Ditto.
24682 (_mm256_mask_i64scatter_pd): Ditto.
24683 (_mm_i64scatter_pd): Ditto.
24684 (_mm_mask_i64scatter_pd): Ditto.
24685 (_mm256_i32scatter_epi32): Ditto.
24686 (_mm256_mask_i32scatter_epi32): Ditto.
24687 (_mm_i32scatter_epi32): Ditto.
24688 (_mm_mask_i32scatter_epi32): Ditto.
24689 (_mm256_i32scatter_epi64): Ditto.
24690 (_mm256_mask_i32scatter_epi64): Ditto.
24691 (_mm_i32scatter_epi64): Ditto.
24692 (_mm_mask_i32scatter_epi64): Ditto.
24693 (_mm256_i64scatter_epi32): Ditto.
24694 (_mm256_mask_i64scatter_epi32): Ditto.
24695 (_mm_i64scatter_epi32): Ditto.
24696 (_mm_mask_i64scatter_epi32): Ditto.
24697 (_mm256_i64scatter_epi64): Ditto.
24698 (_mm256_mask_i64scatter_epi64): Ditto.
24699 (_mm_i64scatter_epi64): Ditto.
24700 (_mm_mask_i64scatter_epi64): Ditto.
24701 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24702 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24703 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24704 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24705 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24706 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24707 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24708 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24709 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24710 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24711 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24712 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24713 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24714 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24715 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24716 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24717 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24718 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24719 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24720 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24721 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24722 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24723 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24724 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24725 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24726 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24727 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24728 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24729 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24730 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24731 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24732 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24733 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24734 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24735 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24736 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24737 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24738 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24739 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24740 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24741 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24742 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24743 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24744 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24745 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24746 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24747 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24748 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24749 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24750 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24751 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24752 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24753 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24754 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24755 definitions accordingly.
24757 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24758 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24761 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24764 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24767 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24768 hard regs for inner regno.
24770 2017-01-17 Martin Liska <mliska@suse.cz>
24773 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24774 assumption and add comment.
24776 2017-01-17 Nathan Sidwell <nathan@acm.org>
24778 * ipa-visibility.c (localize_node): New function, broken out of ...
24779 (function_and_variable_visibility): ... here. Call it.
24781 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24783 PR middle-end/77445
24784 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24785 correctly set frequency of oudgoing edge.
24786 (duplicate_thread_path): Fix profile updating.
24788 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24791 * configure.ac: Add GCC_BASE_VER.
24792 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24793 version from BASE-VER file.
24794 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24795 (gcc.o): Depend on $(BASEVER).
24796 * common.opt (dumpfullversion): New option.
24797 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24798 * doc/invoke.texi: Document -dumpfullversion.
24799 * doc/install.texi: Document --with-gcc-major-version-only.
24800 * configure: Regenerated.
24802 2017-01-17 Richard Biener <rguenther@suse.de>
24804 PR tree-optimization/71433
24805 * tree-vrp.c (register_new_assert_for): Merge same asserts
24806 on all incoming edges.
24807 (process_assert_insertions_for): Handle insertions at the
24810 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24812 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24813 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24815 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24818 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24821 2017-01-17 Alan Modra <amodra@gmail.com>
24824 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24825 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24826 symbolic stack limit when pic.
24828 2017-01-16 Martin Sebor <msebor@redhat.com>
24830 PR tree-optimization/78608
24831 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24833 2017-01-16 Jeff Law <law@redhat.com>
24836 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24837 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24838 for several include directories that may be relative to sysroot.
24839 * config/i386/x-mingw32 (gplus_includedir): Define.
24840 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24841 (native_system_includedir): Likewise.
24842 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24843 override if TARGET_SYSTEM_ROOT is defined.
24844 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24846 PR tree-optimization/79090
24847 PR tree-optimization/33562
24848 PR tree-optimization/61912
24849 PR tree-optimization/77485
24850 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24851 and computed trims into the dump file.
24853 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24855 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24857 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24860 * gimplify.c (gimplify_init_constructor): If want_value and
24861 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24865 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24866 sequence. Formatting fixes.
24867 (doloop_optimize): Formatting fixes.
24870 * gcc.c (debug_level_greater_than_spec_func): New function.
24871 (static_spec_functions): Add debug-level-gt spec function.
24872 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24874 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24875 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24876 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24877 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24878 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24879 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24881 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24883 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24884 QImode fixups to general and mask registers only.
24886 2017-01-16 Carl Love <cel@us.ibm.com>
24888 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24889 for built-in functions
24890 vector signed char vec_nabs (vector signed char)
24891 vector signed short vec_nabs (vector signed short)
24892 vector signed int vec_nabs (vector signed int)
24893 vector signed long long vec_nabs (vector signed long long)
24894 vector float vec_nabs (vector float)
24895 vector double vec_nabs (vector double)
24896 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24898 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24899 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24900 * doc/extend.texi: Update the documentation file for the new built-in
24903 2017-01-16 Martin Sebor <msebor@redhat.com>
24905 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24908 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24910 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24911 UNSPEC_VSX__XXSPLTD to require special splat handling.
24913 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24916 * system.h: Poison strndup.
24918 2017-01-16 Alan Modra <amodra@gmail.com>
24921 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24924 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24926 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24928 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24930 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24931 call recog here. Assert that INSN_CODE (insn) is non-negative.
24933 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24936 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24938 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24939 in the currently scheduled RTL fragment.
24941 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24943 PR rtl-optimization/78751
24944 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24947 2017-01-14 Jeff Law <law@redhat.com>
24949 PR tree-optimization/79090
24950 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24951 variable length stores.
24952 (compute_trims): Delete dead assignment to *trim_tail.
