1 2017-09-12 Martin Liska <mliska@suse.cz>
3 * attribs.c (private_lookup_attribute): New function.
4 * attribs.h (private_lookup_attribute): Declared here.
5 (lookup_attribute): Called from this place.
7 2017-09-12 Richard Biener <rguenther@suse.de>
9 PR tree-optimization/82157
10 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11 stmts with side-effects.
13 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14 Alan Hayward <alan.hayward@arm.com>
15 David Sherwood <david.sherwood@arm.com>
17 * target.def (hard_regno_nregs): New hook.
18 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
19 * targhooks.h (default_hard_regno_nregs): Declare.
20 * targhooks.c (default_hard_regno_nregs): New function.
21 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
22 (TARGET_HARD_REGNO_NREGS): ...this hook.
23 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
24 (CLASS_MAX_NREGS): Likewise.
25 * doc/tm.texi: Regenerate.
26 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
27 instead of HARD_REGNO_NREGS.
28 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
29 HARD_REGNO_NREGS in the comment.
30 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
31 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
32 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
33 Return an unsigned int.
34 (TARGET_HARD_REGNO_NREGS): Redefine.
35 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
36 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
37 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
38 (arc_hard_regno_nregs): New function.
39 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
40 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
41 (arm_hard_regno_nregs): New function.
42 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
43 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
44 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
45 (TARGET_HARD_REGNO_NREGS): Redefine.
46 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
47 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
48 (HARD_REGNO_NREGS): Delete.
49 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
50 (cr16_hard_regno_nregs): New function.
51 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
52 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
53 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
54 (cris_hard_regno_nregs): New function.
55 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
56 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
57 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
58 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
59 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
60 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
61 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
62 (frv_hard_regno_nregs): Make static. Take and return an
64 (frv_class_max_nregs): Remove outdated copy of documentation.
65 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
66 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
67 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
68 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
69 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
70 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
71 (TARGET_HARD_REGNO_NREGS): Redefine.
72 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
73 (CLASS_MAX_NREGS): Update comment.
74 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
75 (ia64_hard_regno_nregs): New function.
76 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
77 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
78 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
79 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
80 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
82 (m32c_hard_regno_nregs): Likewise. Make static.
83 (TARGET_HARD_REGNO_NREGS): Redefine.
84 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
85 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
86 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
87 (m68k_hard_regno_nregs): New function.
88 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
89 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
90 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
91 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
92 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
93 Take and return an unsigned int.
94 (TARGET_HARD_REGNO_NREGS): Redefine.
95 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
96 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
97 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
98 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
99 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
100 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
101 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
102 (msp430_hard_regno_nregs): Make static. Take and return an
104 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
105 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
106 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
107 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
108 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
109 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
110 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
111 (TARGET_HARD_REGNO_NREGS): Redefine.
112 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
113 (PA_HARD_REGNO_NREGS): ...this.
114 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
115 (PA_HARD_REGNO_NREGS): ...this.
116 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
117 (pa_hard_regno_nregs): New function.
118 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
119 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
120 (pdp11_hard_regno_nregs): New function.
121 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
122 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
123 (rs6000_hard_regno_nregs_hook): New function.
124 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
125 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
126 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
127 Take and return an unsigned int. Move earlier in file.
128 (TARGET_HARD_REGNO_NREGS): Redefine.
129 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
130 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
131 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
132 (rl78_hard_regno_nregs): Make static. Take and return an
134 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
135 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
136 (rs6000_hard_regno_nregs_hook): New function.
137 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
138 * config/rx/rx.c (rx_hard_regno_nregs): New function.
139 (TARGET_HARD_REGNO_NREGS): Redefine.
140 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
141 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
142 instead of HARD_REGNO_NREGS.
143 (s390_hard_regno_nregs): New function.
144 (s390_hard_regno_mode_ok): Add comment from s390.h.
145 (TARGET_HARD_REGNO_NREGS): Redefine.
146 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
147 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
148 (sh_hard_regno_nregs): New function.
149 (sh_pass_in_reg_p): Use it.
150 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
151 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
152 (sparc_hard_regno_nregs): New function.
153 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
154 * config/spu/spu.c (spu_hard_regno_nregs): New function.
155 (spu_function_arg_advance): Use it, supplying a valid register number.
156 (TARGET_HARD_REGNO_NREGS): Redefine.
157 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
158 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
159 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
160 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
161 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
162 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
163 (CLASS_MAX_NREGS): Remove copy of old documentation.
164 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
165 (visium_hard_regno_nregs): New function.
166 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
167 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
168 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
169 xtensa_hard_regno_nregs): New function.
170 * system.h (HARD_REGNO_NREGS): Poison.
172 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
174 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
175 hard_regno_nregs instead of HARD_REGNO_NREGS.
176 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
177 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
178 (c6x_expand_epilogue): Likewise.
179 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
180 (frv_read_iacc_argument): Likewise.
181 * config/sh/sh.c: Include regs.h.
182 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
183 (regs_used): Likewise.
184 (output_stack_adjust): Likewise.
185 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
186 * expmed.c: Include regs.h.
187 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
188 * ree.c: Include regs.h.
189 (combine_reaching_defs): Use hard_regno_nregs instead of
191 (add_removable_extension): Likewise.
193 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
195 * regs.h (hard_regno_nregs): Turn into a function.
196 (end_hard_regno): Update accordingly.
197 * caller-save.c (setup_save_areas): Likewise.
198 (save_call_clobbered_regs): Likewise.
199 (replace_reg_with_saved_mem): Likewise.
200 (insert_restore): Likewise.
201 (insert_save): Likewise.
202 * combine.c (can_change_dest_mode): Likewise.
203 (move_deaths): Likewise.
204 (distribute_notes): Likewise.
205 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
206 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
207 (rs6000_split_multireg_move): Likewise.
208 (rs6000_register_move_cost): Likewise.
209 (rs6000_memory_move_cost): Likewise.
210 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
211 (rs6000_split_multireg_move): Likewise.
212 (rs6000_register_move_cost): Likewise.
213 (rs6000_memory_move_cost): Likewise.
214 * cselib.c (cselib_reset_table): Likewise.
215 (cselib_lookup_1): Likewise.
216 * emit-rtl.c (set_mode_and_regno): Likewise.
217 * function.c (aggregate_value_p): Likewise.
218 * ira-color.c (setup_profitable_hard_regs): Likewise.
219 (check_hard_reg_p): Likewise.
220 (calculate_saved_nregs): Likewise.
221 (assign_hard_reg): Likewise.
222 (improve_allocation): Likewise.
223 (calculate_spill_cost): Likewise.
224 * ira-emit.c (modify_move_list): Likewise.
225 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
226 (ira_hard_reg_in_set_p): Likewise.
227 * ira.c (setup_reg_mode_hard_regset): Likewise.
228 (clarify_prohibited_class_mode_regs): Likewise.
229 (check_allocation): Likewise.
230 * lra-assigns.c (find_hard_regno_for_1): Likewise.
231 (lra_setup_reg_renumber): Likewise.
232 (setup_try_hard_regno_pseudos): Likewise.
233 (spill_for): Likewise.
234 (assign_hard_regno): Likewise.
235 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
236 * lra-constraints.c (in_class_p): Likewise.
237 (lra_constraint_offset): Likewise.
238 (simplify_operand_subreg): Likewise.
239 (lra_constraints): Likewise.
240 (split_reg): Likewise.
241 (split_if_necessary): Likewise.
242 (invariant_p): Likewise.
243 (inherit_in_ebb): Likewise.
244 * lra-lives.c (process_bb_lives): Likewise.
245 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
246 (get_hard_regs): Likewise.
247 (do_remat): Likewise.
248 * lra-spills.c (assign_spill_hard_regs): Likewise.
249 * mode-switching.c (create_pre_exit): Likewise.
250 * postreload.c (reload_combine_recognize_pattern): Likewise.
251 * recog.c (peep2_find_free_register): Likewise.
252 * regcprop.c (kill_value_regno): Likewise.
253 (set_value_regno): Likewise.
254 (copy_value): Likewise.
255 (maybe_mode_change): Likewise.
256 (find_oldest_value_reg): Likewise.
257 (copyprop_hardreg_forward_1): Likewise.
258 * regrename.c (check_new_reg_p): Likewise.
259 (regrename_do_replace): Likewise.
260 * reload.c (push_reload): Likewise.
261 (combine_reloads): Likewise.
262 (find_dummy_reload): Likewise.
263 (operands_match_p): Likewise.
264 (find_reloads): Likewise.
265 (find_equiv_reg): Likewise.
266 (reload_adjust_reg_for_mode): Likewise.
267 * reload1.c (count_pseudo): Likewise.
268 (count_spilled_pseudo): Likewise.
269 (find_reg): Likewise.
270 (clear_reload_reg_in_use): Likewise.
271 (free_for_value_p): Likewise.
272 (allocate_reload_reg): Likewise.
273 (choose_reload_regs): Likewise.
274 (reload_adjust_reg_for_temp): Likewise.
275 (emit_reload_insns): Likewise.
276 (delete_output_reload): Likewise.
277 * rtlanal.c (subreg_get_info): Likewise.
278 * sched-deps.c (sched_analyze_reg): Likewise.
279 * sel-sched.c (init_regs_for_mode): Likewise.
280 (mark_unavailable_hard_regs): Likewise.
281 (choose_best_reg_1): Likewise.
282 (verify_target_availability): Likewise.
283 * valtrack.c (dead_debug_insert_temp): Likewise.
284 * var-tracking.c (track_loc_p): Likewise.
285 (emit_note_insn_var_location): Likewise.
286 * varasm.c (make_decl_rtl): Likewise.
287 * reginfo.c (choose_hard_reg_mode): Likewise.
288 (init_reg_modes_target): Refer directly to
289 this_target_regs->x_hard_regno_nregs.
291 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
293 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
294 instead of hard_regno_nregs.
296 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
298 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
299 end_hard_regno instead of hard_regno_nregs.
300 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
301 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
302 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
303 * ira-color.c (improve_allocation): Likewise.
304 * lra-assigns.c (find_hard_regno_for_1): Likewise.
305 * lra-lives.c (mark_regno_live): Likewise.
306 (mark_regno_dead): Likewise.
307 * lra-remat.c (operand_to_remat): Likewise.
308 * lra.c (collect_non_operand_hard_regs): Likewise.
309 * postreload.c (reload_combine_note_store): Likewise.
310 (move2add_valid_value_p): Likewise.
311 * reload.c (regno_clobbered_p): Likewise.
313 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
315 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
317 * config/v850/v850.c (v850_reorg): Likewise.
318 * reload.c (refers_to_regno_for_reload_p): Likewise.
319 (find_equiv_reg): Likewise.
320 * reload1.c (reload_reg_reaches_end_p): Likewise.
322 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
324 * caller-save.c (add_used_regs): Use REG_NREGS instead of
326 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
327 * config/arm/arm.c (output_move_neon): Likewise.
328 (arm_attr_length_move_neon): Likewise.
329 (neon_split_vcombine): Likewise.
330 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
331 (c6x_mark_reg_written): Likewise.
332 (c6x_dwarf_register_span): Likewise.
333 * config/i386/i386.c (ix86_save_reg): Likewise.
334 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
335 (rws_access_reg): Likewise.
336 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
337 * mode-switching.c (create_pre_exit): Likewise.
338 * ree.c (combine_reaching_defs): Likewise.
339 (add_removable_extension): Likewise.
340 * regcprop.c (find_oldest_value_reg): Likewise.
341 (copyprop_hardreg_forward_1): Likewise.
342 * reload.c (reload_inner_reg_of_subreg): Likewise.
343 (push_reload): Likewise.
344 (combine_reloads): Likewise.
345 (find_dummy_reload): Likewise.
346 (reload_adjust_reg_for_mode): Likewise.
347 * reload1.c (find_reload_regs): Likewise.
348 (forget_old_reloads_1): Likewise.
349 (reload_reg_free_for_value_p): Likewise.
350 (reload_adjust_reg_for_temp): Likewise.
351 (emit_reload_insns): Likewise.
352 (delete_output_reload): Likewise.
353 * sel-sched.c (choose_best_reg_1): Likewise.
354 (choose_best_pseudo_reg): Likewise.
356 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
357 Alan Hayward <alan.hayward@arm.com>
358 David Sherwood <david.sherwood@arm.com>
360 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
361 * target.def (slow_unaligned_access): New hook.
362 * targhooks.h (default_slow_unaligned_access): Declare.
363 * targhooks.c (default_slow_unaligned_access): New function.
364 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
365 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
366 * doc/tm.texi: Regenerate.
367 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
368 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
369 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
371 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
372 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
374 (rs6000_slow_unaligned_access): New function.
375 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
376 (expand_block_compare): Likewise.
377 (expand_strn_compare): Likewise.
378 (rs6000_rtx_costs): Likewise.
379 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
380 (riscv_slow_unaligned_access): Likewise.
381 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
382 (riscv_slow_unaligned_access_p): ...this and make static.
383 (riscv_option_override): Update accordingly.
384 (riscv_slow_unaligned_access): New function.
385 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
386 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
387 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
388 (rs6000_slow_unaligned_access): New function.
389 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
390 (rs6000_rtx_costs): Likewise.
391 * config/rs6000/rs6000-string.c (expand_block_compare)
392 (expand_strn_compare): Use targetm.slow_unaligned_access instead
393 of SLOW_UNALIGNED_ACCESS.
394 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
395 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
396 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
397 of SLOW_UNALIGNED_ACCESS.
398 * expmed.c (simple_mem_bitfield_p): Likewise.
399 * expr.c (alignment_for_piecewise_move): Likewise.
400 (emit_group_load_1): Likewise.
401 (emit_group_store): Likewise.
402 (copy_blkmode_from_reg): Likewise.
403 (emit_push_insn): Likewise.
404 (expand_assignment): Likewise.
405 (store_field): Likewise.
406 (expand_expr_real_1): Likewise.
407 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
408 * lra-constraints.c (simplify_operand_subreg): Likewise.
409 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
410 * gimple-ssa-store-merging.c: Likewise in block comment at start
412 * tree-ssa-strlen.c: Include target.h.
413 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
414 of SLOW_UNALIGNED_ACCESS.
415 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
417 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
419 PR rtl-optimization/82185
420 * expmed.c (emit_store_flag_int): Only test tem if it has been
423 2017-09-12 Richard Biener <rguenther@suse.de>
426 * match.pd ((FTYPE) N CMP CST): Fix typo.
428 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
430 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
432 (mips_near_type_p): Add 'short_call' attribute as a synonym
434 * doc/extend.texi (short_call): Document new function attribute.
436 2017-09-12 Jakub Jelinek <jakub@redhat.com>
439 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
440 assertion check that in the condition.
441 (get_atomic_generic_size): Likewise. Before testing if parameter
442 has pointer type, if it has array type, call for C++
443 default_conversion to perform array-to-pointer conversion.
445 2017-09-12 Richard Biener <rguenther@suse.de>
447 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
448 for operations we cannot scalarize.
450 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
452 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
453 vectors heap vectors. Clean up comments.
454 Make visited_bbs a reference.
455 (profitable_jump_thread_path): Make GC
456 vectors heap vectors. Clean up comments.
458 (convert_and_register_jump_thread_path): Make GC vectors heap
460 (check_subpath_and_update_thread_path): Same. Clean up comments.
461 Make visited_bbs a reference.
462 (handle_phi): Abstract common code to to
463 register_jump_thread_path_if_profitable.
464 Rename VAR_BB to DEF_BB.
466 Make GC vectors heap vectors.
467 Make visited_bbs a reference.
468 (handle_assignment): Same.
469 (register_jump_thread_path_if_profitable): New.
470 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
472 Make GC vectors heap vectors. Clean up comments.
473 Make visited_bbs a reference.
474 (find_jump_threads_backwards): Make visited_bbs live in the stack.
475 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
478 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
481 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
482 words of E_DImode object are reachable by xtensa_uimm8x4 access.
484 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
486 Revert r251800 and r251799.
488 2017-09-11 Martin Jambor <mjambor@suse.cz>
491 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
492 arguments in advance.
493 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
494 use it to find predecessor edges.
495 (naive_outof_ssa): Collect vector of predecessors.
497 2017-09-08 Jason Merrill <jason@redhat.com>
499 PR c++/70029 - ICE with ref-qualifier and -flto
500 * langhooks.h (struct lang_hooks_for_types): Add
501 copy_lang_qualifiers.
502 * attribs.c (build_type_attribute_qual_variant): Use it.
503 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
505 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
506 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
508 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
511 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
512 (*mulsi3_sp64): New instruction.
513 (mulsi3): New expander.
515 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
517 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
519 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
521 * sancov.c: Include memmodel.h.
523 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
526 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
529 2017-09-07 Carl Love <cel@us.ibm.com>
531 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
532 the sldi instruction.
534 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
536 * sancov.c: Include tm_p.h.
538 2017-09-07 Jakub Jelinek <jakub@redhat.com>
541 * output.h (switch_to_other_text_partition): New declaration.
542 * varasm.c (switch_to_other_text_partition): New function.
543 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
544 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
545 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
546 to the other text partition before emitting LCL label and switch back
547 after emitting the word after it.
549 2017-09-07 Richard Biener <rguenther@suse.de>
551 * passes.def (pass_split_crit_edges): Remove instance before PRE.
552 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
553 critical edges here, after loop init.
554 (pass_data_pre): Remove PROP_no_crit_edges flags.
555 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
556 for valueization of call args to avoid leaking VN_TOP.
557 (visit_use): Assert we do not visit default defs.
558 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
559 Use error_mark_node to more easily detect leaking VN_TOP.
560 All default-defs are varying, not VN_TOP. Mark them visited.
561 (run_scc_vn): Make code match comment.
563 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
565 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
566 OPTION_MASK_FLOAT128_KEYWORD.
567 (POWERPC_MASKS): Likewise.
568 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
569 support for the -mfloat128-type option, and make -mfloat128
570 default on PowerPC Linux systems. Define or undefine
571 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
572 Define __float128 to be __ieee128 if IEEE 128-bit support is
573 enabled, or undefine it.
574 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
575 Delete defining __FLOAT128_TYPE__.
576 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
577 -mfloat128-type option and make -mfloat128 default on PowerPC
579 (TARGET_FLOAT128_TYPE): Likewise.
580 (-mfloat128-type): Likewise.
581 * config/rs6000/rs6000.c (rs6000_option_override_internal):
582 Delete the -mfloat128-type option and make -mfloat128 default on
583 PowerPC Linux systems. Always use __ieee128 to be the keyword for
584 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
585 128-bit floating point is enabled. Change tests from using
586 -mfloat128-type to -mfloat128.
587 (rs6000_mangle_type): Use the correct mangling for the __float128
588 type even if normal long double is restricted to 64-bits.
589 (floatn_mode): Enable the _Float128 type by default on VSX Linux
591 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
592 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
593 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
595 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
596 documentation for -mfloat128.
598 2017-09-06 Olivier Hainque <hainque@adacore.com>
600 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
602 2017-09-06 Wish Wu <wishwu007@gmail.com>
603 Jakub Jelinek <jakub@redhat.com>
605 * asan.c (initialize_sanitizer_builtins): Add
606 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
607 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
608 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
609 BT_FN_VOID_UINT64_PTR variables.
610 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
611 (BT_FN_VOID_UINT16_UINT16): Likewise.
612 (BT_FN_VOID_UINT32_UINT32): Likewise.
613 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
614 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
615 (BT_FN_VOID_UINT64_PTR): Likewise.
616 * common.opt (flag_sanitize_coverage): New variable.
617 (fsanitize-coverage=trace-pc): Remove.
618 (fsanitize-coverage=): Add.
619 * flag-types.h (enum sanitize_coverage_code): New enum.
620 * fold-const.c (fold_range_test): Disable non-short-circuit
621 optimization if flag_sanitize_coverage.
622 (fold_truth_andor): Likewise.
623 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
624 * opts.c (COVERAGE_SANITIZER_OPT): Define.
625 (coverage_sanitizer_opts): New array.
626 (get_closest_sanitizer_option): Add OPTS argument, handle also
627 OPT_fsanitize_coverage_.
628 (parse_sanitizer_options): Adjusted to also handle
629 OPT_fsanitize_coverage_.
630 (common_handle_option): Add OPT_fsanitize_coverage_.
631 * sancov.c (instrument_comparison, instrument_switch): New function.
632 (sancov_pass): Add trace-cmp support.
633 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
634 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
635 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
636 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
637 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
638 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
639 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
640 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
641 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
642 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
644 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
646 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
647 error. Only quit immediately if parsing is complete.
648 (BEGIN): Initialize fatal_err and parse_done.
649 (begin fpu, end fpu): Check number of arguments.
650 (begin arch, end arch): Likewise.
651 (begin cpu, end cpu): Likewise.
652 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
653 (optalias): Likewise.
655 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
657 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
658 * config/arm/arm-isa.h: Delete. Move definitions to ...
659 * arm-cpus.in: ... here. Use new feature and fgroup values.
660 * config/arm/arm.c (arm_option_override): Use lower case for feature
662 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
663 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
664 * config/arm/parsecpu.awk (END): Add new command 'isa'.
666 (print_isa_bits_for): New function.
667 (gen_isa): New function.
668 (gen_comm_data): Use print_isa_bits_for.
669 (define feature): New keyword.
670 (define fgroup): New keyword.
671 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
672 (arm-isa.h): Add rule to generate file.
673 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
674 case for feature bit names.
676 2017-09-06 Richard Biener <rguenther@suse.de>
678 * tree-ssa-pre.c (NECESSARY): Remove.
679 (create_expression_by_pieces): Do not touch pass-local flags.
680 (insert_into_preds_of_block): Likewise.
681 (do_pre_regular_insertion): Likewise.
682 (eliminate_insert): Likewise.
683 (eliminate_dom_walker::before_dom_children): Likewise.
684 (fini_eliminate): Do not look at inserted_exprs.
685 (mark_operand_necessary): Remove.
686 (remove_dead_inserted_code): Replace with simple work-list
687 algorithm based on inserted_exprs and SSA uses.
688 (pass_pre::execute): Re-order fini_eliminate and
689 remove_dead_inserted_code.
691 2017-09-06 Olivier Hainque <hainque@adacore.com>
693 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
694 for VxWorks 7. Adjust surrounding comments.
696 2017-09-06 Richard Biener <rguenther@suse.de>
698 * gimple-ssa-strength-reduction.c
699 (find_candidates_dom_walker::before_dom_children): Also allow
702 2017-09-06 Richard Biener <rguenther@suse.de>
704 PR tree-optimization/82108
705 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
706 for gap in the non-permutation SLP case.
708 2017-09-06 Martin Jambor <mjambor@suse.cz>
710 PR tree-optimization/82078
711 * tree-sra.c (sort_and_splice_var_accesses): Move call to
712 add_access_to_work_queue...
713 (build_accesses_from_assign): ...here.
714 (propagate_all_subaccesses): Make sure racc is the group
715 representative, if there is one.
717 2017-09-06 Jakub Jelinek <jakub@redhat.com>
720 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
723 2017-09-06 Richard Biener <rguenther@suse.de>
725 * gimple-ssa-strength-reduction.c
726 (find_candidates_dom_walker::before_doom_children): Use a
727 type and not a mode check.
729 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
732 * config/arm/predicates.md (arm_general_adddi_operand): Create new
734 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
736 2017-09-05 Jeff Law <law@redhat.com>
738 PR tree-optimization/64910
739 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
740 cases where we have 3 or more operands.
742 2017-09-05 Jakub Jelinek <jakub@redhat.com>
745 * omp-low.c (lower_omp_for): Recompute tree invariant if
746 gimple_omp_for_initial/final is ADDR_EXPR.
749 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
750 into gimple val before gimplification fo the COND_EXPR.
752 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
754 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
755 REGION_COPY argument.
756 (thread_through_all_blocks): Remove unused argument to
757 duplicate_thread_path.
759 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
760 Alan Hayward <alan.hayward@arm.com>
761 David Sherwood <david.sherwood@arm.com>
763 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
764 Take a scalar_mode rather than a machine_mode.
765 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
766 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
767 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
768 (aarch64_gen_adjusted_ldpstp): Likewise.
769 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
772 Alan Hayward <alan.hayward@arm.com>
773 David Sherwood <david.sherwood@arm.com>
775 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
776 Take a scalar_int_mode instead of a machine_mode.
777 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
778 (aarch64_output_scalar_simd_mov_immediate): Likewise.
779 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
780 (aarch64_simd_attr_length_rglist): Delete.
781 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
782 a scalar_int_mode instead of a machine_mode.
783 (aarch64_add_offset): Likewise.
784 (aarch64_internal_mov_immediate): Likewise
785 (aarch64_add_constant_internal): Likewise.
786 (aarch64_add_constant): Likewise.
787 (aarch64_movw_imm): Likewise.
788 (aarch64_rtx_arith_op_extract_p): Likewise.
789 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
790 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
791 Remove assert that the mode isn't a vector.
792 (aarch64_output_scalar_simd_mov_immediate): Likewise.
793 (aarch64_expand_mov_immediate): Update calls after above changes.
794 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
795 (aarch64_and_bitmask_imm): Check for scalar integer modes.
796 (aarch64_move_imm): Likewise.
797 (aarch64_can_const_movi_rtx_p): Likewise.
798 (aarch64_strip_extend): Likewise.
799 (aarch64_extr_rtx_p): Likewise.
800 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
801 a CONST_INT when the mode parameter is VOIDmode.
802 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
804 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
806 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
807 * stor-layout.c (bitwise_mode_for_mode): Likewise.
808 (bitwise_type_for_mode): Update accordingly.
810 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
812 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
813 * stor-layout.c (mode_for_size_tree): Likewise.
814 (mode_for_array): Update accordingly.
815 (layout_decl): Likewise.
816 (compute_record_mode): Likewise. Only set the mode once.
818 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
820 * target.def (get_mask_mode): Change return type to opt_mode.
822 * doc/tm.texi: Regenerate.
823 * targhooks.h (default_get_mask_mode): Return an opt_mode.
824 * targhooks.c (default_get_mask_mode): Likewise.
825 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
826 * optabs-query.c (can_vec_mask_load_store_p): Update use of
827 targetm.get_mask_mode.
828 * tree.c (build_truth_vector_type): Likewise.
830 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
832 * machmode.h (mode_for_vector): Return an opt_mode.
833 * stor-layout.c (mode_for_vector): Likewise.
834 (mode_for_int_vector): Update accordingly.
835 (layout_type): Likewise.
836 * config/i386/i386.c (emit_memmov): Likewise.
837 (ix86_expand_set_or_movmem): Likewise.
838 (ix86_expand_vector_init): Likewise.
839 (ix86_get_mask_mode): Likewise.
840 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
842 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
843 * expmed.c (extract_bit_field_1): Likewise.
844 * expr.c (expand_expr_real_2): Likewise.
845 * optabs-query.c (can_vec_perm_p): Likewise.
846 (can_vec_mask_load_store_p): Likewise.
847 * optabs.c (expand_vec_perm): Likewise.
848 * targhooks.c (default_get_mask_mode): Likewise.
849 * tree-vect-stmts.c (vectorizable_store): Likewise.
850 (vectorizable_load): Likewise.
851 (get_vectype_for_scalar_type_and_size): Likewise.
853 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
855 * machmode.h (mode_for_int_vector): New function.
856 * stor-layout.c (mode_for_int_vector): Likewise.
857 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
858 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
859 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
860 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
861 (s390_expand_vcond): Likewise.
863 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
865 * machmode.h (opt_machine_mode): New type.
866 (opt_mode<T>): Allow construction from anything that can be
868 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
869 (mode_for_size): Return an opt_machine_mode.
870 * stor-layout.c (mode_for_size): Likewise.
871 (mode_for_size_tree): Update call accordingly.
872 (bitwise_mode_for_mode): Likewise.
873 (make_fract_type): Likewise.
874 (make_accum_type): Likewise.
875 * caller-save.c (replace_reg_with_saved_mem): Update call
877 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
878 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
879 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
880 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
881 * expmed.c (extract_bit_field_1): Likewise.
882 * reload.c (get_secondary_mem): Likewise.
883 * varasm.c (assemble_integer): Likewise.
884 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
887 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
889 * machmode.h (decimal_float_mode_for_size): New function.
890 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
891 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
892 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
893 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
894 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
895 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
897 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
899 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
900 (get_builtin_sync_mode): Likewise.
901 (expand_ifn_atomic_compare_exchange): Likewise.
902 (expand_builtin_atomic_clear): Likewise.
903 (expand_builtin_atomic_test_and_set): Likewise.
904 (fold_builtin_atomic_always_lock_free): Likewise.
905 * calls.c (compute_argument_addresses): Likewise.
906 (emit_library_call_value_1): Likewise.
907 (store_one_arg): Likewise.
908 * combine.c (combine_instructions): Likewise.
909 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
910 * config/arm/arm.c (arm_function_value): Likewise.
911 (aapcs_allocate_return_reg): Likewise.
912 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
913 * config/i386/i386.c (construct_container): Likewise.
914 (ix86_gimplify_va_arg): Likewise.
915 (ix86_expand_sse_cmp): Likewise.
916 (emit_memmov): Likewise.
917 (emit_memset): Likewise.
918 (expand_small_movmem_or_setmem): Likewise.
919 (ix86_expand_pextr): Likewise.
920 (ix86_expand_pinsr): Likewise.
921 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
922 * config/microblaze/microblaze.c (microblaze_block_move_straight):
924 * config/mips/mips.c (mips_function_value_1) Likewise.
925 (mips_block_move_straight): Likewise.
926 (mips_expand_ins_as_unaligned_store): Likewise.
927 * config/powerpcspe/powerpcspe.c
928 (rs6000_darwin64_record_arg_advance_flush): Likewise.
929 (rs6000_darwin64_record_arg_flush): Likewise.
930 * config/rs6000/rs6000.c
931 (rs6000_darwin64_record_arg_advance_flush): Likewise.
932 (rs6000_darwin64_record_arg_flush): Likewise.
933 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
934 (sparc_function_value_1): Likewise.
935 * config/spu/spu.c (adjust_operand): Likewise.
936 (spu_emit_branch_or_set): Likewise.
937 (arith_immediate_p): Likewise.
938 * emit-rtl.c (gen_lowpart_common): Likewise.
939 * expr.c (expand_expr_real_1): Likewise.
940 * function.c (assign_parm_setup_block): Likewise.
941 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
942 * reload1.c (alter_reg): Likewise.
943 * stor-layout.c (mode_for_vector): Likewise.
944 (layout_type): Likewise.
946 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
948 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
949 (spu_convert_move): Likewise.
950 * lower-subreg.c (resolve_simple_move): Likewise.
952 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
955 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
956 define_insn to a define_expand.
957 (altivec_vsum2sws_direct): New define_insn.
958 (altivec_vsumsws): Convert from a define_insn to a define_expand.
960 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
962 * config/arm/arm.c (arm_option_params_internal): Improve setting of
965 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
970 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
971 realign stack if stack alignment needed is less than incoming
974 2017-09-05 Marek Polacek <polacek@redhat.com>
977 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
980 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
982 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
984 2017-09-05 Richard Biener <rguenther@suse.de>
986 PR tree-optimization/82084
987 * fold-const.c (can_native_encode_string_p): Handle wide characters.
989 2017-09-05 Richard Biener <rguenther@suse.de>
991 PR tree-optimization/82102
992 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
994 2017-09-05 Martin Liska <mliska@suse.cz>
996 PR tree-optimization/82032
997 * tree-cfg.c (generate_range_test): New function.
998 * tree-cfg.h (generate_range_test): Declared here.
999 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1000 (cleanup_control_expr_graph): Use it.
1001 * tree-switch-conversion.c (try_switch_expansion): Remove
1003 (emit_case_nodes): Use generate_range_test.
1005 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1008 * config/i386/i386.md (*<btsc><mode>_mask): Add
1009 TARGET_USE_BT to insn constraint.
1010 (*btr<mode>_mask): Ditto.
1012 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1014 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1015 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1017 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1020 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1021 TARGET_NEON and TARGET_IWMMXT.
1022 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1023 TARGET_NEON and TARGET_IWMMXT.
1024 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1026 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1028 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1029 (ix86_rewrite_tls_address): Ditto.
1030 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1031 (ix86_rewrite_tls_address_1): Ditto.
1032 (ix86_rewrite_tls_address): Ditto.
1033 * config/i386/predicates.md (tls_address_pattern): New predicate.
1034 * config/i386/i386.md (TLS address splitter): New splitter.
1036 2017-09-04 Richard Biener <rguenther@suse.de>
1038 PR tree-optimization/82084
1039 * fold-const.h (can_native_encode_string_p): Declare.
1040 * fold-const.c (can_native_encode_string_p): Factor out from ...
1041 (native_encode_string): ... here.
1042 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1043 vectorizing stores from constants we later cannot handle.
1045 2017-09-04 Marek Polacek <polacek@redhat.com>
1048 * doc/invoke.texi: Update -Wtautological-compare documentation.
1050 2017-09-04 Jeff Law <law@redhat.com>
1052 PR tree-optimization/64910
1053 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1054 swap the first and last operand if the last is a constant.
1056 2017-09-04 Marek Polacek <polacek@redhat.com>
1059 * convert.c (do_narrow): When sanitizing signed integer overflows,
1060 bail out for signed types.
1061 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1063 2017-09-04 Richard Biener <rguenther@suse.de>
1065 PR tree-optimization/82060
1066 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1067 Move devirtualization after stmt folding and before EH/AB/noreturn
1068 cleanup to get the stmt refs canonicalized. Use a bool instead
1069 of gimple_modified_p since that doesn't work for NOPs. Schedule
1070 NOPs generated by folding for removal.
1072 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1073 Alan Hayward <alan.hayward@arm.com>
1074 David Sherwood <david.sherwood@arm.com>
1076 * coretypes.h (pad_direction): New enum.
1077 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1078 (FUNCTION_ARG_PADDING): Likewise.
1079 * target.def (function_arg_padding): New hook.
1080 * targhooks.h (default_function_arg_padding): Declare.
1081 * targhooks.c (default_function_arg_padding): New function.
1082 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1083 (TARGET_FUNCTION_ARG_PADDING): ...this.
1084 * doc/tm.texi: Regenerate.
1085 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1086 instead of direction.
1087 (compute_argument_addresses): Likewise.
1088 (load_register_parameters): Likewise.
1089 (emit_library_call_value_1): Likewise.
1090 (store_one_arg): Use targetm.calls.function_arg_padding instead
1091 of FUNCTION_ARG_PADDING.
1092 (must_pass_in_stack_var_size_or_pad): Likewise.
1093 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1094 (emit_group_store): Likewise.
1095 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1096 instead of FUNCTION_ARG_PADDING.
1097 (emit_push_insn): Likewise, and propagate enum change throughout
1099 * function.h (direction): Delete.
1100 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1102 * function.c (assign_parm_find_stack_rtl): Likewise.
1103 (assign_parm_setup_block_p): Likewise.
1104 (assign_parm_setup_block): Likewise.
1105 (gimplify_parameters): Likewise.
1106 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1107 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1109 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1110 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1111 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1112 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1113 (aarch64_function_arg_padding): ...this new function.
1114 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1115 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1116 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1117 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1118 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1119 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1120 (arm_pad_arg_upward): Replace with...
1121 (arm_function_arg_padding): ...this new function.
1122 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1124 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1125 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1126 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1127 (ia64_hpux_function_arg_padding): Replace with...
1128 (ia64_function_arg_padding): ...this new function. Use pad_direction
1129 instead of direction. Check for TARGET_HPUX.
1130 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1131 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1132 (iq2000_function_arg_padding): New function.
1133 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1134 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1135 (mips_function_arg_padding): ...this new function.
1136 (mips_pad_reg_upward): Update accordingly.
1137 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1138 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1139 targetm.calls.function_arg_padding.
1140 (FUNCTION_ARG_PADDING): Delete.
1141 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1142 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1143 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1144 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1145 (nios2_block_reg_padding): Return pad_direction instead of direction.
1146 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1147 instead of direction.
1148 (nios2_function_arg_padding): Likewise. Make static.
1149 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1150 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1151 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1152 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1153 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1154 (pa_function_arg_padding): Make static. Return pad_direction instead
1156 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1157 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1158 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1159 instead of direction. Use targetm.calls.function_arg_padding.
1160 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1161 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1162 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1163 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1164 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1166 (function_arg_padding): Rename to...
1167 (rs6000_function_arg_padding): ...this. Make static. Return
1168 pad_direction instead of direction.
1169 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1170 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1171 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1172 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1173 instead of direction. Use targetm.calls.function_arg_padding.
1174 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1175 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1176 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1177 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1178 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1179 (function_arg_padding): Rename to...
1180 (rs6000_function_arg_padding): ...this. Make static. Return
1181 pad_direction instead of direction.
1182 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1183 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1184 * config/s390/s390.c (s390_function_arg_padding): New function.
1185 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1186 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1187 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1188 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1189 (function_arg_padding): Rename to...
1190 (sparc_function_arg_padding): ...this. Make static. Return
1191 pad_direction instead of direction.
1192 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1193 * config/spu/spu.c (spu_function_arg_padding): New function.
1194 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1195 * system.h (FUNCTION_ARG_PADDING): Poison.
1197 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1198 Alan Hayward <alan.hayward@arm.com>
1199 David Sherwood <david.sherwood@arm.com>
1201 * target.def (modes_tieable_p): New hook.
1202 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1203 (TARGET_MODES_TIEABLE_P): ...this.
1204 * doc/tm.texi.in: Regenerate.
1205 * hooks.h (hook_bool_mode_mode_true): Declare.
1206 * hooks.c (hook_bool_mode_mode_true): New function.
1207 * combine.c (subst): Use targetm.modes_tieable_p instead of
1209 * dse.c (find_shift_sequence): Likewise.
1210 * expmed.c (extract_low_bits): Likewise.
1211 * lower-subreg.c: Include target.h.
1212 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1214 * rtlanal.c (rtx_cost): Likewise.
1215 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1216 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1217 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1218 (TARGET_MODES_TIEABLE_P): Redefine.
1219 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1220 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1221 (TARGET_MODES_TIEABLE_P): Redefine.
1222 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1223 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1224 (arc_modes_tieable_p): New function.
1225 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1226 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1227 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1228 (arm_modes_tieable_p): Make static.
1229 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1230 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1231 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1232 (TARGET_MODES_TIEABLE_P): Redefine.
1233 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1234 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1235 (TARGET_MODES_TIEABLE_P): Redefine.
1236 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1237 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1238 (cr16_modes_tieable_p): New function.
1239 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1240 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1241 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1242 (TRULY_NOOP_TRUNCATION): Update comment.
1243 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1244 (TRULY_NOOP_TRUNCATION): Update comment.
1245 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1246 (frv_modes_tieable_p): New function.
1247 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1248 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1249 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1250 (TARGET_MODES_TIEABLE_P): Redefine.
1251 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1252 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1253 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1254 (TARGET_MODES_TIEABLE_P): Redefine.
1255 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1256 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1257 (ia64_modes_tieable_p): New function.
1258 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1259 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1260 (iq2000_modes_tieable_p): New function.
1261 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1262 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1263 (lm32_modes_tieable_p): New function.
1264 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1265 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1266 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1267 (TARGET_MODES_TIEABLE_P): Redefine.
1268 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1269 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1270 (m32r_modes_tieable_p): New function.
1271 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1272 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1273 (m68k_modes_tieable_p): New function.
1274 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1275 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1276 (mcore_modes_tieable_p): New function.
1277 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1278 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1280 (TARGET_MODES_TIEABLE_P): Redefine.
1281 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1282 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1283 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1284 (TARGET_MODES_TIEABLE_P): Redefine.
1285 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1286 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1287 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1288 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1289 (mn10300_modes_tieable_p): ...this and make static.
1290 (TARGET_MODES_TIEABLE_P): Redefine.
1291 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1292 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1293 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1294 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1295 (msp430_modes_tieable_p): Make static.
1296 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1297 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1298 (TARGET_MODES_TIEABLE_P): Redefine.
1299 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1300 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1301 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1302 (TARGET_MODES_TIEABLE_P): Redefine.
1303 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1304 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1305 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1306 (TARGET_MODES_TIEABLE_P): Redefine.
1307 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1308 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1309 (pdp11_modes_tieable_p): New function.
1310 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1311 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1312 (rs6000_modes_tieable_p): New function.
1313 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1314 * config/powerpcspe/powerpcspe.md: Update comment.
1315 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1316 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1317 (TARGET_MODES_TIEABLE_P): Redefine.
1318 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1319 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1320 (rl78_modes_tieable_p): New function.
1321 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1322 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1323 (rs6000_modes_tieable_p): New function.
1324 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1325 * config/rs6000/rs6000.md: Update comment.
1326 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1327 * config/rx/rx.c (rx_modes_tieable_p): New function.
1328 (TARGET_MODES_TIEABLE_P): Redefine.
1329 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1330 * config/s390/s390.c (s390_modes_tieable_p): New function.
1331 (TARGET_MODES_TIEABLE_P): Redefine.
1332 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1333 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1334 (sh_modes_tieable_p): New function.
1335 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1336 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1337 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1338 (sparc_modes_tieable_p): Make static.
1339 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1340 * config/spu/spu.c (spu_modes_tieable_p): New function.
1341 (TARGET_MODES_TIEABLE_P): Redefine.
1342 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1343 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1344 (TARGET_MODES_TIEABLE_P): Redefine.
1345 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1346 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1347 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1348 * config/v850/v850.c (v850_modes_tieable_p): New function.
1349 (TARGET_MODES_TIEABLE_P): Redefine.
1350 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1351 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1352 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1353 (visium_modes_tieable_p): New function.
1354 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1355 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1356 (xtensa_modes_tieable_p): New function.
1357 * system.h (MODES_TIEABLE_P): Poison.
1359 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * target.def (hard_regno_mode_ok): New hook.
1364 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1365 (TARGET_HARD_REGNO_MODE_OK): ...this.
1366 * doc/tm.texi.in: Regenerate.
1367 * hooks.h (hook_bool_uint_mode_true): Declare.
1368 * hooks.c (hook_bool_uint_mode_true): New function.
1369 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1371 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1372 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1373 instead of HARD_REGNO_MODE_OK.
1374 * caller-save.c: Include target.h.
1375 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1377 * combine.c (can_combine_p): Likewise.
1378 (combinable_i3pat): Likewise.
1379 (can_change_dest_mode): Likewise.
1380 * expr.c (init_expr_target): Likewise.
1381 (convert_move): Likewise.
1382 (convert_modes): Likewise.
1383 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1384 (setup_prohibited_mode_move_regs): Likewise.
1385 * ira.h (target_ira): Likewise.
1386 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1387 * lra-constraints.c (process_alt_operands): Likewise.
1388 (split_reg): Likewise.
1389 * recog.c (peep2_find_free_register): Likewise.
1390 * ree.c (combine_reaching_defs): Likewise.
1391 * regcprop.c (maybe_mode_change): Likewise.
1392 * reginfo.c (init_reg_sets_1): Likewise.
1393 (choose_hard_reg_mode): Likewise.
1394 (simplifiable_subregs): Likewise.
1395 * regrename.c (check_new_reg_p): Likewise.
1396 * reload.c (find_valid_class): Likewise.
1397 (find_valid_class_1): Likewise.
1398 (reload_inner_reg_of_subreg): Likewise.
1399 (push_reload): Likewise.
1400 (combine_reloads): Likewise.
1401 (find_dummy_reload): Likewise.
1402 (find_reloads): Likewise.
1403 * reload1.c (find_reg): Likewise.
1404 (set_reload_reg): Likewise.
1405 (allocate_reload_reg): Likewise.
1406 (choose_reload_regs): Likewise.
1407 (reload_adjust_reg_for_temp): Likewise.
1408 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1409 (simplify_subreg_regno): Likewise.
1410 * sel-sched.c (init_regs_for_mode): Likewise.