24953 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24956 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24958 PR rtl-optimization/78626
24959 PR rtl-optimization/78727
24960 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24961 of a block, and split such blocks after everything else is finished.
24963 2017-01-14 Alan Modra <amodra@gmail.com>
24966 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24967 target legitimate_combined_insn.
24968 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24969 (rs6000_legitimate_combined_insn): New function.
24970 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24972 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24973 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24974 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24976 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24978 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24980 2017-01-13 Jeff Law <law@redhat.com>
24982 PR tree-optimization/33562
24983 PR tree-optimization/61912
24984 PR tree-optimization/77485
24985 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24987 (delete_dead_assignment): Likewise.
24988 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24989 statement to delete_dead_call and delete_dead_assignment.
24991 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24994 * substring-locations.c (format_warning_va): Strengthen case 1 so
24995 that both endpoints of the substring must be within the format
24996 range for just the substring to be printed.
24998 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25000 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25001 * config/i386/i386.c (ix86_target_string): Add missing options
25002 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25003 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25004 flags_other and ix86_target_other to flags2_other. Display unknown
25006 (ix86_valid_target_attribute_inner_p): Add missing options and
25007 reorder options by implied ISAs, as in ix86_target_string.
25009 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25011 * hash-table.h (hash_table::too_empty_p): New function.
25012 (hash_table::expand): Use it.
25013 (hash_table::traverse): Likewise.
25014 (hash_table::empty_slot): Use sizeof (value_type) instead of
25015 sizeof (PTR) to convert bytes to elements. Shrink the table
25016 if the current size is excessive for the current number of
25019 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25021 * ira-costs.c (record_reg_classes): Break from the inner loop
25022 early once alt_fail is known to be true. Update outer loop
25023 handling accordingly.
25025 2017-01-13 Jeff Law <law@redhat.com>
25027 * tree-ssa-dse.c (decrement_count): New function.
25028 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25029 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25030 when we know the partially dead statement is a mem* function.
25032 PR tree-optimization/61912
25033 PR tree-optimization/77485
25034 * tree-ssa-dse.c: Include expr.h.
25035 (maybe_trim_constructor_store): New function.
25036 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25038 PR tree-optimization/33562
25039 PR tree-optimization/61912
25040 PR tree-optimization/77485
25041 * doc/invoke.texi: Document new dse-max-object-size param.
25042 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25043 * tree-ssa-dse.c: Include params.h.
25044 (dse_store_status): New enum.
25045 (initialize_ao_ref_for_dse): New, partially extracted from
25047 (valid_ao_ref_for_dse, normalize_ref): New.
25048 (setup_live_bytes_from_ref, compute_trims): Likewise.
25049 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25050 (maybe_trim_partially_dead_store): Likewise.
25051 (maybe_trim_complex_store): Likewise.
25052 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25053 Track what bytes live from the original store. Return tri-state
25054 for dead, partially dead or live.
25055 (dse_dom_walker): Add constructor, destructor and new private members.
25056 (delete_dead_call, delete_dead_assignment): New extracted from
25058 (dse_optimize_stmt): Make a member of dse_dom_walker.
25059 Use initialize_ao_ref_for_dse.
25061 PR tree-optimization/33562
25062 PR tree-optimization/61912
25063 PR tree-optimization/77485
25064 * sbitmap.h (bitmap_count_bits): Prototype.
25065 (bitmap_clear_range, bitmap_set_range): Likewise.
25066 * sbitmap.c (bitmap_clear_range): New function.
25067 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25069 2017-01-13 Martin Liska <mliska@suse.cz>
25072 * function.c (set_cfun): Add new argument force.
25073 * function.h (set_cfun): Likewise.
25074 * ipa-inline-transform.c (inline_call): Use the function when
25075 strict alising from is dropped for function we inline to.
25077 2017-01-13 Richard Biener <rguenther@suse.de>
25079 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25080 for dumping GIMPLE INTEGER_CSTs.
25082 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25084 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25085 to 201112L since C++17.
25087 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25090 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25091 if -fsanitize=kernel-address is present.
25093 2017-01-13 Richard Biener <rguenther@suse.de>
25095 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25096 as _Literal ( type ) number in case usual suffixes do not
25097 preserve all information.
25099 2017-01-13 Richard Biener <rguenther@suse.de>
25101 PR tree-optimization/77283
25102 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25103 and ssa-iterators.h.
25104 (is_feasible_trace): Implement a cost model based on joiner
25107 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25110 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25111 char or short to __float128/_Float128 directly.
25113 2017-01-12 Martin Sebor <msebor@redhat.com>
25115 to -Wformat-overflow.
25116 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25117 (min_bytes_remaining): Same.
25118 (get_string_length): Same.
25119 (format_string): Same.
25120 (format_directive): Same.
25122 (pass_sprintf_length::handle_gimple_call): Same.
25124 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25126 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25127 info.nowrite calls with no lhs that can't throw. Return bool
25128 whether gsi_remove has been called or not.
25129 (pass_sprintf_length::handle_gimple_call): Return bool whether
25130 try_substitute_return_value called gsi_remove. Formatting fix.
25131 (pass_sprintf_length::execute): Don't use gsi_remove if
25132 handle_gimple_call returned true.
25135 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25136 be removed due to side-effects, don't remove following barrier nor
25137 turn the successor edge into fallthru edge.
25139 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25142 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25143 element-reversing loads and stores as not swappable.
25145 2017-01-12 Nathan Sidwell <nathan@acm.org>
25146 Nicolai Stange <nicstange@gmail.com>
25148 * combine.c (try_combine): Don't ignore result of overlap checking
25149 loop. Combine overlap & asm check into single loop.
25151 2017-01-12 Richard Biener <rguenther@suse.de>
25153 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25154 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25156 2017-01-12 Richard Biener <rguenther@suse.de>
25158 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25159 and TS_TARGET_OPTION directly derive from TS_BASE.
25160 * tree-core.h (tree_optimization_option): Derive from tree_base.
25161 (tree_target_option): Likewise.
25163 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25165 * config/i386/i386.c (memory_address_length): Increase len
25166 only when rip_relative_addr_p returns false.
25168 2017-01-11 Julia Koval <julia.koval@intel.com>
25170 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25171 (OPTION_MASK_ISA_SGX_SET): New.
25172 (ix86_handle_option): Handle OPT_msgx.
25173 * config.gcc: Added sgxintrin.h.
25174 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25175 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25176 * config/i386/i386.c (ix86_target_string): Add -msgx.
25178 (ix86_option_override_internal): Handle new options.
25179 (ix86_valid_target_attribute_inner_p): Add sgx.
25180 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25181 * config/i386/i386.opt: Add msgx.
25182 * config/i386/sgxintrin.h: New file.
25183 * config/i386/x86intrin.h: Add sgxintrin.h.
25185 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25188 * fold-const.c (maybe_nonzero_address): Return 1 for function
25190 (tree_single_nonzero_warnv_p): Don't handle function local objects
25194 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25197 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25200 * opts.c: Include "spellcheck.h"
25201 (struct string_fragment): New struct.
25202 (struct edit_distance_traits<const string_fragment &>): New
25204 (get_closest_sanitizer_option): New function.
25205 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25207 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25209 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25211 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25212 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25213 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25214 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25215 for initial die_offset if dwarf_split_debug_info.
25216 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25217 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25219 (output_skeleton_debug_sections): Formatting fix. Use
25220 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25221 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25223 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25225 * config/arm/cortex-a53.md: Add bypasses for
25226 cortex_a53_r2f_cvt.
25227 (cortex_a53_r2f): Only use for transfers.
25228 (cortex_a53_f2r): Likewise.
25229 (cortex_a53_r2f_cvt): Add reservation for conversions.
25230 (cortex_a53_f2r_cvt): Likewise.
25232 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25234 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25235 to all inlined functions, change static to extern.
25237 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25240 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25242 (arm_assemble_integer): Likewise.
25244 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25246 * config.gcc: Use new awk script to check CPU, FPU and architecture
25247 parameters for --with-... options.
25248 * config/arm/parsecpu.awk: New file
25249 * config/arm/arm-cpus.in: New file.
25250 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25252 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25254 * config/arm/t-arm: Update dependency rules.
25255 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25256 of processing .def files.
25257 * config/arm/genopt.sh: Deleted.
25258 * config/arm/gentune.sh: Deleted.
25259 * config/arm/arm-cores.def: Deleted.
25260 * config/arm/arm-arches.def: Deleted.
25261 * config/arm/arm-fpus.def: Deleted.
25262 * config/arm/arm-tune.md: Regenerated.
25263 * config/arm/arm-tables.opt: Regenerated.
25264 * config/arm/arm-cpu.h: New generated file.
25265 * config/arm/arm-cpu-data.h: New generated file.
25266 * config/arm/arm-cpu-cdata.h: New generated file.
25268 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25271 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25273 (input_varpool_node): Unpack dynamically_initialized bit.
25275 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25277 PR rtl-optimization/79032
25278 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25279 the alignment of the adjusted memory reference against that of MODE,
25280 instead of the alignment of the original memory reference.
25282 2017-01-11 Martin Jambor <mjambor@suse.cz>
25284 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25286 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25287 decorated functions.
25289 2017-01-11 Richard Biener <rguenther@suse.de>
25291 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25292 set range/nonnull info for PHI results. Do not set it on
25293 stmts marked for removal.
25295 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25297 * expr.c (store_field): In the bitfield case, fetch the return value
25298 from the registers before applying a single big-endian adjustment.
25299 Always do a final load for a BLKmode value not larger than a word.
25301 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25304 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25305 that we correctly handle column numbers greater than
25306 LINE_MAP_MAX_COLUMN_NUMBER.
25308 2017-01-10 Martin Sebor <msebor@redhat.com>
25310 PR middle-end/78245
25311 * gimple-ssa-sprintf.c (get_destination_size): Call
25312 {init,fini}object_sizes.
25313 * tree-object-size.c (addr_object_size): Adjust.
25314 (pass_through_call): Adjust.
25315 (pass_object_sizes::execute): Adjust.
25316 * tree-object-size.h (fini_object_sizes): Declare.
25318 2017-01-10 Martin Sebor <msebor@redhat.com>
25320 PR tree-optimization/78775
25321 * builtins.c (get_size_range): Move...
25322 * calls.c: ...to here.
25323 (alloc_max_size): Accept zero argument.
25324 (operand_signed_p): Remove.
25325 (maybe_warn_alloc_args_overflow): Call get_size_range.
25326 * calls.h (get_size_range): Declare.
25328 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25330 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25331 from TI's devices.csv file as of September 2016.
25332 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25334 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25336 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25337 * doc/invoke.texi: Likewise.
25338 * doc/md.texi: Likewise.
25339 * doc/objc.texi: Likewise.
25341 2017-01-10 Joshua Conner <joshconner@google.com>
25343 * config/arm/fuchsia-elf.h: New file.
25344 * config/fuchsia.h: New file.
25345 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25346 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25348 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25350 2016-01-10 Richard Biener <rguenther@suse.de>
25352 PR tree-optimization/79034
25353 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25354 Propagate out degenerate PHIs in the joiner.