1411 * varasm.c (make_decl_rtl): Likewise.
1412 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1413 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1415 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1416 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1418 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1419 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1421 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1422 (arc_mode_class): Delete.
1423 (HARD_REGNO_MODE_OK): Delete.
1424 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1425 (arc_hard_regno_mode_ok): Rename old array to...
1426 (arc_hard_regno_mode_ok_modes): ...this.
1427 (arc_conditional_register_usage): Update accordingly.
1428 (arc_mode_class): Make static.
1429 (arc_hard_regno_mode_ok): New function.
1430 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1431 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1432 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1433 (arm_hard_regno_mode_ok): Make static.
1434 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1436 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1437 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1438 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1440 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1441 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1442 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1443 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1444 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1445 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1446 * config/bfin/predicates.md (valid_reg_operand): Use
1447 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1448 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1449 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1450 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1451 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1452 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1453 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1454 (cr16_hard_regno_mode_ok): Make static and return a bool.
1455 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1456 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1457 (cris_hard_regno_mode_ok): New function.
1458 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1459 (epiphany_mode_class): Delete.
1460 (HARD_REGNO_MODE_OK): Delete.
1461 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1462 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1463 (hard_regno_mode_ok): Rename to...
1464 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1465 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1466 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1468 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1469 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1470 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1471 (frv_hard_regno_mode_ok): Make static and return a bool.
1472 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1474 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1475 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1476 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1477 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1480 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1481 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1482 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1484 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1485 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1486 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1487 (ia64_hard_regno_mode_ok): New function.
1488 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1489 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1490 (iq2000_hard_regno_mode_ok): New function.
1491 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1492 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1493 (lm32_hard_regno_mode_ok): New function.
1494 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1495 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1496 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1497 instead of HARD_REGNO_MODE_OK.
1498 (m32c_hard_regno_ok): Rename to...
1499 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1500 (m32c_cannot_change_mode_class): Update accordingly.
1501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1502 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1503 (m32r_mode_class): Delete.
1504 (HARD_REGNO_MODE_OK): Delete.
1505 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1506 (m32r_hard_regno_mode_ok): Rename to...
1507 (m32r_hard_regno_modes): ...this.
1508 (m32r_mode_class): Make static.
1509 (m32r_hard_regno_mode_ok): New function.
1510 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1511 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1512 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1513 (m68k_hard_regno_mode_ok): Make static.
1514 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1515 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1516 (mcore_hard_regno_mode_ok): New function.
1517 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1518 (HARD_REGNO_MODE_OK): Delete.
1519 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1521 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1522 (microblaze_hard_regno_mode_ok): New function.
1523 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1524 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1525 (mips_hard_regno_mode_ok): Delete.
1526 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1527 (mips_hard_regno_mode_ok_p): ...this and make static.
1528 (mips_hard_regno_mode_ok_p): Rename to...
1529 (mips_hard_regno_mode_ok_uncached): ...this.
1530 (mips_hard_regno_mode_ok): New function.
1531 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1532 of HARD_REGNO_MODE_OK.
1533 (mips_option_override): Update after above name changes.
1534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1535 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1536 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1537 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1538 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1539 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1540 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1541 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1542 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1543 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1544 (msp430_hard_regno_mode_ok): Make static and return a bool.
1545 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1546 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1547 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1550 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1551 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1552 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1553 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1554 (PA_HARD_REGNO_MODE_OK): ...this
1555 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1556 (PA_HARD_REGNO_MODE_OK): ...this.
1557 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1558 (pa_hard_regno_mode_ok): New function.
1559 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1560 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1561 (pdp11_hard_regno_mode_ok): New function.
1562 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1563 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1565 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1567 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1568 (rs6000_hard_regno_mode_ok): Rename to...
1569 (rs6000_hard_regno_mode_ok_uncached): ...this.
1570 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1571 (rs6000_hard_regno_mode_ok): New function.
1572 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1573 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1574 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1575 (riscv_hard_regno_mode_ok): ...this and make static.
1576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1577 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1578 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1579 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1580 (rl78_hard_regno_mode_ok): Make static and return bool.
1581 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1582 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1584 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1586 (rs6000_hard_regno_mode_ok): Rename to...
1587 (rs6000_hard_regno_mode_ok_uncached): ...this.
1588 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1589 (rs6000_hard_regno_mode_ok): New function.
1590 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1591 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1592 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1593 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1594 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1595 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1596 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1597 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1598 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1599 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1600 (sh_hard_regno_mode_ok): Make static.
1601 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1602 instead of HARD_REGNO_MODE_OK.
1603 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1604 (sparc_mode_class): Delete.
1605 (HARD_REGNO_MODE_OK): Delete.
1606 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1607 (hard_regno_mode_classes): Make static.
1608 (sparc_mode_class): Likewise.
1609 (sparc_hard_regno_mode_ok): New function.
1610 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1611 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1612 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1614 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1615 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1616 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1617 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1618 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1620 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1621 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1622 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1623 (visium_hard_regno_mode_ok): New function.
1624 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1625 instead of HARD_REGNO_MODE_OK.
1626 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1627 (HARD_REGNO_MODE_OK): Delete.
1628 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1629 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1630 (xtensa_option_override): Update accordingly.
1631 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1632 (xtensa_hard_regno_mode_ok): New function.
1633 * system.h (HARD_REGNO_MODE_OK): Poison.
1635 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1636 Alan Hayward <alan.hayward@arm.com>
1637 David Sherwood <david.sherwood@arm.com>
1639 * target.def (hard_regno_call_part_clobbered): New hook.
1640 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1641 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1642 * doc/tm.texi: Regenerate.
1643 * hooks.h (hook_bool_uint_mode_false): Declare.
1644 * hooks.c (hook_bool_uint_mode_false): New function.
1645 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1646 * cselib.c (cselib_process_insn): Use
1647 targetm.hard_regno_call_part_clobbered instead of
1648 HARD_REGNO_CALL_PART_CLOBBERED.
1649 * ira-conflicts.c (ira_build_conflicts): Likewise.
1650 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1651 * lra-constraints.c (need_for_call_save_p): Likewise.
1652 * lra-lives.c: Include target.h.
1653 (check_pseudos_live_through_calls): Use
1654 targetm.hard_regno_call_part_clobbered instead of
1655 HARD_REGNO_CALL_PART_CLOBBERED.
1656 * regcprop.c: Include target.h.
1657 (copyprop_hardreg_forward_1): Use
1658 targetm.hard_regno_call_part_clobbered instead of
1659 HARD_REGNO_CALL_PART_CLOBBERED.
1660 * reginfo.c (choose_hard_reg_mode): Likewise.
1661 * regrename.c (check_new_reg_p): Likewise.
1662 * reload.c (find_equiv_reg): Likewise.
1663 * reload1.c (emit_reload_insns): Likewise.
1664 * sched-deps.c (deps_analyze_insn): Likewise.
1665 * sel-sched.c (init_regs_for_mode): Likewise.
1666 (mark_unavailable_hard_regs): Likewise.
1667 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1668 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1669 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1671 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1672 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1673 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1675 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1677 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1678 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1679 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1681 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1682 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1683 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1685 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1686 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1688 * config/powerpcspe/powerpcspe.c
1689 (rs6000_hard_regno_call_part_clobbered): New function.
1690 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1691 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1692 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1694 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1695 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1696 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1699 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1700 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1702 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1703 Alan Hayward <alan.hayward@arm.com>
1704 David Sherwood <david.sherwood@arm.com>
1706 * rtl.h (subreg_memory_offset): Declare.
1707 * emit-rtl.c (subreg_memory_offset): New function.
1708 * expmed.c (store_bit_field_1): Use it.
1709 * expr.c (undefined_operand_subword_p): Likewise.
1710 * simplify-rtx.c (simplify_subreg): Likewise.
1712 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1714 PR rtl-optimization/57448
1717 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1718 using atomic_load pattern.
1719 (expand_atomic_store): Likewise.
1721 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1724 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1725 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1726 handling. Use replace_call_with_value with NULL instead of
1727 gsi_replace, unlink_stmt_vdef and release_defs.
1729 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1732 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1734 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1737 * rtl.h (emit_library_call_value_1): Declare.
1738 (emit_library_call): Replace declaration with a series of overloads.
1739 Remove the parameter count argument.
1740 (emit_library_call_value): Likewise.
1741 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1742 with an "rtx_mode_t *".
1743 (emit_library_call_value): Delete.
1744 (emit_library_call): Likewise.
1745 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1746 (asan_emit_allocas_unpoison): Likewise.
1747 * builtins.c (expand_builtin_powi): Likewise.
1748 (expand_asan_emit_allocas_unpoison): Likewise.
1749 * cfgexpand.c (expand_main_function): Likewise.
1750 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1751 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1752 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1753 * config/arm/arm.c (arm_trampoline_init): Likewise.
1754 (arm_call_tls_get_addr): Likewise.
1755 (arm_expand_divmod_libfunc): Likewise.
1756 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1757 (smulsi3_highpart): Likewise.
1758 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1759 (c6x_expand_compare): Likewise.
1760 (c6x_expand_movmem): Likewise.
1761 * config/frv/frv.c (frv_trampoline_init): Likewise.
1762 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1763 (ix86_expand_divmod_libfunc): Likewise.
1764 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1765 (ia64_expand_compare): Likewise.
1766 (ia64_profile_hook): Likewise.
1767 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1768 (nonlocal_goto): Likewise.
1769 (restore_stack_nonlocal): Likewise.
1770 * config/m32r/m32r.c (block_move_call): Likewise.
1771 (m32r_trampoline_init): Likewise.
1772 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1773 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1774 (m68k_call_m68k_read_tp): Likewise.
1775 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1776 (microblaze_expand_divide): Likewise.
1777 * config/mips/mips.h (mips_args): Likewise.
1778 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1779 (MIPS_ICACHE_SYNC): Likewise.
1780 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1781 (nios2_trampoline_init): Likewise.
1782 * config/pa/pa.c (hppa_tls_call): Likewise.
1783 (pa_trampoline_init): Likewise.
1784 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1785 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1786 (expand_strn_compare): Likewise.
1787 (rs6000_generate_compare): Likewise.
1788 (rs6000_expand_float128_convert): Likewise.
1789 (output_profile_hook): Likewise.
1790 (rs6000_trampoline_init): Likewise.
1791 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1792 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1793 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1794 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1795 (rs6000_generate_compare): Likewise.
1796 (rs6000_expand_float128_convert): Likewise.
1797 (output_profile_hook): Likewise.
1798 (rs6000_trampoline_init): Likewise.
1799 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1800 * config/sh/sh.c (sh_trampoline_init): Likewise.
1801 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1802 (sparc_emit_float_lib_cmp): Likewise.
1803 (sparc32_initialize_trampoline): Likewise.
1804 (sparc64_initialize_trampoline): Likewise.
1805 (sparc_profile_hook): Likewise.
1806 * config/spu/spu.c (ea_load_store): Likewise.
1807 * config/spu/spu.md (floatunssidf2): Likewise.
1808 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1809 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1810 * config/visium/visium.c (expand_block_move_4): Likewise.
1811 (expand_block_move_2): Likewise.
1812 (expand_block_move_1): Likewise.
1813 (expand_block_set_4): Likewise.
1814 (expand_block_set_2): Likewise.
1815 (expand_block_set_1): Likewise.
1816 (visium_trampoline_init): Likewise.
1817 (visium_profile_hook): Likewise.
1818 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1819 (xtensa_setup_frame_addresses): Likewise.
1820 (xtensa_trampoline_init): Likewise.
1821 * except.c (sjlj_emit_function_enter): Likewise.
1822 (sjlj_emit_function_exit): Likewise.
1823 * explow.c (allocate_dynamic_stack_space): Likewise.
1824 (probe_stack_range): Likewise.
1825 * expr.c (convert_mode_scalar): Likewise.
1826 * optabs.c (expand_binop): Likewise.
1827 (expand_twoval_binop_libfunc): Likewise.
1828 (expand_unop): Likewise.
1829 (prepare_cmp_insn): Likewise.
1830 (prepare_float_lib_cmp): Likewise.
1831 (expand_float): Likewise.
1832 (expand_fix): Likewise.
1833 (expand_fixed_convert): Likewise.
1834 (maybe_emit_sync_lock_test_and_set): Likewise.
1835 (expand_atomic_compare_and_swap): Likewise.
1836 (expand_mem_thread_fence): Likewise.
1837 (expand_atomic_fetch_op): Likewise.
1839 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1841 * doc/generic.texi (OpenACC): Adjust URL.
1842 * doc/invoke.texi (C Dialect Options): Ditto.
1844 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1846 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1847 predicate for operand 1. Add (m,<S>) constraint.
1848 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1849 Prevent memory operand 1 with register operand 2.
1851 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1853 PR rtl-optimization/82024
1854 * combine.c (try_combine): If the combination result is a PARALLEL,
1855 and we only need to retain the SET in there that would be placed
1856 at I2, check that we can place that at I3 instead, before doing so.
1858 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1861 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1863 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1864 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1867 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1868 Jeff Law <law@redhat.com>
1870 * varasm.c (bss_initializer_p): Do not put constants into .bss
1871 (categorize_decl_for_section): Handle bss_initializer_p returning
1872 false when DECL_INITIAL is NULL.
1874 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1877 * config/s390/s390.c (s390_can_inline_p): New function.
1879 2017-09-01 Jeff Law <law@redhat.com>
1881 PR tree-optimization/82052
1882 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1883 Always initialize the returned slot after a hash table miss
1884 when INSERT is true.
1886 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1888 * config/s390/s390.md (mem_signal_fence): Remove.
1889 * doc/md.texi (mem_signal_fence): Remove.
1890 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1892 * target-insns.def (mem_signal_fence): Remove.
1894 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1897 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1900 * asan.c (create_odr_indicator): Strip name encoding from assembler
1901 name before appending it after __odr_asan_.
1903 2017-09-01 Martin Liska <mliska@suse.cz>
1905 PR tree-optimization/82059
1906 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1907 frequency only when an edge is redirected.
1909 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1911 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1912 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1913 (arc_conditional_register_usage): Remove ARC600 lp_count
1915 (arc_file_start): Emit Tag_ARC_CPU_variation.
1916 (arc_can_use_doloop_p): New conditions to use ZOLs.
1917 (hwloop_fail): New function.
1918 (hwloop_optimize): Likewise.
1919 (hwloop_pattern_reg): Likewise.
1920 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1921 (arc_reorg_loops): New function, calls reorg_loops.
1922 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1923 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1925 (arc_loop_hazard): Remove function, functionality moved into
1927 (arc_hazard): Remove arc_loop_hazard call.
1928 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1929 into hwloop_optimize.
1930 (arc_label_align): Remove ZOL handling.
1931 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1932 * config/arc/arc.md (doloop_begin): Remove pattern.
1933 (doloop_begin_i): Likewise.
1934 (doloop_end_i): Likewise.
1935 (doloop_fallback): Likewise.
1936 (doloop_fallback_m): Likewise.
1937 (doloop_end): Reimplement expand.
1938 (arc_lp): New pattern for LP instruction.
1939 (loop_end): New pattern.
1940 (loop_fail): Likewise.
1941 (decrement_and_branch_until_zero): Likewise.
1942 * config/arc/arc.opt (mlpc-width): New option.
1943 * doc/invoke.texi (mlpc-width): Document option.
1945 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1947 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1948 (arc_ccfsm_advance): Fix checking for delay slots.
1949 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1951 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1953 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1955 (movhi_insn): Update store instruction constraint which are saving
1956 6-bit short immediates.
1957 (movsi_insn): Consider also short scaled load operations.
1958 (zero_extendhisi2_i): Use Usd constraint instead of T.
1959 (extendhisi2_i): Add q constraint.
1960 (arc_clzsi2): Add type and length attributes.
1961 (arc_ctzsi2): Likewise.
1962 * config/arc/constraints.md (Usc): Update constraint, the
1963 assembler can parse two relocations for a single instruction.
1965 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1967 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1968 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1970 2017-08-31 Olivier Hainque <hainque@adacore.com>
1972 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1973 match as powerpc-wrs-vxworks*.
1975 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1977 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1978 register constraint for by-element operand.
1979 (aarch64_mls_elt_merge<mode>): Likewise.
1981 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1983 * config/arc/arc.c (arc_can_follow_jump): Check for short
1986 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1988 * config.gcc: Use g.opt for arc.
1989 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1990 functionality moved to ...
1991 (legitimate_scaled_address_p): New function, ...here.
1992 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1993 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1994 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1996 (arc_override_options): Handle G option.
1997 (arc_output_pic_addr_const): Correct function definition.
1998 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1999 (arc_decl_anon_ns_mem_p): Delete.
2000 (arc_in_small_data_p): Overhaul this function to take into
2001 consideration the value given via G option.
2002 (arc_rewrite_small_data_1): Renamed and corrected old
2003 arc_rewrite_small_data function.
2004 (arc_rewrite_small_data): New function.
2005 (small_data_pattern): Don't use pic_offset_table_rtx.
2006 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2007 * config/arc/simdext.md (movmisalignv2hi): Use
2008 prepare_move_operands function.
2010 (movmisalign*): Likewise.
2011 * doc/invoke.texi (ARC options): Document -G option.
2013 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2015 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2017 * config/arc/arc.c (arc_print_operand): Output scalled address for
2018 sdata whenever is possible.
2019 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2020 load/stores are available.
2021 (compact_sda_memory_operand): Check for the alignment required by
2022 code density instructions.
2023 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2025 * config/arc/constraints.md (Usd): Update constraint.
2026 (Us0): New constraint.
2027 (Usc): Update constraint.
2029 2017-08-31 Richard Biener <rguenther@suse.de>
2032 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2035 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2037 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2038 Resize type_signature.
2040 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2041 Alan Hayward <alan.hayward@arm.com>
2042 David Sherwood <david.sherwood@arm.com>
2044 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2045 subregs whose inner modes can be stored in GPRs.
2046 (aarch64_classify_index): Likewise.
2048 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2049 Alan Hayward <alan.hayward@arm.com>
2050 David Sherwood <david.sherwood@arm.com>
2052 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2053 (V_INT_EQUIV): ...this.
2054 (v_cmp_result): Rename to...
2055 (v_int_equiv): ...this.
2056 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2057 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2058 (copysign<mode>3): Likewise.
2059 (aarch64_simd_bsl<mode>_internal): Likewise.
2060 (aarch64_simd_bsl<mode>): Likewise.
2061 (vec_cmp<mode><mode>): Likewise.
2062 (vcond<mode><mode>): Likewise.
2063 (vcond<v_cmp_mixed><mode>): Likewise.
2064 (vcondu<mode><v_cmp_mixed>): Likewise.
2065 (aarch64_cm<optab><mode>): Likewise.
2066 (aarch64_cmtst<mode>): Likewise.
2067 (aarch64_fac<optab><mode>): Likewise.
2068 (vec_perm_const<mode>): Likewise.
2069 (vcond_mask_<mode><v_cmp_result>): Rename to...
2070 (vcond_mask_<mode><v_int_equiv>): ...this.
2071 (vec_cmp<mode><v_cmp_result>): Rename to...
2072 (vec_cmp<mode><v_int_equiv>): ...this.
2074 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2075 Alan Hayward <alan.hayward@arm.com>
2076 David Sherwood <david.sherwood@arm.com>
2078 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2080 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2081 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2082 (UNSPEC_LD4_DREG): New unspecs.
2083 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2084 (aarch64_ld2<mode>_dreg_be): Replace with...
2085 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2087 (aarch64_ld3<mode>_dreg_le)
2088 (aarch64_ld3<mode>_dreg_be): Replace with...
2089 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2091 (aarch64_ld4<mode>_dreg_le)
2092 (aarch64_ld4<mode>_dreg_be): Replace with...
2093 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2096 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2098 PR tree-optimization/81987
2099 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2100 insert an initializer in a location not dominated by the stride
2103 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2105 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2106 on the entire header of the finally block in the fallthru case.
2108 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2110 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2112 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2114 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2115 rs6000_emit_move_from_cr and call renamed function.
2116 (rs6000_emit_prologue): Call renamed functions.
2117 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2118 movesi_from_cr, remove volatile CRs.
2120 2017-08-30 Jon Beniston <jon@beniston.com>
2121 Richard Biener <rguenther@suse.de>
2123 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2124 of VECTOR_MODE_P check.
2125 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2126 element vector types.
2128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2130 * df.h (df_read_modify_subreg_p): Remove in favor of...
2131 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2132 const_rtx instead of an rtx.
2133 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2134 * df-problems.c (df_word_lr_mark_ref): Likewise.
2135 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2136 (mark_pseudo_reg_dead): Likewise.
2137 (mark_ref_dead): Likewise.
2138 * reginfo.c (init_subregs_of_mode): Likewise.
2139 * sched-deps.c (sched_analyze_1): Likewise.
2140 * df-scan.c (df_def_record_1): Likewise.
2141 (df_uses_record): Likewise.
2142 (df_read_modify_subreg_p): Remove in favor of...
2143 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2144 const_rtx instead of an rtx.
2146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2147 Alan Hayward <alan.hayward@arm.com>
2148 David Sherwood <david.sherwood@arm.com>
2150 * rtl.h (partial_subreg_p): New function.
2151 * caller-save.c (save_call_clobbered_regs): Use it.
2152 * calls.c (expand_call): Likewise.
2153 * combine.c (combinable_i3pat): Likewise.
2154 (simplify_set): Likewise.
2155 (make_extraction): Likewise.
2156 (make_compound_operation_int): Likewise.
2157 (gen_lowpart_or_truncate): Likewise.
2158 (force_to_mode): Likewise.
2159 (make_field_assignment): Likewise.
2160 (reg_truncated_to_mode): Likewise.
2161 (record_truncated_value): Likewise.
2162 (move_deaths): Likewise.
2163 * cse.c (record_jump_cond): Likewise.
2164 (cse_insn): Likewise.
2165 * cselib.c (cselib_lookup_1): Likewise.
2166 * expmed.c (extract_bit_field_using_extv): Likewise.
2167 * function.c (assign_parm_setup_reg): Likewise.
2168 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2169 * ira-build.c (create_insn_allocnos): Likewise.
2170 * lra-coalesce.c (merge_pseudos): Likewise.
2171 * lra-constraints.c (match_reload): Likewise.
2172 (simplify_operand_subreg): Likewise.
2173 (curr_insn_transform): Likewise.
2174 * lra-lives.c (process_bb_lives): Likewise.
2175 * lra.c (new_insn_reg): Likewise.
2176 (lra_substitute_pseudo): Likewise.
2177 * regcprop.c (mode_change_ok): Likewise.
2178 (maybe_mode_change): Likewise.
2179 (copyprop_hardreg_forward_1): Likewise.
2180 * reload.c (push_reload): Likewise.
2181 (find_reloads): Likewise.
2182 (find_reloads_subreg_address): Likewise.
2183 * reload1.c (alter_reg): Likewise.
2184 (eliminate_regs_1): Likewise.
2185 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2187 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2189 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2190 back to if statements, including unpack.
2192 2017-08-30 Martin Liska <mliska@suse.cz>
2195 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2197 (func_checker::compare_asm_inputs_outputs): ... this function.
2198 (func_checker::compare_gimple_asm): Use the function to compare
2199 also ASM constrains.
2200 * ipa-icf-gimple.h: Rename the function.
2202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2203 Alan Hayward <alan.hayward@arm.com>
2204 David Sherwood <david.sherwood@arm.com>
2206 * coretypes.h (complex_mode): New type.
2207 * gdbhooks.py (build_pretty_printer): Handle it.
2208 * machmode.h (complex_mode): New class.
2209 (complex_mode::includes_p): New function.
2210 (is_complex_int_mode): Likewise.
2211 (is_complex_float_mode): Likewise.
2212 * genmodes.c (get_mode_class): Handle complex mode classes.
2213 * function.c (expand_function_end): Use is_complex_int_mode.
2215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2216 Alan Hayward <alan.hayward@arm.com>
2217 David Sherwood <david.sherwood@arm.com>
2219 * coretypes.h (scalar_mode_pod): New typedef.
2220 * gdbhooks.py (build_pretty_printer): Handle it.
2221 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2222 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2223 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2224 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2228 Alan Hayward <alan.hayward@arm.com>
2229 David Sherwood <david.sherwood@arm.com>
2231 * machmode.h (mode_for_vector): Take a scalar_mode instead
2233 * stor-layout.c (mode_for_vector): Likewise.
2234 * explow.c (promote_mode): Use as_a <scalar_mode>.
2235 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
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 * target.def (preferred_simd_mode): Take a scalar_mode
2242 instead of a machine_mode.
2243 * targhooks.h (default_preferred_simd_mode): Likewise.
2244 * targhooks.c (default_preferred_simd_mode): Likewise.
2245 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2246 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2247 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2248 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2249 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2250 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2251 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2252 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2254 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2255 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2256 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2257 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2258 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2259 * doc/tm.texi: Regenerate.
2260 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2264 Alan Hayward <alan.hayward@arm.com>
2265 David Sherwood <david.sherwood@arm.com>
2267 * target.def (scalar_mode_supported_p): Take a scalar_mode
2268 instead of a machine_mode.
2269 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2270 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2271 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2272 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2273 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2274 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2275 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2276 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2277 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2278 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2279 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2280 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2281 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2282 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2284 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2285 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2286 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2287 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2288 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2290 * doc/tm.texi: Regenerate.
2292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2293 Alan Hayward <alan.hayward@arm.com>
2294 David Sherwood <david.sherwood@arm.com>
2296 * coretypes.h (opt_scalar_mode): New typedef.
2297 * gdbhooks.py (build_pretty_printers): Handle it.
2298 * machmode.h (mode_iterator::get_2xwider): Add overload for
2300 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2302 * expr.c (convert_mode_scalar): Likewise.
2303 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2304 * optabs.c (expand_float): Likewise.
2305 (expand_fix): Likewise.
2306 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2309 Alan Hayward <alan.hayward@arm.com>
2310 David Sherwood <david.sherwood@arm.com>
2312 * optabs.c (expand_float): Explicitly check for scalars before
2313 using a branching expansion.
2314 (expand_fix): Likewise.
2316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2317 Alan Hayward <alan.hayward@arm.com>
2318 David Sherwood <david.sherwood@arm.com>
2320 * expr.c (convert_mode): Split scalar handling out into...
2321 (convert_mode_scalar): ...this new function. Treat the modes
2324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2325 Alan Hayward <alan.hayward@arm.com>
2326 David Sherwood <david.sherwood@arm.com>
2328 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2330 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2333 Alan Hayward <alan.hayward@arm.com>
2334 David Sherwood <david.sherwood@arm.com>
2336 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2337 rather than a machine_mode.
2338 (fixed_from_string): Likewise.
2339 (fixed_convert): Likewise.
2340 (fixed_convert_from_int): Likewise.
2341 (fixed_convert_from_real): Likewise.
2342 (real_convert_from_fixed): Likewise.
2343 * fixed-value.c (fixed_from_double_int): Likewise.
2344 (fixed_from_string): Likewise.
2345 (fixed_convert): Likewise.
2346 (fixed_convert_from_int): Likewise.
2347 (fixed_convert_from_real): Likewise.
2348 (real_convert_from_fixed): Likewise.
2349 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2352 Alan Hayward <alan.hayward@arm.com>
2353 David Sherwood <david.sherwood@arm.com>
2355 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2356 of separate mode class checks. Do not allow vector modes here.
2357 (immed_wide_int_const): Use as_a <scalar_mode>.
2358 * explow.c (trunc_int_for_mode): Likewise.
2359 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2360 (wi::shwi): Likewise.
2361 (wi::min_value): Likewise.
2362 (wi::max_value): Likewise.
2363 * dwarf2out.c (loc_descriptor): Likewise.
2364 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2368 Alan Hayward <alan.hayward@arm.com>
2369 David Sherwood <david.sherwood@arm.com>
2371 * tree.h (SCALAR_TYPE_MODE): New macro.
2372 * expr.c (expand_expr_addr_expr_1): Use it.
2373 (expand_expr_real_2): Likewise.
2374 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2375 (fold_convert_const_fixed_from_int): Likewise.
2376 (fold_convert_const_fixed_from_real): Likewise.
2377 (native_encode_fixed): Likewise
2378 (native_encode_complex): Likewise
2379 (native_encode_vector): Likewise.
2380 (native_interpret_fixed): Likewise.
2381 (native_interpret_real): Likewise.
2382 (native_interpret_complex): Likewise.
2383 (native_interpret_vector): Likewise.
2384 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2385 (simd_clone_adjust_argument_types): Likewise.
2386 (simd_clone_init_simd_arrays): Likewise.
2387 (simd_clone_adjust): Likewise.
2388 * stor-layout.c (layout_type): Likewise.
2389 * tree.c (build_minus_one_cst): Likewise.
2390 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2391 * tree-inline.c (estimate_move_cost): Likewise.
2392 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2393 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2394 (vectorizable_reduction): Likewise.
2395 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2396 (vect_recog_mixed_size_cond_pattern): Likewise.
2397 (check_bool_pattern): Likewise.
2398 (adjust_bool_pattern): Likewise.
2399 (search_type_for_mask_1): Likewise.
2400 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2401 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2402 (vectorizable_load): Likewise.
2403 (vectorizable_store): Likewise.
2404 * ubsan.c (ubsan_encode_value): Likewise.
2405 * varasm.c (output_constant): 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 * coretypes.h (scalar_mode): New class.
2412 * machmode.h (scalar_mode): Likewise.
2413 (scalar_mode::includes_p): New function.
2414 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2415 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2416 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2417 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2418 * expmed.c (store_bit_field_1): Likewise.
2419 (extract_bit_field_1): Likewise.
2420 * expr.c (write_complex_part): Likewise.
2421 (read_complex_part): Likewise.
2422 (emit_move_complex_push): Likewise.
2423 (expand_expr_real_2): Likewise.
2424 * function.c (assign_parm_setup_reg): Likewise.
2425 (assign_parms_unsplit_complex): Likewise.
2426 * optabs.c (expand_binop): Likewise.
2427 * rtlanal.c (subreg_get_info): Likewise.
2428 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2429 * varasm.c (output_constant_pool_2): Likewise.
2431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2432 Alan Hayward <alan.hayward@arm.com>
2433 David Sherwood <david.sherwood@arm.com>
2435 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2437 (expmed_mult_highpart_optab): Likewise.
2438 (expmed_mult_highpart): Likewise.
2440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2441 Alan Hayward <alan.hayward@arm.com>
2442 David Sherwood <david.sherwood@arm.com>
2444 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2445 instead of a machine_mode.
2446 (builtin_memset_read_str): Likewise.
2447 * builtins.c (c_readstr): Likewise.
2448 (builtin_memcpy_read_str): Likewise.
2449 (builtin_strncpy_read_str): Likewise.
2450 (builtin_memset_read_str): Likewise.
2451 (builtin_memset_gen_str): Likewise.
2452 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2453 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2454 instead of a machine_mode.
2455 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2457 (make_extraction): Likewise.
2458 (try_widen_shift_mode): Take and return scalar_int_modes instead
2460 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2461 a scalar_int_mode instead of a machine_mode.
2462 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2463 (avr_addr_space_pointer_mode): Likewise.
2464 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2465 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2466 (msp430_unwind_word_mode): Likewise.
2467 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2468 (spu_addr_space_pointer_mode): Likewise.
2469 (spu_addr_space_address_mode): Likewise.
2470 (spu_libgcc_cmp_return_mode): Likewise.
2471 (spu_libgcc_shift_count_mode): Likewise.
2472 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2473 (rl78_addr_space_pointer_mode): Likewise.
2474 (fl78_unwind_word_mode): Likewise.
2475 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2477 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2478 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2479 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2480 (mips_valid_pointer_mode): Likewise.
2481 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2482 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2483 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2485 (ft32_addr_space_address_mode): Likewise.
2486 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2487 scalar_int_mode instead of a machine_mode.
2488 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2490 (m32c_addr_space_address_mode): Likewise.
2491 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2492 (rs6000_eh_return_filter_mode): Likewise.
2493 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2494 (rs6000_eh_return_filter_mode): Likewise.
2495 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2496 (s390_libgcc_shift_count_mode): Likewise.
2497 (s390_unwind_word_mode): Likewise.
2498 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2500 * target.def (mode_rep_extended): Likewise.
2501 (valid_pointer_mode): Likewise.
2502 (addr_space.valid_pointer_mode): Likewise.
2503 (eh_return_filter_mode): Return a scalar_int_mode rather than
2505 (libgcc_cmp_return_mode): Likewise.
2506 (libgcc_shift_count_mode): Likewise.
2507 (unwind_word_mode): Likewise.
2508 (addr_space.pointer_mode): Likewise.
2509 (addr_space.address_mode): Likewise.
2510 * doc/tm.texi: Regenerate.
2511 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2513 (do_jump): Use scalar_int_mode for local variables.
2514 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2515 rather than a machine_mode.
2516 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2517 (scompare_loc_descriptor_wide): Likewise.
2518 (scompare_loc_descriptor_narrow): Likewise.
2519 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2521 * except.c (sjlj_emit_dispatch_table): Likewise.
2522 (expand_builtin_eh_copy_values): Likewise.
2523 * explow.c (convert_memory_address_addr_space_1): Likewise.
2524 Take a scalar_int_mode rather than a machine_mode.
2525 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2526 than a machine_mode.
2527 (memory_address_addr_space): Use scalar_int_mode for local variables.
2528 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2529 rather than a machine_mode.
2530 * expmed.c (mask_rtx): Likewise.
2531 (init_expmed_one_conv): Likewise.
2532 (expand_mult_highpart_adjust): Likewise.
2533 (extract_high_half): Likewise.
2534 (expmed_mult_highpart_optab): Likewise.
2535 (expmed_mult_highpart): Likewise.
2536 (expand_smod_pow2): Likewise.
2537 (expand_sdiv_pow2): Likewise.
2538 (emit_store_flag_int): Likewise.
2539 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2541 (extract_low_bits): Likewise.
2542 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2544 * expr.c (pieces_addr::adjust): Likewise.
2545 (can_store_by_pieces): Likewise.
2546 (store_by_pieces): Likewise.
2547 (clear_by_pieces_1): Likewise.
2548 (expand_expr_addr_expr_1): Likewise.
2549 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2550 (expand_expr_real_1): Likewise.
2551 (try_casesi): Likewise.
2552 * final.c (shorten_branches): Likewise.
2553 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2554 type of "mode" to machine_mode.
2555 * internal-fn.c (expand_arith_overflow_result_store): Take a
2556 scalar_int_mode rather than a machine_mode.
2557 (expand_mul_overflow): Use scalar_int_mode for local variables.
2558 * loop-doloop.c (doloop_modify): Likewise.
2559 (doloop_optimize): Likewise.
2560 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2561 than a machine_mode.
2562 (expand_doubleword_shift_condmove): Likewise.
2563 (expand_doubleword_shift): Likewise.
2564 (expand_doubleword_clz): Likewise.
2565 (expand_doubleword_popcount): Likewise.
2566 (expand_doubleword_parity): Likewise.
2567 (expand_absneg_bit): Use scalar_int_mode for local variables.
2568 (prepare_float_lib_cmp): Likewise.
2569 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2570 rather than a machine_mode.
2571 (convert_memory_address_addr_space): Likewise.
2572 (get_mode_bounds): Likewise.
2573 (get_address_mode): Return a scalar_int_mode rather than a
2575 * rtlanal.c (get_address_mode): Likewise.
2576 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2577 than a machine_mode.
2578 * targhooks.c (default_mode_rep_extended): Likewise.
2579 (default_valid_pointer_mode): Likewise.
2580 (default_addr_space_valid_pointer_mode): Likewise.
2581 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2582 than a machine_mode.
2583 (default_libgcc_cmp_return_mode): Likewise.
2584 (default_libgcc_shift_count_mode): Likewise.
2585 (default_unwind_word_mode): Likewise.
2586 (default_addr_space_pointer_mode): Likewise.
2587 (default_addr_space_address_mode): Likewise.
2588 * targhooks.h (default_eh_return_filter_mode): Likewise.
2589 (default_libgcc_cmp_return_mode): Likewise.
2590 (default_libgcc_shift_count_mode): Likewise.
2591 (default_unwind_word_mode): Likewise.
2592 (default_addr_space_pointer_mode): Likewise.
2593 (default_addr_space_address_mode): Likewise.
2594 (default_mode_rep_extended): Take a scalar_int_mode rather than
2596 (default_valid_pointer_mode): Likewise.
2597 (default_addr_space_valid_pointer_mode): Likewise.
2598 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2600 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2601 rather than a machine_mode.
2602 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2603 for local variables.
2604 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2605 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2606 than a machine_mode.
2608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2609 Alan Hayward <alan.hayward@arm.com>
2610 David Sherwood <david.sherwood@arm.com>
2612 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2613 the mode argument to scalar_int_mode.
2614 (do_jump_by_parts_zero_rtx): Likewise.
2615 (do_jump_by_parts_equality_rtx): Likewise.
2616 (do_jump_by_parts_greater): Take a mode argument.
2617 (do_jump_by_parts_equality): Likewise.
2618 (do_jump_1): Update calls accordingly.
2620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2621 Alan Hayward <alan.hayward@arm.com>
2622 David Sherwood <david.sherwood@arm.com>
2624 * is-a.h (safe_dyn_cast): New function.
2625 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2626 (jump_table_for_label): Likewise.
2627 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2628 instead of an rtx_insn *.
2629 (shorten_branches): Use dyn_cast instead of LABEL_P and
2630 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2631 rtx_jump_table_data::get_data_mode.
2632 (final_scan_insn): Likewise.
2634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2635 Alan Hayward <alan.hayward@arm.com>
2636 David Sherwood <david.sherwood@arm.com>
2638 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2639 trying to combine a full-register integer set with a subreg
2642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2643 Alan Hayward <alan.hayward@arm.com>
2644 David Sherwood <david.sherwood@arm.com>
2646 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2647 that is always either address_mode or pointer_mode.
2649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2650 Alan Hayward <alan.hayward@arm.com>
2651 David Sherwood <david.sherwood@arm.com>
2653 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2654 when the two are known to be equal.
2656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * simplify-rtx.c (simplify_const_unary_operation): Use
2661 is_a <scalar_int_mode> instead of checking for a nonzero
2662 precision. Forcibly convert op_mode to a scalar_int_mode
2663 in that case. More clearly differentiate the operand and
2664 result modes and use the former when deciding what the value
2665 of a count-bits operation should be. Use is_int_mode instead
2666 of checking for a MODE_INT. Remove redundant check for whether
2667 this mode has a zero precision.
2669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2670 Alan Hayward <alan.hayward@arm.com>
2671 David Sherwood <david.sherwood@arm.com>
2673 * optabs.c (widen_leading): Change the type of the mode argument
2674 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2675 (widen_bswap): Likewise.
2676 (expand_parity): Likewise.
2677 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2678 (expand_ffs): Likewise.
2679 (epand_unop): Check for scalar integer modes before calling the
2682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2683 Alan Hayward <alan.hayward@arm.com>
2684 David Sherwood <david.sherwood@arm.com>
2686 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2688 (expand_expr_real_1): Update call accordingly.
2690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2691 Alan Hayward <alan.hayward@arm.com>
2692 David Sherwood <david.sherwood@arm.com>
2694 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2695 value_mode arguments. Use scalar_int_mode internally.
2696 (store_bit_field_1): Rename the new integer mode from imode
2697 to op0_mode and use it instead of GET_MODE (op0). Update calls
2698 to store_split_bit_field, store_bit_field_using_insv and
2699 store_fixed_bit_field.
2700 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2701 Use scalar_int_mode internally. Use a bit count rather than a mode
2702 when calculating the largest bit size for get_best_mode.
2703 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2704 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2705 Remove assertion that OP0 has a scalar integer mode.
2706 (store_split_bit_field): Add op0_mode and value_mode arguments.
2707 Update calls to extract_fixed_bit_field.
2708 (extract_bit_field_using_extv): Add an op0_mode argument.
2709 Use scalar_int_mode internally.
2710 (extract_bit_field_1): Rename the new integer mode from imode to
2711 op0_mode and use it instead of GET_MODE (op0). Update calls to
2712 extract_split_bit_field, extract_bit_field_using_extv and
2713 extract_fixed_bit_field.
2714 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2715 to extract_split_bit_field and extract_fixed_bit_field_1.
2716 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2717 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2719 (extract_split_bit_field): Add an op0_mode argument. Update call
2720 to extract_fixed_bit_field.
2722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2723 Alan Hayward <alan.hayward@arm.com>
2724 David Sherwood <david.sherwood@arm.com>
2726 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2727 * explow.c (hard_function_value): Likewise.
2728 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2729 convert_to_mode call outside the loop.
2730 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2731 for the mode iterator. Require the mode specified by max_pieces
2733 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2735 (copy_blkmode_to_reg): Likewise.
2736 (set_storage_via_setmem): Likewise.
2737 * optabs.c (prepare_cmp_insn): Likewise.
2738 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2739 * stor-layout.c (finish_bitfield_representative): Likewise.
2741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2742 Alan Hayward <alan.hayward@arm.com>
2743 David Sherwood <david.sherwood@arm.com>
2745 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2746 * expr.c (convert_move): Use them.
2747 (convert_modes): Likewise.
2748 (store_expr_with_bounds): Likewise.
2750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2751 Alan Hayward <alan.hayward@arm.com>
2752 David Sherwood <david.sherwood@arm.com>
2754 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2755 parameter for the mode of "x". Remove the "known_x", "known_mode"
2756 and "known_ret" arguments. Change the type of the mode argument
2758 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2759 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2760 (reg_num_sign_bit_copies_for_combine): Likewise.
2761 * rtlanal.c (nonzero_bits1): Likewise.
2762 (num_sign_bit_copies1): Likewise.
2763 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2764 (reg_num_sign_bit_copies_general): Likewise.
2765 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2766 (reg_nonzero_bits_general): Likewise.
2768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2769 Alan Hayward <alan.hayward@arm.com>
2770 David Sherwood <david.sherwood@arm.com>
2772 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2773 than in subroutines. Return 1 for non-integer modes.
2774 (cached_num_sign_bit_copies): Change the type of the mode parameter
2776 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2777 classes. Handle CONST_INT_P first and then check whether X also
2778 has a scalar integer mode. Check the same thing for inner registers
2779 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2786 in subroutines. Return the mode mask for non-integer modes.
2787 (cached_nonzero_bits): Change the type of the mode parameter
2789 (nonzero_bits1): Likewise. Remove early exit for other mode
2790 classes. Handle CONST_INT_P first and then check whether X
2791 also has a scalar integer mode.
2793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2794 Alan Hayward <alan.hayward@arm.com>
2795 David Sherwood <david.sherwood@arm.com>
2797 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2798 Return a scalar_int_mode and assert that the size is greater than
2800 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2801 (op_by_pieces_d::op_by_pieces_d): Likewise.
2802 (op_by_pieces_d::run): Likewise.
2803 (can_store_by_pieces): Likewise.
2805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2806 Alan Hayward <alan.hayward@arm.com>
2807 David Sherwood <david.sherwood@arm.com>
2809 * combine.c (extract_left_shift): Add a mode argument and update
2811 (make_compound_operation_int): Change the type of the mode parameter
2812 to scalar_int_mode and update the call to extract_left_shift.
2814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2815 Alan Hayward <alan.hayward@arm.com>
2816 David Sherwood <david.sherwood@arm.com>
2818 * combine.c (simplify_and_const_int): Change the type of the mode
2819 parameter to scalar_int_mode.
2820 (simplify_and_const_int_1): Likewise. Update recursive call.
2822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2823 Alan Hayward <alan.hayward@arm.com>
2824 David Sherwood <david.sherwood@arm.com>
2826 * combine.c (simplify_compare_const): Check that the mode is a
2827 scalar_int_mode (rather than VOIDmode) before testing its
2829 (simplify_comparison): Move COMPARISON_P handling out of the
2830 loop and restrict the latter part of the loop to scalar_int_modes.
2831 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2832 and when considering SUBREG_REGs. Use is_int_mode instead of
2833 checking GET_MODE_CLASS against MODE_INT.
2835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2836 Alan Hayward <alan.hayward@arm.com>
2837 David Sherwood <david.sherwood@arm.com>
2839 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2840 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2841 shift_unit_mode and for modes involved in scalar shifts.
2843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2844 Alan Hayward <alan.hayward@arm.com>
2845 David Sherwood <david.sherwood@arm.com>
2847 * combine.c (force_int_to_mode): New function, split out from...
2848 (force_to_mode): ...here. Keep xmode up-to-date and use it
2849 instead of GET_MODE (x).
2851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2852 Alan Hayward <alan.hayward@arm.com>
2853 David Sherwood <david.sherwood@arm.com>
2855 * optabs-query.h (extraction_insn::struct_mode): Change type to
2856 opt_scalar_int_mode and update comment.
2857 (extraction_insn::field_mode): Change type to scalar_int_mode.
2858 (extraction_insn::pos_mode): Likewise.
2859 * combine.c (make_extraction): Update accordingly.
2860 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2861 (get_optab_extraction_insn): Likewise.
2862 * recog.c (simplify_while_replacing): Likewise.
2863 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2864 parameter to opt_scalar_int_mode.
2866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2867 Alan Hayward <alan.hayward@arm.com>
2868 David Sherwood <david.sherwood@arm.com>
2870 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2871 to a scalar_int_mode instead of a machine_mode.
2872 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2873 (get_best_mode): Return a boolean and use a pointer argument to store
2874 the selected mode. Replace the limit mode parameter with a bit limit.
2875 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2876 for the values returned by bit_field_mode_iterator::next_mode.
2877 (store_bit_field): Update call to get_best_mode.
2878 (store_fixed_bit_field): Likewise.
2879 (extract_fixed_bit_field): Likewise.
2880 * expr.c (optimize_bitfield_assignment_op): Likewise.
2881 * fold-const.c (optimize_bit_field_compare): Likewise.
2882 (fold_truth_andor_1): Likewise.
2883 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2884 Update for new type of m_mode.
2885 (get_best_mode): As above.
2887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2888 Alan Hayward <alan.hayward@arm.com>
2889 David Sherwood <david.sherwood@arm.com>
2891 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2892 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2893 (store_bit_field): Check is_a <scalar_int_mode> before calling
2894 strict_volatile_bitfield_p.
2895 (extract_bit_field): Likewise.
2897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2898 Alan Hayward <alan.hayward@arm.com>
2899 David Sherwood <david.sherwood@arm.com>
2901 * target.def (cstore_mode): Return a scalar_int_mode.
2902 * doc/tm.texi: Regenerate.
2903 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2904 * targhooks.h (default_cstore_mode): Likewise.
2905 * targhooks.c (default_cstore_mode): Likewise, using a forced
2907 * expmed.c (emit_cstore): Expect the target of the cstore to be
2910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2911 Alan Hayward <alan.hayward@arm.com>
2912 David Sherwood <david.sherwood@arm.com>
2914 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2916 (niter_desc): Likewise mode.
2917 (iv_analyze): Add a mode parameter.
2919 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2920 and change its type to scalar_int_mode.
2921 * loop-iv.c: Update commentary at head of file.
2922 (iv_constant): Pass the mode paraeter before the rtx it describes
2923 and change its type to scalar_int_mode. Remove VOIDmode handling.
2924 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2925 (iv_extend): Likewise.
2926 (shorten_into_mode): Likewise.
2927 (iv_add): Use scalar_int_mode.
2928 (iv_mult): Likewise.
2929 (iv_shift): Likewise.
2930 (canonicalize_iv_subregs): Likewise.
2931 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2932 it describes and change its mode to scalar_int_mode. Also change
2933 the type of the returned inner_mode to scalar_int_mode.
2934 (get_biv_step): Likewise, turning outer_mode from a pointer
2935 into a direct parameter. Update call to get_biv_step_1.
2936 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2937 iv_constant and get_biv_step.
2938 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2939 and change its type to scalar_int_mode. Don't initialise iv->mode
2940 to VOIDmode and remove later checks for its still being VOIDmode.
2941 Update calls to iv_analyze_op and iv_analyze_expr. Check
2942 is_a <scalar_int_mode> when changing the mode under consideration.
2943 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2944 Update call to iv_analyze_expr.
2945 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2946 inner register is not also a scalar_int_mode. Update call to
2948 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2949 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2950 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2951 separate mode class checks. Update calls to iv_analyze. Remove
2952 fix-up of VOIDmodes after iv_analyze_biv.
2953 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2954 don't have a scalar_int_mode. Update call to biv_p.
2956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2957 Alan Hayward <alan.hayward@arm.com>
2958 David Sherwood <david.sherwood@arm.com>
2960 * cfgexpand.c (convert_debug_memory_address): Use
2961 as_a <scalar_int_mode>.
2962 * combine.c (expand_compound_operation): Likewise.
2963 (make_extraction): Likewise.
2964 (change_zero_ext): Likewise.
2965 (simplify_comparison): Likewise.
2966 * cse.c (cse_insn): Likewise.
2967 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2968 (mem_loc_descriptor): Likewise.
2969 (loc_descriptor): Likewise.
2970 * expmed.c (init_expmed_one_mode): Likewise.
2971 (synth_mult): Likewise.
2972 (emit_store_flag_1): Likewise.
2973 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2974 of a comparison with size.
2975 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2976 (reduce_to_bit_field_precision): Likewise.
2977 * function.c (expand_function_end): Likewise.
2978 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2979 * loop-doloop.c (doloop_modify): Likewise.
2980 * optabs.c (expand_binop): Likewise.
2981 (expand_unop): Likewise.
2982 (expand_copysign_absneg): Likewise.
2983 (prepare_cmp_insn): Likewise.
2984 (maybe_legitimize_operand): Likewise.
2985 * recog.c (const_scalar_int_operand): Likewise.
2986 * rtlanal.c (get_address_mode): Likewise.
2987 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2988 (simplify_cond_clz_ctz): Likewise.
2989 * tree-nested.c (get_nl_goto_field): Likewise.
2990 * tree.c (build_vector_type_for_mode): Likewise.
2991 * var-tracking.c (use_narrower_mode): Likewise.
2993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2994 Alan Hayward <alan.hayward@arm.com>
2995 David Sherwood <david.sherwood@arm.com>
2997 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2998 * builtins.c (expand_builtin_signbit): Use it.
2999 * cfgexpand.c (expand_debug_expr): Likewise.
3000 * dojump.c (do_jump): Likewise.
3001 (do_compare_and_jump): Likewise.
3002 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3003 * expmed.c (make_tree): Likewise.
3004 * expr.c (expand_expr_real_2): Likewise.
3005 (expand_expr_real_1): Likewise.
3006 (try_casesi): Likewise.
3007 * fold-const-call.c (fold_const_call_ss): Likewise.
3008 * fold-const.c (unextend): Likewise.
3009 (extract_muldiv_1): Likewise.
3010 (fold_single_bit_test): Likewise.
3011 (native_encode_int): Likewise.
3012 (native_encode_string): Likewise.
3013 (native_interpret_int): Likewise.
3014 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3015 * internal-fn.c (expand_addsub_overflow): Likewise.
3016 (expand_neg_overflow): Likewise.
3017 (expand_mul_overflow): Likewise.
3018 (expand_arith_overflow): Likewise.
3019 * match.pd: Likewise.
3020 * stor-layout.c (layout_type): Likewise.
3021 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3022 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3023 * tree-ssanames.c (get_range_info): Likewise.
3024 * tree-switch-conversion.c (array_value_type) Likewise.
3025 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3026 (vect_recog_divmod_pattern): Likewise.
3027 (vect_recog_mixed_size_cond_pattern): Likewise.
3028 * tree-vrp.c (extract_range_basic): Likewise.
3029 (simplify_float_conversion_using_ranges): Likewise.
3030 * tree.c (int_fits_type_p): Likewise.
3031 * ubsan.c (instrument_bool_enum_load): Likewise.
3032 * varasm.c (mergeable_string_section): Likewise.
3033 (narrowing_initializer_constant_valid_p): Likewise.
3034 (output_constant): Likewise.
3036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3037 Alan Hayward <alan.hayward@arm.com>
3038 David Sherwood <david.sherwood@arm.com>
3040 * machmode.h (NARROWEST_INT_MODE): New macro.
3041 * expr.c (alignment_for_piecewise_move): Use it instead of
3042 GET_CLASS_NARROWEST_MODE (MODE_INT).
3043 (push_block): Likewise.
3044 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3046 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3049 Alan Hayward <alan.hayward@arm.com>
3050 David Sherwood <david.sherwood@arm.com>
3052 * postreload.c (move2add_valid_value_p): Change the type of the
3053 mode parameter to scalar_int_mode.
3054 (move2add_use_add2_insn): Add a mode parameter and use it instead
3056 (move2add_use_add3_insn): Likewise.
3057 (reload_cse_move2add): Update accordingly.
3059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3060 Alan Hayward <alan.hayward@arm.com>
3061 David Sherwood <david.sherwood@arm.com>
3063 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3065 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3066 * optabs.c (expand_unop): Likewise.
3068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3069 Alan Hayward <alan.hayward@arm.com>
3070 David Sherwood <david.sherwood@arm.com>
3072 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3073 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3074 (popcount_loc_descriptor): Likewise.
3075 (bswap_loc_descriptor): Likewise.
3076 (rotate_loc_descriptor): Likewise.
3077 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3078 calling the functions above.
3080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3081 Alan Hayward <alan.hayward@arm.com>
3082 David Sherwood <david.sherwood@arm.com>
3084 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3086 (try_combine): Likewise.
3087 (simplify_if_then_else): Likewise.
3088 * cse.c (cse_insn): Likewise.
3089 * dwarf2out.c (mem_loc_descriptor): Likewise.
3090 * emit-rtl.c (gen_lowpart_common): Likewise.
3091 * simplify-rtx.c (simplify_truncation): Likewise.
3092 (simplify_binary_operation_1): Likewise.
3093 (simplify_const_relational_operation): Likewise.
3094 (simplify_ternary_operation): Likewise.
3095 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3098 Alan Hayward <alan.hayward@arm.com>
3099 David Sherwood <david.sherwood@arm.com>
3101 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3102 * reload.c (push_reload): Likewise.
3103 (find_reloads): Likewise.
3105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3106 Alan Hayward <alan.hayward@arm.com>
3107 David Sherwood <david.sherwood@arm.com>
3109 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3110 (make_compound_operation_int): Likewise.
3111 (change_zero_ext): Likewise.
3112 * expr.c (convert_move): Likewise.
3113 (convert_modes): Likewise.
3114 * fwprop.c (forward_propagate_subreg): Likewise.
3115 * loop-iv.c (get_biv_step_1): Likewise.
3116 * optabs.c (widen_operand): Likewise.
3117 * postreload.c (move2add_valid_value_p): Likewise.
3118 * recog.c (simplify_while_replacing): Likewise.
3119 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3120 (simplify_binary_operation_1): Likewise. Remove redundant
3121 mode equality check.
3123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3124 Alan Hayward <alan.hayward@arm.com>
3125 David Sherwood <david.sherwood@arm.com>
3127 * combine.c (combine_simplify_rtx): Add checks for
3128 is_a <scalar_int_mode>.
3129 (simplify_if_then_else): Likewise.
3130 (make_field_assignment): Likewise.
3131 (simplify_comparison): Likewise.
3132 * ifcvt.c (noce_try_bitop): Likewise.
3133 * loop-invariant.c (canonicalize_address_mult): Likewise.
3134 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3137 Alan Hayward <alan.hayward@arm.com>
3138 David Sherwood <david.sherwood@arm.com>
3140 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3141 is_a <scalar_int_mode> instead of != BLKmode.
3143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3144 Alan Hayward <alan.hayward@arm.com>
3145 David Sherwood <david.sherwood@arm.com>
3147 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3148 instead of != VOIDmode.
3149 * combine.c (if_then_else_cond): Likewise.
3150 (change_zero_ext): Likewise.
3151 * dwarf2out.c (mem_loc_descriptor): Likewise.
3152 (loc_descriptor): Likewise.
3153 * rtlanal.c (canonicalize_condition): Likewise.
3154 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3157 Alan Hayward <alan.hayward@arm.com>
3158 David Sherwood <david.sherwood@arm.com>
3160 * simplify-rtx.c (simplify_binary_operation_1): Use
3161 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3164 Alan Hayward <alan.hayward@arm.com>
3165 David Sherwood <david.sherwood@arm.com>
3167 * wide-int.h (int_traits<unsigned char>) New class.
3168 (int_traits<unsigned short>) Likewise.
3169 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3170 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3172 * combine.c (set_nonzero_bits_and_sign_copies): Use
3173 is_a <scalar_int_mode>.
3174 (find_split_point): Likewise.
3175 (combine_simplify_rtx): Likewise.
3176 (simplify_logical): Likewise.
3177 (expand_compound_operation): Likewise.
3178 (expand_field_assignment): Likewise.
3179 (make_compound_operation): Likewise.
3180 (extended_count): Likewise.
3181 (change_zero_ext): Likewise.
3182 (simplify_comparison): Likewise.
3183 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3184 (ucompare_loc_descriptor): Likewise.
3185 (minmax_loc_descriptor): Likewise.
3186 (mem_loc_descriptor): Likewise.
3187 (loc_descriptor): Likewise.
3188 * expmed.c (init_expmed_one_mode): Likewise.
3189 * lra-constraints.c (lra_constraint_offset): Likewise.
3190 * optabs.c (prepare_libcall_arg): Likewise.
3191 * postreload.c (move2add_note_store): Likewise.
3192 * reload.c (operands_match_p): Likewise.
3193 * rtl.h (load_extend_op): Likewise.
3194 * rtlhooks.c (gen_lowpart_general): Likewise.
3195 * simplify-rtx.c (simplify_truncation): Likewise.
3196 (simplify_unary_operation_1): Likewise.
3197 (simplify_binary_operation_1): Likewise.
3198 (simplify_const_binary_operation): Likewise.
3199 (simplify_const_relational_operation): Likewise.
3200 (simplify_subreg): Likewise.
3201 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3202 * var-tracking.c (adjust_mems): Likewise.
3203 (prepare_call_arguments): Likewise.
3205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3206 Alan Hayward <alan.hayward@arm.com>
3207 David Sherwood <david.sherwood@arm.com>
3209 * machmode.h (is_int_mode): New fuction.
3210 * combine.c (find_split_point): Use it.
3211 (combine_simplify_rtx): Likewise.
3212 (simplify_if_then_else): Likewise.
3213 (simplify_set): Likewise.
3214 (simplify_shift_const_1): Likewise.
3215 (simplify_comparison): Likewise.
3216 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3217 * cse.c (notreg_cost): Likewise.
3218 (cse_insn): Likewise.
3219 * cselib.c (cselib_lookup_1): Likewise.
3220 * dojump.c (do_jump_1): Likewise.
3221 (do_compare_rtx_and_jump): Likewise.
3222 * dse.c (get_call_args): Likewise.
3223 * dwarf2out.c (rtl_for_decl_init): Likewise.
3224 (native_encode_initializer): Likewise.
3225 * expmed.c (emit_store_flag_1): Likewise.
3226 (emit_store_flag): Likewise.
3227 * expr.c (convert_modes): Likewise.
3228 (store_field): Likewise.
3229 (expand_expr_real_1): Likewise.
3230 * fold-const.c (fold_read_from_constant_string): Likewise.
3231 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3232 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3233 * optabs.c (expand_binop): Likewise.
3234 (expand_unop): Likewise.
3235 (expand_abs_nojump): Likewise.
3236 (expand_one_cmpl_abs_nojump): Likewise.
3237 * simplify-rtx.c (mode_signbit_p): Likewise.
3238 (val_signbit_p): Likewise.
3239 (val_signbit_known_set_p): Likewise.
3240 (val_signbit_known_clear_p): Likewise.
3241 (simplify_relational_operation_1): Likewise.
3242 * tree.c (vector_type_mode): Likewise.
3244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3245 Alan Hayward <alan.hayward@arm.com>
3246 David Sherwood <david.sherwood@arm.com>
3248 * machmode.h (smallest_mode_for_size): Fix formatting.
3249 (smallest_int_mode_for_size): New function.
3250 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3251 instead of smallest_mode_for_size.
3252 * combine.c (make_extraction): Likewise.
3253 * config/arc/arc.c (arc_expand_movmem): Likewise.
3254 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3255 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3256 * config/s390/s390.c (s390_expand_insv): Likewise.
3257 * config/sparc/sparc.c (assign_int_registers): Likewise.
3258 * config/spu/spu.c (spu_function_value): Likewise.
3259 (spu_function_arg): Likewise.
3260 * coverage.c (get_gcov_type): Likewise.
3261 (get_gcov_unsigned_t): Likewise.
3262 * dse.c (find_shift_sequence): Likewise.
3263 * expmed.c (store_bit_field_1): Likewise.
3264 * expr.c (convert_move): Likewise.
3265 (store_field): Likewise.
3266 * internal-fn.c (expand_arith_overflow): Likewise.
3267 * optabs-query.c (get_best_extraction_insn): Likewise.
3268 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3269 * stor-layout.c (layout_type): Likewise.
3270 (initialize_sizetypes): Likewise.
3271 * targhooks.c (default_get_mask_mode): Likewise.
3272 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3275 Alan Hayward <alan.hayward@arm.com>
3276 David Sherwood <david.sherwood@arm.com>
3278 * machmode.h (opt_mode::else_blk): New function.
3279 (int_mode_for_mode): Declare.
3280 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3281 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3283 * cfgexpand.c (expand_debug_expr): Likewise.
3284 * combine.c (gen_lowpart_or_truncate): Likewise.
3285 (gen_lowpart_for_combine): Likewise.
3286 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3287 * config/avr/avr.c (avr_to_int_mode): Likewise.
3288 (avr_out_plus_1): Likewise.
3289 (avr_out_plus): Likewise.
3290 (avr_out_round): Likewise.
3291 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3292 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3293 (s390_expand_vcond): Likewise.
3294 * config/spu/spu.c (spu_split_immediate): Likewise.
3295 (spu_expand_mov): Likewise.
3296 * dse.c (get_stored_val): Likewise.
3297 * expmed.c (store_bit_field_1): Likewise.
3298 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3300 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3301 (extract_low_bits): Likewise.
3302 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3303 handling rather than repeating the check.
3304 (emit_group_store): Likewise.
3305 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3306 * optabs.c (expand_absneg_bit): Likewise.
3307 (expand_copysign_absneg): Likewise.
3308 (expand_copysign_bit): Likewise.
3309 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3310 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3311 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3312 * var-tracking.c (prepare_call_arguments): Likewise.
3313 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3314 int_mode_for_mode instead of mode_for_size.
3315 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3318 Alan Hayward <alan.hayward@arm.com>
3319 David Sherwood <david.sherwood@arm.com>
3321 * machmode.h (int_mode_for_size): New function.
3322 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3323 instead of mode_for_size.
3324 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3326 * combine.c (expand_field_assignment): Use int_mode_for_size
3327 instead of mode_for_size.
3328 (make_extraction): Likewise.
3329 (simplify_shift_const_1): Likewise.
3330 (simplify_comparison): Likewise.
3331 * dojump.c (do_jump): Likewise.
3332 * dwarf2out.c (mem_loc_descriptor): Likewise.
3333 * emit-rtl.c (init_derived_machine_modes): Likewise.
3334 * expmed.c (flip_storage_order): Likewise.
3335 (convert_extracted_bit_field): Likewise.
3336 * expr.c (copy_blkmode_from_reg): Likewise.
3337 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3338 * internal-fn.c (expand_mul_overflow): Likewise.
3339 * lower-subreg.c (simple_move): Likewise.
3340 * optabs-libfuncs.c (init_optabs): Likewise.
3341 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3342 * tree.c (vector_type_mode): Likewise.
3343 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3344 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3345 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3346 * tree-vect-stmts.c (vectorizable_load): Likewise.
3347 (vectorizable_store): Likewise.
3349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3350 Alan Hayward <alan.hayward@arm.com>
3351 David Sherwood <david.sherwood@arm.com>
3353 * coretypes.h (pod_mode): New type.
3354 (scalar_int_mode_pod): New typedef.
3355 * machmode.h (pod_mode): New class.
3356 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3357 * genmodes.c (emit_mode_int_n): Update accordingly.
3358 * lower-subreg.h (target_lower_subreg): Change type to
3359 scalar_int_mode_pod.
3360 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3361 scalar_int_mode_pod.
3363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3364 Alan Hayward <alan.hayward@arm.com>
3365 David Sherwood <david.sherwood@arm.com>
3367 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3368 machine_mode to scalar_int_mode.
3369 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3370 (rs6000_option_override_internal): Remove cast to int.
3371 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3372 machine_mode to scalar_int_mode.
3373 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3374 (rs6000_option_override_internal): Remove cast to int.
3375 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3376 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3378 * config/s390/s390.c (s390_expand_builtin): Likewise.
3379 * coretypes.h (scalar_int_mode): New type.
3380 (opt_scalar_int_mode): New typedef.
3381 * machmode.h (scalar_int_mode): New class.
3382 (scalar_int_mode::includes_p): New function.
3383 (byte_mode): Change type to scalar_int_mode.
3384 (word_mode): Likewise.
3385 (ptr_mode): Likewise.
3386 * emit-rtl.c (byte_mode): Likewise.
3387 (word_mode): Likewise.
3388 (ptr_mode): Likewise.
3389 (init_derived_machine_modes): Update accordingly.
3390 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3391 and MODE_PARTIAL_INT.
3392 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3393 opt_scalar_int_mode.
3395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3396 Alan Hayward <alan.hayward@arm.com>
3397 David Sherwood <david.sherwood@arm.com>
3399 * target.def (libgcc_floating_mode_supported_p): Take a
3401 * doc/tm.texi: Regenerate.
3402 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3404 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3405 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3413 * doc/tm.texi: Regenerate.
3414 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3415 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3416 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3417 * targhooks.h (default_floatn_mode): Likewise.
3418 * targhooks.c (default_floatn_mode): Likewise.
3419 * tree.c (build_common_tree_nodes): Update accordingly.
3421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3422 Alan Hayward <alan.hayward@arm.com>
3423 David Sherwood <david.sherwood@arm.com>
3425 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3426 (mode_iterator::iterate_p): Likewise.
3427 (mode_iterator::get_wider): Likewise.
3428 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3431 Alan Hayward <alan.hayward@arm.com>
3432 David Sherwood <david.sherwood@arm.com>
3434 * coretypes.h (opt_scalar_float_mode): New typedef.
3435 * machmode.h (float_mode_for_size): New function.
3436 * emit-rtl.c (double_mode): Delete.
3437 (init_emit_once): Use float_mode_for_size.
3438 * stor-layout.c (layout_type): Likewise.
3439 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3442 Alan Hayward <alan.hayward@arm.com>
3443 David Sherwood <david.sherwood@arm.com>
3445 * output.h (assemble_real): Take a scalar_float_mode.
3446 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3447 * config/arm/arm.md (consttable_4): Likewise.
3448 (consttable_8): Likewise.
3449 (consttable_16): Likewise.
3450 * config/mips/mips.md (consttable_float): Likewise.
3451 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3452 * varasm.c (assemble_real): Take a scalar_float_mode.
3453 (output_constant_pool_2): Update accordingly.
3454 (output_constant): Likewise.
3456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3457 Alan Hayward <alan.hayward@arm.com>
3458 David Sherwood <david.sherwood@arm.com>
3460 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3461 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3462 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3463 (native_encode_real): Likewise.
3464 (native_interpret_real): Likewise.
3465 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3466 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3469 Alan Hayward <alan.hayward@arm.com>
3470 David Sherwood <david.sherwood@arm.com>
3472 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3473 <scalar_float_mode>. Simplify.
3474 (gen_extend_conv_libfunc): Likewise.
3476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3477 Alan Hayward <alan.hayward@arm.com>
3478 David Sherwood <david.sherwood@arm.com>
3480 * coretypes.h (scalar_float_mode): New type.
3481 * machmode.h (mode_traits::from_int): Use machine_mode if
3482 USE_ENUM_MODES is defined.
3483 (is_a): New function.
3485 (dyn_cast): Likewise.
3486 (scalar_float_mode): New class.
3487 (scalar_float_mode::includes_p): New function.
3488 (is_float_mode): Likewise.
3489 * gdbhooks.py (MachineModePrinter): New class.
3490 (build_pretty_printer): Use it for scalar_float_mode.
3491 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3492 (format_helper::format_helper): Turn into a template.
3493 * genmodes.c (get_mode_class): New function.
3494 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3495 or machine_mode if none.
3496 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3497 as_a <scalar_float_mode>.
3498 * dwarf2out.c (mem_loc_descriptor): Likewise.
3499 (insert_float): Likewise.
3500 (add_const_value_attribute): Likewise.
3501 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3502 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3503 (expand_unop): Update accordingly.
3504 (expand_abs_nojump): Likewise.
3505 (expand_copysign_absneg): Take a scalar_float_mode.
3506 (expand_copysign_bit): Likewise.
3507 (expand_copysign): Update accordingly.
3509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3510 Alan Hayward <alan.hayward@arm.com>
3511 David Sherwood <david.sherwood@arm.com>
3513 * coretypes.h (opt_mode): New class.
3514 * machmode.h (opt_mode): Likewise.
3515 (opt_mode::else_void): New function.
3516 (opt_mode::require): Likewise.
3517 (opt_mode::exists): Likewise.
3518 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3519 (GET_MODE_2XWIDER_MODE): Likewise.
3520 (mode_iterator::get_wider): Update accordingly.
3521 (mode_iterator::get_2xwider): Likewise.
3522 (mode_iterator::get_known_wider): Likewise, turning into a template.
3523 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3524 forcing a wider mode to exist.
3525 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3526 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3527 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3528 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3529 * lower-subreg.c (init_lower_subreg): Likewise.
3530 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3531 on the final iteration.
3532 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3533 a wider mode exists before asking for a move pattern.
3534 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3535 forcing a wider mode to exist.
3536 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3537 returning false if no such mode exists.
3538 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3539 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3540 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3541 Avoid checking for a MODE_INT if we already know the mode is not a
3543 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3544 forcing a wider mode to exist.
3545 (expmed_mult_highpart_optab): Likewise.
3546 (expmed_mult_highpart): Likewise.
3547 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3549 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3550 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3551 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3552 * internal-fn.c (expand_mul_overflow): Update use of
3553 GET_MODE_2XWIDER_MODE.
3554 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3555 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3556 GET_MODE_WIDER_MODE.
3557 (convert_plusminus_to_widen): Likewise.
3558 * tree-switch-conversion.c (array_value_type): Likewise.
3559 * var-tracking.c (emit_note_insn_var_location): Likewise.
3560 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3561 Return false inside rather than outside the loop if no wider mode
3563 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3564 and GET_MODE_2XWIDER_MODE
3565 (can_compare_p): Use else_void.
3566 * gdbhooks.py (OptMachineModePrinter): New class.
3567 (build_pretty_printer): Use it for opt_mode.
3569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3570 Alan Hayward <alan.hayward@arm.com>
3571 David Sherwood <david.sherwood@arm.com>
3573 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3574 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3577 Alan Hayward <alan.hayward@arm.com>
3578 David Sherwood <david.sherwood@arm.com>
3580 * machmode.h (mode_traits): New structure.
3581 (get_narrowest_mode): New function.
3582 (mode_iterator::start): Likewise.
3583 (mode_iterator::iterate_p): Likewise.
3584 (mode_iterator::get_wider): Likewise.
3585 (mode_iterator::get_known_wider): Likewise.
3586 (mode_iterator::get_2xwider): Likewise.
3587 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3588 (FOR_EACH_MODE): Likewise.
3589 (FOR_EACH_MODE_FROM): Likewise.
3590 (FOR_EACH_MODE_UNTIL): Likewise.
3591 (FOR_EACH_WIDER_MODE): Likewise.
3592 (FOR_EACH_2XWIDER_MODE): Likewise.
3593 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3594 * combine.c (simplify_comparison): Likewise
3595 * config/i386/i386.c (type_natural_mode): Likewise.
3596 * cse.c (cse_insn): Likewise.
3597 * dse.c (find_shift_sequence): Likewise.
3598 * emit-rtl.c (init_derived_machine_modes): Likewise.
3599 (init_emit_once): Likewise.
3600 * explow.c (hard_function_value): Likewise.
3601 * expmed.c (extract_fixed_bit_field_1): Likewise.
3602 (extract_bit_field_1): Likewise.
3603 (expand_divmod): Likewise.
3604 (emit_store_flag_1): Likewise.
3605 * expr.c (init_expr_target): Likewise.
3606 (convert_move): Likewise.
3607 (alignment_for_piecewise_move): Likewise.
3608 (widest_int_mode_for_size): Likewise.
3609 (emit_block_move_via_movmem): Likewise.
3610 (copy_blkmode_to_reg): Likewise.
3611 (set_storage_via_setmem): Likewise.
3612 (compress_float_constant): Likewise.
3613 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3614 * optabs-query.c (get_best_extraction_insn): Likewise.
3615 * optabs.c (expand_binop): Likewise.
3616 (expand_twoval_unop): Likewise.
3617 (expand_twoval_binop): Likewise.
3618 (widen_leading): Likewise.
3619 (widen_bswap): Likewise.
3620 (expand_parity): Likewise.
3621 (expand_unop): Likewise.
3622 (prepare_cmp_insn): Likewise.
3623 (prepare_float_lib_cmp): Likewise.
3624 (expand_float): Likewise.
3625 (expand_fix): Likewise.
3626 (expand_sfix_optab): Likewise.
3627 * postreload.c (move2add_use_add2_insn): Likewise.
3628 * reg-stack.c (reg_to_stack): Likewise.
3629 * reginfo.c (choose_hard_reg_mode): Likewise.
3630 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3631 * stor-layout.c (mode_for_size): Likewise.
3632 (smallest_mode_for_size): Likewise.
3633 (mode_for_vector): Likewise.
3634 (finish_bitfield_representative): Likewise.
3635 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3636 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3637 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3638 * var-tracking.c (prepare_call_arguments): Likewise.
3640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3641 Alan Hayward <alan.hayward@arm.com>
3642 David Sherwood <david.sherwood@arm.com>
3644 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3645 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3646 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3647 * machmode.h (mode_size): Move earlier in file.
3648 (mode_precision): Likewise.
3649 (mode_inner): Likewise.
3650 (mode_nunits): Likewise.
3651 (mode_unit_size): Likewise.
3652 (unit_unit_precision): Likewise.
3653 (mode_wider): Likewise.
3654 (mode_2xwider): Likewise.
3655 (machine_mode): New class.
3656 (mode_to_bytes): New function.
3657 (mode_to_bits): Likewise.
3658 (mode_to_precision): Likewise.
3659 (mode_to_inner): Likewise.
3660 (mode_to_unit_size): Likewise.
3661 (mode_to_unit_precision): Likewise.
3662 (mode_to_nunits): Likewise.
3663 (GET_MODE_SIZE): Use mode_to_bytes.
3664 (GET_MODE_BITSIZE): Use mode_to_bits.
3665 (GET_MODE_PRECISION): Use mode_to_precision.
3666 (GET_MODE_INNER): Use mode_to_inner.
3667 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3668 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3669 (GET_MODE_NUNITS): Use mode_to_nunits.
3670 * system.h (ALWAYS_INLINE): New macro.
3671 * config/powerpcspe/powerpcspe-c.c
3672 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3673 int for arg1_mode and arg2_mode.
3675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3676 Alan Hayward <alan.hayward@arm.com>
3677 David Sherwood <david.sherwood@arm.com>
3679 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3680 Prefix mode names with E_ in case statements.
3681 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3682 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3683 (aarch64_split_simd_move): Likewise.
3684 (aarch64_gen_storewb_pair): Likewise.
3685 (aarch64_gen_loadwb_pair): Likewise.
3686 (aarch64_gen_store_pair): Likewise.
3687 (aarch64_gen_load_pair): Likewise.
3688 (aarch64_get_condition_code_1): Likewise.
3689 (aarch64_constant_pool_reload_icode): Likewise.
3690 (get_rsqrte_type): Likewise.
3691 (get_rsqrts_type): Likewise.
3692 (get_recpe_type): Likewise.
3693 (get_recps_type): Likewise.
3694 (aarch64_gimplify_va_arg_expr): Likewise.
3695 (aarch64_simd_container_mode): Likewise.
3696 (aarch64_emit_load_exclusive): Likewise.
3697 (aarch64_emit_store_exclusive): Likewise.
3698 (aarch64_expand_compare_and_swap): Likewise.
3699 (aarch64_gen_atomic_cas): Likewise.
3700 (aarch64_emit_bic): Likewise.
3701 (aarch64_emit_atomic_swap): Likewise.
3702 (aarch64_emit_atomic_load_op): Likewise.
3703 (aarch64_evpc_trn): Likewise.
3704 (aarch64_evpc_uzp): Likewise.
3705 (aarch64_evpc_zip): Likewise.
3706 (aarch64_evpc_ext): Likewise.
3707 (aarch64_evpc_rev): Likewise.
3708 (aarch64_evpc_dup): Likewise.
3709 (aarch64_gen_ccmp_first): Likewise.
3710 (aarch64_gen_ccmp_next): Likewise.
3711 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3712 (alpha_emit_xfloating_libcall): Likewise.
3713 (emit_insxl): Likewise.
3714 (alpha_arg_type): Likewise.
3715 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3716 (arc_preferred_simd_mode): Likewise.
3717 (arc_secondary_reload): Likewise.
3718 (get_arc_condition_code): Likewise.
3719 (arc_print_operand): Likewise.
3720 (arc_legitimate_constant_p): Likewise.
3721 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3722 * config/arc/arc.md (casesi_load): Likewise.
3723 (casesi_compact_jump): Likewise.
3724 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3725 (cc_use_register): Likewise.
3726 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3727 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3728 (arm_init_iwmmxt_builtins): Likewise.
3729 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3730 (neon_expand_vector_init): Likewise.
3731 (arm_attr_length_move_neon): Likewise.
3732 (maybe_get_arm_condition_code): Likewise.
3733 (arm_emit_vector_const): Likewise.
3734 (arm_preferred_simd_mode): Likewise.
3735 (arm_output_iwmmxt_tinsr): Likewise.
3736 (thumb1_output_casesi): Likewise.
3737 (thumb2_output_casesi): Likewise.
3738 (arm_emit_load_exclusive): Likewise.
3739 (arm_emit_store_exclusive): Likewise.
3740 (arm_expand_compare_and_swap): Likewise.
3741 (arm_evpc_neon_vuzp): Likewise.
3742 (arm_evpc_neon_vzip): Likewise.
3743 (arm_evpc_neon_vrev): Likewise.
3744 (arm_evpc_neon_vtrn): Likewise.
3745 (arm_evpc_neon_vext): Likewise.
3746 (arm_validize_comparison): Likewise.
3747 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3748 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3749 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3750 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3751 (c6x_preferred_simd_mode): Likewise.
3752 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3753 (epiphany_rtx_costs): Likewise.
3754 * config/epiphany/predicates.md (proper_comparison_operator):
3756 * config/frv/frv.c (condexec_memory_operand): Likewise.
3757 (frv_emit_move): Likewise.
3758 (output_move_single): Likewise.
3759 (output_condmove_single): Likewise.
3760 (frv_hard_regno_mode_ok): Likewise.
3761 (frv_matching_accg_mode): Likewise.
3762 * config/h8300/h8300.c (split_adds_subs): Likewise.
3763 (h8300_rtx_costs): Likewise.
3764 (h8300_print_operand): Likewise.
3765 (compute_mov_length): Likewise.
3766 (output_logical_op): Likewise.
3767 (compute_logical_op_length): Likewise.
3768 (compute_logical_op_cc): Likewise.
3769 (h8300_shift_needs_scratch_p): Likewise.
3770 (output_a_shift): Likewise.
3771 (compute_a_shift_length): Likewise.
3772 (compute_a_shift_cc): Likewise.
3773 (expand_a_rotate): Likewise.
3774 (output_a_rotate): Likewise.
3775 * config/i386/i386.c (classify_argument): Likewise.
3776 (function_arg_advance_32): Likewise.
3777 (function_arg_32): Likewise.
3778 (function_arg_64): Likewise.
3779 (function_value_64): Likewise.
3780 (ix86_gimplify_va_arg): Likewise.
3781 (ix86_legitimate_constant_p): Likewise.
3782 (put_condition_code): Likewise.
3783 (split_double_mode): Likewise.
3784 (ix86_avx256_split_vector_move_misalign): Likewise.
3785 (ix86_expand_vector_logical_operator): Likewise.
3786 (ix86_split_idivmod): Likewise.
3787 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3788 (ix86_build_const_vector): Likewise.
3789 (ix86_build_signbit_mask): Likewise.
3790 (ix86_match_ccmode): Likewise.
3791 (ix86_cc_modes_compatible): Likewise.
3792 (ix86_expand_branch): Likewise.
3793 (ix86_expand_sse_cmp): Likewise.
3794 (ix86_expand_sse_movcc): Likewise.
3795 (ix86_expand_int_sse_cmp): Likewise.
3796 (ix86_expand_vec_perm_vpermi2): Likewise.
3797 (ix86_expand_vec_perm): Likewise.
3798 (ix86_expand_sse_unpack): Likewise.
3799 (ix86_expand_int_addcc): Likewise.
3800 (ix86_split_to_parts): Likewise.
3801 (ix86_vectorize_builtin_gather): Likewise.
3802 (ix86_vectorize_builtin_scatter): Likewise.
3803 (avx_vpermilp_parallel): Likewise.
3804 (inline_memory_move_cost): Likewise.
3805 (ix86_tieable_integer_mode_p): Likewise.
3806 (x86_maybe_negate_const_int): Likewise.
3807 (ix86_expand_vector_init_duplicate): Likewise.
3808 (ix86_expand_vector_init_one_nonzero): Likewise.
3809 (ix86_expand_vector_init_one_var): Likewise.
3810 (ix86_expand_vector_init_concat): Likewise.
3811 (ix86_expand_vector_init_interleave): Likewise.
3812 (ix86_expand_vector_init_general): Likewise.
3813 (ix86_expand_vector_set): Likewise.
3814 (ix86_expand_vector_extract): Likewise.
3815 (emit_reduc_half): Likewise.
3816 (ix86_emit_i387_round): Likewise.
3817 (ix86_mangle_type): Likewise.
3818 (ix86_expand_round_sse4): Likewise.
3819 (expand_vec_perm_blend): Likewise.
3820 (canonicalize_vector_int_perm): Likewise.
3821 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3822 (expand_vec_perm_1): Likewise.
3823 (expand_vec_perm_interleave3): Likewise.
3824 (expand_vec_perm_even_odd_pack): Likewise.
3825 (expand_vec_perm_even_odd_1): Likewise.
3826 (expand_vec_perm_broadcast_1): Likewise.
3827 (ix86_vectorize_vec_perm_const_ok): Likewise.
3828 (ix86_expand_vecop_qihi): Likewise.
3829 (ix86_expand_mul_widen_hilo): Likewise.
3830 (ix86_expand_sse2_abs): Likewise.
3831 (ix86_expand_pextr): Likewise.
3832 (ix86_expand_pinsr): Likewise.
3833 (ix86_preferred_simd_mode): Likewise.
3834 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3835 * config/i386/sse.md (*andnot<mode>3): Likewise.
3836 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3837 (*<code><mode>3): Likewise.
3838 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3839 (ia64_expand_atomic_op): Likewise.
3840 (ia64_arg_type): Likewise.
3841 (ia64_mode_to_int): Likewise.
3842 (ia64_scalar_mode_supported_p): Likewise.
3843 (ia64_vector_mode_supported_p): Likewise.
3844 (expand_vec_perm_broadcast): Likewise.
3845 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3846 (iq2000_function_arg_advance): Likewise.
3847 (iq2000_function_arg): Likewise.
3848 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3849 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3850 (m68k_libcall_value): Likewise.
3851 (m68k_function_value): Likewise.
3852 (sched_attr_op_type): Likewise.
3853 * config/mcore/mcore.c (mcore_output_move): Likewise.
3854 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3856 (microblaze_function_arg): Likewise.
3857 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3858 (mips_print_operand): Likewise.
3859 (mips_mode_ok_for_mov_fmt_p): Likewise.
3860 (mips_vector_mode_supported_p): Likewise.
3861 (mips_preferred_simd_mode): Likewise.
3862 (mips_expand_vpc_loongson_even_odd): Likewise.
3863 (mips_expand_vec_unpack): Likewise.
3864 (mips_expand_vi_broadcast): Likewise.
3865 (mips_expand_vector_init): Likewise.
3866 (mips_expand_vec_reduc): Likewise.
3867 (mips_expand_msa_cmp): Likewise.
3868 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3869 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3870 (cc_flags_for_mode): Likewise.
3871 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3872 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3873 (nds32_output_casesi_pc_relative): Likewise.
3874 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3875 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3876 (nvptx_gen_unpack): Likewise.
3877 (nvptx_gen_pack): Likewise.
3878 (nvptx_gen_shuffle): Likewise.
3879 (nvptx_gen_wcast): Likewise.
3880 (nvptx_preferred_simd_mode): Likewise.
3881 * config/pa/pa.c (pa_secondary_reload): Likewise.
3882 * config/pa/predicates.md (base14_operand): Likewise.
3883 * config/powerpcspe/powerpcspe-c.c
3884 (altivec_resolve_overloaded_builtin): Likewise.
3885 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3887 (rs6000_preferred_simd_mode): Likewise.
3888 (output_vec_const_move): Likewise.
3889 (rs6000_expand_vector_extract): Likewise.
3890 (rs6000_split_vec_extract_var): Likewise.
3891 (reg_offset_addressing_ok_p): Likewise.
3892 (rs6000_legitimate_offset_address_p): Likewise.
3893 (rs6000_legitimize_address): Likewise.
3894 (rs6000_emit_set_const): Likewise.
3895 (rs6000_const_vec): Likewise.
3896 (rs6000_emit_move): Likewise.
3897 (spe_build_register_parallel): Likewise.
3898 (rs6000_darwin64_record_arg_recurse): Likewise.
3899 (swap_selector_for_mode): Likewise.
3900 (spe_init_builtins): Likewise.
3901 (paired_init_builtins): Likewise.
3902 (altivec_init_builtins): Likewise.
3903 (do_load_for_compare): Likewise.
3904 (rs6000_generate_compare): Likewise.
3905 (rs6000_expand_float128_convert): Likewise.
3906 (emit_load_locked): Likewise.
3907 (emit_store_conditional): Likewise.
3908 (rs6000_output_function_epilogue): Likewise.
3909 (rs6000_handle_altivec_attribute): Likewise.
3910 (rs6000_function_value): Likewise.
3911 (emit_fusion_gpr_load): Likewise.
3912 (emit_fusion_p9_load): Likewise.
3913 (emit_fusion_p9_store): Likewise.
3914 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3915 (fusion_gpr_mem_load): Likewise.
3916 (fusion_addis_mem_combo_load): Likewise.
3917 (fusion_addis_mem_combo_store): Likewise.
3918 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3919 (fusion_gpr_mem_load): Likewise.
3920 (fusion_addis_mem_combo_load): Likewise.
3921 (fusion_addis_mem_combo_store): Likewise.
3922 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3924 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3925 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3926 (rs6000_preferred_simd_mode): Likewise.
3927 (output_vec_const_move): Likewise.
3928 (rs6000_expand_vector_extract): Likewise.
3929 (rs6000_split_vec_extract_var): Likewise.
3930 (reg_offset_addressing_ok_p): Likewise.
3931 (rs6000_legitimate_offset_address_p): Likewise.
3932 (rs6000_legitimize_address): Likewise.