25356 2017-01-10 Martin Liska <mliska@suse.cz>
25358 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25359 (sort_congruence_classes_by_decl_uid): Likewise.
25360 (sort_congruence_class_groups_by_decl_uid): Likewise.
25361 (sem_item_optimizer::merge_classes): Sort class, groups in these
25362 classes and members in the groups by DECL_UID of declarations.
25363 This would make merge operations stable.
25365 2017-01-10 Martin Liska <mliska@suse.cz>
25367 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25368 usage of m_classes_vec.
25369 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25370 (sem_item_optimizer::get_group_by_hash): Likewise.
25371 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25372 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25373 (sem_item_optimizer::verify_classes): Likewise.
25374 (sem_item_optimizer::process_cong_reduction): Likewise.
25375 (sem_item_optimizer::dump_cong_classes): Likewise.
25376 (sem_item_optimizer::merge_classes): Likewise.
25377 * ipa-icf.h (congruence_class_hash): Rename from
25378 congruence_class_group_hash. Remove declaration of m_classes_vec.
25380 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25382 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25383 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25384 * config.gcc: Add avx512vpopcntdqintrin.h.
25385 * config/i386/avx512vpopcntdqintrin.h: New.
25386 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25387 * config/i386/i386-builtin-types.def: Add new types.
25388 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25389 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25390 __builtin_ia32_vpopcountq_v8di_mask): New.
25391 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25392 __AVX512VPOPCNTDQ__.
25393 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25394 (PTA_AVX512VPOPCNTDQ): Define.
25395 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25396 TARGET_AVX512VPOPCNTDQ_P): Define.
25397 * config/i386/i386.opt: Add mavx512vpopcntdq.
25398 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25399 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25401 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25403 PR middle-end/77484
25404 * predict.def (PRED_CALL): Set to 67.
25406 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25408 * expr.c (store_field): In the bitfield case, if the value comes from
25409 a function call and is of an aggregate type returned in registers, do
25410 not modify the field mode; extract the value in all cases if the mode
25411 is BLKmode and the size is not larger than a word.
25413 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25416 * config/i386/cpuid.h: Fix undefined behavior.
25418 2017-01-04 Jeff Law <law@redhat.com>
25420 PR tree-optimization/79007
25421 PR tree-optimization/67955
25422 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25423 conservative for pt.null when flag_non_call_exceptions is on.
25425 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25427 PR translation/79019
25428 PR translation/79020
25429 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25430 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25431 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25433 * config/avr/avr.opt (maccumulate-args): Likewise.
25434 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25435 * common.opt (freport-bug): Likewise.
25436 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25437 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25438 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25439 translatable string.
25440 * config/i386/i386.c (function_value_32): Likewise.
25441 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25442 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25444 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25445 * common/config/msp430/msp430-common.c (msp430_handle_option):
25447 * symtab.c (symtab_node::verify_base): Likewise.
25448 * opts.c (set_debug_level): Likewise.
25449 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25450 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25451 missing whitespace to translatable strings.
25452 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25453 * config/sh/superh.h: Likewise.
25454 * config/i386/xopintrin.h: Likewise.
25455 * config/i386/znver1.md: Likewise.
25456 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25457 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25458 * double-int.h (struct double_int): Likewise.
25459 * double-int.c (div_and_round_double): Likewise.
25460 * wide-int.cc: Likewise.
25461 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25462 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25463 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25464 (crossjumps_occurred): ... this.
25465 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25468 PR tree-optimization/78899
25469 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25470 returning bool return struct loop *, NULL for failure and the new
25472 (versionable_outer_loop_p): Don't version outer loop if it has
25473 dont_vectorized bit set.
25474 (tree_if_conversion): When versioning outer loop, ensure
25475 tree_if_conversion is performed also on the inner loop of the
25476 non-vectorizable outer loop copy.
25477 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25478 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25479 prevent vectorization of it.
25480 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25481 the outer loop vectorization of the non-scalar version is attempted
25482 before vectorization of the inner loop in scalar version. If
25483 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25484 vectorization of its inner loop.
25485 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25486 has 2 inner loops, rename also on edges from bb whose single pred
25487 is outer_loop->header. Fix typo in function comment.
25489 2017-01-09 Martin Sebor <msebor@redhat.com>
25492 * asan.c (asan_emit_stack_protection): Increase local buffer size
25493 to avoid snprintf truncation warning.
25495 2017-01-09 Andrew Pinski <apinski@cavium.com>
25497 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25498 to reference thunderx2t99 for the tuning structure
25499 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25501 (thunderx2t99_extra_costs): This.
25502 * config/aarch64/aarch64-tune.md: Regenerate.
25503 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25504 (vulcan_addrcost_table): This.
25505 (vulcan_regmove_cost): Rename to ...
25506 (thunderx2t99_regmove_cost): This.
25507 (vulcan_vector_cost): Rename to ...
25508 (thunderx2t99_vector_cost): this.
25509 (vulcan_branch_cost): Rename to ...
25510 (thunderx2t99_branch_cost): This.
25511 (vulcan_tunings): Rename to ...
25512 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25513 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25515 2017-01-09 Martin Jambor <mjambor@suse.cz>
25519 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25520 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25521 (propagate_vr_accross_jump_function): Use the above function for all
25522 value range computations for pass-through jump functions and type
25523 converasion from explicit value range values.
25524 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25525 parameters from TYPE_ARG_TYPES.
25526 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25527 (ipa_write_node_info): Stream type of the actual argument.
25528 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25530 2017-01-09 Martin Liska <mliska@suse.cz>
25533 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25534 (lookup_compiler): Do not show error message with have_E.