3933 (rs6000_emit_set_const): Likewise.
3934 (rs6000_const_vec): Likewise.
3935 (rs6000_emit_move): Likewise.
3936 (rs6000_darwin64_record_arg_recurse): Likewise.
3937 (swap_selector_for_mode): Likewise.
3938 (paired_init_builtins): Likewise.
3939 (altivec_init_builtins): Likewise.
3940 (rs6000_expand_float128_convert): Likewise.
3941 (emit_load_locked): Likewise.
3942 (emit_store_conditional): Likewise.
3943 (rs6000_output_function_epilogue): Likewise.
3944 (rs6000_handle_altivec_attribute): Likewise.
3945 (rs6000_function_value): Likewise.
3946 (emit_fusion_gpr_load): Likewise.
3947 (emit_fusion_p9_load): Likewise.
3948 (emit_fusion_p9_store): Likewise.
3949 * config/rx/rx.c (rx_gen_move_template): Likewise.
3950 (flags_from_mode): Likewise.
3951 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3952 (s390_slb_comparison): Likewise.
3953 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3954 (s390_vector_mode_supported_p): Likewise.
3955 (s390_cc_modes_compatible): Likewise.
3956 (s390_match_ccmode_set): Likewise.
3957 (s390_canonicalize_comparison): Likewise.
3958 (s390_emit_compare_and_swap): Likewise.
3959 (s390_branch_condition_mask): Likewise.
3960 (s390_rtx_costs): Likewise.
3961 (s390_secondary_reload): Likewise.
3962 (__SECONDARY_RELOAD_CASE): Likewise.
3963 (s390_expand_cs): Likewise.
3964 (s390_preferred_simd_mode): Likewise.
3965 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3966 * config/sh/sh.c (sh_print_operand): Likewise.
3967 (dump_table): Likewise.
3968 (sh_secondary_reload): Likewise.
3969 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3970 * config/sh/sh.md (casesi_worker_1): Likewise.
3971 (casesi_worker_2): Likewise.
3972 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3973 (fcc_comparison_operator): Likewise.
3974 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3975 (emit_soft_tfmode_cvt): Likewise.
3976 (sparc_preferred_simd_mode): Likewise.
3977 (output_cbranch): Likewise.
3978 (sparc_print_operand): Likewise.
3979 (sparc_expand_vec_perm_bmask): Likewise.
3980 (vector_init_bshuffle): Likewise.
3981 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3982 (spu_vector_mode_supported_p): Likewise.
3983 (spu_expand_insv): Likewise.
3984 (spu_emit_branch_or_set): Likewise.
3985 (spu_handle_vector_attribute): Likewise.
3986 (spu_builtin_splats): Likewise.
3987 (spu_builtin_extract): Likewise.
3988 (spu_builtin_promote): Likewise.
3989 (spu_expand_sign_extend): Likewise.
3990 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3991 (tilegx_simd_int): Likewise.
3992 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3993 (tilepro_simd_int): Likewise.
3994 * config/v850/v850.c (const_double_split): Likewise.
3995 (v850_print_operand): Likewise.
3996 (ep_memory_offset): Likewise.
3997 * config/vax/vax.c (vax_rtx_costs): Likewise.
3998 (vax_output_int_move): Likewise.
3999 (vax_output_int_add): Likewise.
4000 (vax_output_int_subtract): Likewise.
4001 * config/visium/predicates.md (visium_branch_operator): Likewise.
4002 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4003 (visium_print_operand_address): Likewise.
4004 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4005 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4006 (xtensa_expand_conditional_branch): Likewise.
4007 (xtensa_copy_incoming_a7): Likewise.
4008 (xtensa_output_literal): Likewise.
4009 * dfp.c (decimal_real_maxval): Likewise.
4010 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4013 Alan Hayward <alan.hayward@arm.com>
4014 David Sherwood <david.sherwood@arm.com>
4016 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4017 (mode_nunits_inline): Likewise.
4018 (mode_inner_inline): Likewise.
4019 (mode_unit_size_inline): Likewise.
4020 (mode_unit_precision_inline): Likewise.
4021 (emit_insn_modes_h): Likewise. Also emit a #define of the
4023 (emit_mode_wider): Add an E_ prefix to mode names.
4024 (emit_mode_complex): Likewise.
4025 (emit_mode_inner): Likewise.
4026 (emit_mode_adjustments): Likewise.
4027 (emit_mode_int_n): Likewise.
4028 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4029 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4030 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4031 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4032 (CRC32_BUILTIN, ENTRY): Likewise.
4033 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4034 (aarch64_pop_regs): Likewise.
4035 (aarch64_process_components): Likewise.
4036 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4037 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4038 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4039 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4040 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4041 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4043 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4044 (pcmpistr): Likewise.
4045 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4046 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4047 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4049 * config/rl78/rl78.c (mduc_regs): Likewise.
4050 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4051 (htm_expand_builtin): Likewise.
4052 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4053 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4054 * config/xtensa/xtensa.c (print_operand): Likewise.
4055 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4056 (NUM_MODE_VECTOR_INT): Likewise.
4057 * genoutput.c (null_operand): Likewise.
4058 (output_operand_data): Likewise.
4059 * genrecog.c (print_parameter_value): Likewise.
4060 * lra.c (debug_operand_data): Likewise.
4062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4064 * dwarf2out.c (scompare_loc_descriptor_wide)
4065 (scompare_loc_descriptor_narrow): New functions, split out from...
4066 (scompare_loc_descriptor): ...here.
4067 * expmed.c (emit_store_flag_int): New function, split out from...
4068 (emit_store_flag): ...here.
4070 2017-08-30 Richard Biener <rguenther@suse.de>
4072 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4073 (dwarf2out_early_finish): Move setting of AT_pubnames from
4074 early debug output to early finish.
4076 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4078 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4079 and -mdata-region to the assembler.
4081 2017-08-30 Richard Biener <rguenther@suse.de>
4083 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4085 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4086 (dwarf2out_early_global_decl): Only generate a DIE for the
4087 abstract origin if it doesn't already exist or is a declaration DIE.
4088 (resolve_addr): Do not add the linkage name twice when
4089 generating a stub DIE for the DW_TAG_GNU_call_site target.
4091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4093 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4094 Use machine_mode rather than int for arg1_mode.
4096 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4099 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4100 that the second argument of the built-in functions to unpack
4101 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4102 switch statement instead a lot of if statements.
4103 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4104 Allow 64-bit values to be in Altivec registers as well as
4105 traditional floating point registers.
4106 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4108 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4110 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4111 MAX_REGS_PER_ADDRESS == 1.
4113 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4115 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4116 * config/i386/i386.c (ix86_option_override_internal): Simplify
4117 setting of opts->x_flag_entry.
4119 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4120 Jakub Jelinek <jakub@redhat.com>
4121 Richard Biener <rguenther@suse.de>
4123 PR tree-optimization/81503
4124 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4125 folded constant fits in the target type; reorder tests for clarity.
4127 2017-08-29 Martin Liska <mliska@suse.cz>
4129 * passes.def: Include pass_lower_switch.
4130 * stmt.c (dump_case_nodes): Remove and move to
4131 tree-switch-conversion.
4132 (case_values_threshold): Likewise.
4133 (expand_switch_as_decision_tree_p): Likewise.
4134 (emit_case_decision_tree): Likewise.
4135 (expand_case): Likewise.
4136 (balance_case_nodes): Likewise.
4137 (node_has_low_bound): Likewise.
4138 (node_has_high_bound): Likewise.
4139 (node_is_bounded): Likewise.
4140 (emit_case_nodes): Likewise.
4141 (struct simple_case_node): New struct.
4142 (add_case_node): Remove.
4143 (emit_case_dispatch_table): Use vector instead of case_list.
4144 (reset_out_edges_aux): Remove.
4145 (compute_cases_per_edge): Likewise.
4146 (expand_case): Build list of simple_case_node.
4147 (expand_sjlj_dispatch_table): Use it.
4148 * tree-switch-conversion.c (struct case_node): Moved from
4149 stmt.c and adjusted.
4150 (emit_case_nodes): Likewise.
4151 (node_has_low_bound): Likewise.
4152 (node_has_high_bound): Likewise.
4153 (node_is_bounded): Likewise.
4154 (case_values_threshold): Likewise.
4155 (reset_out_edges_aux): Likewise.
4156 (compute_cases_per_edge): Likewise.
4157 (add_case_node): Likewise.
4158 (dump_case_nodes): Likewise.
4159 (balance_case_nodes): Likewise.
4160 (expand_switch_as_decision_tree_p): Likewise.
4161 (emit_jump): Likewise.
4162 (emit_case_decision_tree): Likewise.
4163 (try_switch_expansion): Likewise.
4164 (do_jump_if_equal): Likewise.
4165 (emit_cmp_and_jump_insns): Likewise.
4166 (fix_phi_operands_for_edge): New function.
4167 (record_phi_operand_mapping): Likewise.
4168 (class pass_lower_switch): New pass.
4169 (pass_lower_switch::execute): New function.
4170 (make_pass_lower_switch): Likewise.
4171 (conditional_probability):
4172 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4173 * tree-pass.h: Add make_pass_lower_switch.
4175 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4178 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4181 2017-08-29 Richard Biener <rguenther@suse.de>
4183 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4184 we do not add a DW_AT_inline attribute twice.
4185 (gen_subprogram_die): Remove code setting DW_AT_inline on
4186 DECL_ABSTRACT_P nodes.
4188 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4190 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4191 calls to internal functions.
4192 (gimplify_modify_expr): Likewise.
4193 * tree-call-cdce.c (use_internal_fn): Likewise.
4194 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4195 (convert_to_divmod): Set the nothrow flag.
4196 * tree-if-conv.c (predicate_mem_writes): Likewise.
4197 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4198 (vectorizable_call): Likewise.
4199 (vectorizable_store): Likewise.
4200 (vectorizable_load): Likewise.
4201 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4202 (vect_recog_mask_conversion_pattern): Likewise.
4204 2017-08-29 Martin Liska <mliska@suse.cz>
4207 * gcc.c (driver_handle_option): Add new argument.
4208 * opts-common.c (handle_option): Pass
4209 target_option_override_hook.
4210 * opts-global.c (lang_handle_option): Add new option.
4211 (set_default_handlers): Add new argument.
4212 (decode_options): Likewise.
4213 * opts.c (target_handle_option): Likewise.
4214 (common_handle_option): Call target_option_override_hook.
4215 * opts.h (struct cl_option_handler_func): Add hook for
4216 target option override.
4217 (struct cl_option_handlers): Likewise.
4218 (set_default_handlers): Add new argument.
4219 (decode_options): Likewise.
4220 (common_handle_option): Likewise.
4221 (target_handle_option): Likewise.
4222 * toplev.c (toplev::main): Pass targetm.target_option.override
4225 2017-08-29 Richard Biener <rguenther@suse.de>
4226 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4228 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4229 life to the active subtree.
4231 2017-08-28 Jeff Law <law@redhat.com>
4233 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4234 derive_equivalences.
4235 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4237 (edge_info::derive_equivalences): New private member function
4239 * tree-ssa-dom.c (class edge_info): Changed from a struct
4240 to a class. Add ctor/dtor, methods and data members.
4241 (edge_info::edge_info): Renamed from allocate_edge_info.
4242 Initialize additional members.
4243 (edge_info::~edge_info): New.
4244 (free_dom_edge_info): Delete the edge info.
4245 (record_edge_info): Use new class & associated member functions.
4246 Tighten forms for testing for edge equivalences.
4247 (record_temporary_equivalences): Iterate over the simple
4248 equivalences rather than assuming there's only one per edge.
4249 (cprop_into_successor_phis): Iterate over the simple
4250 equivalences rather than assuming there's only one per edge.
4251 (optimize_stmt): Use operand_equal_p rather than pointer
4252 equality for mini-DSE code.
4254 2017-08-28 Nathan Sidwell <nathan@acm.org>
4256 * gcc.c (execute): Fold SIGPIPE handling into switch
4257 statement. Adjust internal error message.
4259 2017-08-28 Richard Biener <rguenther@suse.de>
4262 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4263 Do nothing for removed DIEs.
4265 2017-08-28 Richard Biener <rguenther@suse.de>
4267 PR tree-optimization/81977
4268 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4271 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4274 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4275 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4276 using targetm.gen_mem_thread_fence.
4278 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4281 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4282 predicate to register_operand. Reorder operands.
4283 (*btr<mode>): Ditto.
4284 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4285 (*btr<mode>_mask): Ditto.
4287 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4289 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4290 * config/rs6000/xmmintrin.h: New file.
4291 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4293 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4296 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4297 parameter and_insn and return it.
4298 (recombine_lvx_pattern): Insert a copy to ensure availability of
4299 the base register of the copied masking operation at the point of
4300 the instruction replacement.
4301 (recombine_stvx_pattern): Likewise.
4303 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4305 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4306 undocumented switches.
4307 (-mpower9-dform-vector): Likewise.
4308 (-mpower9-dform): Likewise.
4309 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4310 comments to delete references to -mpower9-dform* switches.
4311 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4312 Delete reference to -mpower9-dform* switches, test for
4313 -mpower9-vector instead.
4314 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4315 (OTHER_P9_VECTOR_MASKS): Likewise.
4316 (POWERPC_MASKS): Likewise.
4317 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4318 tests against -mpower9-dform* to -mpower9-vector. Delete code
4319 that checked for -mpower9-dform* consistancy with other options.
4320 Add test for -mpower9-misc to enable other power9 switches.
4321 (rs6000_init_hard_regno_mode_ok): Likewise.
4322 (rs6000_option_override_internal): Likewise.
4323 (rs6000_emit_prologue): Likewise.
4324 (rs6000_emit_epilogue): Likewise.
4325 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4326 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4327 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4329 (emit_fusion_p9_store): Likewise.
4330 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4331 resetting these macros if the assembler does not support ISA 3.0
4333 (TARGET_P9_DFORM_VECTOR): Likewise.
4334 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4335 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4337 2017-08-25 Alan Modra <amodra@gmail.com>
4339 PR rtl-optimization/81747
4340 * cse.c (cse_extended_basic_block): Don't attempt to record
4341 equivalences for degenerate conditional branches that branch
4342 to their fall-through.
4344 2017-08-24 Martin Sebor <msebor@redhat.com>
4347 * gimple-fold.c (size_must_be_zero_p): New function.
4348 (gimple_fold_builtin_memory_op): Call it.
4350 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4352 * config/rs6000/mm_malloc.h: New file.
4354 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4356 PR tree-optimization/81913
4357 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4358 analysis when either IVs in condition can wrap.
4360 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4362 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4363 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4365 2017-08-24 Richard Biener <rguenther@suse.de>
4368 * targhooks.c (default_target_can_inline_p): Properly
4369 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4370 is present and always compare.
4371 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4372 infer -mfpmath=sse from TARGET_SSE_P.
4373 (ix86_can_inline_p): Properly use target_option_default_node when
4374 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4376 2017-08-24 Richard Biener <rguenther@suse.de>
4379 * dwarf2out.c (output_die): Handle flag_generate_offload like
4381 (output_comp_unit): Likewise.
4382 (gen_array_type_die): Likewise.
4383 (dwarf2out_early_finish): Likewise.
4384 (note_variable_value_in_expr): Likewise.
4385 (dwarf2out_finish): Likewise. Adjust assert.
4386 * cgraphunit.c (symbol_table::compile): Move setting of
4387 flag_generate_offload earlier ...
4388 (symbol_table::finalize_compilation_unit): ... here, before
4389 early debug finalization.
4391 2017-08-24 Richard Biener <rguenther@suse.de>
4393 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4394 and ipa-fnsummary.h.
4395 (ix86_can_inline_p): When ix86_fpmath flags do not match
4396 check whether the callee uses FP math at all.
4398 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4401 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4402 instead of TYPE_PRECISION.
4404 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4405 Alan Hayward <alan.hayward@arm.com>
4406 David Sherwood <david.sherwood@arm.com>
4408 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4410 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4412 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4414 (ix86_frame::stack_realign_allocate): New field.
4415 (struct machine_frame_state): Modify comments.
4416 (machine_frame_state::sp_realigned_fp_end): New field.
4417 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4419 (sp_valid_at): Add assertion to assure no attempt to access invalid
4420 offset of a realigned stack.
4421 (fp_valid_at): Likewise.
4422 (choose_baseaddr): Modify comments.
4423 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4424 ix86_expand_prologue.
4425 (ix86_expand_prologue): Modify stack realignment and allocation.
4426 (ix86_expand_epilogue): Modify comments.
4427 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4428 avx2_runtime, avx512f, and avx512f_runtime.
4430 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4432 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4433 (mstackrealign): Do not init to -1.
4434 * config/i386/i386.c (ix86_option_override_internal):
4435 Check opts_set, not opts when setting default value of
4436 opts->x_ix86_force_align_arg_pointer.
4438 2017-08-23 Richard Biener <rguenther@suse.de>
4440 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4441 lang_hooks.decl_printable_name.
4442 * print-rtl-function.c (print_rtx_function): Likewise.
4443 * tree-pretty-print.c (dump_function_header): Likewise.
4445 2017-08-23 Richard Biener <rguenther@suse.de>
4448 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4449 -g0 at compile-time.
4451 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4454 * doc/sourcebuild.texi (Other hardware attributes):
4457 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4460 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4461 Fix single-use check.
4463 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4465 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4467 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4469 * doc/install.texi: Modify to add more details on running selected
4472 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4474 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4475 is combined with -mabi=ms.
4476 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4479 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4481 PR tree-optimization/81488
4482 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4483 and cached_basis fields.
4484 (MAX_SPREAD): New constant.
4485 (alloc_cand_and_find_basis): Initialize new fields.
4486 (clear_visited): New function.
4487 (create_phi_basis_1): Rename from create_phi_basis, set visited
4488 and cached_basis fields.
4489 (create_phi_basis): New wrapper function.
4490 (phi_add_costs_1): Rename from phi_add_costs, add spread
4491 parameter, set visited field, short-circuit when limits reached.
4492 (phi_add_costs): New wrapper function.
4493 (record_phi_increments_1): Rename from record_phi_increments, set
4495 (record_phi_increments): New wrapper function.
4496 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4497 (phi_incr_cost): New wrapper function.
4498 (all_phi_incrs_profitable_1): Rename from
4499 all_phi_incrs_profitable, set visited field.
4500 (all_phi_incrs_profitable): New wrapper function.
4502 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4503 Alan Hayward <alan.hayward@arm.com>
4504 David Sherwood <david.sherwood@arm.com>
4506 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4507 that takes the outer and inner modes.
4508 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4509 comparison as the canonical test for a paradoxical subreg.
4510 * combine.c (simplify_set): Use paradoxical_subreg_p.
4511 (make_extraction): Likewise.
4512 (force_to_mode): Likewise.
4513 (rtx_equal_for_field_assignment_p): Likewise.
4514 (gen_lowpart_for_combine): Likewise.
4515 (simplify_comparison): Likewise.
4516 * cse.c (equiv_constant): Likewise.
4517 * expmed.c (store_bit_field_1): Likewise.
4518 * final.c (alter_subreg): Likewise.
4519 * fwprop.c (propagate_rtx): Likewise.
4520 (forward_propagate_subreg): Likewise.
4521 * ira-conflicts.c (ira_build_conflicts): Likewise.
4522 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4523 * lra-constraints.c (curr_insn_transform): Likewise.
4524 (split_reg): Likewise.
4525 * lra-eliminations.c (move_plus_up): Likewise.
4526 (lra_eliminate_regs_1): Likewise.
4527 * recog.c (general_operand): Likewise.
4528 * ree.c (combine_reaching_defs): Likewise.
4529 * reload.c (push_reload): Likewise.
4530 (find_reloads): Likewise.
4531 * reload1.c (elimination_effects): Likewise.
4532 (compute_reload_subreg_offset): Likewise.
4533 (choose_reload_regs): Likewise.
4534 * rtlanal.c (subreg_lsb_1): Likewise.
4535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4536 (simplify_subreg): Likewise.
4537 * var-tracking.c (track_loc_p): Likewise.
4538 * emit-rtl.c (byte_lowpart_offset): Likewise.
4539 (paradoxical_subreg_p): Delete out-of-line definition.
4541 2017-08-22 Jeff Law <law@redhat.com>
4543 PR tree-optimization/81741
4544 PR tree-optimization/71947
4545 * tree-ssa-dom.c: Include tree-inline.h.
4546 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4547 equivalences if one is more expensive to compute than the other.
4548 * tree-ssa-scopedtables.h (class const_or_copies): Make
4549 record_const_or_copy_raw method private.
4550 (class avail_exprs_stack): New method simplify_binary_operation.
4551 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4552 avail_exprs_stack::simplify_binary_operation as needed.
4553 (avail_exprs_stack::simplify_binary_operation): New function.
4555 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4557 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4558 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4559 (DOT_SYMBOLS): Likewise.
4560 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4561 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4562 (RS6000_ABI_NAME): Likewise.
4563 (TARGET_CMODEL): Likewise.
4564 (TOC_SECTION_ASM_OP): Likewise.
4565 (SET_CMODEL): New macro.
4566 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4568 2017-08-22 Richard Biener <rguenther@suse.de>
4570 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4571 to allow for free-lang-data replacements similar to verify_type_variant.
4573 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4574 Alan Hayward <alan.hayward@arm.com>
4575 David Sherwood <david.sherwood@arm.com>
4577 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4578 VOIDmode for the LABEL_REF.
4580 2017-08-22 Richard Biener <rguenther@suse.de>
4582 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4584 2017-08-22 Richard Biener <rguenther@suse.de>
4586 * common.opt (feliminate-dwarf2-dups): Ignore.
4587 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4588 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4589 same_die_p_wrap, compute_section_prefix,
4590 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4591 (comdat_symbol_id, comdat_symbol_number): Likewise.
4592 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4594 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4595 (output_die): Mark unreachable path unreachable.
4596 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4597 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4598 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4599 (dwarf2out_early_finish): Likewise.
4601 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4603 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4605 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4608 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4609 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4610 (avr_attribute_table) <io, io_low, address>: Initialize
4611 .decl_required with true.
4613 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4615 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4616 undocumented debugging options.
4617 (-mvsx-scalar-double): Likewise.
4618 (-mallow-df-permute): Likewise.
4619 (-mvectorize-builtins): Likewise.
4620 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4621 (rs6000_builtin_vectorized_function): Likewise.
4622 (rs6000_builtin_md_vectorized_function): Likewise.
4623 (rs6000_opt_vars): Likewise.
4625 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4628 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4629 (*btrq_imm): Rename from *btrq.
4630 (*btcq_imm): Rename from *btcq.
4631 (btsc): New code attribute.
4632 (*<btsc><mode>): New insn pattern.
4633 (*btr<mode>): Ditto.
4634 (*<btsc><mode>_mask): New insn_and_split pattern.
4635 (*btr<mode>_mask): Ditto.
4637 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4638 Alan Hayward <alan.hayward@arm.com>
4639 David Sherwood <david.sherwood@arm.com>
4641 * function.c (pad_below): Simplify padding calculation.
4643 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4644 Alan Hayward <alan.hayward@arm.com>
4645 David Sherwood <david.sherwood@arm.com>
4647 * target.def (function_prologue): Remove frame size argument.
4648 (function_epilogue): Likewise.
4649 * doc/tm.texi: Regenerate.
4650 * final.c (final_start_function): Update call to function_prologue.
4651 (final_end_function): Update call to function_epilogue.
4652 (default_function_pro_epilogue): Remove frame size argument.
4653 * output.h (default_function_pro_epilogue): Likewise.
4654 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4655 (arm_output_function_prologue): Likewise.
4656 * config/frv/frv.c (frv_function_prologue): Likewise.
4657 (frv_function_epilogue): Likewise.
4658 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4659 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4660 (ia64_output_function_epilogue): Likewise.
4661 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4662 (m32r_output_function_epilogue): Likewise.
4663 * config/microblaze/microblaze.c (microblaze_function_prologue)
4664 (microblaze_function_epilogue): Likewise.
4665 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4666 (mips_output_function_epilogue): Likewise.
4667 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4668 (mmix_target_asm_function_epilogue): Likewise.
4669 * config/msp430/msp430.c (msp430_start_function): Likewise.
4670 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4671 (nds32_asm_function_epilogue): Likewise.
4672 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4673 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4674 (pa_output_function_epilogue): Likewise.
4675 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4676 (rs6000_output_function_epilogue): Likewise.
4677 * config/rl78/rl78.c (rl78_start_function): Likewise.
4678 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4679 (rs6000_output_function_epilogue): Likewise.
4680 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4681 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4682 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4683 (sparc_asm_function_epilogue): Likewise.
4685 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4687 * tree.h (type_has_mode_precision_p): New function.
4688 * convert.c (convert_to_integer_1): Use it.
4689 * expr.c (expand_expr_real_2): Likewise.
4690 (expand_expr_real_1): Likewise.
4691 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4692 * match.pd: Likewise.
4693 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4694 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4695 * tree-tailcall.c (process_assignment): Likewise.
4696 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4697 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4698 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4699 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4700 (vectorizable_assignment): Likewise.
4701 (vectorizable_shift): Likewise.
4702 (vectorizable_operation): Likewise.
4703 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4705 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4707 * match.pd: Add pow (C, x) simplification.
4709 2017-08-21 Richard Biener <rguenther@suse.de>
4711 PR tree-optimization/81900
4712 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4713 for blocks with abnormal predecessors.
4714 (compute_antic): Do not set visited flag prematurely.
4716 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4719 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4721 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4723 * stor-layout.h (vector_type_mode): Move to...
4724 * tree.h (vector_type_mode): ...here.
4725 * stor-layout.c (vector_type_mode): Move to...
4726 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4728 2017-08-21 Richard Biener <rguenther@suse.de>
4730 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4731 register_external_die hooks.
4732 (debug_false_tree_charstarstar_uhwistar): Declare.
4733 (debug_nothing_tree_charstar_uhwi): Likewise.
4734 * debug.c (do_nothing_debug_hooks): Adjust.
4735 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4736 (debug_nothing_tree_charstar_uhwi): Likewise.
4737 * dbxout.c (dbx_debug_hooks): Adjust.
4738 (xcoff_debug_hooks): Likewise.
4739 * sdbout.c (sdb_debug_hooks): Likewise.
4740 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4741 * dwarf2out.c (macinfo_label_base): New global.
4742 (dwarf2out_register_external_die): New function for the
4743 register_external_die hook.
4744 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4745 (dwarf2_debug_hooks): Use them.
4746 (dwarf2_lineno_debug_hooks): Adjust.
4747 (struct die_struct): Add with_offset flag.
4748 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4749 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4750 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4751 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4752 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4753 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4754 defining section names for the early LTO debug variants.
4755 (reset_indirect_string): New helper.
4756 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4757 (print_dw_val): Add support for offsetted symbol references.
4758 (get_ultimate_context): Split out from is_cxx.
4759 (is_cxx): Use get_ultimate_context.
4760 (is_fortran): Add decl overload.
4761 (compute_comp_unit_symbol): Split out worker from
4762 compute_section_prefix.
4763 (compute_section_prefix): Call compute_comp_unit_symbol and
4764 set comdat_type_p here.
4765 (output_die): Skip DIE symbol output for the LTO added one.
4766 Handle DIE symbol references with offset.
4767 (output_comp_unit): Guard section name mangling properly.
4768 For LTO debug sections emit a symbol at the section beginning
4769 which we use to refer to its DIEs.
4770 (add_abstract_origin_attribute): For DIEs registered via
4771 dwarf2out_register_external_die directly refer to the early
4772 DIE rather than indirectly through the shadow one we created.
4773 Remove obsolete call to dwarf2out_abstract_function for
4774 non-function/block origins.
4775 (gen_array_type_die): When generating early LTO debug do
4776 not emit DW_AT_string_length.
4777 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4778 late when in LTO. As suggested place a gcc_unreachable for
4779 the DECL_ABSTRACT_P case.
4780 (gen_subprogram_die): Avoid another specification DIE
4781 for early built declarations/definitions for the late LTO case.
4782 (gen_variable_die): Add type references for late duplicated VLA dies
4784 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4785 we have the abstract instance already.
4786 (process_scope_var): Adjust decl DIE contexts in LTO which
4787 first puts them in limbo.
4788 (gen_decl_die): Do not generate type DIEs late apart from
4789 types for VLAs or for decls we do not yet have a DIE. Do not
4790 call dwarf2out_abstract_function late.
4791 (dwarf2out_early_global_decl): Make sure to create DIEs
4792 for abstract instances of a decl first.
4793 (dwarf2out_late_global_decl): Adjust comment.
4794 (output_macinfo_op): With multiple macro sections use
4795 macinfo_label_base to distinguish labels.
4796 (output_macinfo): Likewise. Update macinfo_label_base.
4797 Pass in the line info label.
4798 (note_variable_value_in_expr): When generating LTO resolve
4799 all variable values here by generating DIEs as needed.
4800 (init_sections_and_labels): Add early LTO debug flag parameter
4801 and generate different sections and names if set. Add generation
4802 counter for the labels so we can have multiple of them.
4803 (reset_dies): Helper to allow DIEs to be output multiple times.
4804 (dwarf2out_finish): When outputting DIEs to the fat part of an
4805 LTO object first reset DIEs.
4806 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4807 (modified_type_die): Check for decl_ultimate_origin being self
4809 (gen_type_die_with_usage): Likewise.
4810 (gen_typedef_die): Allow decl_ultimate_origin being self.
4811 (set_decl_abstract_flags): Remove.
4812 (set_block_abstract_flags): Likewise.
4813 (dwarf2out_abstract_function): Treat the early generated DIEs
4814 as the abstract copy and only add DW_AT_inline and
4815 DW_AT_artificial here and call set_decl_origin_self.
4816 If the DIE has an abstract origin don't do anything.
4817 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4818 if we have none yet (Go fails to build one, PR78628).
4819 (variably_modified_type_p): Prevent endless recursion for Ada
4820 cyclic pointer types.
4821 * lto-streamer-in.c: Include debug.h.
4822 (dref_queue): New global.
4823 (lto_read_tree_1): Stream in DIE references.
4824 (lto_input_tree): Register DIE references.
4825 (input_function): Stream DECL_DEBUG_ARGS.
4826 * lto-streamer-out.c: Include debug.h.
4827 (lto_write_tree_1): Output DIE references.
4828 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4829 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4830 (output_function): Stream DECL_DEBUG_ARGS.
4831 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4832 Stream DECL_ABSTRACT_ORIGIN.
4833 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4834 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4835 DECL_CONTEXT for file-scope decls.
4836 * lto-streamer.h (struct dref_entry): Declare.
4837 (dref_queue): Likewise.
4838 * cfgexpand.c (pass_expand::execute): Do not call the
4839 outlining_inline_function hook here.
4840 * lto-wrapper.c (debug_obj): New global.
4841 (tool_cleanup): Unlink it if required.
4842 (debug_objcopy): New function.
4843 (run_gcc): Handle early debug sections in the IL files by
4844 extracting them to separate files, partially linkin them and
4845 feeding the result back as result to the linker.
4846 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4847 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4848 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4849 sections into a separate segment.
4850 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4852 (darwin_asm_dwarf_section): Likewise.
4853 (darwin_asm_output_dwarf_offset): Likewise.
4854 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4856 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4857 Alan Hayward <alan.hayward@arm.com>
4858 David Sherwood <david.sherwood@arm.com>
4860 * read-md.h (md_reader::record_potential_iterator_use): Replace
4861 pointer argument with an rtx and an index.
4862 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4863 (apply_mode_iterator): Likewise.
4864 (apply_code_iterator): Likewise.
4865 (apply_int_iterator): Likewise.
4866 (apply_subst_iterator): Likewise.
4867 (record_iterator_use): Likewise.
4868 (record_attribute_use): Likewise.
4869 (md_reader::record_potential_iterator_use): Likewise. Update calls
4870 to record_iterator_use and apply_iterator.
4871 (iterator_use): Replace ptr with x and index.
4872 (attribute_use): Likewise.
4873 (apply_attribute_uses): Update calls to apply_iterator.
4874 (apply_iterators): Likewise. Update initialization of iterator_use.
4875 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4876 and record_potential_iterator_use.
4877 (rtx_reader::read_rtx_operand): Likewise.
4879 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4880 Alan Hayward <alan.hayward@arm.com>
4881 David Sherwood <david.sherwood@arm.com>
4883 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4886 2017-08-21 Richard Biener <rguenther@suse.de>
4889 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4890 at struct end conservatively when comparing common bases.
4892 2017-08-21 Richard Biener <rguenther@suse.de>
4894 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4895 (mem_ref_in_stmt): Remove.
4896 (determine_max_movement): Use ref index to get at the reference.
4897 (invariantness_dom_walker::before_dom_children): Deal with
4898 lim data already initialized.
4899 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4901 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4903 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4904 (TARGET_ISA_ROUND): Ditto.
4905 (TARGET_ROUND): Ditto.
4906 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4907 * config/i386/i386.md: Ditto.
4908 * config/i386/sse.md: Ditto.
4909 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4910 with OPTION_MASK_ISA_SSE4_1.
4912 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4915 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4916 __builtin_ia32_lzcnt_u16.
4918 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4921 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4922 (rs6000_set_current_function): Rewrite function to use it.
4924 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4927 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4928 and TYPE_WARN_IF_NOT_ALIGN.
4929 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4930 (handle_warn_if_not_align): New.
4931 (place_union_field): Call handle_warn_if_not_align.
4932 (place_field): Call handle_warn_if_not_align.
4933 Copy TYPE_WARN_IF_NOT_ALIGN.
4934 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4935 (layout_type): Likewise.
4936 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4938 (tree_decl_common): Add warn_if_not_align.
4939 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4940 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4941 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4942 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4943 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4944 * doc/extend.texi: Document warn_if_not_aligned attribute.
4945 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4947 2017-08-17 Martin Liska <mliska@suse.cz>
4950 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4951 (get_data_dependence): Use it as pointer type.
4952 (distribute_loop): Likewise.
4954 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4956 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4957 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4958 (p8_vmrgow_<mode>_direct): New define_insn.
4959 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4960 handle endianness for vmrgew and vmrgow permute patterns.
4962 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4964 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4965 * config/rs6000/rs6000-cpus.def: Remove comment.
4966 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4967 (POWERPC_MASKS): Likewise.
4968 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4969 use of TARGET_VSX_TIMODE.
4970 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4971 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4972 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4973 (rs6000_option_override_internal): Remove dead code.
4974 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4975 (rs6000_legitimize_reload_address): Likewise.
4976 (rs6000_legitimate_address_p): Likewise.
4977 (rs6000_opt_masks): Delete "vsx-timode".
4978 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4979 from function comment.
4980 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4981 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4982 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4984 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4985 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4986 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4989 (define_peephole2): Likewise.
4991 2017-08-17 Martin Sebor <msebor@redhat.com>
4994 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4995 past the end of an array.
4996 (test_pp_format): Add test cases.
4998 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5000 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5001 missing ECF_NOTHROW flags.
5003 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5006 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5007 operands residing in integer registers.
5008 (*vsx_le_perm_load_<mode>): Likewise.
5009 (*vsx_le_perm_store_<mode>): Likewise.
5010 (define_peephole2): Add peepholes to optimize the above.
5012 2017-08-17 Marek Polacek <polacek@redhat.com>
5015 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5016 to mimic what shorten_compare did. Change the return type to bool.
5017 (fold_cond_expr_with_comparison): Update call to
5018 operand_equal_for_comparison_p.
5019 (fold_ternary_loc): Likewise.
5021 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5023 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5025 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5027 2017-08-17 Richard Biener <rguenther@suse.de>
5029 * tree-ssa-structalias.c (solve_graph): When propagating
5030 to successors update the graphs succ edges and avoid duplicate work.
5032 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5035 * config/i386/i386.c (ix86_option_override_internal): Save target
5036 specific options after ix86_stack_protector_guard_reg was changed.
5038 2017-08-17 Richard Biener <rguenther@suse.de>
5040 PR tree-optimization/81827
5041 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5042 (new_var_info): Initialize it conservatively.
5043 (get_call_vi): Mark register vars.
5044 (new_scalar_tmp_constraint_exp): Likewise.
5045 (handle_rhs_call): Likewise.
5046 (handle_const_call): Likewise.
5047 (create_function_info_for): Likewise.
5048 (solve_constraints): Sort varinfos to separate register from
5049 non-register vars to pack points-to solution bitmaps during
5052 2017-08-17 Marek Polacek <polacek@redhat.com>
5054 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5056 2017-08-17 Richard Biener <rguenther@suse.de>
5058 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5059 to true when overflow is undefined and we saturated the result.
5061 2017-08-17 Alan Modra <amodra@gmail.com>
5064 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5065 Don't use store multiple if only one reg needs saving.
5066 (interesting_frame_related_regno): New function.
5067 (rs6000_frame_related): Don't emit frame info for regs that
5069 (rs6000_emit_epilogue): Likewise.
5071 2017-08-16 Nathan Sidwell <nathan@acm.org>
5073 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5074 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5075 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5076 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5077 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5078 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5079 (verify_type): Adjust for TYPE_BINFO move.
5080 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5081 process TYPE_BINFO directly.
5082 (hash_tree): Likewise.
5083 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5085 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5088 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5090 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5092 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5095 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5096 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5097 Add ix86_binary_operator_ok to insn constraint.
5098 (*iordi_1_bts): Ditto.
5099 (*xordi_1_btc): Ditto.
5100 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5101 Update corresponding peephole2 pattern.
5105 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5107 PR tree-optimization/81832
5108 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5109 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5111 2017-08-16 Marek Polacek <polacek@redhat.com>
5114 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5115 regarding size_zero_node.
5117 2017-08-16 Martin Liska <mliska@suse.cz>
5120 * config.gcc: Respect previously set extra_objs in case
5123 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5125 PR tree-optimization/81835
5126 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5127 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5128 not depend on the phi.
5130 2017-08-16 Alan Modra <amodra@gmail.com>
5132 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5135 2017-08-16 Alan Modra <amodra@gmail.com>
5137 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5138 (save_reg_p): ..into this. Update all callers.
5139 (first_reg_to_save): Simplify.
5141 2017-08-16 Alan Modra <amodra@gmail.com>
5143 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5146 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5150 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5151 constant count if that count is less than 32.
5153 2017-08-15 Nathan Sidwell <nathan@acm.org>
5155 * gcc.c (execute): Emit friendlier message if inferior is killed
5156 by an external cause.
5158 2017-08-15 Richard Biener <rguenther@suse.de>
5160 PR tree-optimization/81790
5161 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5162 CONSTRUCTORs from simplifying and VN.
5164 2017-08-14 Martin Sebor <msebor@redhat.com>
5166 * builtin-attrs.def: Add comments.
5168 2017-08-14 Martin Sebor <msebor@redhat.com>
5171 * doc/extend.texi (attribute nonstring): Document new attribute.
5173 2017-08-14 Martin Sebor <msebor@redhat.com>
5176 * tree-diagnostic.c (default_tree_printer): Handle %G.
5177 * gimple-pretty-print.h (percent_G_format): Declare new function.
5178 * gimple-pretty-print.c (percent_G_format): Define.
5179 * tree-pretty-print.c (percent_K_format): Add argument.
5181 2017-08-14 Martin Sebor <msebor@redhat.com>
5183 PR translation/79998
5184 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5185 Remove a stray space.
5187 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5190 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5191 (*iordi_1_bts): Ditto.
5192 (*xordi_1_btc): Ditto.
5194 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5197 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5198 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5200 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5201 quoted strings, and make more translator-friendly.
5202 (darwin_rs6000_override_options): Likewise.
5203 (rs6000_option_override_internal): Likewise.
5204 (rs6000_return_in_memory): Fix overlong line.
5205 (init_cmulative_args): Use quoted strings, and make more
5206 translator-friendly.
5207 (rs6000_pass_by_reference): Fix overlong line.
5208 (def_builtin): Use quoted strings.
5209 (altivec_expand_predicate_builtin): Use quoted strings, and make
5210 more translator-friendly.
5211 (htm_expand_builtin): Use quoted strings.
5212 (cpu_expand_builtin): Use quoted strings, and make more
5213 translator-friendly.
5214 (altivec_expand_builtin): Likewise.
5215 (paired_expand_predicate_builtin): Likewise.
5216 (rs6000_invalid_builtin): Likewise.
5217 (builtin_function_type): Use quoted strings.
5218 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5219 more translator-friendly.
5220 (rs6000_trampoline_init): Likewise.
5221 (rs6000_handle_altivec_attribute): Likewise.
5222 (rs6000_inner_target_options): Use quoted strings.
5223 (rs6000_disable_incompatible_switches): Likewise.
5224 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5225 strings, and make more translator-friendly.
5226 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5228 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5230 PR tree-optimization/81799
5231 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5232 cond_expr to simple gimple operand.
5234 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5237 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5239 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5243 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5244 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5245 TARGET_GASISR_PROLOGUES.
5246 * config/avr/avr.c (avr_option_override): Same.
5247 (avr_pass_pre_proep::execute): Same.
5249 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5252 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5253 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5255 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5257 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5258 define_insn_and_split. Split to a memory load from 0 in
5259 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5260 using PTR mode iterator.
5261 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5262 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5263 (*add_tp_<mode>): Redefine as define_insn_and_split.
5264 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5265 address space. Merge with *add_tp_x32 using PTR mode iterator.
5266 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5267 Split to an add with a memory load from 0 in
5268 DEFAULT_TLS_SEG_REG address space.
5270 2017-08-12 Andrew Pinski <apinski@cavium.com>
5272 * config/aarch64/aarch64-option-extensions.def (rdma):
5273 Fix feature string to what Linux prints out in /proc/cpuinfo.
5275 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5278 * dwarf2out.c (modified_type_die): For C typedef types that have
5279 an ultimate origin, process the ultimate origin instead of the
5281 (gen_typedef_die): Assert that input DECLs have no ultimate
5283 (gen_type_die_with_usage): For typedef variants that have an
5284 ultimate origin, just call gen_decl_die on the original DECL.
5285 (process_scope_var): Avoid creating DIEs for local typedefs and
5286 concrete static variables.
5288 2017-08-12 Alan Modra <amodra@gmail.com>
5292 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5293 match gnu-user.h startfile.
5294 (ENDFILE_LINUX_SPEC): Similarly.
5296 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5299 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5301 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5303 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5304 * config/aarch64/aarch64.md (mov<mode>): Change.
5305 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5306 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5307 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5309 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5311 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5312 for storage order barriers.
5314 2017-08-11 Martin Liska <mliska@suse.cz>
5317 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5318 variables of void type.
5320 2017-08-11 Martin Liska <mliska@suse.cz>
5322 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5323 TARGET_SUPPORTS_ALIASES.
5324 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5325 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5326 (optimize_weakref): Likewise.
5327 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5328 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5329 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5331 2017-08-11 Martin Liska <mliska@suse.cz>
5334 * config/i386/i386.c (make_resolver_func): Do complete
5335 refactoring of the function.
5337 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5340 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5341 * config/i386/i386.c (ix86_stack_protect_guard): Use
5342 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5343 * doc/invoke.texi (x86 Options): Document
5344 -mstack-protector-guard-symbol= option.
5346 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5348 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5349 * config/i386/i386.c (ix86_split_stack_guard): New function.
5350 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5351 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5352 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5353 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5354 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5355 (split_stack_space_check): Call ix86_split_stack_guard.
5357 2017-08-10 Martin Sebor <msebor@redhat.com>
5359 * print-tree.c (print_node): Print location using the established
5361 Replace spaces with colons.
5362 (debug_raw, debug): Ditto.
5364 2017-08-10 Martin Sebor <msebor@redhat.com>
5367 * pretty-print.c (pp_format): Correct the handling of %s precision.
5369 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5372 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5374 (ix86_finalize_stack_frame_flags): This. Also clear
5375 frame_pointer_needed if -fno-omit-frame-pointer is used without
5377 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5378 with ix86_finalize_stack_frame_flags.
5379 (ix86_expand_epilogue): Likewise.
5380 (ix86_expand_split_stack_prologue): Likewise.
5381 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5383 2017-08-10 Martin Liska <mliska@suse.cz>
5386 * c-attribs.c (handle_target_attribute):
5387 Report warning for an empty string argument of target attribute.
5389 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5392 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5394 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5395 or DECL_NONLOCAL labels.
5396 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5397 or DECL_NONLOCAL labels here.
5399 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5401 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5402 to indicate when early gimple folding has been disabled.
5403 (rs6000_gimple_fold_builtin): Add debug content.
5404 (rs6000_invalid_builtin): Fix whitespace.
5405 (rs6000_expand_builtin): Fix whitespace.
5406 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5408 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5411 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5412 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5414 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5416 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5417 (qdf24xx): Likewise.
5418 * config/aarch64/aarch64.md: Include falkor.md.
5419 * config/aarch64/falkor.md: New.
5421 2017-08-09 Marek Polacek <polacek@redhat.com>
5424 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5425 * diagnostic.c (emit_diagnostic): Add a comment.
5426 (emit_diagnostic_valist): New function.
5428 2017-08-09 Marek Polacek <polacek@redhat.com>
5431 * input.c (make_location): New overload.
5432 * input.h (make_location): Declare.
5434 2017-08-08 Alan Modra <amodra@gmail.com>
5435 H.J. Lu <hongjiu.lu@intel.com>
5438 * gcc.c (NO_PIE_SPEC): Delete.
5439 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5441 (LINK_PIE_SPEC): ..to here.
5442 (LINK_COMMAND_SPEC): Support -no-pie.
5443 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5444 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5446 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5447 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5448 (ENDFILE_CRTEND_SPEC): Similarly.
5450 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5453 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5454 (mstack-protector-guard-offset=): Ditto.
5455 * config/i386/i386.c (ix86_option_override): Handle
5456 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5458 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5459 ix86_stack_protect_guard_offset variables.
5460 (TARGET_STACK_PROTECT_GUARD): Always define.
5461 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5462 and -mstack-protector-guard-offset= options.
5464 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5466 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5467 boundary case for the last candidate.
5469 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5471 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5472 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5474 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5477 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5478 * config/aarch64/aarch64-builtins.c
5479 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5480 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5481 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5483 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5484 Andrew Pinski <pinskia@gmail.com>
5487 * internal-fn.def (XORSIGN): New.
5488 * optabs.def (xorsign_optab): New.
5489 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5490 (convert_expand_mult_copysign): New.
5491 (pass_optimize_widening_mul::execute): Call
5492 convert_expand_mult_copysign.
5494 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5496 PR tree-optimization/81354
5497 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5498 Insert on edges rather than explicitly creating landing pads.
5499 (analyze_candidates_and_replace): Commit edge inserts.
5501 2017-08-08 Richard Biener <rguenther@suse.de>
5504 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5505 (expand_simple_operations): Also look through ADDR_EXPRs with
5506 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5508 2017-08-08 Richard Biener <rguenther@suse.de>
5510 PR tree-optimization/81723
5511 * tree-vect-slp.c (struct bst_traits): New hash traits.
5512 (bst_fail): New global.
5513 (vect_build_slp_tree_2): New worker, split out from ...
5514 (vect_build_slp_tree): ... this now wrapping it with using
5515 bst_fail set to cache SLP tree build fails. Properly handle
5517 (vect_analyze_slp_instance): Allocate and free bst_fail.
5519 2017-08-08 Martin Liska <mliska@suse.cz>
5522 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5523 LABEL_DECLs that can be from a different function.
5525 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5527 PR tree-optimization/81744
5528 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5529 loop's number of iterations.
5531 2017-08-08 Martin Liska <mliska@suse.cz>
5533 * asan.c: Include header files.
5534 * attribs.c (build_decl_attribute_variant): New function moved
5536 (build_type_attribute_qual_variant): Likewise.
5537 (cmp_attrib_identifiers): Likewise.
5538 (simple_cst_list_equal): Likewise.
5539 (omp_declare_simd_clauses_equal): Likewise.
5540 (attribute_value_equal): Likewise.
5541 (comp_type_attributes): Likewise.
5542 (build_type_attribute_variant): Likewise.
5543 (lookup_ident_attribute): Likewise.
5544 (remove_attribute): Likewise.
5545 (merge_attributes): Likewise.
5546 (merge_type_attributes): Likewise.
5547 (merge_decl_attributes): Likewise.
5548 (merge_dllimport_decl_attributes): Likewise.
5549 (handle_dll_attribute): Likewise.
5550 (attribute_list_equal): Likewise.
5551 (attribute_list_contained): Likewise.
5552 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5553 (lookup_attribute_by_prefix): Likewise.
5554 * bb-reorder.c: Include header files.
5555 * builtins.c: Likewise.
5556 * calls.c: Likewise.
5557 * cfgexpand.c: Likewise.
5558 * cgraph.c: Likewise.
5559 * cgraphunit.c: Likewise.
5560 * convert.c: Likewise.
5561 * dwarf2out.c: Likewise.
5562 * final.c: Likewise.
5563 * fold-const.c: Likewise.
5564 * function.c: Likewise.
5565 * gimple-expr.c: Likewise.
5566 * gimple-fold.c: Likewise.
5567 * gimple-pretty-print.c: Likewise.
5568 * gimple.c: Likewise.
5569 * gimplify.c: Likewise.
5570 * hsa-common.c: Likewise.
5571 * hsa-gen.c: Likewise.
5572 * internal-fn.c: Likewise.
5573 * ipa-chkp.c: Likewise.
5574 * ipa-cp.c: Likewise.
5575 * ipa-devirt.c: Likewise.
5576 * ipa-fnsummary.c: Likewise.
5577 * ipa-inline.c: Likewise.
5578 * ipa-visibility.c: Likewise.
5580 * lto-cgraph.c: Likewise.
5581 * omp-expand.c: Likewise.
5582 * omp-general.c: Likewise.
5583 * omp-low.c: Likewise.
5584 * omp-offload.c: Likewise.
5585 * omp-simd-clone.c: Likewise.
5586 * opts-global.c: Likewise.
5587 * passes.c: Likewise.
5588 * predict.c: Likewise.
5589 * sancov.c: Likewise.
5590 * sanopt.c: Likewise.
5591 * symtab.c: Likewise.
5592 * toplev.c: Likewise.
5593 * trans-mem.c: Likewise.
5594 * tree-chkp.c: Likewise.
5595 * tree-eh.c: Likewise.
5596 * tree-into-ssa.c: Likewise.
5597 * tree-object-size.c: Likewise.
5598 * tree-parloops.c: Likewise.
5599 * tree-profile.c: Likewise.
5600 * tree-ssa-ccp.c: Likewise.
5601 * tree-ssa-live.c: Likewise.
5602 * tree-ssa-loop.c: Likewise.
5603 * tree-ssa-sccvn.c: Likewise.
5604 * tree-ssa-structalias.c: Likewise.
5605 * tree-ssa.c: Likewise.
5606 * tree-streamer-in.c: Likewise.
5607 * tree-vectorizer.c: Likewise.
5608 * tree-vrp.c: Likewise.
5610 * ubsan.c: Likewise.
5611 * varasm.c: Likewise.
5612 * varpool.c: Likewise.
5613 * tree.c: Remove functions moved to attribs.[ch].
5615 * config/aarch64/aarch64.c: Add attrs.h header file.
5616 * config/alpha/alpha.c: Likewise.
5617 * config/arc/arc.c: Likewise.
5618 * config/arm/arm.c: Likewise.
5619 * config/avr/avr.c: Likewise.
5620 * config/bfin/bfin.c: Likewise.
5621 * config/c6x/c6x.c: Likewise.
5622 * config/cr16/cr16.c: Likewise.
5623 * config/cris/cris.c: Likewise.
5624 * config/darwin.c: Likewise.
5625 * config/epiphany/epiphany.c: Likewise.
5626 * config/fr30/fr30.c: Likewise.
5627 * config/frv/frv.c: Likewise.
5628 * config/ft32/ft32.c: Likewise.
5629 * config/h8300/h8300.c: Likewise.
5630 * config/i386/winnt.c: Likewise.
5631 * config/ia64/ia64.c: Likewise.
5632 * config/iq2000/iq2000.c: Likewise.
5633 * config/lm32/lm32.c: Likewise.
5634 * config/m32c/m32c.c: Likewise.
5635 * config/m32r/m32r.c: Likewise.
5636 * config/m68k/m68k.c: Likewise.
5637 * config/mcore/mcore.c: Likewise.
5638 * config/microblaze/microblaze.c: Likewise.
5639 * config/mips/mips.c: Likewise.
5640 * config/mmix/mmix.c: Likewise.
5641 * config/mn10300/mn10300.c: Likewise.
5642 * config/moxie/moxie.c: Likewise.
5643 * config/msp430/msp430.c: Likewise.
5644 * config/nds32/nds32-isr.c: Likewise.
5645 * config/nds32/nds32.c: Likewise.
5646 * config/nios2/nios2.c: Likewise.
5647 * config/nvptx/nvptx.c: Likewise.
5648 * config/pa/pa.c: Likewise.
5649 * config/pdp11/pdp11.c: Likewise.
5650 * config/powerpcspe/powerpcspe.c: Likewise.
5651 * config/riscv/riscv.c: Likewise.
5652 * config/rl78/rl78.c: Likewise.
5653 * config/rx/rx.c: Likewise.
5654 * config/s390/s390.c: Likewise.
5655 * config/sh/sh.c: Likewise.
5656 * config/sol2.c: Likewise.
5657 * config/sparc/sparc.c: Likewise.
5658 * config/spu/spu.c: Likewise.
5659 * config/stormy16/stormy16.c: Likewise.
5660 * config/tilegx/tilegx.c: Likewise.
5661 * config/tilepro/tilepro.c: Likewise.
5662 * config/v850/v850.c: Likewise.
5663 * config/vax/vax.c: Likewise.
5664 * config/visium/visium.c: Likewise.
5665 * config/xtensa/xtensa.c: Likewise.
5667 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5670 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5671 constraints since the -mupper-regs-* switches have been
5673 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5674 into a vector from a double word element that was extracted from
5675 another vector, and eliminate extra XXPERMDI instructions.
5676 (vsx_concat_<mode>_2): Likewise.
5677 (vsx_concat_<mode>_3): Likewise.
5678 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5679 concat to allow optimizing inserts from previous extracts.
5681 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5683 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5684 memory reference to a SSP offset in TLS address space.
5685 (ix86_print_operand) <case '@'>: Remove.
5686 (ix86_print_operand_punct_valid_p): Remove '@' code.
5687 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5689 (stack_tls_protect_set_<mode>): Remove.
5690 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5691 (stack_tls_protect_test_<mode>): Remove.
5692 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5694 2017-08-07 Olivier Hainque <hainque@adacore.com>
5697 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5699 2017-08-07 Douglas Rupp <rupp@adacore.com>
5701 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5702 variable was referenced as multidir in command.
5704 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5707 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5710 2017-08-07 Martin Liska <mliska@suse.cz>
5712 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5713 * config/rl78/rl78.c: Add include of attribs.h.
5714 * config/sh/sh.c: Likewise.
5715 * config/v850/v850.c: Likewise.
5717 2017-08-07 Tom de Vries <tom@codesourcery.com>
5720 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5722 2017-08-07 Martin Liska <mliska@suse.cz>
5724 * config/mips/mips.c: Include attribs.h.
5726 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5729 * doc/invoke.texi: Document change in behvaior for -Ofast for
5732 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5734 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5736 (aarch64_pop_regs): Likewise.
5737 (aarch64_gen_load_pair): Likewise.
5738 (aarch64_save_callee_saves): Likewise.
5739 (aarch64_restore_callee_saves): Likewise.
5741 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5743 * config/i386/i386.c: Revert the last change.
5745 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5748 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5750 (ix86_finalize_stack_frame_flags): This. Also clear
5751 frame_pointer_needed if -fno-omit-frame-pointer is used without
5753 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5754 with ix86_finalize_stack_frame_flags.
5755 (ix86_expand_epilogue): Likewise.
5756 (ix86_expand_split_stack_prologue): Likewise.
5758 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5761 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5762 to P_AES for Westmere.
5764 2017-08-07 Jonathan Yong <10walls@gmail.com>
5766 * config/i386/mingw.opt (fset-stack-executable): Removed.
5767 * config/i386/cygming.opt (fset-stack-executable): Moved
5769 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5771 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5773 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5775 2017-08-07 Marek Polacek <polacek@redhat.com>
5778 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5780 2017-08-07 Martin Liska <mliska@suse.cz>
5782 * attribs.h (canonicalize_attr_name): New function.
5783 (cmp_attribs): Move from c-format.c and adjusted.
5784 (is_attribute_p): Moved from tree.h.
5785 * tree-inline.c: Add new includes.
5786 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5787 (private_is_attribute_p): Remove.
5788 (private_lookup_attribute): Likewise.
5789 (private_lookup_attribute_by_prefix): Simplify.
5790 (remove_attribute): Use is_attribute_p.
5791 * tree.h: Remove removed declarations.
5793 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5796 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5797 instead of computing it in the function. Formatting fix.
5798 (expand_case): Don't rely on default_edge being the first edge,
5799 clear it if removing it, pass default_edge to
5800 emit_case_dispatch_table.
5801 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5804 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5806 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5807 insn in the function, emit NOP after the insn.
5809 2017-08-06 Tom de Vries <tom@codesourcery.com>
5811 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5814 2017-08-06 Tom de Vries <tom@codesourcery.com>
5816 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5819 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5821 PR tree-optimization/57371
5822 * match.pd: New pattern.
5824 2017-08-04 Marek Polacek <polacek@redhat.com>
5827 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5828 perform the computation in offset_int.
5830 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5832 PR tree-optimization/81136
5833 * tree-vectorizer.h: Include tree-hash-traits.h.
5834 (vec_base_alignments): New typedef.
5835 (vec_info): Add a base_alignments field.
5836 (vect_record_base_alignments): Declare.
5837 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5839 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5840 (create_data_ref): Add an is_conditional_in_stmt argument.
5841 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5842 the is_conditional_in_stmt field.
5843 (data_ref_loc): Add an is_conditional_in_stmt field.
5844 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5845 (find_data_references_in_stmt): Update call to create_data_ref.
5846 (graphite_find_data_references_in_stmt): Likewise.
5847 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5848 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5849 (vect_record_base_alignment): New function.
5850 (vect_record_base_alignments): Likewise.
5851 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5852 for nested statements even if we fail to compute a misalignment.
5853 Use pooled base alignments for unconditional references.
5854 (vect_find_same_alignment_drs): Compare base addresses instead
5856 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5857 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5859 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5861 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5862 Add an explicit name for the enum. Use auto_vec for slp_instances
5864 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5866 (_bb_vec_info): Add a constructor and destructor.
5867 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5868 (destroy_loop_vec_info): Delete.
5869 (vect_destroy_datarefs): Likewise.
5870 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5871 (vec_info::vec_info): New function.
5872 (vec_info::~vec_info): Likewise.
5873 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5874 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5875 destroy_loop_vec_info.
5876 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5877 (_loop_vec_info::_loop_vec_info): ...this.
5878 (destroy_loop_vec_info): Replace with...
5879 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5880 the stmt_vec_infos. Leave handling of vec_info information to its
5881 destructor. Remove explicit vector releases.
5882 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5883 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5884 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5885 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5886 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5887 (destroy_bb_vec_info): Replace with...
5888 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5889 information to its destructor.
5890 (vect_slp_analyze_bb_1): Use new and delete instead of
5891 new_bb_vec_info and destroy_bb_vec_info.
5892 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5895 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5897 * tree-data-ref.h (subscript): Add access_fn field.
5898 (data_dependence_relation): Add could_be_independent_p.
5899 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5900 (same_access_functions): Move to tree-data-ref.c.
5901 * tree-data-ref.c (ref_contains_union_access_p): New function.
5902 (access_fn_component_p): Likewise.
5903 (access_fn_components_comparable_p): Likewise.
5904 (dr_analyze_indices): Add a reference to access_fn_component_p.
5905 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5907 (constant_access_functions): Likewise.
5908 (add_other_self_distances): Likewise.
5909 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5910 (initialize_data_dependence_relation): Use XCNEW and remove
5911 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5912 of access functions that have the same type. Allow the
5913 subsequence to end with different bases in some circumstances.
5914 Record the chosen access functions in SUB_ACCESS_FN.
5915 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5916 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5917 (subscript_dependence_tester_1): Likewise dra and drb.
5918 (build_classic_dist_vector): Update calls accordingly.
5919 (subscript_dependence_tester): Likewise.
5920 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5921 DDR_COULD_BE_INDEPENDENT_P.
5922 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5923 comp_alias_ddrs instead of may_alias_ddrs.
5924 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5926 (vect_analyze_data_ref_dependence): Use it if
5927 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5928 distance vectors if that fails.
5929 (dependence_distance_ge_vf): New function.
5930 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5931 LOOP_VINFO_MAY_ALIAS_DDRS.
5933 2017-08-04 Richard Biener <rguenther@suse.de>
5936 * fold-const.c (fold_binary_loc): Properly restrict
5937 minus_var0 && minus_var1 case when associating undefined overflow
5940 2017-08-04 Tom de Vries <tom@codesourcery.com>
5942 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5944 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5946 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5947 Don't start diagnostic messages with a capital letter.
5948 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5950 (rs6000_invalid_builtin): Likewise.
5951 (rs6000_trampoline_init): Likewise.
5953 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5956 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5957 after setting changeable df flags.
5959 2017-08-03 Richard Biener <rguenther@suse.de>
5961 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5962 up if the use is in USE - X.
5964 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5966 * toplev.c (dumpfile.h): New include.
5967 (internal_error_reentered): New static function. Use it...
5968 (internal_error_function): ...here to handle reentered internal_error.
5970 2017-08-03 Richard Biener <rguenther@suse.de>
5973 * fold-const.c (split_tree): Add minus_var and minus_con
5974 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5975 here but always use minus_*.
5976 (associate_trees): Assert we never associate with MINUS_EXPR
5977 and NULL first operand. Do not recurse for PLUS_EXPR operands
5978 when associating as MINUS_EXPR either.
5979 (fold_binary_loc): Track minus_var and minus_con.
5981 2017-08-03 Tom de Vries <tom@codesourcery.com>
5984 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5985 ACCEL_COMPILER, apply finish_options on
5986 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5988 2017-08-03 Tom de Vries <tom@codesourcery.com>
5991 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5992 function_entry_patch_area_size > 0.
5994 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5997 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5998 instead of 10??LU, perform unit multiplication in wide_int,
5999 don't change alloc_object_size_limit if the limit is larger
6002 PR tree-optimization/81655
6003 PR tree-optimization/81588
6004 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6005 the case when ranges[i].low and high are 1 for unsigned type with
6009 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6010 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6012 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6014 * tree-vrp.h: Add include guard.
6016 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6019 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6020 (ud2): New insn pattern.
6021 * config/i386/i386.c (ix86_expand_epilogue):
6022 For naked functions, generate ud2 instead of trap insn.
6024 2017-08-02 Marek Polacek <polacek@redhat.com>
6027 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6029 2017-08-02 Tom de Vries <tom@codesourcery.com>
6030 Cesar Philippidis <cesar@codesourcery.com>
6032 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6033 Add missing edge probabilities.
6035 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6037 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6040 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6043 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6044 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6045 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6046 prologue_seq sequences - if any.
6048 2017-08-02 Richard Biener <rguenther@suse.de>
6050 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6051 via vectors if supported, integer extracts via punning if supported
6052 or otherwise vector extracts.
6054 2017-08-02 Richard Biener <rguenther@suse.de>
6056 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6058 (bitmap_insert_into_set): ... this.
6060 2017-08-02 Richard Biener <rguenther@suse.de>
6062 PR tree-optimization/81633
6064 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6066 PR tree-optimization/71752
6067 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6069 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6071 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6072 (machine_function::call_ms2sysv_pad_out): Remove field.
6073 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6074 (ix86_compute_frame_layout): Likewise.
6076 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6079 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6080 attribute with interrupt attribute.
6082 2017-08-01 Andrew Pinski <apinski@cavium.com>
6084 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6085 BIT_INSERT_EXPR's operand 1
6086 to see if the types precision matches.
6088 2017-08-01 Martin Liska <mliska@suse.cz>
6091 * builtins.c (expand_builtin_memcpy_args): Remove.
6092 (expand_builtin_memcpy): Call newly added function
6093 expand_builtin_memory_copy_args.
6094 (expand_builtin_memcpy_with_bounds): Likewise.
6095 (expand_builtin_mempcpy): Remove last argument.
6096 (expand_builtin_mempcpy_with_bounds): Likewise.
6097 (expand_builtin_memory_copy_args): New function created from
6098 expand_builtin_mempcpy_args with small modifications.
6099 (expand_builtin_mempcpy_args): Remove.
6100 (expand_builtin_stpcpy): Remove unused argument.
6101 (expand_builtin): Likewise.
6102 (expand_builtin_with_bounds): Likewise.
6104 2017-08-01 Martin Liska <mliska@suse.cz>
6107 Make mempcpy more optimal (PR middle-end/70140).
6109 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6112 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6113 __builtin_vec_cmpne verify both arguments are compatible vectors
6114 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6115 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6116 move computation of aligned to after checking the argument types.
6120 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6121 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6124 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6125 Doug Rupp <rupp@adacore.com>
6126 Olivier Hainque <hainque@adacore.com>
6128 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6129 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6131 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6132 for TARGET_OS_CPP_BUILTIN.
6133 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6134 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6136 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6137 passing required abi options to the assembler for EABI configurations.
6138 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6139 of .text.hot and .text.unlikely sections for kernel modules when
6140 using ARM style exceptions.
6141 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6142 options. Add EXTRA_CC1_SPEC.
6143 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6145 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6147 (ARM_TARGET2_DWARF_FORMAT): Define.
6148 * config/arm/t-vxworks: Adjust multilib control to removal of the
6149 Diab command line options.
6151 2017-08-01 Martin Liska <mliska@suse.cz>
6153 PR gcov-profile/81561
6154 * gcov.c (unblock): Make unblocking safe as we need to preserve
6155 index correspondence of blocks and block_lists.
6157 2017-08-01 Richard Biener <rguenther@suse.de>
6159 PR tree-optimization/81181
6160 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6161 (compute_antic): ... end of iteration here.
6163 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6165 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6166 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6167 (ftree-slp-vectorize): Likewise.
6168 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6169 can no longer be set independent of flag_tree_loop_vectorize.
6170 * omp-general.c (emp_max_vf): Likewise.
6171 * opts.c (enable_fdo_optimizations): Remove references to
6172 flag_tree_vectorize, these are now implicit.
6173 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6174 and leave it for the options machinery.
6176 2017-08-01 Martin Liska <mliska@suse.cz>
6179 * builtins.c (expand_builtin_memcpy_args): Remove.
6180 (expand_builtin_memcpy): Call newly added function
6181 expand_builtin_memory_copy_args.
6182 (expand_builtin_memcpy_with_bounds): Likewise.
6183 (expand_builtin_mempcpy): Remove last argument.
6184 (expand_builtin_mempcpy_with_bounds): Likewise.
6185 (expand_builtin_memory_copy_args): New function created from
6186 expand_builtin_mempcpy_args with small modifications.
6187 (expand_builtin_mempcpy_args): Remove.
6188 (expand_builtin_stpcpy): Remove unused argument.
6189 (expand_builtin): Likewise.
6190 (expand_builtin_with_bounds): Likewise.
6192 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6195 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6196 print "ds:" only for immediates in generic address space.
6198 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6201 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6202 (ix86_function_ok_for_sibcall): Return false for naked functions.
6204 2017-08-01 Richard Biener <rguenther@suse.de>
6206 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6207 (compute_antic): Seed worklist with exit block predecessors.
6208 * cfganal.c (dfs_find_deadend): For a cycle return the source
6209 of the edge closing it.
6211 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6213 * config/aarch64/aarch64.c
6214 (aarch64_can_const_movi_rtx_p): Move 0 check.
6216 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6218 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6219 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6221 * match.pd: Ditto in address comparison pattern.
6223 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6225 PR tree-optimization/81627
6226 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6227 closed ssa form for store-store chain.
6229 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6231 PR tree-optimization/81620
6232 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6233 for store-store chain.
6235 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6237 PR tree-optimization/81588
6238 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6239 ranges[i].in_p, invert comparison code ccode. For >/>=,
6240 swap rhs1 and rhs2 and comparison code unconditionally,
6241 for </<= don't do that. Don't swap rhs1/rhs2 again if
6242 ranges[i].in_p, instead invert comparison code ccode if
6243 opcode or oe->rank is BIT_IOR_EXPR.
6246 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6247 a direct optab to conversion optab.
6248 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6249 with GET_MODE_INNER as last argument instead of optab_handler.
6250 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6251 vector extraction if possible and optab is available.
6252 * expr.c (store_constructor): Use convert_optab_handler instead
6253 of optab_handler. Use vector initialization from smaller
6254 vectors if possible and optab is available.
6255 * tree-vect-stmts.c (vectorizable_load): Likewise.
6256 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6258 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6259 of vec_init from half-sized vectors with the same element mode.
6260 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6261 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6262 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6263 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6264 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6265 after mode in gen_vec_extract* calls.
6266 (vec_extract<mode>): Renamed to ...
6267 (vec_extract<mode><ssescalarmodelower>): ... this.
6268 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6269 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6270 element mode after mode in gen_vec_init* calls.
6271 (VEC_INIT_HALF_MODE): New mode iterator.
6272 (vec_init<mode>): Renamed to ...
6273 (vec_init<mode><ssescalarmodelower>): ... this.
6274 (vec_init<mode><ssehalfvecmodelower>): New expander.
6275 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6276 (vec_extractv2sfsf): ... this.
6277 (vec_initv2sf): Renamed to ...
6278 (vec_initv2sfsf): ... this.
6279 (vec_extractv2si): Renamed to ...
6280 (vec_extractv2sisi): ... this.
6281 (vec_initv2si): Renamed to ...
6282 (vec_initv2sisi): ... this.
6283 (vec_extractv4hi): Renamed to ...
6284 (vec_extractv4hihi): ... this.
6285 (vec_initv4hi): Renamed to ...
6286 (vec_initv4hihi): ... this.
6287 (vec_extractv8qi): Renamed to ...
6288 (vec_extractv8qiqi): ... this.
6289 (vec_initv8qi): Renamed to ...
6290 (vec_initv8qiqi): ... this.
6291 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6292 (vec_init<mode>): Renamed to ...
6293 (vec_init<mode><VEC_base_l>): ... this.
6294 (vec_extract<mode>): Renamed to ...
6295 (vec_extract<mode><VEC_base_l>): ... this.
6296 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6297 (vec_initv2sfsf): ... this.
6298 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6299 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6300 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6301 element mode after mode in gen_vec_init* calls.
6302 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6303 (vec_init<mode><Vel>): ... this.
6304 (vec_extract<mode>): Renamed to ...
6305 (vec_extract<mode><Vel>): ... this.
6306 * config/aarch64/iterators.md (Vel): New mode attribute.
6307 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6308 Add element mode after mode in gen_vec_extract* calls.
6309 * config/s390/vector.md (non_vec_l): New mode attribute.
6310 (vec_extract<mode>): Renamed to ...
6311 (vec_extract<mode><non_vec_l>): ... this.
6312 (vec_init<mode>): Renamed to ...
6313 (vec_init<mode><non_vec_l>): ... this.
6314 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6315 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6317 * config/arm/iterators.md (V_elem_l): New mode attribute.
6318 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6319 (vec_extract<mode><V_elem_l>): ... this.
6320 (vec_extractv2di): Renamed to ...
6321 (vec_extractv2didi): ... this.
6322 (vec_init<mode>): Renamed to ...
6323 (vec_init<mode><V_elem_l>): ... this.
6324 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6325 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6326 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6327 Add element mode after gen_vec_extract* calls.
6328 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6329 (vec_init<mode><unitmode>): ... this.
6330 (vec_extract<mode>): Renamed to ...
6331 (vec_extract<mode><unitmode>): ... this.
6332 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6333 (vec_init<mode><unitmode>): ... this.
6334 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6335 (vec_initv2sfsf): ... this.
6336 (vec_extractv2sf): Renamed to ...
6337 (vec_extractv2sfsf): ... this.
6338 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6339 Add element mode after gen_vec_extract* calls.
6340 * config/mips/mips.md (unitmode): New mode iterator.
6341 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6342 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6343 * config/spu/spu.md (inner_l): New mode attribute.
6344 (vec_init<mode>): Renamed to ...
6345 (vec_init<mode><inner_l>): ... this.
6346 (vec_extract<mode>): Renamed to ...
6347 (vec_extract<mode><inner_l>): ... this.
6348 * config/sparc/sparc.md (veltmode): New mode iterator.
6349 (vec_init<VMALL:mode>): Renamed to ...
6350 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6351 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6352 (vec_initv2sisi): ... this.
6353 (vec_initv2sf): Renamed to ...
6354 (vec_initv2sfsf): ... this.
6355 (vec_extractv2sf): Renamed to ...
6356 (vec_extractv2sfsf): ... this.
6357 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6358 (vec_init<mode>): Renamed to ...
6359 (vec_init<mode><VEC_base_l>): ... this.
6360 (vec_extract<mode>): Renamed to ...
6361 (vec_extract<mode><VEC_base_l>): ... this.
6362 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6363 (vec_initv2sfsf): ... this.
6364 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6365 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6366 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6367 gen_vec_init* calls.
6369 2017-08-01 Richard Biener <rguenther@suse.de>
6371 PR tree-optimization/81297
6372 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6373 TREE_OVERFLOW from INTEGER_CSTs.
6375 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6377 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6379 2017-07-31 Carl Love <cel@us.ibm.com>
6381 * config/rs6000/rs6000-c: Add support for built-in functions
6382 vector signed char vec_xl_be (signed long long, signed char *);
6383 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6384 vector signed int vec_xl_be (signed long long, signed int *);
6385 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6386 vector signed long long vec_xl_be (signed long long, signed long long *);
6387 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6388 vector signed short vec_xl_be (signed long long, signed short *);
6389 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6390 vector double vec_xl_be (signed long long, double *);
6391 vector float vec_xl_be (signed long long, float *);
6392 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6393 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6394 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6396 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6397 (altivec_expand_builtin): Add switch statement to call
6398 altivec_expand_xl_be for each builtin.
6399 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6400 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6401 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6402 __builtin_vsx_le_be_v16qi.
6403 * doc/extend.texi: Update the built-in documentation file for the
6404 new built-in functions.
6406 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6409 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6411 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6413 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6415 * config.gcc: Add z14.
6416 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6417 CPU model numbers for z13s and z14.
6418 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6420 * config/s390/s390-opts.h (enum processor_type): Rename
6421 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6422 * config/s390/s390.c (processor_table): Add field for CPU name to
6423 be passed to Binutils.
6424 (s390_asm_output_machine_for_arch): Use the new field in
6425 processor_table for Binutils.
6426 (s390_expand_builtin): Replace arch12 with z14.
6427 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6428 (s390_get_sched_attrmask): Likewise.
6429 (s390_get_unit_mask): Likewise.
6430 * config/s390/s390.opt: Add z14 to processor_type enum.
6432 2017-07-31 Martin Jambor <mjambor@suse.cz>
6435 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6436 regardless of optimization level.
6438 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6439 Martin Liska <mliska@suse.cz>
6441 * predict.def: Remove old comment and adjust probability.
6442 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6445 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6448 * config/i386/i386.c (ix86_function_naked): New function.
6449 (ix86_can_use_return_insn_p): Return false for naked functions.
6450 (ix86_expand_prologue): Skip prologue for naked functions.
6451 (ix86_expand_epilogue): Skip epilogue for naked functions
6452 and emit trap instruction.
6453 (ix86_warn_func_return): New function.
6454 (ix86_attribute_table): Add "naked" attribute specification.
6455 (TARGET_WARN_FUNC_RETURN): Define.
6456 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6458 2017-07-31 Martin Liska <mliska@suse.cz>
6460 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6461 (dump_gimple_bb_header): Always dump BB info.
6462 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6464 2017-07-31 Martin Liska <mliska@suse.cz>
6467 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6468 also with current_function_decl non-null equality.
6470 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6473 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6474 change type to the element type, instead add eltype variable and
6475 use it where we are interested in the element type.
6477 PR tree-optimization/81603
6478 * ipa-polymorphic-call.c
6479 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6480 offset arithmetic in offset_int, bail out if the resulting bit offset
6481 doesn't fit into shwi.
6483 2017-07-31 Martin Liska <mliska@suse.cz>
6485 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6486 (gimplify_save_expr): Fix comment.
6488 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6491 * config/i386/i386.c (ix86_function_arg): Update arguments for
6493 (ix86_compute_frame_layout): Set the initial stack offset to
6494 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6495 INCOMING_FRAME_SP_OFFSET.
6496 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6497 stack before exception handler returns.
6498 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6499 the 'ERROR_CODE' for exception handler.
6501 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6503 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6504 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6505 (ASM_OUTPUT_REG_POP): Ditto.
6506 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6507 instead of asm_fprintf to output pure string.
6509 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6511 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6512 to imported_module_or_decl hook.
6513 (debug_nothing_tree_tree_tree_bool): Remove.
6514 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6515 * debug.c (do_nothing_debug_hooks): Use
6516 debug_nothing_tree_tree_tree_bool_bool instead of
6517 debug_nothing_tree_tree_tree_bool.
6518 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6519 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6520 * sdbout.c (sdb_debug_hooks): Likewise.
6521 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6522 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6524 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6525 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6526 attribute, don't add anything.
6528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6530 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6531 (fold_build2_stat_loc): Likewise.
6532 (fold_build3_stat_loc): Likewise.
6533 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6534 (fold_build1_loc): Remove macro.
6535 (fold_build2_loc): Likewise.
6536 (fold_build3_loc): Likewise.
6538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6540 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6541 (gimple_build_debug_bind_source_stat): Likewise.
6542 * gimple.h (gimple_build_debug_bind): Remove macro.
6543 (gimple_build_debug_bind_source): Likewise.
6545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6547 * bitmap.c (bitmap_alloc): Adjust.
6548 (bitmap_gc_alloc): Likewise.
6549 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6551 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6553 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6554 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6555 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6556 (bitmap_gc_alloc_stat): Likewise.
6557 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6561 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6562 * rtl.h (shallow_copy_rtx): Remove macro.
6564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6566 * emit-rtl.c (gen_raw_REG): Adjust.
6567 * gengenrtl.c (gendef): Likewise.
6568 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6569 * rtl.h (rtx_alloc): Remove macro.
6571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6573 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6574 (build_tree_list_stat): Likewise.
6575 * tree.h (build_tree_list): Remove macro.
6576 (build_tree_list_vec): Likewise.
6578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6580 * tree.c (make_vector_stat): Remove _stat from name.
6581 (build_vector_stat): Likewise.
6582 * tree.h (make_vector_stat): Remove macro.
6583 (build_vector_stat): Likewise.
6585 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6587 * tree.h (build_var_debug_value): Remove prototype.
6589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6591 * tree.c (tree_cons_stat): Remove _stat from name.
6592 * tree.h (tree_cons): Remove macro.
6594 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6596 * tree.c (build_vl_exp_stat): Remove _stat from name.
6597 * tree.h (build_vl_exp): Remove macro.
6599 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6601 * tree.c (build_decl_stat): Remove _stat from name.
6602 * tree.h (build_decl): Remove macro.
6604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6606 * gimple.c (gimple_build_with_ops_stat): Adjust.
6607 (gimple_alloc_stat): Remove _stat from name.
6608 * gimple.h (gimple_alloc): Remove macro.
6610 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6612 * tree.c (make_tree_vec_stat): Remove _stat from name.
6613 (grow_tree_vec_stat): Likewise.
6614 * tree.h (make_tree_vec_stat): Adjust prototype.
6615 (grow_tree_vec_stat): Likewise.
6616 (make_tree_vec): Remove macro.
6617 (grow_tree_vec): Likewise.
6619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6621 * fold-const.c (fold_build1_stat_loc): Adjust.
6622 (fold_build2_stat_loc): Likewise.
6623 (fold_build3_stat_loc): Likewise.
6624 * tree.c (build0_stat): Remove _stat from name.
6625 (build1_stat): Likewise.
6626 (build2_stat): Likewise.
6627 (build3_stat): Likewise.
6628 (build4_stat): Likewise.
6629 (build5_stat): Likewise.
6630 * tree.h (build1_loc): Remove macro, and rename _stat function
6632 (build2_loc): Likewise.
6633 (build3_loc): Likewise.
6634 (build4_loc): Likewise.
6635 (build5_loc): Likewise.
6637 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6639 * tree.c (make_int_cst_stat): Remove _stat from name.
6640 * tree.h (make_int_cst_stat): Adjust prototype.
6641 (make_int_cst): Remove macro.
6643 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6645 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6646 * tree.h (make_tree_binfo_stat): Adjust prototype.
6647 (make_tree_binfo): Remove.
6649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6651 * tree.c (copy_node_stat): Rename to copy_node.
6652 (build_distinct_type_copy): Adjust.
6653 * tree.h (copy_node_stat): Adjust prototype.
6654 (copy_node): Remove macro.
6656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6658 * tree.c (make_node_stat): rename to make_node.
6659 (build_tree_list_stat): Adjust.
6660 (build0_stat): Likewise.
6661 (build2_stat): Likewise.
6662 (build3_stat): Likewise.
6663 (build4_stat): Likewise.
6664 (build5_stat): Likewise.
6665 (build_decl_stat): Likewise.
6666 * tree.h (make_node_stat): Adjust prototype.
6667 (make_node): remove macro.
6669 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6671 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6672 (PPC_FEATURE2_SCV): Likewise.
6673 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6675 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6677 * config/aarch64/aarch64.c
6678 (aarch64_internal_mov_immediate): Add new special pattern.
6679 * config/aarch64/aarch64.md (*movdi_aarch64):
6680 Add reg/32bit const mov case.
6682 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6683 Richard Sandiford <richard.sandiford@linaro.org>
6685 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6686 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6687 Add integer and movi cases.
6688 (movi-split-hf-df-sf split, fp16): New.
6689 (enabled): Added TARGET_FP_F16INST.
6690 * config/aarch64/iterators.md (GPF_HF): New.
6691 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6693 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6695 * config/aarch64/aarch64.c
6696 (aarch64_simd_container_mode): Add prototype.
6697 (aarch64_expand_mov_immediate): Add HI support.
6698 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6699 (aarch64_can_const_movi_rtx_p): New.
6700 (aarch64_preferred_reload_class):
6701 Remove restrictions of using FP registers for certain SIMD operations.
6702 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6703 (aarch64_valid_floating_const): Add integer move validation.
6704 (aarch64_simd_imm_scalar_p): Remove.
6705 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6706 (aarch64_legitimate_constant_p): Expand list of supported cases.
6707 * config/aarch64/aarch64-protos.h
6708 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6709 (aarch64_reinterpret_float_as_int): New.
6710 (aarch64_simd_imm_scalar_p): Remove.
6711 * config/aarch64/constraints.md (Uvi): New.
6712 (Dd): Split into Ds and new Dd.
6713 * config/aarch64/aarch64.md (*movsi_aarch64):
6715 (*movdi_aarch64): Add SIMD mov case.
6717 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6719 * tree-predcom.c: (struct chain): Handle store-store chain in which
6720 stores for elimination only store loop invariant values.
6721 (execute_pred_commoning_chain): Ditto.
6722 (prepare_initializers_chain_store_elim): Ditto.
6723 (prepare_finalizers): Ditto.
6724 (is_inv_store_elimination_chain): New function.
6725 (initialize_root_vars_store_elim_1): New function.
6727 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6729 * tree-predcom.c: Revise general description of the pass.
6730 (enum chain_type): New enum type for store elimination.
6731 (struct chain): New field supporting store elimination.
6732 (struct component): Ditto.
6733 (dump_chain): Dump store-stores chain.
6734 (release_chain): Release resources.
6735 (split_data_refs_to_components): Compute and create component
6736 contains only stores for elimination.
6737 (get_chain_last_ref_at): New function.
6738 (make_invariant_chain): Initialization.
6739 (make_rooted_chain): Specify chain type in parameter and record it.
6740 (add_looparound_copies): Skip for store-stores chain.
6741 (determine_roots_comp): Compute type of chain and pass it to
6743 (initialize_root_vars_store_elim_2): New function.
6744 (finalize_eliminated_stores): New function.
6745 (remove_stmt): Handle store for elimination.
6746 (execute_pred_commoning_chain): Execute predictive commoning on
6748 (determine_unroll_factor): Skip unroll for store-stores chain.
6749 (prepare_initializers_chain_store_elim): New function.
6750 (prepare_initializers_chain): Hanlde store-store chain.
6751 (prepare_finalizers_chain, prepare_finalizers): New function.
6752 (tree_predictive_commoning_loop): Return integer value indicating
6753 if loop is unrolled or lcssa form is corrupted.
6754 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6756 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6758 * tree-predcom.c (initialize_root): Delete.
6759 (execute_pred_commoning_chain): Initialize root vars and replace
6760 reference of non-combined chain directly, rather than call above
6763 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6765 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6766 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6768 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6770 * tree-predcom.c (struct chain): New field init_seq.
6771 (release_chain): Release init_seq.
6772 (prepare_initializers_chain): Record intialization stmts in above
6774 (insert_init_seqs): New function.
6775 (tree_predictive_commoning_loop): Call insert_init_seqs.
6777 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6779 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6781 2017-07-28 Richard Biener <rguenther@suse.de>
6783 * match.pd: Remove superfluous :c.
6784 * genmatch.c (simplify::id): Add member.
6785 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6787 (current_id): New global.
6788 (dt_node::parent): Move from ...
6789 (dt_operand::parent): ... here. Add for_id member.
6790 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6791 (decision_tree::find_node): Relax order requirement when
6792 merging DT_TRUE nodes to ones inbetween the current simplify
6793 and the one we try to merge with. Add diagnostic whenever
6794 we need to enforce pattern order by not merging.
6795 (decision_tree::insert): Set current_id.
6796 (decision_tree::print_node): Dump parent node and for_id.
6797 (parser::last_id): Add member.
6798 (parser::push_simplify): Assign unique id.
6799 (parser::parser): Initialize last_id.
6801 2017-07-28 Martin Liska <mliska@suse.cz>
6804 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6805 gimple_build_debug_bind.
6807 2017-07-28 Richard Biener <rguenther@suse.de>
6809 PR tree-optimization/81502
6810 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6812 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6814 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6815 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6816 for BIT_FIELD_REF args.
6817 * fold-const.c (make_bit_field_ref): Likewise.
6818 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6820 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6823 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6824 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6825 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6826 Or it into SANITIZER_UNDEFINED.
6827 * ubsan.c: Include gimple-fold.h and varasm.h.
6828 (ubsan_expand_ptr_ifn): New function.
6829 (instrument_pointer_overflow): New function.
6830 (maybe_instrument_pointer_overflow): New function.
6831 (instrument_object_size): Formatting fix.
6832 (pass_ubsan::execute): Call instrument_pointer_overflow
6833 and maybe_instrument_pointer_overflow.
6834 * internal-fn.c (expand_UBSAN_PTR): New function.
6835 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6836 * sanitizer.def (__ubsan_handle_pointer_overflow,
6837 __ubsan_handle_pointer_overflow_abort): New builtins.
6838 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6839 * internal-fn.def (UBSAN_PTR): New internal function.
6840 * opts.c (sanitizer_opts): Add pointer-overflow.
6841 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6842 * fold-const.c (build_range_check): Compute pointer range check in
6843 integral type if pointer arithmetics would be needed. Formatting
6846 2017-07-28 Martin Liska <mliska@suse.cz>
6849 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6850 parameters that are of a variable-length.