25536 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25538 PR tree-optimization/78938
25539 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25540 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25541 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25542 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25545 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25547 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25550 2017-01-09 Richard Biener <rguenther@suse.de>
25552 PR tree-optimization/78997
25553 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25554 name condition properly.
25556 2017-01-09 Richard Biener <rguenther@suse.de>
25559 * dwarf2out.c (is_cxx): New overload with context.
25560 (is_naming_typedef_decl): Use it.
25562 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25564 * invoke.texi (Option Summary): Correct spacing in option lists
25565 and add line breaks to fix over-long lines.
25567 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25569 PR middle-end/17660
25571 * extend.texi (Common Variable Attributes): Add xref to GCC
25572 Internals manual to explain mode attribute keywords.
25574 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25577 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25578 and Preprocessor Options.
25579 (Options for Linking): Document -pthread here....
25580 (RS/6000 and PowerPC Options): ...not here.
25581 (Solaris 2 Options): ...or here.
25582 * doc/cppopts.texi: Document -pthread.
25584 2017-01-08 Martin Sebor <msebor@redhat.com>
25586 PR middle-end/77708
25587 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25588 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25589 New member functions.
25590 (format_directive): Used them.
25592 (pass_sprintf_length::handle_gimple_call): Same.
25593 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25594 to avoid truncation for any argument.
25595 (extract_affine_mul): Same.
25596 * tree.c (get_file_function_name): Same.
25598 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25600 PR middle-end/77484
25601 * predict.def (PRED_INDIR_CALL): Set to 86.
25603 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25605 PR preprocessor/54124
25606 * doc/cppopts.texi: Reformat -d subtable to list the full name
25607 of the options. Add cross-reference to the docs for the general
25608 compiler -d options.
25609 * doc/invoke.texi (Developer Options): Add cross-reference to the
25610 preprocessor-specific -d option documentation.
25612 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25614 PR preprocessor/13498
25615 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25616 redudant material, and reflect new command-line options.
25617 (System Headers): Likewise.
25619 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25621 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25622 -isystem, and -idirafter. Copy-edit.
25623 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25624 default for -ftrack-macro-expansion. Delete obsolete and
25625 badly-formatted implementation details about -fdebug-cpp output.
25626 * doc/cppwarnopts.texi: Copy-edit.
25628 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25631 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25632 that the transition from a max line width >= 1<<10 to narrower
25633 lines works correctly.
25635 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25637 * doc/options.texi (PerFunction): New.
25638 * opt-functions.awk (switch_flags): Map both Optimization and
25639 PerFunction to CL_OPTIMIZATION.
25640 * opth-gen.awk: Test for PerFunction flag along with
25642 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25643 it only when the latter is present. Skip those that don't in
25644 the hash function generator.
25645 * common.opt (fvar-tracking): Mark as PerFunction instead of
25647 (fvar-tracking-assignments): Likewise.
25648 (fvar-tracking-assignments-toggle): Likewise.
25649 (fvar-tracking-uninit): Likewise.
25651 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25653 PR translation/79018
25654 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25657 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25660 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25661 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25662 TARGET_LONG_JUMP_TABLE_OFFSETS.
25663 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25664 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25665 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25666 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25667 * config/m68k/m68k.md (tablejump expander): Likewise.
25668 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25669 TARGET_LONG_JUMP_TABLE_OFFSETS.
25670 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25671 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25673 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25674 David Holsgrove <david.holsgrove@xilinx.com>
25676 * common/config/microblaze/microblaze-common.c
25677 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25678 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25680 * config/microblaze/microblaze.c (microblaze_must_save_register)
25681 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25683 (microblaze_eh_return): New function.
25684 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25685 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25686 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25687 * config/microblaze/microblaze.md (eh_return): New pattern.
25689 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25691 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25692 GCC_DIAGNOSTIC_STRINGIFY): Define.
25694 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25696 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25698 * config/arm/arm.md (<mcrr>): New.
25700 * config/arm/arm.c (arm_arch5te): New.
25701 (arm_option_override): Set arm_arch5te.
25702 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25704 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25705 (arm_mcrr_qualifiers): ... this. New.
25706 (MRRC_QUALIFIERS): Define to...
25707 (arm_mrrc_qualifiers): ... this. New.
25708 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25710 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25711 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25712 (MRRCI, mrrc, MRRC): New.
25713 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25714 VUNSPEC_MRRC2): New.
25716 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25718 * config/arm/arm.md (<mcr>): New.
25720 * config/arm/arm.c (arm_coproc_builtin_available): Add
25721 support for mcr, mrc, mcr2 and mrc2.
25722 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25723 (arm_mcr_qualifiers): ... this. New.
25724 (MRC_QUALIFIERS): Define to ...
25725 (arm_mrc_qualifiers): ... this. New.
25726 (MCR_QUALIFIERS): Define to ...
25727 (arm_mcr_qualifiers): ... this. New.
25728 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25730 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25731 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25732 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25733 VUNSPEC_MRC2): New.
25735 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25737 * config/arm/arm.md (*ldc): New.
25741 * config/arm/arm.c (arm_coproc_builtin_available): Add
25742 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25743 (arm_coproc_ldc_stc_legitimate_address): New.
25744 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25745 'qualifier_const_pointer'.
25746 (LDC_QUALIFIERS): Define to...
25747 (arm_ldc_qualifiers): ... this. New.
25748 (STC_QUALIFIERS): Define to...
25749 (arm_stc_qualifiers): ... this. New.
25750 * config/arm/arm-protos.h
25751 (arm_coproc_ldc_stc_legitimate_address): New.
25752 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25753 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25754 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25755 stc2, stcl, stc2l): New.
25756 * config/arm/constraints.md (Uz): New.
25757 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25758 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25759 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25760 VUNSPEC_STC2L): New.