6852 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6854 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6856 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6857 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6858 (CRT_CALL_STATIC_FUNCTION): Likewise.
6859 (ASM_DEFAULT_SPEC): New define.
6860 (ASM_SPEC32): Likewise.
6861 (ASM_SPEC64): Likewise.
6862 (ASM_SPEC_COMMON): Likewise.
6863 (ASM_SPEC): Likewise.
6864 (INVALID_64BIT): Likewise.
6865 (LINK_OS_DEFAULT_SPEC): Likewise.
6866 (LINK_OS_SPEC32): Likewise.
6867 (LINK_OS_SPEC64): Likewise.
6868 (POWERPC_LINUX): Likewise.
6869 (PTRDIFF_TYPE): Likewise.
6870 (RESTORE_FP_PREFIX): Likewise.
6871 (RESTORE_FP_SUFFIX): Likewise.
6872 (SAVE_FP_PREFIX): Likewise.
6873 (SAVE_FP_SUFFIX): Likewise.
6874 (SIZE_TYPE): Likewise.
6875 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6876 (TARGET_64BIT): Likewise.
6877 (TARGET_64BIT): Likewise.
6878 (TARGET_AIX): Likewise.
6879 (WCHAR_TYPE_SIZE): Likewise.
6880 (WCHAR_TYPE): Undefine.
6881 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6882 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6883 (CPP_OS_RTEMS_SPEC): Delete.
6884 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6885 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6887 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6889 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6891 PR tree-optimization/81578
6892 * tree-parloops.c (build_new_reduction): Bail out if
6893 reduction_code isn't one of the standard OpenMP reductions.
6894 Move the details printing after that decision.
6896 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6898 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6899 related to reload_in_progress.
6900 (splat_input_operand): Likewise.
6901 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6903 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6905 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6906 (TARGET_INSTANTIATE_DECLS): Likewise.
6907 (legitimate_indexed_address_p): Delete reload_in_progress code.
6908 (rs6000_debug_legitimate_address_p): Likewise.
6909 (rs6000_eliminate_indexed_memrefs): Likewise.
6910 (rs6000_emit_le_vsx_store): Likewise.
6911 (rs6000_emit_move_si_sf_subreg): Likewise.
6912 (rs6000_emit_move): Likewise.
6913 (register_to_reg_type): Likewise.
6914 (rs6000_pre_atomic_barrier): Likewise.
6915 (rs6000_machopic_legitimize_pic_address): Likewise.
6916 (rs6000_allocate_stack_temp): Likewise.
6917 (rs6000_address_for_fpconvert): Likewise.
6918 (rs6000_address_for_altivec): Likewise.
6919 (rs6000_secondary_memory_needed_rtx): Delete function.
6920 (rs6000_check_sdmode): Likewise.
6921 (rs6000_alloc_sdmode_stack_slot): Likewise.
6922 (rs6000_instantiate_decls): Likewise.
6923 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6924 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6925 Delete reload_in_progress.
6926 (*vec_reload_and_plus_<mptrsize>): Likewise.
6927 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6928 (vsx_div_v2di): Likewise.
6929 (vsx_udiv_v2di): Likewise.
6931 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6933 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6934 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6935 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6936 (rs6000_debug_reg_global): Delete print of LRA status.
6937 (rs6000_option_override_internal): Delete dead LRA related code.
6938 (rs6000_lra_p): Delete function.
6939 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6941 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6943 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6944 * config/riscv/rtems.h: New file.
6946 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6947 Sudakshina Das <sudi.das@arm.com>
6949 * config/aarch64/aarch64.md
6950 (define_split for and<mode>3nr_compare): Move
6951 non aarch64_logical_operand to a register.
6952 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6953 register immediate operand to a register.
6954 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6956 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6959 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6961 2017-07-27 Richard Biener <rguenther@suse.de>
6963 PR tree-optimization/81573
6964 PR tree-optimization/81494
6965 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6966 multi defuse cycle case.
6968 2017-07-27 Richard Biener <rguenther@suse.de>
6970 PR tree-optimization/81571
6971 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6974 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6976 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6977 earlier and only if MASK_FPU is set. Adjust formatting.
6979 2017-07-27 Martin Liska <mliska@suse.cz>
6981 * opt-functions.awk: Add validation of value of Init.
6982 * optc-gen.awk: Pass new argument.
6984 2017-07-27 Martin Liska <mliska@suse.cz>
6986 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6987 Fix wrong condition.
6989 2017-07-27 Martin Liska <mliska@suse.cz>
6991 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6992 BBs and edges seen by autoFDO.
6994 2017-07-27 Richard Biener <rguenther@suse.de>
6996 PR tree-optimization/81502
6997 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6998 with incompatible but same sized type.
6999 (execute_update_addresses_taken): Likewise.
7001 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7003 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7004 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7006 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7009 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7010 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7011 Change s_operand to memory_operand.
7013 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7015 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7016 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7017 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7018 Emit instructions rather than returning an expression. Handle TFmode
7019 and KFmode by casting to TImode.
7020 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7021 (rs6000_emit_le_vsx_store): Likewise.
7022 * config/rs6000/vsx.md (VSX_TI): New iterator.
7023 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7024 (*vsx_le_undo_permute_<mode>): Likewise.
7025 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7026 emit the split sequence.
7027 (*vsx_le_perm_store_<mode>): Likewise.
7029 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7031 PR tree-optimization/81555
7032 PR tree-optimization/81556
7033 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7034 if true, force CHANGED for the recursive invocation.
7035 (reassociate_bb): Remember original length of ops array, pass
7036 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7038 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7039 attributes for noipa attribute. For naked attribute use
7040 lookup_attribute first before lookup_attribute_spec.
7041 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7043 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7044 for functions with noipa attribute.
7045 (cgraph_externally_visible_p): Return true for functions with noipa
7047 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7048 for functions with noipa attribute.
7049 * doc/extend.texi: Document noipa function attribute.
7050 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7051 also for functions with noipa attribute.
7052 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7054 2017-07-26 Andrew Pinski <apinski@cavium.com>
7056 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7057 vec_unalign_load_cost and vec_unalign_store_cost.
7059 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7061 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7062 -mvsx-small-integer option.
7063 (ISA_3_0_MASKS_IEEE): Likewise.
7064 (OTHER_VSX_VECTOR_MASKS): Likewise.
7065 (POWERPC_MASKS): Likewise.
7066 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7067 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7068 code, only testing for DImode being allowed in non-VSX floating
7070 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7071 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7073 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7074 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7075 TARGET_P8_VECTOR test.
7076 (rs6000_secondary_reload_simple_move): Likewise.
7077 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7078 since TARGET_P9_VECTOR was already tested.
7079 (rs6000_opt_masks): Remove -mvsx-small-integer.
7080 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7081 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7083 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7084 test for TARGET_VEXTRACTUB was used, and that uses
7086 (p9 extract splitter): Likewise.
7087 (vsx_extract_<mode>_di_p9): Likewise.
7088 (vsx_extract_<mode>_store_p9): Likewise.
7089 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7090 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7091 the elimination of TARGET_VSX_SMALL_INTEGER.
7092 (vsx_extract_<mode>_p8): Likewise.
7093 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7094 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7095 (vsx_set_<mode>_p9): Likewise.
7096 (vsx_set_v4sf_p9): Likewise.
7097 (vsx_set_v4sf_p9_zero): Likewise.
7098 (vsx_insert_extract_v4sf_p9): Likewise.
7099 (vsx_insert_extract_v4sf_p9_2): Likewise.
7100 * config/rs6000/rs6000.md (sign extend splitter): Change
7101 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7102 (floatsi<mode>2_lfiwax_mem): Likewise.
7103 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7104 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7105 since a test for TARGET_P9_VECTOR was used.
7106 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7107 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7108 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7109 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7110 TARGET_P8_VECTOR test.
7111 (fix_trunc<mode>si2_stfiwx): Likewise.
7112 (fix_trunc<mode>si2_internal): Likewise.
7113 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7114 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7116 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7117 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7118 TARGET_P8_VECTOR test.
7119 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7120 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7121 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7123 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7124 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7125 since a test for TARGET_P9_VECTOR was used.
7126 (splitter for loading small constants): Likewise.
7128 2017-07-26 Andrew Pinski <apinski@cavium.com>
7130 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7133 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7136 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7137 (fp_valid_at): Likewise.
7139 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7141 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7142 (qdf24xx_addrcost_table): Likewise.
7143 (cortexa57_tunings): Update to use generic_branch_cost.
7144 (cortexa72_tunings): Likewise.
7145 (cortexa73_tunings): Likewise.
7146 (qdf24xx_tunings): Likewise.
7148 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7150 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7151 (thunderx2t99_branch_cost): Likewise.
7152 (cortexa35_tunings): Update to use generic_branch_cost.
7153 (cortexa53_tunings): Likewise.
7154 (cortexa57_tunings): Likewise.
7155 (cortexa72_tunings): Likewise.
7156 (cortexa73_tunings): Likewise.
7157 (thunderx2t99_tunings): Likewise.
7159 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7161 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7162 (sparc_option_override): Honour MASK_FSMULD.
7163 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7164 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7165 * config/sparc/sparc.opt (mfsmuld): New option.
7166 * doc/invoke.texi (mfsmuld): Document option.
7168 2017-07-26 Marek Polacek <polacek@redhat.com>
7171 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7173 2017-07-26 Richard Biener <rguenther@suse.de>
7175 * gimple-match-head.c (do_valueize): Return OP if valueize
7177 (get_def): New helper to get at the def stmt of a SSA name
7179 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7180 do_valueize to get at the def stmt.
7181 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7183 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7186 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7188 2017-07-26 Martin Liska <mliska@suse.cz>
7191 * function.c (expand_function_start): Make expansion of
7192 nonlocal_goto_save_area after parm_birth_insn.
7194 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7196 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7197 from all CPU target flags enable members.
7199 2017-07-26 Richard Biener <rguenther@suse.de>
7201 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7202 (decision_tree::gen): Make 'type' const.
7203 (write_predicate): Likewise.
7205 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7207 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7208 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7209 (rs6000_option_override_internal): Likewise.
7210 (rs6000_expand_vector_set): Likewise.
7211 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7212 (TARGET_UPPER_REGS_SF): Likewise.
7213 (TARGET_UPPER_REGS_DI): Likewise.
7214 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7215 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7216 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7217 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7218 (Splitters for DI constants in Altivec registers): Likewise.
7219 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7220 (vsx_set_v4sf_p9): Likewise.
7221 (vsx_set_v4sf_p9_zero): Likewise.
7222 (vsx_insert_extract_v4sf_p9): Likewise.
7223 (vsx_insert_extract_v4sf_p9_2): Likewise.
7225 2017-07-25 Carl Love <cel@us.ibm.com>
7227 * doc/extend.texi: Update the built-in documentation file for the
7228 existing built-in functions
7229 vector signed char vec_cnttz (vector signed char);
7230 vector unsigned char vec_cnttz (vector unsigned char);
7231 vector signed short vec_cnttz (vector signed short);
7232 vector unsigned short vec_cnttz (vector unsigned short);
7233 vector signed int vec_cnttz (vector signed int);
7234 vector unsigned int vec_cnttz (vector unsigned int);
7235 vector signed long long vec_cnttz (vector signed long long);
7236 vector unsigned long long vec_cnttz (vector unsigned long long);
7238 2017-07-25 Andrew Pinski <apinski@cavium.com>
7240 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7241 accesses where the use is for the first operand of a BIT_INSERT.
7243 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7246 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7247 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7249 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7251 * config/i386/gstabs.h: Delete.
7252 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7254 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7256 * config/i386/i386.c (ix86_decompose_address): Do not check for
7257 register RTX when looking at index_reg or base_reg.
7258 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7260 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7262 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7263 to update EH info here.
7265 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7267 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7269 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7271 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7273 2017-07-25 Torsten Duwe <duwe@suse.de>
7275 * common.opt: Introduce -fpatchable-function-entry
7276 command line option, and its variables function_entry_patch_area_size
7277 and function_entry_patch_area_start.
7278 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7279 including a two-value parser.
7280 * target.def (print_patchable_function_entry): New target hook.
7281 * targhooks.h (default_print_patchable_function_entry): New function.
7282 * targhooks.c (default_print_patchable_function_entry): Likewise.
7283 * toplev.c (process_options): Switch off IPA-RA if
7284 patchable function entries are being generated.
7285 * varasm.c (assemble_start_function): Look at the
7286 patchable-function-entry command line switch and current
7287 function attributes and maybe generate NOP instructions by
7288 calling the print_patchable_function_entry hook.
7289 * doc/extend.texi: Document patchable_function_entry attribute.
7290 * doc/invoke.texi: Document -fpatchable_function_entry
7291 command line option.
7292 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7294 * doc/tm.texi: Re-generate.
7296 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7299 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7300 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7302 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7304 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7306 2017-07-25 Richard Biener <rguenther@suse.de>
7308 PR tree-optimization/81455
7309 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7310 not walk in cycles when looking for guards.
7312 2017-07-25 Richard Biener <rguenther@suse.de>
7314 PR tree-optimization/81529
7315 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7316 when optimizing backedge uses.
7318 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7320 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7322 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7323 to dl_section_ref. On AIX, append an expression to subtract
7324 the size of the section length to dl_section_ref.
7326 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7328 * configure.ac: If any of the config.* scripts fail, exit 1.
7329 * configure: Regenerate.
7331 2017-07-25 Richard Biener <rguenther@suse.de>
7334 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7335 of immediate uses, be more verbose on errors.
7337 2017-07-25 Richard Biener <rguenther@suse.de>
7339 PR tree-optimization/81510
7340 * tree-vect-loop.c (vect_is_simple_reduction): When the
7341 reduction stmt is not inside the loop bail out.
7343 2017-07-25 Richard Biener <rguenther@suse.de>
7345 PR tree-optimization/81303
7346 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7347 profitability check against LOOP_VINFO_NITERSM1.
7349 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7351 * domwalk.c (cmp_bb_postorder): Simplify.
7352 (sort_bbs_postorder): New function. Use it...
7353 (dom_walker::walk): ...here to optimize common cases.
7355 2017-07-25 Martin Liska <mliska@suse.cz>
7358 * ipa-visibility.c (function_and_variable_visibility): Make the
7359 redirection just on target that supports aliasing.
7360 Fix GNU coding style.
7362 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7365 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7367 * config/arm/rtems.h: Likewise.
7368 * config/bfin/rtems.h: Likewise.
7369 * config/i386/rtemself.h: Likewise.
7370 * config/lm32/rtems.h: Likewise.
7371 * config/m32c/rtems.h: Likewise.
7372 * config/m68k/rtemself.h: Likewise.
7373 * config/microblaze/rtems.h: Likewise.
7374 * config/mips/rtems.h: Likewise.
7375 * config/moxie/rtems.h: Likewise.
7376 * config/nios2/rtems.h: Likewise.
7377 * config/powerpcspe/rtems.h: Likewise.
7378 * config/rs6000/rtems.h: Likewise.
7379 * config/rtems.h: Likewise.
7380 * config/sh/rtems.h: Likewise.
7381 * config/sh/rtemself.h: Likewise.
7382 * config/sparc/rtemself.h: Likewise.
7384 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7387 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7388 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7389 * tree-ssa-structalias.c (alias_get_name): Same.
7391 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7394 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7395 instructions if no du chain is found.
7397 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7399 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7401 2017-07-25 Richard Biener <rguenther@suse.de>
7404 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7407 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7409 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7411 (ISA_2_7_MASKS_SERVER): Likewise.
7412 (ISA_3_0_MASKS_IEEE): Likewise.
7413 (OTHER_P8_VECTOR_MASKS): Likewise.
7414 (OTHER_VSX_VECTOR_MASKS): Likewise.
7415 (POWERPC_MASKS): Likewise.
7416 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7417 duplicate list of options.
7418 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7419 explicit -mupper-regs options.
7420 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7421 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7422 alias for -mupper-regs-df.
7423 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7424 (rs6000_init_hard_regno_mode_ok): Likewise.
7425 (rs6000_option_override_internal): Likewise.
7426 (rs6000_opt_masks): Likewise.
7427 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7428 options in terms of whether -mvsx or -mpower8-vector was used.
7429 (TARGET_UPPER_REGS_DI): Likewise.
7430 (TARGET_UPPER_REGS_SF): Likewise.
7431 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7432 -mupper-regs-* options.
7434 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7436 * passes.c (emergency_dump_function): Print some empty lines and a
7437 header before the RTL dump.
7439 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7441 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7443 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7446 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7447 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7449 2017-07-24 Carl Love <cel@us.ibm.com>
7451 * config/rs6000/rs6000-c.c: Add support for built-in functions
7452 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7453 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7454 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7455 vec_extract_fp_from_shortl): Add defines for the two builtins.
7456 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7457 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7459 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7460 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7461 * doc/extend.texi: Update the built-in documentation file for the
7462 new built-in function.
7464 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7467 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7469 * doc/generic.texi: Likewise.
7470 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7471 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7473 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7475 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7476 (aarch64_mls_elt_merge<mode>): Likewise.
7478 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7480 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7481 having __cxa_atexit.
7483 2017-07-23 Michael Collison <michael.collison@arm.com>
7485 * config/arm/arm.c (arm_option_override): Deprecate
7486 use of -mstructure-size-boundary.
7487 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7488 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7490 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7493 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7494 Reduce cost estimate for direct moves.
7496 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7499 * config/i386/i386.c (ix86_option_override_internal): Disable
7500 BMI, BMI2 and TBM instructions for -m16.
7502 2017-07-21 Carl Love <cel@us.ibm.com>
7504 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7505 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7506 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7508 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7509 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7510 VMULOSW): New enum "unspec" values.
7511 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7512 altivec_vmulosw): New patterns.
7513 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7514 VMULOSW): Add definitions.
7516 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7518 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7519 (qdf24xx): Likewise.
7520 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7521 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7522 (AARCH64_FL_V8_1): Renumber.
7523 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7524 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7525 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7526 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7527 rdma to feature modifiers list.
7529 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7532 * ipa-visibility (function_and_variable_visibility): Convert
7533 recursive PLT call to direct call if appropriate.
7535 2017-07-21 Andrew Pinski <apinski@cavium.com>
7537 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7538 operand 1 to see if the types precision matches.
7539 * fold-const.c (operand_equal_p): Likewise.
7541 2017-07-21 Richard Biener <rguenther@suse.de>
7543 PR tree-optimization/81303
7544 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7545 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7546 (vect_peeling_hash_get_lowest_cost): Adjust.
7547 (vect_enhance_data_refs_alignment): Likewise. Use
7548 vect_get_peeling_costs_all_drs to compute the penalty for no
7549 peeling to match up costs.
7551 2017-07-21 Richard Biener <rguenther@suse.de>
7553 PR tree-optimization/81500
7554 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7555 we didn't identify a reduction path.
7557 2017-07-21 Tom de Vries <tom@codesourcery.com>
7558 Cesar Philippidis <cesar@codesourcery.com>
7560 PR gcov-profile/81442
7561 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7564 2017-07-21 Tom de Vries <tom@codesourcery.com>
7567 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7569 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7570 nvptx_override_options_after_change.
7572 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7574 * dwarf2out.c (output_file_names): Avoid double testing for
7577 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7579 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7581 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7583 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7585 (try_crossjump_to_edge): Do not punt on partitioned functions.
7587 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7589 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7590 Put all BBs reachable only via paths crossing cold region to cold
7592 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7594 2016-07-21 Richard Biener <rguenther@suse.de>
7596 PR tree-optimization/81303
7597 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7598 into account prologue and epilogue iterations when raising
7599 min_profitable_iters to sth at least covering one vector iteration.
7601 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7603 * config/arm/arm.c (arm_test_cpu_arch_dat):
7606 2017-07-20 Nathan Sidwell <nathan@acm.org>
7608 Remove TYPE_METHODS.
7609 * tree.h (TYPE_METHODS): Delete.
7610 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7611 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7612 (dbxout_type_methods): Scan TYPE_FIELDS.
7613 (dbxout_type): Don't check TYPE_METHODS here.
7614 * function.c (use_register_for_decl): Always ignore register for
7615 class types when not optimizing.
7616 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7617 * tree.c (free_lang_data_in_type): Stitch out member functions and
7618 templates from TYPE_FIELDS.
7619 (build_distinct_type_copy, verify_type_variant,
7620 verify_type): Member fns are on TYPE_FIELDS.
7621 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7622 * tree-pretty-print.c (dump_generic_node): Likewise.
7624 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7627 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7628 V2TImode and V4TImode.
7629 (ix86_expand_vector_extract): Likewise.
7630 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7631 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7632 (ssescalarmode): Handle V4TImode and V2TImode.
7633 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7634 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7635 (VEXTRACTI128_MODE): New mode iterator.
7636 (splitter for *vec_extractv?ti first element): New.
7637 (VEC_INIT_MODE): New mode iterator.
7638 (vec_init<mode>): Consolidate 3 expanders into one using
7639 VEC_INIT_MODE mode iterator.
7641 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7643 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7644 non_spilled_static_chain_regno_p.
7646 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7648 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7650 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7652 * bb-reorder.c (connect_traces): Allow copying of blocks within
7655 2017-07-20 Richard Biener <rguenther@suse.de>
7657 * gimple.h (gimple_phi_result): Add gphi * overload.
7658 (gimple_phi_result_ptr): Likewise.
7659 (gimple_phi_arg): Likewise. Adjust index assert to only
7660 allow actual argument accesses rather than all slots available
7662 (gimple_phi_arg_def): Add gphi * overload.
7663 * tree-phinodes.c (make_phi_node): Initialize only actual
7665 (resize_phi_node): Clear memory not covered by old node,
7666 do not initialize excess argument slots.
7667 (reserve_phi_args_for_new_edge): Initialize new argument slot
7670 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7672 PR tree-optimization/81388
7674 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7676 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7677 by removing computation of may_be_zero.
7679 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7680 Tom de Vries <tom@codesourcery.com>
7683 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7684 when gimple level profile disagrees with what RTL expander did.
7686 2017-07-20 Richard Biener <rguenther@suse.de>
7688 PR tree-optimization/61171
7689 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7690 (vect_analyze_stmt): Add slp instance parameter.
7691 (vectorizable_reduction): Likewise.
7692 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7693 (vect_is_simple_reduction): Deal with chains not detected
7694 as SLP reduction chain, specifically not properly associated
7695 chains containing a mix of plus/minus.
7696 (get_reduction_op): Remove.
7697 (get_initial_defs_for_reduction): Simplify, pass in whether
7698 this is a reduction chain, pass in the SLP node for the PHIs.
7699 (vect_create_epilog_for_reduction): Get the SLP instance as
7701 (vectorizable_reduction): Get the SLP instance as arg.
7702 During analysis remember the SLP node with the PHIs in the
7703 instance. Simplify getting at the vectorized reduction PHIs.
7704 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7705 through SLP instance.
7706 (vect_slp_analyze_operations): Likewise.
7707 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7708 (vect_transform_stmt): Likewise.
7710 2017-07-20 Tom de Vries <tom@codesourcery.com>
7712 PR tree-optimization/81489
7713 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7714 read of phi arg location to before loop that modifies phi.
7716 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7718 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7721 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7724 * except.c (execute): Fix ordering issue.
7726 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7728 PR rtl-optimization/81423
7729 * combine.c (make_compound_operation_int): Don't try to optimize
7730 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7732 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7734 PR rtl-optimization/81423
7735 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7736 with a constant that is -1 in the truncated to mode.
7738 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7740 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7741 (determine_unlikely_bbs): ... here.
7742 * predict.h (propagate_unlikely_bbs_forward): Declare.
7743 * cfgexpand.c (pass_expand::execute): Use it.
7744 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7746 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7747 propagate_unlikely_bbs_forward.
7749 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7752 * except.c (maybe_add_nop_after_section_switch): New function.
7755 2017-07-19 Tom de Vries <tom@codesourcery.com>
7757 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7759 2017-07-19 Tom de Vries <tom@codesourcery.com>
7761 * gimple.h (gimple_phi_arg): Make assert more strict.
7763 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7765 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7766 * config/rs6000/mmintrin.h: New file.
7767 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7769 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7771 PR tree-optimization/81346
7772 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7774 2017-07-19 Tom de Vries <tom@codesourcery.com>
7776 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7778 2017-07-19 Tom de Vries <tom@codesourcery.com>
7780 * config/nvptx/nvptx-modes.def: Add V2DImode.
7781 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7782 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7783 (nvptx_output_mov_insn): Handle lack of mov.b128.
7784 (nvptx_print_operand): Handle 'H' and 'L' codes.
7785 (nvptx_vector_mode_supported): Allow V2DImode.
7786 (nvptx_preferred_simd_mode): New function.
7787 (nvptx_data_alignment): New function.
7788 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7789 nvptx_preferred_simd_mode.
7790 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7792 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7794 2017-07-19 Tom de Vries <tom@codesourcery.com>
7796 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7797 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7798 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7799 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7800 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7801 (mov<VECIM>_insn): New define_insn.
7802 (define_expand "mov<VECIM>): New define_expand.
7804 2017-07-19 Tom de Vries <tom@codesourcery.com>
7806 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7808 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7810 PR tree-optimization/81346
7811 * fold-const.h (fold_div_compare, range_check_type): Declare.
7812 * fold-const.c (range_check_type): New function.
7813 (build_range_check): Use range_check_type.
7814 (fold_div_compare): No longer static, rewritten into
7815 a match.pd helper function.
7816 (fold_comparison): Don't call fold_div_compare here.
7817 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7820 2017-07-19 Nathan Sidwell <nathan@acm.org>
7822 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7823 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7824 * tree.c (find_decls_types_r, verify_type): Use
7825 TYPE_{MIN,MAX}_VALUE_RAW.
7826 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7827 (hash_tree): Likewise.
7828 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7830 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7833 2017-07-18 Tom de Vries <tom@codesourcery.com>
7836 * omp-expand.c (expand_omp_for_static_chunk): Handle
7837 equal-argument loop exit phi.
7839 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7842 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7843 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7844 operand 2 predicate.
7845 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7846 operand 2 predicate.
7847 (ror,rol -> rorx splitters): Use const_int_operand as
7848 operand 2 predicate.
7850 2017-06-18 Richard Biener <rguenther@suse.de>
7852 PR tree-optimization/81410
7853 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7854 the gap in the ! slp_perm SLP case after each group.
7856 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7859 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7862 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7865 * predict.c (set_even_probabilities): Cleanup; do not affect
7866 probabilities that are already known.
7867 (combine_predictions_for_bb): Call even when count is set.
7869 2017-07-18 Nathan Sidwell <nathan@acm.org>
7871 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7874 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7877 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7878 optimization for loop niter analysis.
7880 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7883 * config/avr/avr.c (avr_optimize_casesi): Don't use
7884 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7886 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7888 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7889 body_cost_vec from _vect_peel_extended_info.
7890 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7891 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7894 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7896 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7898 2017-07-18 Richard Biener <rguenther@suse.de>
7900 PR tree-optimization/80620
7901 PR tree-optimization/81403
7902 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7903 info when re-using a VN table entry.
7905 2017-07-18 Richard Biener <rguenther@suse.de>
7907 PR tree-optimization/81418
7908 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7909 vectype_in. Verify that with lane-reducing reduction operations
7910 we have a single def-use cycle.
7912 2017-07-17 Carl Love <cel@us.ibm.com>
7914 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7916 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7917 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7918 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7919 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7920 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7921 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7922 VMULOSW): New enum "unspec" values.
7923 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7924 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7925 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7926 altivec_vmulosw): New patterns.
7927 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7928 VMULOSW): Add definitions.
7929 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7931 * config/alpha/alpha.c: Include predict.h.
7933 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7935 * tree-vrp.c (compare_assert_loc): Fix comparison function
7936 to return predictable results.
7938 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7940 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7943 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7944 * doc/invoke.texi (mexpand-adddi): Update text.
7946 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7948 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7949 that also clobbers the CC register. The old expand code is moved
7951 (*arc_clzsi2): ... here.
7952 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7953 the CC register. The old expand code is moved to ...
7954 (arc_ctzsi2): ... here.
7956 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7958 * config/arc/arc.opt (mindexed-loads): Use initial value
7959 TARGET_INDEXED_LOADS_DEFAULT.
7960 (mauto-modify-reg): Use initial value
7961 TARGET_AUTO_MODIFY_REG_DEFAULT.
7962 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7963 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7964 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7965 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7967 2017-07-17 Martin Liska <mliska@suse.cz>
7970 * opts.c (finish_options): Do not allow -fgnu-tm
7971 w/ -fsanitize={kernel-,}address. Say sorry.
7973 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7976 * tree-loop-distribution.c (classify_partition): Only assert on
7977 numer of iterations.
7978 (merge_dep_scc_partitions): Delete prameter. Update function call.
7979 (distribute_loop): Remove code handling loop with unknown niters.
7980 (pass_loop_distribution::execute): Skip loop with unknown niters.
7982 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7985 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7986 function sort_partitions_by_post_order.
7988 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7990 PR tree-optimization/81374
7991 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7992 the max index of basic blocks, rather than number of basic blocks.
7994 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7996 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7998 (arc_legitimate_pic_operand_p): Likewise.
7999 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8001 (arc_needs_pcl_p): Likewise.
8002 (arc_legitimate_pc_offset_p): Likewise.
8003 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8004 function is also used in constrains.md.
8005 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8006 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8007 PLUS. Only return true/false in known cases, otherwise assert.
8008 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8009 is already called in arc_legitimate_constant_p.
8010 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8012 (LEGITIMATE_PIC_OPERAND_P): Use
8013 arc_raw_symbolic_reference_mentioned_p function.
8014 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8019 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8020 Andrew Burgess <andrew.burgess@embecosm.com>
8022 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8023 (arc_return_address_register): New function.
8024 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8025 (arc_handle_fndecl_attribute): Add naked attribute.
8026 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8027 (TARGET_WARN_FUNC_RETURN): Likewise.
8028 (arc_allocate_stack_slots_for_args): New function.
8029 (arc_warn_func_return): Likewise.
8030 (machine_function): Change type fn_type.
8031 (arc_compute_function_type): Consider new naked function type,
8032 change function return type.
8033 (arc_must_save_register): Adapt to handle new
8034 arc_compute_function_type's return type.
8035 (arc_expand_prologue): Likewise.
8036 (arc_expand_epilogue): Likewise.
8037 (arc_return_address_regs): Delete.
8038 (arc_return_address_register): New function.
8039 (arc_epilogue_uses): Use above function.
8040 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8041 (arc_function_type): Change encoding, add naked type.
8042 (ARC_INTERRUPT_P): Change to handle the new encoding.
8043 (ARC_FAST_INTERRUPT_P): Likewise.
8044 (ARC_NORMAL_P): Define.
8045 (ARC_NAKED_P): Likewise.
8046 (arc_compute_function_type): Delete prototype.
8047 * config/arc/arc.md (in_ret_delay_slot): Use
8048 arc_return_address_register function.
8049 (simple_return): Likewise.
8050 (p_return_i): Likewise.
8052 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8054 PR tree-optimization/81428
8055 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8056 can't be built for those types.
8058 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8060 Remove stuff dead since r239246.
8062 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8063 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8064 (avr_inform_devices): Remove dead stuff.
8066 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8068 * config/arm/arm_neon.h: Fix softp typo.
8070 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8072 PR tree-optimization/81365
8073 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8074 aggregate moves onto bb predecessor edges, make sure there are no
8075 loads that could alias the lhs in between the start of bb and the
8078 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8081 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8082 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8083 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8085 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8087 PR tree-optimization/81396
8088 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8089 (init_symbolic_number): Initialize it to 1.
8090 (perform_symbolic_merge): Add n_ops from both operands into the new
8092 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8093 without base_addr as useless if they need more than one operation.
8094 (bswap_replace): Handle !bswap case for NULL base_addr.
8096 2017-07-17 Tom de Vries <tom@codesourcery.com>
8099 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8102 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8104 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8105 conditional builtin define __FIX_LEON3FT_B2BST.
8107 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8109 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8110 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8113 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8115 PR rtl-optimization/81424
8116 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8117 to remove potential trapping from operands if -fnon-call-exceptions.
8119 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8122 profile_proability for scalling.
8123 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8125 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8127 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8129 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8131 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8132 fixpoint arithmetics.
8134 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8136 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8137 fixpoint arithmetics.
8139 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8141 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8142 fixpoint arithmetics.
8144 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8146 * profile-count.h (profile_probability::from_reg_br_prob_note,
8147 profile_probability::to_reg_br_prob_note): New functions.
8148 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8149 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8150 * predict.c (probability_reliable_p): Update.
8151 (edge_probability_reliable_p): Update.
8152 (br_prob_note_reliable_p): Update.
8153 (invert_br_probabilities): Update.
8154 (add_reg_br_prob_note): New function.
8155 (combine_predictions_for_insn): Update.
8156 * asan.c (asan_clear_shadow): Update.
8157 * cfgbuild.c (compute_outgoing_frequencies): Update.
8158 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8159 (update_br_prob_note): Update.
8160 (rtl_verify_edges): Update.
8161 (purge_dead_edges): Update.
8162 (fixup_reorder_chain): Update.
8163 * emit-rtl.c (try_split): Update.
8164 * ifcvt.c (cond_exec_process_insns): Update.
8165 (cond_exec_process_if_block): Update.
8166 (dead_or_predicable): Update.
8167 * internal-fn.c (expand_addsub_overflow): Update.
8168 (expand_neg_overflow): Update.
8169 (expand_mul_overflow): Update.
8170 * loop-doloop.c (doloop_modify): Update.
8171 * loop-unroll.c (compare_and_jump_seq): Update.
8172 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8173 * predict.h: Update.
8174 * reorg.c (mostly_true_jump): Update.
8176 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8177 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8178 * config/arc/arc.c: (emit_unlikely_jump): Update.
8179 * config/arm/arm.c: (emit_unlikely_jump): Update.
8180 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8181 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8182 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8183 (ix86_print_operand): Update.
8184 (ix86_split_fp_branch): Update.
8185 (predict_jump): Update.
8186 * config/ia64/ia64.c (ia64_print_operand): Update.
8187 * config/mmix/mmix.c (mmix_print_operand): Update.
8188 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8189 (rs6000_expand_split_stack_prologue): Update.
8190 * config/rs6000/rs6000.c: Update.
8191 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8192 (s390_expand_vec_movstr): Update.
8193 (s390_expand_cs_tdsi): Update.
8194 (s390_expand_split_stack_prologue): Update.
8195 * config/sh/sh.c (sh_print_operand): Update.
8196 (expand_cbranchsi4): Update.
8197 (expand_cbranchdi4): Update.
8198 * config/sparc/sparc.c (output_v9branch): Update.
8199 * config/spu/spu.c (get_branch_target): Update.
8200 (ea_load_store_inline): Update.
8201 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8202 * config/tilepro/tilepro.c: Update.
8204 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8206 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8207 (gimplify_save_expr): Likewise.
8209 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8211 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8213 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8215 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8217 * timevar.def (TV_IPA_FNSUMMARY): Define.
8219 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8221 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8222 to back store errata sensitive sequence from being generated.
8223 (sqrtdf2_fix): Likewise.
8225 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8227 * tree-ssa-threadupdate.c (compute_path_counts,
8228 update_joiner_offpath_counts): Use profile_probability.
8230 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8233 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8235 * config/arm/arm-c.c (arm_cpu_builtins): Define
8236 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8238 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8240 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8241 array entries to represent __ieee128 versions of the
8242 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8243 scalar_extract_sig, and scalar_insert_exp built-in functions.
8244 (altivec_resolve_overloaded_builtin): Add special case handling
8245 for the __builtin_scalar_insert_exp function, as represented by
8246 the P9V_BUILTIN_VEC_VSIEDP constant.
8247 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8248 exponent support for __ieee128 argument.
8249 (VSESQP): Add scalar extract signature support for __ieee128
8251 (VSTDCNQP): Add scalar test negative support for __ieee128
8253 (VSIEQP): Add scalar insert exponent support for __int128 argument
8254 with __ieee128 result.
8255 (VSIEQPF): Add scalar insert exponent support for __ieee128
8256 argument with __ieee128 result.
8257 (VSTDCQP): Add scalar test data class support for __ieee128
8259 (VSTDCNQP): Add overload support for scalar test negative with
8261 (VSTDCQP): Add overload support for scalar test data class
8263 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8265 (UNSPEC_VSX_SIEXPQP): New constant.
8266 (xsxexpqp): New insn for VSX scalar extract exponent quad
8268 (xsxsigqp): New insn for VSX scalar extract significand quad
8270 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8271 precision with floating point argument.
8272 (xststdcqp): New expand for VSX scalar test data class quad
8274 (xststdcnegqp): New expand for VSX scalar test negative quad
8276 (xststdcqp): New insn to match expansions for VSX scalar test data
8277 class quad precision and VSX scalar test negative quad precision.
8278 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8279 special case operand checking to enforce that second operand of
8280 VSX scalar test data class with quad precision argument is a 7-bit
8282 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8283 prototypes and descriptions of __ieee128 versions of
8284 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8285 scalar_test_data_class, and scalar_test_neg built-in functions.
8287 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8289 PR tree-optimization/81162
8290 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8291 replace a negate with an add.
8293 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8295 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8297 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8299 * config/arm/arm-c.c (arm_cpu_builtins): Define
8300 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8302 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8304 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8305 (armv8-r): Set ARM Cortex-R52 as default CPU.
8306 * config/arm/arm-tables.opt: Regenerate.
8307 * config/arm/arm-tune.md: Regenerate.
8308 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8310 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8311 extension for -mcpu=cortex-r52.
8313 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8315 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8316 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8317 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8318 (fp-armv8): Define it as FP_ARMv8 only.
8319 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8320 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8322 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8323 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8324 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8325 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8326 than TARGET_FPU_ARMV8.
8327 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8328 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8329 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8331 * config/arm/neon.md (neon_vrint): Likewise.
8332 (neon_vcvt): Likewise.
8333 (neon_<fmaxmin_op><mode>): Likewise.
8334 (<fmaxmin><mode>3): Likewise.
8335 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8336 * config/arm/predicates.md (arm_cond_move_operator): Check against
8337 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8339 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8341 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8344 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8346 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8347 loop in comment with memset.
8349 2017-07-14 Martin Liska <mliska@suse.cz>
8351 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8352 * dwarf2out.c (is_java): Remove the function.
8353 (output_pubname): Remove usage of the function.
8354 (lower_bound_default): Remove usage of DW_LANG_Java.
8355 (gen_compile_unit_die): Likewise.
8356 * gcc.c: Remove compiler defaults for .java and .zip files.
8357 * gimple-expr.c (remove_suffix): Change as there's no longer
8358 extension than 4-letter one.
8359 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8360 (gimplify_save_expr): Likewise.
8361 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8362 as it's possible even for other languages than Java.
8363 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8364 * lto-opts.c (lto_write_options): Remove reference to Java.
8365 * opts.c (strip_off_ending): Update file extension handling.
8366 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8367 * tree-eh.c (lower_resx): Likewise.
8368 * tree.c (free_lang_data_in_type): Remove dead code.
8369 (find_decls_types_r): Likewise.
8370 (build_common_builtin_nodes): Remove Java from a comment.
8371 (verify_type): Remove dead code.
8372 * varasm.c (assemble_external): Remove Java from a comment.
8374 2017-07-14 Martin Liska <mliska@suse.cz>
8376 * opts.c (finish_options): Add quotes.
8377 (common_handle_option): Likewise.
8379 2017-07-14 Martin Liska <mliska@suse.cz>
8381 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8382 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8384 * dwarf2out.c (lower_bound_default): Do not handle
8386 (gen_compile_unit_die): Likewise.
8387 * gcc.c: Remove default extension binding for GNU Pascal.
8388 * stmt.c: Remove Pascal language from a comment.
8389 * xcoffout.c: Likewise.
8391 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8394 * diagnostic-show-locus.c (fixit_cmp): New function.
8395 (layout::layout): Sort m_fixit_hints.
8396 (column_range::column_range): Assert that the values are valid.
8397 (struct char_span): New struct.
8398 (correction::overwrite): New method.
8399 (struct source_line): New struct.
8400 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8401 calls in terms of classes source_line and char_span, and
8402 correction::overwrite.
8403 (selftest::test_overlapped_fixit_printing_2): New function.
8404 (selftest::diagnostic_show_locus_c_tests): Call it.
8406 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8408 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8409 early if there is no lhs.
8411 2017-07-13 Martin Liska <mliska@suse.cz>
8413 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8414 (gen_reference_type_die): Likewise.
8415 * stor-layout.c: Remove Pascal-related comment.
8417 2017-07-13 Martin Liska <mliska@suse.cz>
8419 * opts.c (finish_options): Add quotes to error messages.
8420 (parse_sanitizer_options): Likewise.
8422 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8424 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8426 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8428 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8430 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8432 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8434 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8436 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8439 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8440 provides the hardware capability bits, define the macro
8441 __BUILTIN_CPU_SUPPORTS__.
8442 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8443 if GLIBC does not provide the hardware capability bits. Add a
8444 gcc_unreachable call if the built-in cpu function is neither
8445 __builtin_cpu_is nor __builtin_cpu_supports.
8446 (rs6000_get_function_versions_dispatcher): Change the warning
8447 that an old GLIBC is used which does not export the capability
8448 bits to be an error.
8449 * doc/extend.texi (target_clones attribute): Document the
8450 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8451 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8452 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8453 the macros defined by GCC if the newer GLIBC is available.
8455 2017-07-12 Jeff Law <law@redhat.com>
8457 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8458 remaining includes slightly.
8459 * config/riscv/riscv-builtins.c: Include profile-count.h.
8461 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8464 * config/avr/avr.c (avr_set_current_function): In diagnostic
8465 messages: Quote keywords and (parts of) identifiers.
8466 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8469 2017-07-12 Carl Love <cel@us.ibm.com>
8471 * config/rs6000/rs6000-c.c: Add support for built-in functions
8472 vector bool char vec_revb (vector bool char);
8473 vector bool short vec_revb (vector short char);
8474 vector bool int vec_revb (vector bool int);
8475 vector bool long long vec_revb (vector bool long long);
8476 * doc/extend.texi: Update the built-in documentation file for the
8477 new built-in functions.
8479 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8481 * config/s390/s390.md: Remove movcc splitter.
8483 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8485 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8486 load/store on condition.
8488 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8491 * config/avr/avr.c (avr_encode_section_info)
8492 [progmem && !TREE_READONLY]: Error if progmem object needs
8495 2017-07-11 Michael Collison <michael.collison@arm.com>
8497 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8500 2017-07-11 Carl Love <cel@us.ibm.com>
8502 * config/rs6000/rs6000-c.c: Add support for builtins
8503 vector unsigned int vec_parity_lsbb (vector signed int);
8504 vector unsigned int vec_parity_lsbb (vector unsigned int);
8505 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8506 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8507 vector unsigned long long vec_parity_lsbb (vector signed long long);
8508 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8509 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8510 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8511 * doc/extend.texi: Update the built-in documentation file for the
8512 new built-in functions.
8514 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8516 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8517 (layout::m_primary_loc): New field.
8518 (layout::layout): Initialize new field. Move location filtering
8519 logic from here to...
8520 (layout::maybe_add_location_range): ...this new method. Add
8521 support for filtering to just the lines already specified by other
8523 (layout::will_show_line_p): New method.
8524 (gcc_rich_location::add_location_if_nearby): New method.
8525 (selftest::test_add_location_if_nearby): New test function.
8526 (selftest::diagnostic_show_locus_c_tests): Call it.
8527 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8530 2017-07-11 Tom de Vries <tom@codesourcery.com>
8532 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8533 (bb_first_real_insn): New function.
8534 (nvptx_single): Add extra initialization of broadcasted condition
8537 2017-07-11 Nathan Sidwell <nathan@acm.org>
8539 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8541 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8543 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8544 Binutils doc as TEXI will mess them up.
8545 * doc/invoke.texi (AVR Options): Same here.
8547 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8549 * config/sparc/sparc.opt (mfix-ut700): New option.
8550 (mfix-gr712rc): Likewise.
8551 (sparc_fix_b2bst): New variable.
8552 * doc/invoke.texi (SPARC options): Document them.
8553 (ARM options): Fix warnings.
8554 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8555 instructions to prevent sequences that can trigger the store-store
8556 errata for certain LEON3FT processors.
8557 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8558 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8559 * config/sparc/sparc.md (fix_b2bst): New attribute.
8560 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8562 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8565 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8567 (*rsqrtsf2_sse): Ditto.
8569 (div<mode>3): Macroize insn from divdf3 and divsf3
8570 using MODEF mode iterator.
8572 2017-07-10 Martin Sebor <msebor@redhat.com>
8574 PR tree-optimization/80397
8575 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8576 instead of testing for equality to INTEGER_TYPE.
8578 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8580 * config.gcc: Remove uclibc from arc target spec.
8582 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8584 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8586 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8589 * lto-wrapper.c (remove_option): New function.
8590 (merge_and_complain): Merge PIC/PIE options more realistically.
8592 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8594 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8598 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8599 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8600 * config.in: Regenerate.
8601 * configure: Regenerate.
8602 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8603 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8604 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8605 (TARGET_GASISR_PROLOGUES): ...target mask.
8606 * common/config/avr/avr-common.c
8607 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8608 Set -mgas-isr-prologues.
8609 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8610 INSERT_PASS_BEFORE for it.
8611 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8612 * config/avr/avr.c (avr_option_override)
8613 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8614 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8615 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8616 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8617 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8619 (make_avr_pass_pre_proep): New function.
8620 (emit_push_sfr) <treg>: Add argument to function and use it
8622 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8623 and set machine->gasisr.yes.
8624 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8625 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8626 __gcc_isr.n_pushed to .L__stack_usage.
8627 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8628 (avr_asm_final_postscan_insn): ...this new static function.
8629 * config/avr/avr.h (machine_function)
8630 <is_no_gccisr, use_L__stack_usage>: New fields.
8631 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8632 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8633 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8634 (gasisr, *gasisr): New expander and insn.
8635 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8636 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8637 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8639 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8641 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8644 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8646 Move jump-tables out of .text again.
8649 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8650 (ASM_OUTPUT_ADDR_VEC): New function.
8651 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8652 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8653 INSN_ADDRESSes as asm comment.
8654 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8655 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8656 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8657 * config/avr/avr.md (*tablejump): Adjust comment.
8658 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8659 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8661 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8662 (avr_output_addr_vec): New proto.
8663 (avr_log_t) <insn_addresses>: New field.
8665 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8668 * config/i386/i386.c (ix86_function_arg_advance): Set
8669 outgoing_args_on_stack to true if there are outgoing arguments
8671 (ix86_function_arg): Likewise.
8672 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8673 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8674 * config/i386/i386.h (machine_function): Add
8675 outgoing_args_on_stack.
8677 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8679 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8681 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8683 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8685 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8686 (REAL_H): Remove $(MACHMODE_H).
8687 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8689 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8690 $(MACHMODE_H) and double-int.h.
8691 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8693 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8696 2017-07-07 Andrew Pinski <apinski@cavium.com>
8698 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8699 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8701 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8703 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8704 Add warning if GCC was not configured to link against a GLIBC that
8705 exports the hardware capability bits.
8706 (make_resolver_func): Make resolver function private and not a
8707 COMDAT function. Create the name with clone_function_name instead
8708 of make_unique_name.
8711 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8712 correct operand in doing the split.
8714 2017-07-07 Carl Love <cel@us.ibm.com>
8716 * config/rs6000/rs6000-c: Add support for built-in function
8717 vector unsigned short vec_pack_to_short_fp32 (vector float,
8719 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8720 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8721 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8722 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8723 (convert_4f32_8i16): Add define_expand.
8724 * doc/extend.texi: Update the built-in documentation file for the
8725 new built-in function.
8727 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8729 * config/sparc/m8.md: New file.
8730 * config/sparc/sparc.md: Include m8.md.
8732 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8734 * config/sparc/sparc.opt: New option -mvis4b.
8735 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8736 (sparc_option_override): Handle VIS4B.
8737 (enum sparc_builtins): Define
8738 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8739 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8740 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8741 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8742 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8743 (check_constant_argument): New function.
8744 (sparc_vis_init_builtins): Define builtins
8745 __builtin_vis_dictunpack{8,16,32},
8746 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8747 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8748 __builtin_vis_fpcmpde{8,16,32}shl and
8749 __builtin_vis_fpcmpur{8,16,32}shl.
8750 (sparc_expand_builtin): Check that the constant operands to
8751 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8752 constant and in range.
8753 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8755 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8756 (SPARC_IMM5_P): Likewise.
8757 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8758 (enabled): Handle vis4b.
8759 (UNSPEC_DICTUNPACK): New unspec.
8760 (UNSPEC_FPCMPSHL): Likewise.
8761 (UNSPEC_FPUCMPSHL): Likewise.
8762 (UNSPEC_FPCMPDESHL): Likewise.
8763 (UNSPEC_FPCMPURSHL): Likewise.
8764 (cpu_feature): New CPU feature `vis4b'.
8765 (dictunpack{8,16,32}): New insns.
8766 (FPCSMODE): New mode iterator.
8767 (fpcscond): New code iterator.
8768 (fpcsucond): Likewise.
8769 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8770 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8771 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8772 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8773 * config/sparc/constraints.md: Define constraints `q' for unsigned
8774 2-bit integer constants and `t' for unsigned 5-bit integer
8776 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8778 (imm5_operand_dictunpack16): Likewise.
8779 (imm5_operand_dictunpack32): Likewise.
8780 (imm2_operand): Likewise.
8781 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8782 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8783 ditunpack* and fpcmp*shl builtins.
8785 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8787 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8788 * config.in: Add HAVE_AS_SPARC6 define.
8789 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8791 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8793 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8794 (CPP_CPU_SPEC): Handle m8.
8795 (ASM_CPU_SPEC): Likewise.
8796 * config/sparc/sparc-opts.h (enum processor_type): Add
8798 * config/sparc/sparc.c (m8_costs): New struct.
8799 (sparc_option_override): Handle TARGET_CPU_m8.
8800 (sparc32_initialize_trampoline): Likewise.
8801 (sparc64_initialize_trampoline): Likewise.
8802 (sparc_issue_rate): Likewise.
8803 (sparc_register_move_cost): Likewise.
8804 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8805 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8806 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8807 (CPP_CPU_SPEC): Handle M8.
8808 (ASM_CPU_SPEC): Likewise.
8809 (AS_M8_FLAG): Define.
8810 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8811 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8812 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8814 * configure: Regenerate.
8815 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8820 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8822 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8823 ("*movdi_insn_sp32"): Do not set v3pipe.
8824 ("*movsi_insn"): Likewise.
8825 ("*movdi_insn_sp64"): Likewise.
8826 ("*movsf_insn"): Likewise.
8827 ("*movdf_insn_sp32"): Likewise.
8828 ("*movdf_insn_sp64"): Likewise.
8829 ("*zero_extendsidi2_insn_sp64"): Likewise.
8830 ("*sign_extendsidi2_insn"): Likewise.
8831 ("*mov<VM32:mode>_insn"): Likewise.
8832 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8833 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8834 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8835 ("<vlop:code><VL:mode>3"): Likewise.
8836 ("*not_<vlop:code><VL:mode>3"): Likewise.
8837 ("*nand<VL:mode>_vis"): Likewise.
8838 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8839 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8840 ("one_cmpl<VL:mode>2"): Likewise.
8841 ("faligndata<VM64:mode>_vis"): Likewise.
8842 ("alignaddrsi_vis"): Likewise.
8843 ("alignaddrdi_vis"): Likweise.
8844 ("alignaddrlsi_vis"): Likewise.
8845 ("alignaddrldi_vis"): Likewise.
8846 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8847 ("bmaskdi_vis"): Likewise.
8848 ("bmasksi_vis"): Likewise.
8849 ("bshuffle<VM64:mode>_vis"): Likewise.
8850 ("cmask8<P:mode>_vis"): Likewise.
8851 ("cmask16<P:mode>_vis"): Likewise.
8852 ("cmask32<P:mode>_vis"): Likewise.
8853 ("pdistn<P:mode>_vis"): Likewise.
8854 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8856 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8858 * config/sparc/sparc.md ("subtype"): New insn attribute.
8859 ("*wrgsr_sp64"): Set insn subtype.
8860 ("*rdgsr_sp64"): Likewise.
8861 ("alignaddrsi_vis"): Likewise.
8862 ("alignaddrdi_vis"): Likewise.
8863 ("alignaddrlsi_vis"): Likewise.
8864 ("alignaddrldi_vis"): Likewise.
8865 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8866 ("fexpand_vis"): Likewise.
8867 ("fpmerge_vis"): Likewise.
8868 ("faligndata<VM64:mode>_vis"): Likewise.
8869 ("bshuffle<VM64:mode>_vis"): Likewise.
8870 ("cmask8<P:mode>_vis"): Likewise.
8871 ("cmask16<P:mode>_vis"): Likewise.
8872 ("cmask32<P:mode>_vis"): Likewise.
8873 ("fchksm16_vis"): Likewise.
8874 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8875 ("fmean16_vis"): Likewise.
8876 ("fp<plusminus_insn>64_vis"): Likewise.
8877 ("<plusminus_insn>v8qi3"): Likewise.
8878 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8879 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8880 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8881 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8882 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8883 ("*movqi_insn"): Likewise.
8884 ("*movhi_insn"): Likewise.
8885 ("*movsi_insn"): Likewise.
8886 ("movsi_pic_gotdata_op"): Likewise.
8887 ("*movdi_insn_sp32"): Likewise.
8888 ("*movdi_insn_sp64"): Likewise.
8889 ("movdi_pic_gotdata_op"): Likewise.
8890 ("*movsf_insn"): Likewise.
8891 ("*movdf_insn_sp32"): Likewise.
8892 ("*movdf_insn_sp64"): Likewise.
8893 ("*zero_extendhisi2_insn"): Likewise.
8894 ("*zero_extendqihi2_insn"): Likewise.
8895 ("*zero_extendqisi2_insn"): Likewise.
8896 ("*zero_extendqidi2_insn"): Likewise.
8897 ("*zero_extendhidi2_insn"): Likewise.
8898 ("*zero_extendsidi2_insn_sp64"): Likewise.
8899 ("ldfsr"): Likewise.
8900 ("prefetch_64"): Likewise.
8901 ("prefetch_32"): Likewise.
8902 ("tie_ld32"): Likewise.
8903 ("tie_ld64"): Likewise.
8904 ("*tldo_ldub_sp32"): Likewise.
8905 ("*tldo_ldub1_sp32"): Likewise.
8906 ("*tldo_ldub2_sp32"): Likewise.
8907 ("*tldo_ldub_sp64"): Likewise.
8908 ("*tldo_ldub1_sp64"): Likewise.
8909 ("*tldo_ldub2_sp64"): Likewise.
8910 ("*tldo_ldub3_sp64"): Likewise.
8911 ("*tldo_lduh_sp32"): Likewise.
8912 ("*tldo_lduh1_sp32"): Likewise.
8913 ("*tldo_lduh_sp64"): Likewise.
8914 ("*tldo_lduh1_sp64"): Likewise.
8915 ("*tldo_lduh2_sp64"): Likewise.
8916 ("*tldo_lduw_sp32"): Likewise.
8917 ("*tldo_lduw_sp64"): Likewise.
8918 ("*tldo_lduw1_sp64"): Likewise.
8919 ("*tldo_ldx_sp64"): Likewise.
8920 ("*mov<VM32:mode>_insn"): Likewise.
8921 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8922 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8924 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8926 * config/sparc/sparc.md ("type"): New insn type viscmp.
8927 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8929 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8930 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8931 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8932 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8934 ("n7_vis_logical_11cycle"): Likewise.
8935 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8936 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8937 * config/sparc/niagara.md ("niag_vis"): Likewise.
8938 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8939 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8941 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8943 * config/sparc/sparc.md: New instruction type `bmask'.
8944 (bmaskdi_vis): Use the `bmask' type.
8945 (bmasksi_vis): Likewise.
8946 * config/sparc/ultra3.md (us3_array): Likewise.
8947 * config/sparc/niagara7.md (n7_array): Likewise.
8948 * config/sparc/niagara4.md (n4_array): Likewise.
8949 * config/sparc/niagara2.md (niag2_vis): Likewise.
8950 (niag3_vis): Likewise.
8951 * config/sparc/niagara.md (niag_vis): Likewise.
8953 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8955 * ipa-comdats.c: Remove optimize check from gate.
8956 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8957 for functions not optimized.
8958 (ipa_fn_summary_read): Skip optimize check.
8959 (ipa_fn_summary_write): Likewise.
8960 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8962 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8964 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8965 (check_callers): Give up on caller which is not optimized.
8966 (inline_small_functions): Likewise.
8967 (ipa_inline): Do not give up when not optimizing.
8968 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8969 away unoptimizes cdtors.
8970 (whole_program_function_and_variable_visibility): Do
8971 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8972 * ipa.c (process_references): Do not check optimize.
8973 (symbol_table::remove_unreachable_nodes): Update optimize check.
8974 (set_writeonly_bit): Update optimize check.
8975 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8976 (pass_ipa_single_use::gate): Remove.
8978 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8980 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8981 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8982 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8983 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8984 permute_load, permute_store, adjust_extract, adjust_splat,
8985 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8986 replace_swap_with_copy, dump_swap_insn_table,
8987 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8988 recombine_lvx_pattern, recombine_stvx_pattern,
8989 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8990 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8991 to file rs6000-p8swap.c.
8992 * config/rs6000/rs6000-p8swap.c: New file.
8993 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8994 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8995 and rs6000*-*-* targets.
8997 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8999 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9001 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9003 * lto-wrapper.c (merge_and_complain): Do not merge
9004 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9005 fsigned_zeros, ftrapping_math, fwrapv.
9006 (append_compiler_options): Do not track these options.
9007 (append_linker_options): Likewie
9009 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9011 * cgraphunit.c (cgraph_node::finalize_function): When
9012 !flag_toplevel_reorde set no_reorder flag.
9013 (varpool_node::finalize_decl): Likewise.
9014 (symbol_table::compile): Drop no toplevel reorder path.
9016 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9018 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9019 edges; zero probability is not better than uninitialized.
9021 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9023 * asan.h (asan_sanitize_allocas_p): Declare.
9024 * asan.c (asan_sanitize_allocas_p): New function.
9025 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9026 (handle_builtin_alloca): Likewise.
9027 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9028 if !asan_sanitize_allocas_p.
9029 * params.def (asan-instrument-allocas): Add new option.
9030 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9031 * opts.c (common_handle_option): Disable allocas sanitization for
9034 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9036 * asan.c: Include gimple-fold.h.
9037 (get_last_alloca_addr): New function.
9038 (handle_builtin_stackrestore): Likewise.
9039 (handle_builtin_alloca): Likewise.
9040 (asan_emit_allocas_unpoison): Likewise.
9041 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9042 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9043 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9044 (instrument_builtin_call): Pass gimple iterator to
9045 get_mem_refs_of_builtin_call.
9046 (last_alloca_addr): New global.
9047 * asan.h (asan_emit_allocas_unpoison): Declare.
9048 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9049 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9050 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9051 if function calls alloca.
9052 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9053 * gimple-fold.h (replace_call_with_value): Declare.
9054 * internal-fn.c: Include asan.h.
9055 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9056 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9058 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9060 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9061 (C_SELFTEST_FLAGS): New.
9062 (CPP_SELFTEST_FLAGS): New.
9063 (SELFTEST_DEPS): New, from deps of s-selftest.
9064 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9065 (CPP_SELFTEST_DEPS): New.
9066 (selftest): Add dependency on s-selftest-c++.
9067 (s-selftest): Rename to...
9068 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9069 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9070 than SELFTEST_FLAGS.
9071 (selftest-gdb): Rename to...
9072 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9074 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9075 (selftest-valgrind): Rename to...
9076 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9078 (selftest-valgrind): Reintroduce as an alias for
9079 selftest-c-valgrind.
9080 (s-selftest-c++): New.
9081 (selftest-c++-gdb): New.
9082 (selftest-c++-valgrind): New.
9084 2017-07-06 Olivier Hainque <hainque@adacore.com>
9086 * gcc.c (process_command): When deciding if undefined variables
9087 should be ignored when processing specs, accept "gcc -v" as well.
9089 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9091 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9092 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9094 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9096 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9097 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9098 * config/arm/arm-tables.opt: Regenerate.
9099 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9101 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9103 2017-07-06 Carl Love <cel@us.ibm.com>
9105 * ChangeLog: Clean up from mid air collision
9107 2017-07-06 Carl Love <cel@us.ibm.com>
9109 * config/rs6000/rs6000-c.c: Add support for built-in functions
9110 vector signed int vec_subc (vector signed int, vector signed int);
9111 vector signed __int128 vec_subc (vector signed __int128,
9112 vector signed __int128);
9113 vector unsigned __int128 vec_subc (vector unsigned __int128,
9114 vector unsigned __int128);
9115 vector signed int vec_sube (vector signed int, vector signed int,
9117 vector unsigned int vec_sube (vector unsigned int,
9118 vector unsigned int,
9119 vector unsigned int);
9120 vector signed __int128 vec_sube (vector signed __int128,
9121 vector signed __int128,
9122 vector signed__int128);
9123 vector unsigned __int128 vec_sube (vector unsigned __int128,
9124 vector unsigned __int128,
9125 vector unsigned __int128);
9126 vector signed int vec_subec (vector signed int, vector signed int,
9128 vector unsigned int vec_subec (vector unsigned int,
9129 vector unsigned int,
9130 vector unsigned int);
9131 vector signed __int128 vec_subec (vector signed __int128,
9132 vector signed __int128,
9133 vector signed__int128);
9134 vector unsigned __int128 vec_subec (vector unsigned __int128,
9135 vector unsigned __int128,
9136 vector unsigned __int128);
9137 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9138 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9139 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9140 BU_ALTIVEC_OVERLOAD_X definitions.
9141 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9142 * doc/extend.texi: Update the built-in documentation file for the new
9145 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9148 * diagnostic-show-locus.c (layout::layout): Use start and finish
9149 spelling location for the start and finish of each range.
9150 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9151 Add unused aspect param.
9152 * input.c (expand_location_1): Add "aspect" param, and use it
9153 to access the correct part of the location.
9154 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9156 (expand_location_to_spelling_point): Likewise.
9157 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9158 param, and pass it to expand_location_1.
9160 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9162 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9163 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9164 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9165 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9166 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9167 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9168 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9169 _mm_maskz_getmant_ss): New intrinsics.
9170 (__builtin_ia32_getexpss128_mask): Changed to ...
9171 __builtin_ia32_getexpss128_round ... this.
9172 (__builtin_ia32_getexpsd128_mask): Changed to ...
9173 __builtin_ia32_getexpsd128_round ... this.
9174 * config/i386/i386-builtin-types.def
9175 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9176 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9177 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9178 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9179 __builtin_ia32_getmantss_mask_round): New builtins.
9180 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9181 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9182 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9183 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9184 * config/i386/sse.md
9185 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9186 avx512f_sgetexp<mode><mask_scalar_name>
9187 <round_saeonly_scalar_name> ... this.
9188 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9189 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9190 vgetexp<ssescalarmodesuffix>
9191 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9192 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9193 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9194 avx512f_vgetmant<mode><mask_scalar_name>
9195 <round_saeonly_scalar_name> ... this.
9196 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9197 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9198 vgetmant<ssescalarmodesuffix>
9199 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9200 %0<mask_scalar_operand4>, %1, %2
9201 <round_saeonly_scalar_mask_op4>, %3} ... this.
9202 * config/i386/subst.md (mask_scalar_operand4,
9203 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9204 round_saeonly_scalar_nimm_predicate): New subst attributes.
9206 2017-07-06 Julia Koval <julia.koval@intel.com>
9208 * config/i386/i386.c (ix86_erase_embedded_rounding):
9209 Remove code for old rounding pattern.
9211 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9213 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9215 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9217 * doc/sourcebuild.texi (Test Directives, Variants of
9218 dg-require-support): Add documentation for dg-require-stack-check.
9220 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9222 * config/i386/subst.md (mask_scalar, round_scalar,
9223 round_saeonly_scalar): New meta-templates.
9224 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9225 round_scalar_mask_operand3, round_scalar_mask_op3,
9226 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9227 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9228 round_saeonly_scalar_constraint,
9229 round_saeonly_scalar_prefix): New subst attribute.
9230 * config/i386/sse.md
9231 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9232 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9233 <round_scalar_name> ... this.
9234 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9235 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9236 <round_scalar_name> ... this.
9237 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9238 <sse>_vm<code><mode>3<mask_scalar_name>
9239 <round_saeonly_scalar_name> ... this.
9240 (v<plusminus_mnemonic><ssescalarmodesuffix>
9241 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9242 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9243 v<plusminus_mnemonic><ssescalarmodesuffix>
9244 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9245 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9246 (v<multdiv_mnemonic><ssescalarmodesuffix>
9247 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9248 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9249 v<multdiv_mnemonic><ssescalarmodesuffix>
9250 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9251 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9252 (v<maxmin_float><ssescalarmodesuffix>
9253 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9254 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9255 v<maxmin_float><ssescalarmodesuffix>
9256 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9257 %0<mask_scalar_operand3>, %1, %<iptr>2
9258 <round_saeonly_scalar_mask_op3>} ... this.
9260 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9262 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9263 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9265 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9266 Alan Hayward <alan.hayward@arm.com>
9267 David Sherwood <david.sherwood@arm.com>
9269 * combine.c (simplify_if_then_else): Remove "enum" before
9271 * compare-elim.c (can_eliminate_compare): Likewise.
9272 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9274 (aarch64_lookup_simd_builtin_type): Likewise.
9275 (aarch64_simd_builtin_type): Likewise.
9276 (aarch64_init_simd_builtin_types): Likewise.
9277 (aarch64_simd_expand_args): Likewise.
9278 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9280 (aarch64_reverse_mask): Likewise.
9281 (aarch64_simd_emit_reg_reg_move): Likewise.
9282 (aarch64_gen_adjusted_ldpstp): Likewise.
9283 (aarch64_ccmp_mode_to_code): Likewise.
9284 (aarch64_operands_ok_for_ldpstp): Likewise.
9285 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9286 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9288 (aarch64_min_divisions_for_recip_mul): Likewise.
9289 (aarch64_reassociation_width): Likewise.
9290 (aarch64_get_condition_code_1): Likewise.
9291 (aarch64_simd_emit_reg_reg_move): Likewise.
9292 (aarch64_simd_attr_length_rglist): Likewise.
9293 (aarch64_reverse_mask): Likewise.
9294 (aarch64_operands_ok_for_ldpstp): Likewise.
9295 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9296 (aarch64_gen_adjusted_ldpstp): Likewise.
9297 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9299 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9300 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9301 (arm_lookup_simd_builtin_type): Likewise.
9302 (arm_simd_builtin_type): Likewise.
9303 (arm_init_simd_builtin_types): Likewise.
9304 (arm_expand_builtin_args): Likewise.
9305 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9306 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9307 (ft32_setup_incoming_varargs): Likewise.
9308 (ft32_function_arg): Likewise.
9309 (ft32_function_arg_advance): Likewise.
9310 (ft32_pass_by_reference): Likewise.
9311 (ft32_arg_partial_bytes): Likewise.
9312 (ft32_valid_pointer_mode): Likewise.
9313 (ft32_addr_space_pointer_mode): Likewise.
9314 (ft32_addr_space_legitimate_address_p): Likewise.
9315 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9317 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9318 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9319 (iamcu_alignment): Likewise.
9320 (canonicalize_vector_int_perm): Likewise.
9321 (ix86_noce_conversion_profitable_p): Likewise.
9322 (ix86_mpx_bound_mode): Likewise.
9323 (ix86_operands_ok_for_move_multiple): Likewise.
9324 * config/microblaze/microblaze-protos.h
9325 (microblaze_expand_conditional_branch_reg): Likewise.
9326 * config/microblaze/microblaze.c
9327 (microblaze_expand_conditional_branch_reg): Likewise.
9328 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9330 (rs6000_reassociation_width): Likewise.
9331 (rs6000_invalid_binary_op): Likewise.
9332 (fusion_p9_p): Likewise.
9333 (emit_fusion_p9_load): Likewise.
9334 (emit_fusion_p9_store): Likewise.
9335 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9337 (riscv_hard_regno_mode_ok_p): Likewise.
9338 (riscv_address_insns): Likewise.
9339 (riscv_split_symbol): Likewise.
9340 (riscv_legitimize_move): Likewise.
9341 (riscv_function_value): Likewise.
9342 (riscv_hard_regno_nregs): Likewise.
9343 (riscv_expand_builtin): Likewise.
9344 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9345 (riscv_build_integer): Likewise.
9346 (riscv_split_integer): Likewise.
9347 (riscv_legitimate_constant_p): Likewise.
9348 (riscv_cannot_force_const_mem): Likewise.
9349 (riscv_regno_mode_ok_for_base_p): Likewise.
9350 (riscv_valid_base_register_p): Likewise.
9351 (riscv_valid_offset_p): Likewise.
9352 (riscv_valid_lo_sum_p): Likewise.
9353 (riscv_classify_address): Likewise.
9354 (riscv_legitimate_address_p): Likewise.
9355 (riscv_address_insns): Likewise.
9356 (riscv_load_store_insns): Likewise.
9357 (riscv_force_binary): Likewise.
9358 (riscv_split_symbol): Likewise.
9359 (riscv_force_address): Likewise.
9360 (riscv_legitimize_address): Likewise.
9361 (riscv_move_integer): Likewise.
9362 (riscv_legitimize_const_move): Likewise.
9363 (riscv_legitimize_move): Likewise.
9364 (riscv_address_cost): Likewise.
9365 (riscv_subword): Likewise.
9366 (riscv_output_move): Likewise.
9367 (riscv_canonicalize_int_order_test): Likewise.
9368 (riscv_emit_int_order_test): Likewise.
9369 (riscv_function_arg_boundary): Likewise.
9370 (riscv_pass_mode_in_fpr_p): Likewise.
9371 (riscv_pass_fpr_single): Likewise.
9372 (riscv_pass_fpr_pair): Likewise.
9373 (riscv_get_arg_info): Likewise.
9374 (riscv_function_arg): Likewise.
9375 (riscv_function_arg_advance): Likewise.
9376 (riscv_arg_partial_bytes): Likewise.
9377 (riscv_function_value): Likewise.
9378 (riscv_pass_by_reference): Likewise.
9379 (riscv_setup_incoming_varargs): Likewise.
9380 (riscv_print_operand): Likewise.
9381 (riscv_elf_select_rtx_section): Likewise.
9382 (riscv_save_restore_reg): Likewise.
9383 (riscv_for_each_saved_reg): Likewise.
9384 (riscv_register_move_cost): Likewise.
9385 (riscv_hard_regno_mode_ok_p): Likewise.
9386 (riscv_hard_regno_nregs): Likewise.
9387 (riscv_class_max_nregs): Likewise.
9388 (riscv_memory_move_cost): Likewise.
9389 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9390 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9391 (rl78_addr_space_address_mode): Likewise.
9392 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9394 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9395 (rs6000_reassociation_width): Likewise.
9396 (rs6000_invalid_binary_op): Likewise.
9397 (fusion_p9_p): Likewise.
9398 (emit_fusion_p9_load): Likewise.
9399 (emit_fusion_p9_store): Likewise.
9400 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9401 (ok_for_simple_move_operands): Likewise.
9402 (ok_for_simple_move_strict_operands): Likewise.
9403 (ok_for_simple_arith_logic_operands): Likewise.
9404 (visium_legitimize_reload_address): Likewise.
9405 (visium_select_cc_mode): Likewise.
9406 (output_cbranch): Likewise.
9407 (visium_split_double_move): Likewise.
9408 (visium_expand_copysign): Likewise.
9409 (visium_expand_int_cstore): Likewise.
9410 (visium_expand_fp_cstore): Likewise.
9411 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9412 (visium_function_arg): Likewise.
9413 (visium_function_arg_advance): Likewise.
9414 (visium_libcall_value): Likewise.
9415 (visium_setup_incoming_varargs): Likewise.
9416 (visium_legitimate_constant_p): Likewise.
9417 (visium_legitimate_address_p): Likewise.
9418 (visium_legitimize_address): Likewise.
9419 (visium_secondary_reload): Likewise.
9420 (visium_register_move_cost): Likewise.
9421 (visium_memory_move_cost): Likewise.
9422 (prepare_move_operands): Likewise.
9423 (ok_for_simple_move_operands): Likewise.
9424 (ok_for_simple_move_strict_operands): Likewise.
9425 (ok_for_simple_arith_logic_operands): Likewise.
9426 (visium_function_value_1): Likewise.
9427 (rtx_ok_for_offset_p): Likewise.
9428 (visium_legitimize_reload_address): Likewise.
9429 (visium_split_double_move): Likewise.
9430 (visium_expand_copysign): Likewise.
9431 (visium_expand_int_cstore): Likewise.
9432 (visium_expand_fp_cstore): Likewise.
9433 (visium_split_cstore): Likewise.
9434 (visium_select_cc_mode): Likewise.
9435 (visium_split_cbranch): Likewise.
9436 (output_cbranch): Likewise.
9437 (visium_print_operand_address): Likewise.
9438 * expmed.c (flip_storage_order): Likewise.
9439 * expmed.h (emit_cstore): Likewise.
9440 (flip_storage_order): Likewise.
9441 * genrecog.c (validate_pattern): Likewise.
9442 * hsa-gen.c (gen_hsa_addr): Likewise.
9443 * internal-fn.c (expand_arith_overflow): Likewise.
9444 * ira-color.c (allocno_copy_cost_saving): Likewise.
9445 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9446 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9447 (process_invariant_for_inheritance): Likewise.
9448 * lra-eliminations.c (move_plus_up): Likewise.
9449 * omp-low.c (lower_oacc_reductions): Likewise.
9450 * simplify-rtx.c (simplify_subreg): Likewise.
9451 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9452 (TARGET_CHKP_BOUND_MODE): Likewise..
9453 * targhooks.c (default_chkp_bound_mode): Likewise.
9454 (default_setup_incoming_vararg_bounds): Likewise.
9455 * targhooks.h (default_chkp_bound_mode): Likewise.
9456 (default_setup_incoming_vararg_bounds): Likewise.
9457 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9458 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9459 (have_whole_vector_shift): Likewise.
9460 * tree-vect-stmts.c (vectorizable_load): Likewise.
9461 * doc/tm.texi: Regenerate.
9463 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9465 Graceful degrade if Binutils PR21472 is not available.
9468 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9469 .rodata in flash test fails.
9470 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9471 * confgure: Regenerate.
9472 * config.in: Regenerate.
9473 * config/avr/avr.c (avr_asm_named_section)
9474 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9475 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9476 (avr_asm_init_sections): Same.
9478 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9480 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9481 (fma<VH:mode>4_intrinsic): Likewise.
9482 (*fmsub<VCVTF:mode>4): Likewise.
9483 (*fmsub<VH:mode>4_intrinsic): Likewise.
9485 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9488 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9489 Don't depend on "optimize > 0".
9490 (out_movhi_r_mr, out_movqi_mr_r): Same.
9491 (out_movhi_mr_r, out_movqi_r_mr): Same.
9492 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9493 io_address_operand on "optimize > 0".
9495 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9497 * tree-loop-distribution.c: Add general explanantion on the pass.
9498 (generate_loops_for_partition): Mark distributed loop.
9499 (pg_add_dependence_edges): New parameter. Handle alias data
9500 dependence specially and record it in the parameter if asked.
9501 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9502 (init_partition_graph_vertices, add_partition_graph_edge): New.
9503 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9504 (free_partition_graph_vdata, build_partition_graph): New.
9505 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9506 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9507 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9508 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9509 (version_for_distribution_p, finalize_partitions): New.
9510 (distribute_loop): Handle alias data dependence specially. Factor
9511 out loop fusion code as functions and call these functions.
9513 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9515 * tree-loop-distribution.c (classify_partition): New parameter and
9516 better handle reduction statement.
9517 (rdg_build_partitions): Revise comment.
9518 (distribute_loop): Compute statements in all partitions and pass it
9519 to classify_partition.
9521 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9523 * tree-loop-distribution.c (enum partition_type): New.
9524 (struct partition): New field type.
9525 (partition_merge_into): Add parameter. Update partition type.
9526 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9527 (build_rdg_partition_for_vertex): Compute partition type.
9528 (rdg_build_partitions): Dump partition type.
9529 (distribute_loop): Update calls to partition_merge_into.
9531 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9533 * tree-loop-distribution.c (struct ddr_hasher): New.
9534 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9536 (classify_partition): Call get_data_dependence.
9537 (pg_add_dependence_edges): Ditto.
9538 (distribute_loop): Release data dependence hash table.
9540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9542 * tree-loop-distribution.c (ref_base_address): Delete.
9543 (similar_memory_accesses): Rename ...
9544 (share_memory_accesses): ... to this. Check if partitions access
9545 the same memory reference.
9546 (distribute_loop): Call share_memory_accesses.
9548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9550 * tree-loop-distribution.c (struct partition): New field recording
9552 (partition_alloc, partition_free): Init and release data refs.
9553 (partition_merge_into): Merge data refs.
9554 (build_rdg_partition_for_vertex): Collect data refs for partition.
9555 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9557 (distribute_loop): Remve data refs from vertice data of partition
9560 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9562 * tree-loop-distribution.c (params.h): Include header file.
9563 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9564 (datarefs_vec): New global var.
9565 (create_rdg_vertices): Use datarefs_vec directly.
9566 (free_rdg): Don't free data references.
9567 (build_rdg): Update use. Don't free data references.
9568 (distribute_loop): Compute global variable for data references.
9569 Bail out if there are too many data references.
9571 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9573 * tree-loop-distribution.c (loop_nest): New global var.
9574 (build_rdg): Use loop directly, rather than loop nest.
9575 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9577 (distribute_loop): Compute global variable loop nest. Update use.
9579 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9581 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9582 (partition_merge_into): New parameter. Dump reason for fusion.
9583 (distribute_loop): Update use of partition_merge_into.
9585 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9587 * tree-loop-distribution.c (bb_top_order_index): New.
9588 (bb_top_order_index_size, bb_top_order_cmp): New.
9589 (stmts_from_loop): Use topological order.
9590 (pass_loop_distribution::execute): Compute and release topological
9591 order for basic blocks.
9593 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9595 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9598 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9600 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9601 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9602 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9603 * internal-fn.def (LOOP_DIST_ALIAS): New.
9604 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9605 (fold_loop_internal_call): ... this.
9606 (vect_loop_dist_alias_call): New function.
9607 (set_uid_loop_bbs): Call fold_loop_internal_call.
9608 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9611 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9614 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9615 Require dead FLAGS_REG at the beginning of a peephole.
9617 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9620 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9621 arguments in the call to __builtin_ia32_sbb_u32.
9622 (_subborrow_u64): Swap _X and _Y arguments in the call to
9623 __builtin_ia32_sbb_u64.
9625 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9628 * tree-vrp.c (compare_assert_loc): Turn into a function template
9629 with stable template parameter. Only test if a->e is NULL,
9630 !a->e == !b->e has been verified already. Use e == NULL or
9631 e != NULL instead of e or ! e tests. If stable is true, don't use
9632 iterative_hash_expr, on the other side allow a or b or both NULL
9633 and sort the NULLs last.
9634 (process_assert_insertions): Sort using compare_assert_loc<false>
9635 instead of compare_assert_loc, later sort using
9636 compare_assert_loc<true> before calling process_assert_insertions_for
9637 in a loop. Use break instead of continue once seen NULL pointer.
9639 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9641 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9642 Cortex-R7 and Cortex-R8 processors.
9644 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9646 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9647 uninitialized while src is not.
9649 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9651 * common/config/arm/arm-common.c: Adjust include path for
9653 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9654 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9655 (arm-cpu-data.h): Likewise.
9656 (arm-cpu-cdata.h): Likewise.
9657 * config/arm/arm-cpu.h: Delete.
9658 * config/arm/arm-cpu-cdata.h: Delete.
9659 * config/arm/arm-cpu-data.h: Delete.
9661 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9663 * config/arm/arm-cpus.in (cortex-a55): New.
9664 (cortex-a75): Likewise.
9665 (cortex-a75.cortex-a55): Likewise.
9666 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9668 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9669 * config/arm/arm-cpu-cdata.h: Regenerate.
9670 * config/arm/arm-cpu-data.h: Regenerate.
9671 * config/arm/arm-cpu.h: Regenerate.
9672 * config/arm/arm-tables.opt: Regenerate.
9673 * config/arm/arm-tune.md: Regenerate.
9675 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9677 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9679 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9681 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9684 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9686 PR tree-optimization/81292
9687 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9688 full_string_p, also call adjust_related_strinfos if the adjustment
9689 is simple, otherwise invalidate related strinfos.
9691 2017-07-04 Martin Liska <mliska@suse.cz>
9694 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9695 newly created variable as DECL_IGNORED_P.
9697 2017-07-04 Martin Liska <mliska@suse.cz>
9700 * ipa-inline.c (inline_small_functions):
9701 Use xstrdup_for_dump.
9703 2017-07-04 Tom de Vries <tom@codesourcery.com>
9705 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9707 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9710 * config/darwin.c (darwin_function_switched_text_sections):
9713 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9715 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9717 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9719 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9721 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9723 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9724 min_profitable_iters, and th as inclusive lower bounds.
9725 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9726 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9727 for min_profitable_iters and min_profitable_estimate.
9728 (vect_transform_loop): Treat th as an inclusive lower bound.
9729 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9731 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9734 * config/darwin.c (darwin_function_switched_text_sections):
9735 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9736 in two pieces, and suppress the use of buf.
9738 2017-07-03 Nathan Sidwell <nathan@acm.org>
9740 * hash-table.h (hash_table_mod1): Fix indentation.
9742 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9745 * predict.c (force_edge_cold): Be more careful about propagation
9747 * profile-count.h (profile_probability::guessed,
9748 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9750 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9752 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9754 * doc/invoke.texi (rcpc architecture extension): Document it.
9756 2017-07-03 Richard Biener <rguenther@suse.de>
9758 PR tree-optimization/60510
9759 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9760 the scalar reduction PHI and use it.
9761 (vectorizable_reduction): Properly guard the single_defuse_cycle
9762 path for non-SLP reduction chains where we cannot use it.
9763 Rework reduc_def/index and vector type deduction. Rework
9764 vector operand gathering during reduction op code-gen.
9765 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9766 chains dissolve the chain and leave it to non-SLP reduction
9769 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9771 * tree-data-ref.h (dr_alignment): Declare.
9772 * tree-data-ref.c (dr_alignment): New function.
9773 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9774 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9776 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9780 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9781 and base_misalignment fields.
9782 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9783 * tree-data-ref.c: Include builtins.h.
9784 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9785 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9786 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9787 * tree-vect-data-refs.c: Include tree-cfg.h.
9788 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9789 fields instead of calculating an alignment here.
9790 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9791 innermost_loop_behavior fields.
9793 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9795 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9797 (DR_STEP_ALIGNMENT): New macro.
9798 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9799 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9800 (create_data_ref): Print it.
9801 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9802 to tell whether the step preserves vector (mis)alignment.
9803 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9804 Move the check for an integer step and generalise to all INTEGER_CST.
9805 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9806 Print the outer step alignment.
9808 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9810 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9811 with offset_alignment.
9812 (DR_ALIGNED_TO): Delete.
9813 (DR_OFFSET_ALIGNMENT): New macro.
9814 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9815 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9816 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9817 (create_data_ref): Likewise.
9818 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9819 (vect_analyze_data_refs): Likewise.
9820 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9821 creating dummy innermost behavior.
9823 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9825 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9826 with a "innermost_loop_behavior *" and refeence tree.
9827 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9828 (create_data_ref): Update call accordingly.
9829 * tree-predcom.c (find_looparound_phi): Likewise.
9831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9833 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9834 fields with dr_wrt_vec_loop.
9835 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9836 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9837 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9838 (vect_dr_behavior): New function.
9839 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9840 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9841 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9842 track whether the step preserves the misalignment.
9843 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9844 Use vect_dr_behavior.
9845 (vect_setup_realignment): Update call accordingly.
9846 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9847 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9848 call to vect_create_addr_base_for_vector_ref.
9849 (vect_create_cond_for_align_checks): Likewise.
9850 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9851 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9852 (vect_recog_mask_conversion_pattern): Likewise.
9853 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9854 (new_stmt_vec_info): Remove redundant zeroing.
9856 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9858 * common/config/arm/arm-common.c (arm_be8_option): New function.
9859 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9860 (ISA_ARMv6): Add isa_bit_be8.
9861 * config/arm/arm.h (arm_be8_option): Add prototype.
9862 (BE8_SPEC_FUNCTION): New define.
9863 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9864 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9865 (mlittle-endian): Similarly.
9866 (mbe8, mbe32): New options.
9867 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9868 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9870 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9872 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9876 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9877 (cleanup_tree_cfg_bb): Use it.
9878 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9880 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9882 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9885 * loop-doloop.c (add_test): Update profile.
9887 2017-07-03 Martin Liska <mliska@suse.cz>
9890 * sanopt.c (rewrite_usage_of_param): New function.
9891 (sanitize_rewrite_addressable_params): Likewise.
9892 (pass_sanopt::execute): Call rewrite_usage_of_param.
9894 2017-07-03 Richard Biener <rguenther@suse.de>
9896 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9897 back to using VIEW_CONVERT_EXPR.
9899 2017-07-03 Martin Liska <mliska@suse.cz>
9902 * doc/extend.texi: Document when a resolver function is
9903 generated for target_clones.
9905 2017-07-03 Martin Liska <mliska@suse.cz>
9907 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9908 and shadow memory of auto variables which are subject of
9909 use-after-scope sanitization.
9910 (asan_expand_mark_ifn): Add do set only when is_poison.
9912 2016-07-03 Richard Biener <rguenther@suse.de>
9914 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9916 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9917 (vectorizable_reduction): Perform reduction PHI creation when
9918 visiting a reduction PHI and adjust and simplify code generation
9919 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9920 (vect_transform_loop): Visit reduction PHIs.
9921 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9922 defs into the SLP tree.
9923 (vect_build_slp_tree): Reduction defs terminate the recursion.
9924 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9926 (vect_get_vec_defs_for_stmt_copy): Export.
9927 (vect_get_vec_defs): Likewise.
9928 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9930 (vect_get_vec_defs_for_stmt_copy): Declare.
9931 (vect_get_vec_defs): Likewise.
9933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9935 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9936 parameter with a "loop" parameter and use it instead of the
9937 loop containing DR_STMT. Don't check simple_iv when doing
9938 BB analysis. Describe the two analysis modes in the comment.
9940 2017-07-03 Tom de Vries <tom@codesourcery.com>
9942 PR tree-optimization/69468
9943 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9944 (find_same_succ_bb): Handle ignore_edge_flags.
9946 2017-07-03 Tom de Vries <tom@codesourcery.com>
9948 PR tree-optimization/81192
9949 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9951 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9953 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9955 2017-07-03 Tom de Vries <tom@codesourcery.com>
9957 PR tree-optimization/81192
9958 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9959 BB_SAME_SUCC (bb) == NULL.
9961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9963 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9966 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9968 * dumpfile.c: Include profile-count.h
9969 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9971 (insert_cond_bb): Update profile.
9972 * tree-cfg.h (insert_cond_bb): Update prototype.
9973 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9974 * tree-dump.c: Do not include tree-cfg.