25762 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25764 * config/arm/arm.md (<cdp>): New.
25765 * config/arm/arm.c (neon_const_bounds): Rename this ...
25766 (arm_const_bounds): ... this.
25767 (arm_coproc_builtin_available): New.
25768 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25769 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25770 (CDP_QUALIFIERS): Define to...
25771 (arm_cdp_qualifiers): ... this. New.
25773 (arm_expand_builtin_args): Add case for 6 arguments.
25774 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25775 (arm_const_bounds): ... this.
25776 (arm_coproc_builtin_available): New.
25777 * config/arm/arm_acle.h (__arm_cdp): New.
25779 * config/arm/arm_acle_builtins.def (cdp): New.
25781 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25782 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25783 'arm_const_bounds'.
25784 * config/arm/types.md (coproc): New.
25785 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25786 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25787 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25788 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25790 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25792 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25793 (UBINOP_QUALIFIERS): New.
25795 (acle_builtin_data): New. Change comment.
25796 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25797 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25798 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25799 arm_acle_builtins.def.
25800 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25801 (arm_init_acle_builtins): New.
25802 (CRC32_BUILTIN): Remove.
25803 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25804 crc32cb, crc32ch and crc32cw.
25805 (arm_init_crc32_builtins): Remove.
25806 (arm_init_builtins): Use arm_init_acle_builtins rather
25807 than arm_init_crc32_builtins.
25808 (arm_expand_acle_builtin): New.
25809 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25810 * config/arm/arm_acle_builtins.def: New.
25812 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25814 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25815 (arm_builtin_datum): ... this.
25816 (arm_init_neon_builtin): Rename to ...
25817 (arm_init_builtin): ... this. Add a new parameters PREFIX
25818 and USE_SIG_IN_NAME.
25819 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25820 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25821 'arm_builtin_datum'.
25822 (arm_init_vfp_builtins): Likewise.
25823 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25824 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25825 (arm_expand_neon_args): Rename to ...
25826 (arm_expand_builtin_args): ... this. Rename builtin_arg
25827 enum values and differentiate between ARG_BUILTIN_MEMORY
25828 and ARG_BUILTIN_NEON_MEMORY.
25829 (arm_expand_neon_builtin_1): Rename to ...
25830 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25831 values, arm_expand_builtin_args and add bool parameter NEON.
25832 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25833 (arm_expand_vfp_builtin): Likewise.
25834 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25836 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25838 PR middle-end/77484
25839 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25840 * predict.c (tree_estimate_probability_bb): Reverse direction of
25841 polymorphic call predictor.
25843 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25845 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25846 (determine_pass_name_match): ...this new function and...
25847 (should_skip_pass_p): ...this new function.
25849 2017-01-06 Nathan Sidwell <nathan@acm.org>
25851 * ipa-visibility.c (function_and_variable_visibility): Reformat
25852 comments and long lines. Remove extrneous if.
25853 * symtab.c (symtab_node::make_decl_local): Fix code format.
25854 (symtab_node::set_section_for_node): Fix comment typo.
25856 2017-01-06 Martin Liska <mliska@suse.cz>
25859 * lra-constraints.c: Rename invariant to lra_invariant.
25860 * predict.c (set_even_probabilities): Initialize e to NULL.
25862 2017-01-05 Martin Sebor <msebor@redhat.com>
25864 PR tree-optimization/78910
25865 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25866 (format_integer): Correct off-by-one error in the handling
25867 of precision with negative numbers in signed conversions..
25869 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25871 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25873 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25875 PR tree-optimization/71016
25876 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25877 factor_out_conditional_conversion. Formatting fix.
25878 (factor_out_conditional_conversion): Add cond_stmt argument.
25879 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25880 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25883 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25885 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25886 read-rtl-function.o, and selftest-rtl.o.
25887 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25888 (selftest::aarch64_test_loading_full_dump): New function.
25889 (selftest::aarch64_run_selftests): New function.
25890 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25891 selftest::aarch64_run_selftests.
25892 * config/i386/i386.c
25893 (selftest::ix86_test_loading_dump_fragment_1): New function.
25894 (selftest::ix86_test_loading_call_insn): New function.
25895 (selftest::ix86_test_loading_full_dump): New function.
25896 (selftest::ix86_test_loading_unspec): New function.
25897 (selftest::ix86_run_selftests): Call the new functions.
25898 * emit-rtl.c (maybe_set_max_label_num): New function.
25899 * emit-rtl.h (maybe_set_max_label_num): New decl.
25900 * function.c (instantiate_decls): Guard call to
25901 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25902 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25904 * gensupport.c (gen_reader::gen_reader): Pass "false"
25905 for new "compact" param of rtx_reader.
25906 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25907 rather than an empty string for NULL strings.
25908 * read-md.c: Potentially include config.h rather than bconfig.h.
25909 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25910 (have_error): New global, copied from errors.c.
25911 (md_reader::read_name): Rename to...
25912 (md_reader::read_name_1): ...this, adding "out_loc" param,
25913 and converting "missing name or number" to returning false, rather
25915 (md_reader::read_name): Reimplement in terms of read_name_1.
25916 (md_reader::read_name_or_nil): New function.
25917 (md_reader::read_string): Handle "(nil)" by returning NULL.
25918 (md_reader::md_reader): Add new param "compact".
25919 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25920 (md_reader::read_file): New method.
25921 * read-md.h (md_reader::md_reader): Add new param "compact".
25922 (md_reader::read_file): New method.
25923 (md_reader::is_compact): New accessor.
25924 (md_reader::read_name): Convert return type from void to file_location.
25925 (md_reader::read_name_or_nil): New decl.
25926 (md_reader::read_name_1): New decl.
25927 (md_reader::m_compact): New field.
25928 (noop_reader::noop_reader): Pass "false" for new "compact" param
25930 (rtx_reader::rtx_reader): Add new "compact" param.
25931 (rtx_reader::read_rtx_operand): Make virtual and convert return
25932 type from void to rtx.
25933 (rtx_reader::read_until): New decl.
25934 (rtx_reader::handle_any_trailing_information): New virtual function.
25935 (rtx_reader::postprocess): New virtual function.
25936 (rtx_reader::finalize_string): New virtual function.
25937 (rtx_reader::m_in_call_function_usage): New field.
25938 (rtx_reader::m_reuse_rtx_by_id): New field.
25939 * read-rtl-function.c: New file.
25940 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25941 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25942 (selftest::verify_three_block_rtl_cfg): New decl.
25943 * read-rtl-function.h: New file.
25944 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25945 For host, include function.h, memmodel.h, and emit-rtl.h.
25946 (one_time_initialization): New function.
25947 (struct compact_insn_name): New struct.
25948 (compact_insn_names): New array.
25949 (find_code): Handle insn codes in compact dumps.
25950 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25951 (bind_subst_iter_and_attr): Likewise.
25952 (add_condition_to_string): Likewise.
25953 (add_condition_to_rtx): Likewise.
25954 (apply_attribute_uses): Likewise.
25955 (add_current_iterators): Likewise.
25956 (apply_iterators): Likewise.
25957 (initialize_iterators): Guard usage of apply_subst_iterator with
25958 #ifdef GENERATOR_FILE.
25959 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25960 (md_reader::read_mapping): Likewise.
25961 (add_define_attr_for_define_subst): Likewise.
25962 (add_define_subst_attr): Likewise.
25963 (read_subst_mapping): Likewise.
25964 (check_code_iterator): Likewise.
25965 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25967 (one_time_initialization): New function.
25968 (rtx_reader::read_until): New method.
25969 (read_flags): New function.
25970 (parse_reg_note_name): New function.
25971 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25972 Handle reuse_rtx ids.
25973 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25974 Add parsing support for RTL dumps, mirroring the special-cases in
25975 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25976 values, and calling handle_any_trailing_information.
25977 (rtx_reader::read_rtx_operand): Convert return type from void
25978 to rtx, returning return_rtx. Handle case 'e'. Call
25979 finalize_string on XSTR and XTMPL fields.
25980 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25981 "(nil)" values were omitted. Call the postprocess vfunc on the
25983 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25984 class ctor. Initialize m_in_call_function_usage. Call
25985 one_time_initialization.
25986 * rtl-tests.c (selftest::test_uncond_jump): Call
25987 set_new_first_and_last_insn.
25988 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25989 * selftest-rtl.c: New file.
25990 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25991 (selftest::get_insn_by_uid): New decl.
25992 * selftest-run-tests.c (selftest::run_tests): Call
25993 read_rtl_function_c_tests.
25994 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25995 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25998 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26000 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26001 operands in a special way. Assert that pos+len <= mode precision.
26003 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26005 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26006 3 argument Alias with unlimited for the negative form.
26007 (fno-vect-cost-model): Removed.
26009 2017-01-05 Martin Liska <mliska@suse.cz>
26011 * hsa-gen.c (gen_hsa_divmod): New function.
26012 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26014 2017-01-05 Martin Liska <mliska@suse.cz>
26017 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26020 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26022 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26023 small constant length operands.
26025 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26027 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26028 between loop iterations.
26030 2017-01-05 Martin Liska <mliska@suse.cz>
26033 * gimplify.c (gimplify_decl_expr): Compare to
26034 asan_poisoned_variables instread of checking flags.
26035 (gimplify_target_expr): Likewise.
26036 (gimplify_expr): Likewise.
26037 (gimplify_function_tree): Conditionally initialize
26038 asan_poisoned_variables.
26040 2017-01-04 Jeff Law <law@redhat.com>
26042 PR tree-optimizatin/78812
26043 * rtl.h (contains_mem_rtx_p): Prototype.
26044 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26045 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26046 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26047 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26048 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26050 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26052 * input.c (assert_char_at_range): Default-initialize actual_range.
26054 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26056 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26057 to match the caller.
26059 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26061 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26062 insns after final jump in test to emit dummy move.
26064 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26066 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26067 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26069 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26071 * multiple_target.c (create_dispatcher_calls): Init e_next.
26072 * tree-ssa-loop-split.c (split_loop): Init border.
26073 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26076 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26081 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26082 (altivec_register_operand): Do not return true if the operand
26083 contains a SUBREG mixing SImode and SFmode.
26084 (vsx_register_operand): Likewise.
26085 (vsx_reg_sfsubreg_ok): New predicate.
26086 (vfloat_operand): Do not return true if the operand contains a
26087 SUBREG mixing SImode and SFmode.
26088 (vint_operand): Likewise.
26089 (vlogical_operand): Likewise.
26090 (gpc_reg_operand): Likewise.
26091 (int_reg_operand): Likewise.
26092 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26093 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26094 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26096 (rs6000_emit_move_si_sf_subreg): New helper function.
26097 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26098 fixup SUBREGs involving SImode and SFmode.
26099 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26100 numbers for the new peephole2 optimization.
26101 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26102 the GLIBC math library that do AND/IOR/XOR operations on single
26103 precision floating point.
26104 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26105 target macros to say whether we need to avoid SUBREGs mixing
26107 (TARGET_ALLOW_SF_SUBREG): Likewise.
26108 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26109 (UNSPEC_SI_FROM_SF): Likewise.
26110 (iorxor): Change spacing.
26111 (and_ior_xor): New iterator for AND, IOR, and XOR.
26112 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26113 (movdi_from_sf_zero_ext): Likewise.
26114 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26115 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26116 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26117 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26118 (fms<mode>4): Likewise.
26119 (fnma<mode>4): Likewise.
26120 (fnms<mode>4): Likewise.
26121 (nfma<mode>4): Likewise.
26122 (nfms<mode>4): Likewise.
26124 2017-01-04 Marek Polacek <polacek@redhat.com>
26127 * doc/invoke.texi: Document -Wpointer-compare.
26129 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26131 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26134 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26135 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26138 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26141 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26142 documentation of the powerpc_popcntb_ok attribute.
26143 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26144 code to issue warning messages if a requested CPU configuration is
26145 not supported by the binary (assembler and loader) toolchain.
26146 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26147 made to define a built-in function that has been disabled.
26148 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26149 made to define a built-in function that has been disabled.
26150 (altivec_init_builtins): Add comment explaining why definition
26151 of the DST built-in functions is not preceded by an assertion
26152 check. Add assertions to prevent ICE if attempts are made to
26153 define an altivec predicate or an abs* built-in function that has
26155 (htm_init_builtins): Add comment explaining why definition of the
26156 htm built-in functions is not preceded by an assertion check.
26158 2017-01-04 Jeff Law <law@redhat.com>
26160 PR tree-optimizatin/67955
26161 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26162 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26163 the points-to solution does not include pt_null. Use DECL_PT_UID
26166 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26168 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26169 Use gen_int_mode instead of gen_lopwart for const_int operands.
26171 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26173 PR tree-optimization/71563
26174 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26175 out of range value - has low bits known to be zero.
26177 2017-01-04 Alan Modra <amodra@gmail.com>
26179 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26180 * configure: Regenerate.
26181 * config.in: Regenerate.
26183 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26186 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26187 a substring of the message, but strcmp with the whole message. Ifdef
26188 ENABLE_NLS, translate the message first using dgettext.
26190 2017-01-03 Jeff Law <law@redhat.com>
26192 PR tree-optimizatin/78856
26193 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26194 (mark_threaded_blocks): Remove code to truncate thread paths that
26195 cross multiple loop headers. Instead invalidate the cached loop
26196 iteration information and handle case of a thread path walking
26197 into an irreducible region.
26199 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26202 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26203 assertions. Add support for doing the signbit if the IEEE 128-bit
26204 floating point value is in a GPR.
26205 * config/rs6000/rs6000.md (Fsignbit): Delete.
26206 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26207 Update the length attribute if the value is in a GPR.
26208 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26209 the sign or zero extension instruction, since the value is always 0/1.
26210 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26213 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26214 extracting SImode to a GPR register so that we can generate a
26215 store, limit the vector to be in a traditional Altivec register
26216 for the vextuwrx instruction.
26218 2017-01-03 Ian Lance Taylor <iant@google.com>
26220 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26222 2017-01-03 Martin Sebor <msebor@redhat.com>
26224 PR tree-optimization/78696
26225 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26226 precision. Use MPFR for %f for greater fidelity. Correct handling
26228 (pass_sprintf_length::compute_format_length): Set width and precision
26229 specified by asrerisk to void_node for vararg functions.
26230 (try_substitute_return_value): Adjust dump output.
26232 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26234 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26236 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26238 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26239 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26240 -mlra/-mno-lra was passed to the compiler.
26242 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26244 PR rtl-optimization/65618
26245 * emit-rtl.c (try_split): Move initialization of "before" and
26246 "after" to just before the call to emit_insn_after_setloc.
26248 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26250 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26252 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26254 * dwarf2out.c (gen_enumeration_type_die): When
26255 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26257 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26259 PR tree-optimization/78965
26260 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26261 Change first argument from const call_info & to call_info &. For %n
26262 set info.nowrite to false.
26264 PR middle-end/78901
26265 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26266 possibly throwing calls.
26268 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26269 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26270 and fns handling, rather than in a separate case SSA_NAME.
26272 2017-01-02 Jeff Law <law@redhat.com>
26274 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26275 fixes for first_period and second_period variables.
26277 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26280 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26281 (*insvqi_1): New insn pattern.
26282 (*insvqi_1_mem_rex64): Ditto.
26283 (*insvqi_2): Ditto.
26284 (*insvqi_3): Rename from *insvqi.
26286 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26288 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26290 * doc/cfg.texi (Edges): Remove reference to Java.
26291 (Maintaining the CFG): Ditto.
26293 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26295 PR middle-end/77674
26296 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26297 transparent aliases.
26299 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26301 PR middle-end/77484
26302 * predict.def (PRED_CALL): Update hitrate.
26303 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26304 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26305 into direct/indirect/polymorphic variants.
26307 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26309 Update copyright years.
26311 * gcc.c (process_command): Update copyright notice dates.
26312 * gcov-dump.c (print_version): Ditto.
26313 * gcov.c (print_version): Ditto.
26314 * gcov-tool.c (print_version): Ditto.
26315 * gengtype.c (create_file): Ditto.
26316 * doc/cpp.texi: Bump @copying's copyright year.
26317 * doc/cppinternals.texi: Ditto.
26318 * doc/gcc.texi: Ditto.
26319 * doc/gccint.texi: Ditto.
26320 * doc/gcov.texi: Ditto.
26321 * doc/install.texi: Ditto.
26322 * doc/invoke.texi: Ditto.
26324 Copyright (C) 2017 Free Software Foundation, Inc.
26326 Copying and distribution of this file, with or without modification,
26327 are permitted in any medium without royalty provided the copyright
26328 notice and this notice are preserved.