9976 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9978 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9980 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9982 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9985 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9987 * tree-complex.c (expand_complex_div_wide): update profile.
9989 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9990 Alan Hayward <alan.hayward@arm.com>
9991 David Sherwood <david.sherwood@arm.com>
9993 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9994 (CORETYPES_H): New define.
9995 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9996 (insn-modes-inline.h, s-modes-inline-h): New rules.
9997 (generated_files): Add insn-modes-inline.h.
9998 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9999 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10000 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10001 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10002 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10003 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10004 (build/gencodes.o, build/genconditions.o): Likewise.
10005 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10006 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10007 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10008 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10009 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10010 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10011 * coretypes.h: Include everything up to real.h for generators.
10012 Include insn-modes.h first. Include wide-int-print.h after
10013 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10014 * machmode.h: Don't include insn-modes.h here.
10015 * function-tests.c: Remove includes of signop.h, machmode.h,
10016 double-int.h and wide-int.h.
10018 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10020 * optc-save-gen.awk: Likewise.
10021 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10022 * godump.c: Remove include of wide-int-print.h.
10023 * pretty-print.h: Likewise.
10024 * wide-int-print.cc: Likewise.
10025 * wide-int.cc: Likewise.
10026 * hash-map-tests.c: Remove include of signop.h.
10027 * hash-set-tests.c: Likewise.
10028 * rtl-tests.c: Likewise.
10029 * mkconfig.sh: Remove include of machmode.h.
10030 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10032 (emit_insn_modes_inline_h): ...this new function. Emit the code
10033 into an insn-modes-inline.h header file, adding appropriate
10034 include guards and end comments.
10035 (emit_insn_modes_c_header): Remove include of machmode.h.
10036 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10038 (main): Handle -i flag and call emit_insn_modes_inline_h when
10041 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10043 * tree-ssa-strlen.c (strinfo): Rename the length field to
10044 nonzero_chars. Add a full_string_p field.
10045 (compare_nonzero_chars, zero_length_string_p): New functions.
10046 (get_addr_stridx): Add an offset_out parameter.
10047 Use compare_nonzero_chars.
10048 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10049 (new_strinfo): Update after above changes to strinfo.
10050 (set_endptr_and_length): Set full_string_p.
10051 (get_string_length): Update after above changes to strinfo.
10052 (unshare_strinfo): Update call to new_strinfo.
10053 (maybe_invalidate): Likewise.
10054 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10055 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10056 as a uhwi instead of an shwi. Update after above changes to
10057 strinfo and new_strinfo.
10058 (zero_length_string): Assert that chainsi contains full strings.
10059 Use zero_length_string_p. Update call to new_strinfo.
10060 (adjust_related_strinfos): Update after above changes to strinfo.
10061 Copy full_string_p from origsi.
10062 (adjust_last_stmt): Use zero_length_string_p.
10063 (handle_builtin_strlen): Update after above changes to strinfo and
10064 new_strinfo. Install the lhs as the string length if the previous
10065 entry didn't describe a full string.
10066 (handle_builtin_strchr): Update after above changes to strinfo
10068 (handle_builtin_strcpy): Likewise.
10069 (handle_builtin_strcat): Likewise.
10070 (handle_builtin_malloc): Likewise.
10071 (handle_pointer_plus): Likewise.
10072 (handle_builtin_memcpy): Likewise. Track nonzero characters
10073 that aren't necessarily followed by a nul terminator.
10074 (handle_char_store): Likewise.
10076 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10078 PR tree-optimization/80769
10079 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10080 for malloc and calloc. Document the new invariant that all related
10081 strinfos have delayed lengths or none do.
10082 (verify_related_strinfos): Move earlier in file.
10083 (set_endptr_and_length): New function, split out from...
10084 (get_string_length): ...here. Also set the lengths of related
10086 (zero_length_string): Assert that chainsi has known (rather than
10088 (adjust_related_strinfos): Likewise.
10090 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10092 PR tree-optimization/81136
10093 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10094 assert that two references with the same misalignment have the same
10095 compile-time misalignment if those compile-time misalignments
10098 2017-07-01 Andi Kleen <ak@linux.intel.com>
10100 * print-tree.c (print_node): Print all attributes.
10102 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10104 * cfg.c (scale_bbs_frequencies): New function.
10105 * cfg.h (scale_bbs_frequencies): Declare it.
10106 * cfgloopanal.c (single_likely_exit): Cleanup.
10107 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10109 (scale_loop_profile): Likewise.
10110 (loop_version): Likewise.
10111 (create_empty_loop_on_edge): Update.
10112 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10113 scale_loop_frequencies, scale_loop_profile, loopify,
10114 loop_version): Update prototypes.
10115 * modulo-sched.c (sms_schedule): Update.
10116 * predict.c (unlikely_executed_edge_p): Also check probability.
10117 (probably_never_executed_edge_p): Fix typo.
10118 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10119 * tree-parloops.c (gen_parallel_loop): Update.
10120 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10122 * tree-ssa-loop-split.c (split_loop): Update.
10123 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10124 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10125 (vect_loop_versioning): Update.
10126 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10128 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10130 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10132 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10134 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10135 to keep profile consistent.
10137 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10139 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10140 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10141 * profile-count.h (max_safe_multiplier): Make unsigned.
10142 (profile_count::guessed_zero): New.
10144 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10146 * bb-reorder.c (fix_up_crossing_landing_pad,
10147 fix_crossing_conditional_branches): Use make_single_succ_edge
10148 to keep profile consistent.
10150 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10152 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10155 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10158 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10159 the right scopes, make sure cond_jump isn't preserved between multiple
10160 iterations. Search for fallthru edge whenever there are 3+ edges and
10161 use find_fallthru_edge for it.
10163 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10165 Patch by Alexander Monakov <amonakov@ispras.ru>
10166 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10167 probabilities consistently.
10169 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10171 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10172 emit_cmp_and_jump_insns.
10174 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10177 * tree-inline.c (expand_call_inline): Combine profile statuses.
10179 2017-06-30 Andrew Pinski <apinski@cavium.com>
10181 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10182 fold_stmt returned true.
10184 2017-06-30 Nathan Sidwell <nathan@acm.org>
10186 * ggc.h (empty_string): Delete.
10187 * cfgexpand.c (expand_asm_stmt): Use plain "".
10188 * optabs.c (expand_asm_memory_barrier): Likewise.
10189 * stringpool.c (empty_string): Delete.
10190 (digit_vector, digit_string): Delete.
10191 (ggc_alloc_string): Use plain "", don't optimize single digit
10192 strings. Use ggc_alloc_atomic.
10194 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10196 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10197 comparison set and one other set, use the cost of the non-comparison
10200 2017-06-30 Nathan Sidwell <nathan@acm.org>
10202 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10205 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10207 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10208 loops. Remove now unneeded calls to gimple_switch_set_label() that
10209 just set removed labels to NULL_TREE.
10211 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10213 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10214 (set_range_info): ...here. Only call set_range_info_raw if domain
10216 (set_nonzero_bits): Call set_range_info_raw.
10217 * tree-ssanames.h (set_range_info_raw): New.
10219 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10222 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10223 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10224 of nonimmediate_operand and <store_mask_constraint> instead of m
10225 for the input operand. For V8FI iterator, always split if input
10226 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10227 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10228 <store_mask_predicate> instead of register_operand and
10229 <store_mask_constraint> instead of v for the input operand. Make
10230 sure both operands aren't MEMs for if not <mask_applied>.
10232 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10234 * lto-wrapper.c (copy_file) Close both file descriptors before
10237 2017-06-30 Martin Liska <mliska@suse.cz>
10240 * multiple_target.c (create_dispatcher_calls): Make ifunc
10241 also for function that don't have calls or are not referenced.
10243 2017-06-30 Richard Biener <rguenther@suse.de>
10245 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10246 analyze the first scalar stmt. Move vector type computation
10247 for the BB case here from ...
10248 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10249 live operation processing in the SLP case properly.
10251 2017-06-30 Richard Biener <rguenther@suse.de>
10253 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10255 2017-06-30 Martin Liska <mliska@suse.cz>
10258 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10259 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10261 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10263 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10266 2017-06-30 Richard Biener <rguenther@suse.de>
10268 PR tree-optimization/81249
10269 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10270 condition reduction result to original scalar type.
10272 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10274 * profile-count.h (enum profile_quality): Fix typos and whitespace
10277 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10279 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10280 type for branch probabilities.
10282 2017-06-29 Julian Brown <julian@codesourcery.com>
10283 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10285 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10286 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10287 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10288 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10290 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10292 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10293 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10294 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10295 CC usage from generic code to here.
10296 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10297 CC usage into the target macros.
10299 2017-06-29 Maya Rashish <coypu@sdf.org>
10301 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10304 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10306 * arm/arm-builtins.c: Include profile-count.h
10307 * except.c (sjlj_emit_function_enter): Use
10308 profile_probability::unlikely.
10310 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10312 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10313 and tocrel_offset be pointer args rather than implicitly using
10315 (legitimate_constant_pool_address_p, rs6000_emit_move,
10316 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10317 tocrel_offset and use in toc_relative_expr_p call.
10318 (print_operand, print_operand_address): Use static tocrel_base_oac
10319 and tocrel_offset_oac.
10320 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10323 2017-06-29 Maya Rashish <coypu@sdf.org>
10325 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10327 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10329 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10330 objects, take into account only the alignment of 'op0' and 'mode1' if
10333 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10335 * ccmp.c (ccmp_tree_comparison_p): New function.
10336 (ccmp_candidate_p): Update to use above function.
10337 (get_compare_parts): New function.
10338 (expand_ccmp_next): Update to use new functions.
10339 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10341 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10342 take mode as argument.
10343 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10344 * expr.c (expand_expr_real_1): Pass mode as argument.
10346 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10348 * combine.c (combine_instructions): Print insns to dump_file, together
10351 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10353 * asan.c (asan_emit_stack_protection): Update.
10354 (create_cond_insert_point): Update.
10355 * auto-profile.c (afdo_propagate_circuit): Update.
10356 * basic-block.h (struct edge_def): Turn probability to
10357 profile_probability.
10358 (EDGE_FREQUENCY): Update.
10359 * bb-reorder.c (find_traces_1_round): Update.
10360 (better_edge_p): Update.
10361 (sanitize_hot_paths): Update.
10362 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10363 (make_single_succ_edge): Update.
10364 (check_bb_profile): Update.
10365 (dump_edge_info): Update.
10366 (update_bb_profile_for_threading): Update.
10367 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10368 probabilitycount to 0.
10369 * cfgbuild.c (compute_outgoing_frequencies): Update.
10370 * cfgcleanup.c (try_forward_edges): Update.
10371 (outgoing_edges_match): Update.
10372 (try_crossjump_to_edge): Update.
10373 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10374 (expand_gimple_tailcall): Update.
10375 (construct_init_block): Use make_single_succ_edge.
10376 (construct_exit_block): Use make_single_succ_edge.
10377 * cfghooks.c (verify_flow_info): Update.
10378 (redirect_edge_succ_nodup): Update.
10379 (split_edge): Update.
10380 (account_profile_record): Update.
10381 * cfgloopanal.c (single_likely_exit): Update.
10382 * cfgloopmanip.c (scale_loop_profile): Update.
10383 (set_zero_probability): Remove.
10384 (duplicate_loop_to_header_edge): Update.
10385 * cfgloopmanip.h (loop_version): Update prototype.
10386 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10387 (force_nonfallthru_and_redirect): Update.
10388 (update_br_prob_note): Update.
10389 (rtl_verify_edges): Update.
10390 (purge_dead_edges): Update.
10391 (rtl_lv_add_condition_to_bb): Update.
10392 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10393 * cgraphunit.c (init_lowered_empty_function): Update.
10394 (cgraph_node::expand_thunk): Update.
10395 * cilk-common.c: Include profile-count.h
10396 * dojump.c (inv): Remove.
10397 (jumpifnot): Update.
10398 (jumpifnot_1): Update.
10399 (do_jump_1): Update.
10401 (do_jump_by_parts_greater_rtx): Update.
10402 (do_compare_rtx_and_jump): Update.
10403 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10404 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10405 * dwarf2cfi.c: Include profile-count.h
10406 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10407 (sjlj_emit_dispatch_table): Likewise.
10408 * explow.c: Include profile-count.h
10409 * expmed.c (emit_store_flag_force): Update.
10410 (do_cmp_and_jump): Update.
10411 * expr.c (compare_by_pieces_d::generate): Update.
10412 (compare_by_pieces_d::finish_mode): Update.
10413 (emit_block_move_via_loop): Update.
10414 (store_expr_with_bounds): Update.
10415 (store_constructor): Update.
10416 (expand_expr_real_2): Update.
10417 (expand_expr_real_1): Update.
10418 * expr.h (try_casesi, try_tablejump): Update prototypes.
10419 * gimple-pretty-print.c (dump_probability): Update.
10420 (dump_profile): New.
10421 (dump_gimple_label): Update.
10422 (dump_gimple_bb_header): Update.
10423 * graph.c (draw_cfg_node_succ_edges): Update.
10424 * hsa-gen.c (convert_switch_statements): Update.
10425 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10426 (find_if_case_1): Update.
10427 (find_if_case_2): Update.
10428 * internal-fn.c (expand_arith_overflow_result_store): Update.
10429 (expand_addsub_overflow): Update.
10430 (expand_neg_overflow): Update.
10431 (expand_mul_overflow): Update.
10432 (expand_vector_ubsan_overflow): Update.
10433 * ipa-cp.c (good_cloning_opportunity_p): Update.
10434 * ipa-split.c (split_function): Use make_single_succ_edge.
10435 * ipa-utils.c (ipa_merge_profiles): Update.
10436 * loop-doloop.c (add_test): Update.
10437 (doloop_modify): Update.
10438 * loop-unroll.c (compare_and_jump_seq): Update.
10439 (unroll_loop_runtime_iterations): Update.
10440 * lra-constraints.c (lra_inheritance): Update.
10441 * lto-streamer-in.c (input_cfg): Update.
10442 * lto-streamer-out.c (output_cfg): Update.
10443 * mcf.c (adjust_cfg_counts): Update.
10444 * modulo-sched.c (sms_schedule): Update.
10445 * omp-expand.c (expand_omp_for_init_counts): Update.
10446 (extract_omp_for_update_vars): Update.
10447 (expand_omp_ordered_sink): Update.
10448 (expand_omp_for_ordered_loops): Update.
10449 (expand_omp_for_generic): Update.
10450 (expand_omp_for_static_nochunk): Update.
10451 (expand_omp_for_static_chunk): Update.
10452 (expand_cilk_for): Update.
10453 (expand_omp_simd): Update.
10454 (expand_omp_taskloop_for_outer): Update.
10455 (expand_omp_taskloop_for_inner): Update.
10456 * omp-simd-clone.c (simd_clone_adjust): Update.
10457 * optabs.c (expand_doubleword_shift): Update.
10458 (expand_abs): Update.
10459 (emit_cmp_and_jump_insn_1): Update.
10460 (expand_compare_and_swap_loop): Update.
10461 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10462 * predict.c (predictable_edge_p): Update.
10463 (edge_probability_reliable_p): Update.
10464 (set_even_probabilities): Update.
10465 (combine_predictions_for_insn): Update.
10466 (combine_predictions_for_bb): Update.
10467 (propagate_freq): Update.
10468 (estimate_bb_frequencies): Update.
10469 (force_edge_cold): Update.
10470 * profile-count.c (profile_count::dump): Add missing space into dump.
10471 (profile_count::debug): Add newline.
10472 (profile_count::differs_from_p): Explicitly convert to unsigned.
10473 (profile_count::stream_in): Update.
10474 (profile_probability::dump): New member function.
10475 (profile_probability::debug): New member function.
10476 (profile_probability::differs_from_p): New member function.
10477 (profile_probability::differs_lot_from_p): New member function.
10478 (profile_probability::stream_in): New member function.
10479 (profile_probability::stream_out): New member function.
10480 * profile-count.h (profile_count_quality): Rename to ...
10481 (profile_quality): ... this one.
10482 (profile_probability): New.
10483 (profile_count): Update.
10484 * profile.c (compute_branch_probabilities): Update.
10485 * recog.c (peep2_attempt): Update.
10486 * sched-ebb.c (schedule_ebbs): Update.
10487 * sched-rgn.c (find_single_block_region): Update.
10488 (compute_dom_prob_ps): Update.
10489 (schedule_region): Update.
10490 * sel-sched-ir.c (compute_succs_info): Update.
10491 * stmt.c (struct case_node): Update.
10492 (do_jump_if_equal): Update.
10493 (get_outgoing_edge_probs): Update.
10494 (conditional_probability): Update.
10495 (emit_case_dispatch_table): Update.
10496 (expand_case): Update.
10497 (expand_sjlj_dispatch_table): Update.
10498 (emit_case_nodes): Update.
10499 * targhooks.c: Update.
10500 * tracer.c (better_p): Update.
10501 (find_best_successor): Update.
10502 * trans-mem.c (expand_transaction): Update.
10503 * tree-call-cdce.c: Update.
10504 * tree-cfg.c (gimple_split_edge): Upate.
10505 (move_sese_region_to_fn): Upate.
10506 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10507 * tree-eh.c (lower_resx): Upate.
10508 (cleanup_empty_eh_move_lp): Upate.
10509 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10510 * tree-inline.c (copy_edges_for_bb): Update.
10511 (copy_cfg_body): Update.
10512 * tree-parloops.c (gen_parallel_loop): Update.
10513 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10514 (gimple_gen_time_profiler): Update.
10515 * tree-ssa-dce.c (remove_dead_stmt): Update.
10516 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10517 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10518 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10519 (unloop_loops): Update.
10520 (try_peel_loop): Update.
10521 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10522 * tree-ssa-loop-split.c (connect_loops): Update.
10523 (split_loop): Update.
10524 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10525 (hoist_guard): Update.
10526 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10527 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10528 (value_replacement): Update.
10529 * tree-ssa-reassoc.c (branch_fixup): Update.
10530 * tree-ssa-tail-merge.c (replace_block_by): Update.
10531 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10532 (create_edge_and_update_destination_phis): Update.
10533 (compute_path_counts): Update.
10534 (recompute_probabilities): Update.
10535 (update_joiner_offpath_counts): Update.
10536 (freqs_to_counts_path): Update.
10537 (duplicate_thread_path): Update.
10538 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10539 (struct switch_conv_info): Update.
10540 (gen_inbound_check): Update.
10541 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10542 (vect_do_peeling): Update.
10543 (vect_loop_versioning): Update.
10544 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10545 (optimize_mask_stores): Update.
10546 * ubsan.c (ubsan_expand_null_ifn): Update.
10547 * value-prof.c (gimple_divmod_fixed_value): Update.
10548 (gimple_divmod_fixed_value_transform): Update.
10549 (gimple_mod_pow2): Update.
10550 (gimple_mod_pow2_value_transform): Update.
10551 (gimple_mod_subtract): Update.
10552 (gimple_mod_subtract_transform): Update.
10553 (gimple_ic): Update.
10554 (gimple_stringop_fixed_value): Update.
10555 (gimple_stringops_transform): Update.
10556 * value-prof.h: Update.
10558 2017-06-29 Carl Love <cel@us.ibm.com>
10560 * config/rs6000/rs6000-c.c: Add support for built-in functions
10561 vector signed int vec_signed (vector float);
10562 vector signed long long vec_signed (vector double);
10563 vector signed int vec_signed2 (vector double, vector double);
10564 vector signed int vec_signede (vector double);
10565 vector signed int vec_signedo (vector double);
10566 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10567 instruction generator.
10568 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10569 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10570 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10572 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10573 vunsignede_v2df): Add define_expands.
10574 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10575 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10576 VEC_UNSIGNEDO): Add definitions.
10577 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10578 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10579 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10580 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10581 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10582 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10583 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10584 vec_unsignede, vec_unsignedo): Add builtin defines.
10585 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10587 * doc/extend.texi: Update the built-in documentation file for the
10588 new built-in functions.
10590 2017-06-29 Richard Biener <rguenther@suse.de>
10592 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10593 reduction chains to LOOP_VINFO_REDUCTIONS.
10594 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10595 SLP reductions after processing reduction chains.
10597 2017-06-29 Nathan Sidwell <nathan@acm.org>
10599 * builtins.c (fold_builtin_FUNCTION): Use
10600 lang_hooks.decl_printable_name.
10602 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10604 PR middle-end/81194
10605 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10606 with only one label.
10607 * stmt.c (expand_case): Assert NCASES is greater than one.
10609 2017-06-29 Richard Biener <rguenther@suse.de>
10611 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10613 (group_case_labels): Likewise.
10614 (find_taken_edge): Push sanity checking on val to workers...
10615 (find_taken_edge_cond_expr): ... here
10616 (find_taken_edge_switch_expr): ... and here, handle cases
10617 with just a default label.
10618 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10619 (group_case_labels): Likewise.
10620 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10621 group_case_labels does anything cleanup the CFG again.
10623 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10625 PR tree-optimization/81196
10626 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10627 exit condition comparing two IVs.
10629 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10631 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10632 profile to the dummy entry at the end of the list of architectures.
10633 * config/arm/arm-cpu-cdata.h: Regenerated.
10635 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10636 Michael Collison <michael.collison@arm.com>
10639 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10641 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10642 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10643 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10644 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10645 of shift when the shift amount is masked with constant equal to
10646 the size of the mode.
10647 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10650 2017-06-29 Martin Liska <mliska@suse.cz>
10652 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10654 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10657 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10660 (edge_growth_cache_entry::edge_growth_cache_entry): New
10662 (reset_edge_growth_cache): Update to use constructor.
10664 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10666 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10667 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10668 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10670 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10672 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10673 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10675 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10677 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10678 (*-linux-uclibc*): Add t-uclibc tmake_file.
10679 * config/t-musl: New.
10680 * config/t-uclibc: New.
10682 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10684 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10686 (gen_comm_data): Emit architectural setting of arch_prof.
10687 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10689 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10690 (armv8-m.base, armv8-m.main): Likewise.
10691 * arm-protos.h (arm_build_target): Add profile field.
10692 (arch_option): Likewise.
10693 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10695 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10696 arm_active_target.profile.
10698 2017-06-28 Richard Biener <rguenther@suse.de>
10700 PR middle-end/81227
10701 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10702 TYPE_OVERFLOW_WRAPS.
10703 * match.pd (negate_expr_p): Likewise.
10704 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10705 fold_build2, not fold_binary.
10707 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10709 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10710 Convert memory address to Pmode.
10711 (aarch64_print_operand): Assert MEM operands are always Pmode.
10713 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10716 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10717 Remove redundant if.
10718 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10719 * config/arm/aarch-common-protos.h
10720 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10721 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10723 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10726 * multiple_target.c (create_dispatcher_calls): Set the default
10727 clone to be static, not public.
10729 2017-06-28 Richard Biener <rguenther@suse.de>
10731 * tree-vect-loop.c (vectorizable_reduction): Move special
10732 cond reduction IV var creation ...
10733 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10734 parameter. Use STMT_VINFO_VECTYPE.
10735 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10738 2017-06-28 Martin Liska <mliska@suse.cz>
10741 * ipa-visibility.c (non_local_p): Handle visibility.
10743 2017-06-28 Martin Liska <mliska@suse.cz>
10746 * common.opt: Add IntegerRange to various options.
10747 * opt-functions.awk (integer_range_info): New function.
10748 * optc-gen.awk: Add integer_range_info to cl_options struct.
10749 * opts-common.c (decode_cmdline_option): Handle
10750 CL_ERR_INT_RANGE_ARG.
10751 (cmdline_handle_error): Likewise.
10752 * opts.c (print_filtered_help): Show valid interval in
10753 when --help is provided.
10754 * opts.h (struct cl_option): Add range_min and range_max fields.
10755 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10757 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10759 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10760 (x * C EQ/NE y * C): New transformation.
10762 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10764 * genmultilib (combination_space): Accept '+' in option names.
10766 2017-06-28 Martin Liska <mliska@suse.cz>
10769 * asan.c (instrument_derefs): Bail out inner references
10770 that are hard register variables.
10772 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10775 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10776 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10778 2017-06-28 Richard Biener <rguenther@suse.de>
10780 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10781 (vect_get_slp_defs): Adjust.
10782 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10784 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10786 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10787 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10788 (vectorizable_reduction): Adjust.
10789 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10791 (vect_get_slp_defs): Likewise.
10792 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10793 (vectorizable_bswap): Adjust.
10794 (vectorizable_call): Likewise.
10795 (vectorizable_conversion): Likewise.
10796 (vectorizable_assignment): Likewise.
10797 (vectorizable_shift): Likewise.
10798 (vectorizable_operation): Likewise.
10799 (vectorizable_store): Likewise.
10800 (vectorizable_condition): Likewise.
10801 (vectorizable_comparison): Likewise.
10803 2017-06-28 Michael Collison <michael.collison@arm.com>
10806 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10808 (arm_gen_movmemqi): Removed unused variable 'i'.
10809 Convert 'for' loop into 'while' loop.
10810 (arm_expand_prologue): Remove last unnecessary set of insn.
10811 (thumb_pop): Remove unused variable 'pushed_words'.
10812 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10814 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10816 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10817 * config/s390/s390-protos.h: Add prototype of
10818 s390_rel_address_ok_p.
10819 * config/s390/s390.c (s390_got_symbol): New function.
10820 (s390_rel_address_ok_p): New function.
10821 (legitimize_pic_address): Use s390_rel_address_ok_p.
10822 (s390_load_got): Use s390_got_symbol.
10823 (s390_option_override): Issue error if
10824 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10825 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10827 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10829 2017-06-27 Andrew Pinski <apinski@cavium.com>
10831 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10832 (X * copysign (1.0, X)): New pattern.
10833 (X * copysign (1.0, -X)): New pattern.
10834 (copysign (-1.0, CST)): New pattern.
10836 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10838 * genmultilib (combination_space): Remove variable.
10839 Validate reuse rules against regular expression for any sequence
10840 of multilib options in any order.
10842 2017-06-27 Michael Collison <michael.collison@arm.com>
10844 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10845 call aarch64_split_simd_combine.
10846 * (aarch64_combine_internal<mode>): Delete pattern.
10847 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10848 Allow register and subreg operands.
10850 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10852 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10853 specific need, just fallback on defaults.
10854 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10856 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10857 Olivier Hainque <hainque@adacore.com>
10859 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10861 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10862 targets. Pick a default if no particular attempt applied.
10863 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10866 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10868 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10869 (x86_64-wrs-vxworks7): Likewise.
10871 2017-06-27 Marek Polacek <polacek@redhat.com>
10874 * ubsan.c (instrument_null): Check get_base_address's result for null.
10876 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10878 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10880 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10882 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10883 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10884 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10885 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10886 New function types.
10887 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10888 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10889 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10890 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10891 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10892 BUILT_IN_FEUPDATEENV): New builtins.
10893 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10894 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10895 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10896 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10898 (builtin_structptr_types): Adjust size.
10899 * tree.c (builtin_structptr_types): Add four entries.
10901 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10902 Olivier Hainque <hainque@adacore.com>
10904 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10905 (TLS_SYM): New local macro, forcing reference to __tls__ on
10906 link command lines for VxWorks 7 RTPs, triggering initialization
10908 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10909 OS features TLS support, true for RTPs on VxWorks 7.
10910 * config/vxworks.c (vxworks_override_options): Setup emutls
10913 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10915 * predict.c (test_prediction_value_range): Use -1U instead of -1
10916 to avoid narrowing conversion warning.
10917 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10918 to avoid narrowing conversion warning.
10919 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10921 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10923 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10925 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10926 64bit configurations.
10927 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10928 (SIZE_TYPE): Likewise.
10929 * config/vxworks.c (vxworks_emutls_var_fields): Use
10930 long_unsigned_type_node instead of unsigned_type_node as the offset
10931 field type, which is "pointer" mode in emutls.c.
10933 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10936 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10938 PR middle-end/81207
10939 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10940 gimple_vuse copying separately from gimple_vdef copying.
10942 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10944 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10947 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10948 Olivier Hainque <hainque@adacore.com>
10950 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10951 for all vxworks7 targets.
10952 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10953 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10954 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10955 variations for VX6/VX7 and 32/64bits later on in ...
10956 (VXWORKS_LIB_SPEC): Leverage new macros.
10957 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10958 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10960 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10962 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10963 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10965 2017-06-26 Carl Love <cel@us.ibm.com>
10967 * config/rs6000/rs6000-c.c: Add support for built-in functions
10968 vector bool char vec_reve (vector bool char);
10969 vector signed char vec_reve (vector signed char);
10970 vector unsigned char vec_reve (vector unsigned char);
10971 vector bool int vec_reve (vector bool int);
10972 vector signed int vec_reve (vector signed int);
10973 vector unsigned int vec_reve (vector unsigned int);
10974 vector bool long long vec_reve (vector bool long long);
10975 vector signed long long vec_reve (vector signed long long);
10976 vector unsigned long long vec_reve (vector unsigned long long);
10977 vector bool short vec_reve (vector bool short);
10978 vector signed short vec_reve (vector signed short);
10979 vector double vec_reve (vector double);
10980 vector float vec_reve (vector float);
10981 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10982 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10983 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10984 (altivec_vreve): New pattern.
10985 * config/rs6000/altivec.h (vec_reve): New define.
10986 * doc/extend.texi (vec_rev): Update the built-in documentation file
10987 for the new built-in functions.
10989 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10991 PR tree-optimization/71815
10992 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10994 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10996 (slsr_process_phi): Likewise.
10997 (replace_uncond_cands_and_profitable_phis): Don't replace a
10998 multiply candidate with a stride of 1 (copy or cast).
10999 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11001 (lowest_cost_path): Likewise.
11002 (total_savings): Likewise.
11004 2017-06-26 Richard Biener <rguenther@suse.de>
11007 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11008 Use def_builtin_pure for all gather builtins.
11010 2017-06-26 Richard Biener <rguenther@suse.de>
11012 PR tree-optimization/81203
11013 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11014 non-dominating BBs.
11016 2017-06-26 Marek Polacek <polacek@redhat.com>
11019 * doc/invoke.texi: Document -Wmultistatement-macros.
11021 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11023 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11024 arm_neon_ok_no_float_abi effective target.
11026 2017-06-26 Richard Biener <rguenther@suse.de>
11028 PR tree-optimization/80928
11029 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11030 (copy_bbs): Set BB_DUPLICATED flag early.
11031 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11033 (execute_on_shrinking_pred): Likewise.
11034 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11035 BB_DUPLICATED blocks.
11036 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11037 iterate over all PHIs considering removal of *gsi.
11039 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11041 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11044 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11046 * config/rs6000/rs6000-string.c: (expand_block_clear,
11047 do_load_for_compare, select_block_compare_mode,
11048 compute_current_alignment, expand_block_compare,
11049 expand_strncmp_align_check, expand_strn_compare,
11050 expand_block_move, rs6000_output_load_multiple)
11051 Move functions related to string/block move/compare
11052 to a separate file.
11053 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11054 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11055 for this function which is now used in two files.
11056 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11057 * config.gcc: Add rs6000-string.o to extra_objs for
11058 targets powerpc*-*-* and rs6000*-*-*.
11060 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11063 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11064 32-bit, since indexed is not valid for DImode.
11065 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11066 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11067 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11068 (define_peephole2 for Altivec d-form store): Likewise.
11071 * multiple_target.c (create_dispatcher_calls): Only create the
11072 dispatcher call if the function is the default clone of a
11073 versioned function.
11075 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11077 PR middle-end/80902
11078 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11079 a call, force the call to not be a tail call.
11081 2017-06-23 Jeff Law <law@redhat.com>
11083 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11086 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11088 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11089 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11090 handling for early expansion of vector shifts (sl,sr,sra,rl).
11091 (builtin_function_type): Add vector shift right instructions
11092 to the unsigned argument list.
11094 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11096 rtl-optimizatoin/79286
11097 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11098 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11099 trap. PIC register plus a const unspec without offset can never trap.
11101 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11103 * tree.h (builtin_structptr_type): New type.
11104 (builtin_structptr_types): Declare new array.
11105 * tree.c (builtin_structptr_types): New array.
11106 (free_lang_data, build_common_tree_nodes): Use it.
11108 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11111 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11114 2017-06-22 Matt Turner <mattst88@gmail.com>
11116 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11117 Lake models to skylake case. Assume skylake for unknown
11118 models with clflushopt.
11120 2017-06-22 Jeff Law <law@redhat.com>
11122 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11123 frame sizes that do not satisfy aarch64_uimm12_shift.
11125 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11127 * profile-count.h (apply_probability,
11128 apply_scale, probability_in): Fix checks for zero.
11130 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11132 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11133 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11135 2016-06-22 Richard Biener <rguenther@suse.de>
11137 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11138 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11139 REDUC_MAX_EXPR support.
11140 (vectorizable_reduction): Likewise.
11141 (vect_create_epilog_for_reduction): Likewise.
11143 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11145 * match.pd (A / (1 << B) -> A >> B): New.
11146 * generic-match-head.c: Include optabs-tree.h.
11147 * gimple-match-head.c: Likewise.
11148 * optabs-tree.h (target_supports_op_p): New.
11149 * optabs-tree.c (target_supports_op_p): New.
11151 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11153 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11154 $gcc_cv_ld --help output.
11155 (gcc_cv_ld_demangle): Likewise.
11156 (gcc_cv_ld_eh_frame_hdr): Likewise.
11157 (gcc_cv_ld_pie): Likewise.
11158 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11159 (gcc_cv_ld_buildid): Likewise.
11160 (gcc_cv_ld_sysroot): Likewise.
11161 (ld_bndplt_support): Likewise.
11162 (ld_pushpopstate_support): Likewise.
11163 * configure: Regenerate.
11164 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11166 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11169 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11170 operands indexes to avoid gap between operands and match_dups.
11172 2017-06-21 Andrew Pinski <apinski@cavium.com>
11174 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11175 Increment Arith_shift and Arith_shift_reg by 1.
11176 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11178 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11179 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11180 (aarch64_strip_extend): Add new argument and test for it.
11181 (aarch64_cheap_mult_shift_p): New function.
11182 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11183 add a cost if it is true.
11184 Update calls to aarch64_strip_extend.
11185 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11187 2017-06-21 Andrew Pinski <apinski@cavium.com>
11189 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11191 (thunderxt88): Likewise.
11192 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11193 (thunderx_prefetch_tune): New variable.
11194 (thunderx2t99_prefetch_tune): Update for the correct values.
11195 (thunderxt88_tunings): New variable.
11196 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11197 generic_prefetch_tune.
11198 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11200 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11202 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11203 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11204 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11205 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11206 (aarch64_atomic_cas<mode>, GPI): Likewise.
11208 2017-06-21 Martin Liska <mliska@suse.cz>
11210 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11211 statements on cold and hot labels.
11212 * predict.c (tree_estimate_probability_bb): Remove the
11213 prediction from this place.
11215 2017-06-21 Martin Liska <mliska@suse.cz>
11217 PR tree-optimization/79489
11218 * gimplify.c (maybe_add_early_return_predict_stmt): New
11220 (gimplify_return_expr): Call the function.
11221 * predict.c (tree_estimate_probability_bb): Remove handling
11223 * predict.def: Update comment about early return predictor.
11224 * gimple-predict.h (is_gimple_predict): New function.
11225 * predict.def: Change default value of early return to 66.
11226 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11228 * passes.def: Put pass_strip_predict_hints to the beginning of
11231 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11233 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11234 FUNCTION_DECL declarations.
11235 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11237 (dwaf2out_decl): Likewise.
11238 * godump.c (go_early_global_decl): Skip call to the real debug hook
11239 for FUNCTION_DECL declarations.
11240 * passes.c (rest_of_decl_compilation): Skip call to the
11241 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11242 -fdump-go-spec is passed.
11244 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11246 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11247 Reorder for compactness.
11248 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11249 (def_builtin_pure, def_builtin_pure2): New functions.
11250 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11252 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11254 * match.pd (nop_convert): New predicate.
11255 ((A +- CST1) +- CST2): Allow some NOP conversions.
11257 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11260 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11261 with ctors/dtors if GOVD_SHARED is set.
11263 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11265 * config/aarch64/aarch64.md (movti_aarch64):
11266 Emit mov rather than orr.
11267 (movtf_aarch64): Likewise.
11268 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11269 Emit mov rather than orr.
11271 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11273 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11274 Swap alternatives, make integer dup more expensive.
11276 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11278 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11279 Return true for non-tls symbols.
11281 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11283 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11284 (cortex-a75): Likewise.
11285 (cortex-a75.cortex-a55): Likewise.
11286 * config/aarch64/aarch64-tune.md: Regenerate.
11287 * doc/invoke.texi (-mtune): Document new values for -mtune.
11289 2017-06-21 Tom de Vries <tom@codesourcery.com>
11291 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11292 stack_size feature.
11293 (Effective-Target Keywords, Other attributes): Suggest using
11294 dg-add-options stack_size feature to get stack limit in stack_size
11295 effective target documentation.
11297 2017-06-21 Julian Brown <julian@codesourcery.com>
11298 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11300 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11301 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11302 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11304 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11305 attribute type list for neon_multiply.
11306 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11307 attribute type list for neon_multiply.
11308 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11309 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11310 attribute type list for neon_multiply.
11311 * config/arm/types.md (crypto_pmull): Add.
11312 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11313 attribute type list.
11315 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11317 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11320 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11322 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11323 to make sure not to dereference a NULL cost_classes_ptr pointer.
11325 2017-06-20 Carl Love <cel@us.ibm.com>
11327 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11328 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11329 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11330 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11331 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11332 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11333 VMULOSW): New enum "unspec" values.
11334 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11335 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11336 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11337 altivec_vmulosw): New patterns.
11338 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11339 VMULOSW): Add definitions.
11341 2017-06-20 Julia Koval <julia.koval@intel.com>
11343 * config/i386/i386.c: Fix rounding expand for new pattern.
11344 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11346 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11348 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11349 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11351 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11353 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11356 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11358 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11359 architecture, then by implementer ID.
11360 * config/aarch64/aarch64-tune.md: Regenerate.
11362 2017-06-20 Richard Biener <rguenther@suse.de>
11364 PR middle-end/81097
11365 * fold-const.c (split_tree): Fold to type before negating.
11367 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11369 * diagnostic-show-locus.c
11370 (selftest::test_fixit_deletion_affecting_newline): New function.
11371 (selftest::diagnostic_show_locus_c_tests): Call it.
11373 2017-06-20 Andreas Schwab <schwab@suse.de>
11376 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11379 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11381 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11382 __ARM_FEATURE_COPROC according to support.
11384 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11386 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11387 Rewritten to avoid overflow for > 32-bit pointers.
11390 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11391 by removing enum keyword.
11392 (ubsan_type_descriptor): Likewise. Formatting fix.
11395 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11396 splitter): Require TARGET_SSE2 in the condition.
11398 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11401 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11402 for doing vector set of SFmode on ISA 3.0.
11403 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11404 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11406 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11407 SFmode value into a V4SF variable that was extracted from another
11408 V4SF variable without converting the element to double precision
11409 and back to single precision vector format.
11410 (vsx_insert_extract_v4sf_p9_2): Likewise.
11412 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11414 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11415 in UWHI to avoid undefined overflow.
11418 * ubsan.h (enum ubsan_encode_value_phase): New.
11419 (ubsan_encode_value): Change second argument to
11420 enum ubsan_encode_value_phase with default value of
11421 UBSAN_ENCODE_VALUE_GENERIC.
11422 * ubsan.c (ubsan_encode_value): Change second argument to
11423 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11424 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11425 create_tmp_var_raw instead of create_tmp_var and use a
11427 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11428 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11429 ubsan_encode_value callers.
11432 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11433 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11434 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11436 2017-06-19 Richard Biener <rguenther@suse.de>
11438 PR middle-end/81118
11439 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11440 estimates if we changed anything.
11442 2017-06-19 Richard Biener <rguenther@suse.de>
11444 PR tree-optimization/80887
11445 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11446 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11447 simplified lookups, then reset mprts_hook.
11448 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11450 (try_to_simplify): Likewise.
11452 2017-06-19 Martin Liska <mliska@suse.cz>
11455 * gimplify.c (gimplify_switch_expr):
11456 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11458 2017-06-19 Martin Liska <mliska@suse.cz>
11460 * doc/install.texi: Document that PGO runs in 4 stages.
11462 2017-06-19 Martin Liska <mliska@suse.cz>
11465 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11466 to dispatcher function name.
11467 * multiple_target.c (replace_function_decl): New function.
11468 (create_dispatcher_calls): Redirect both edges and references.
11470 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11472 * profile-count.c (profile_count::dump): Dump quality.
11473 (profile_count::differs_from_p): Update for unsigned val.
11474 * profile-count.h (profile_count_quality): New enum.
11475 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11477 2017-06-19 Richard Biener <rguenther@suse.de>
11479 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11480 struct function as arg.
11481 (estimate_numbers_of_iterations): Export overload with loop arg.
11482 (free_numbers_of_iterations_estimates_loop): Use an overload of
11483 free_numbers_of_iterations_estimates instead.
11484 * tree-cfg.c (remove_bb): Adjust.
11485 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11486 * tree-parloops.c (gen_parallel_loop): Likewise.
11487 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11489 (tree_unroll_loops_completely): Likewise.
11490 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11491 Use an overload instead and export.
11492 (estimated_loop_iterations): Adjust.
11493 (max_loop_iterations): Likewise.
11494 (likely_max_loop_iterations): Likewise.
11495 (estimate_numbers_of_iterations): Take struct function as arg
11497 (loop_exits_before_overflow): Adjust.
11498 (free_numbers_of_iterations_estimates_loop): Use an overload.
11499 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11500 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11502 2017-06-19 Richard Biener <rguenther@suse.de>
11505 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11506 RANGE_EXPR conservatively.
11508 2017-06-16 Carl Love <cel@us.ibm.com>
11510 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11511 definitions for vec_float, vec_float2, vec_floato,
11512 vec_floate built-ins.
11513 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11514 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11516 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11517 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11518 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11519 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11520 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11521 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11522 vec_floato): Add builtin defines.
11523 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11524 Update the built-in documentation file for the new built-in
11527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11529 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11530 (mthumb): Mark as the negative of -marm.
11532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11534 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11536 (ARM Options, -mtune): Document that this accepts the same
11537 extension options as -mcpu.
11538 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11542 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11543 permitted extensions.
11545 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11547 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11548 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11549 (armv8-m.main): Add option +nodsp.
11550 * config/arm/arm-cpu-cdata.h: Regenerated.
11552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11554 * config/arm/t-fuchsia: New file.
11555 * config.gcc (arm*-*-fuchsia*): Use it.
11557 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11559 * config/arm/t-symbian: Rewrite for new option infrastructure.
11561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11563 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11564 (MULTILIB_REQUIRED): Likewise.
11566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11568 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11569 (MULTILIB_RESUE): Likewise.
11570 (MULTILIB_MATCHES): Likewise.
11571 (MULTLIB_REQUIRED): Likewise.
11573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11575 * config/arm/t-rtems: Rewrite for new option framework.
11577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11579 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11580 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11581 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11582 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11583 * config/arm/t-multilib: ... here.
11584 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11585 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11586 armv7-a and armv8*-a when A-profile libraries have not been built.
11587 * config/arm/t-rmprofile: Rewrite.
11589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11591 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11592 with a backslash. Remove the backslash after substituting unescaped
11594 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11598 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11599 the driver to pass through the default setting of -mfloat-abi.
11600 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11602 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11603 (all_feat_combs): New rule.
11604 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11606 * config/arm/t-aprofile: Rewrite.
11608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11610 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11611 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11612 fpu is not specified by the user/command-line.
11613 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11614 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11615 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11616 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11617 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11618 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11622 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11623 * config/arm/t-arm-elf: Rewritten.
11625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11627 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11628 have some floating-point instructions.
11629 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11630 (TARGET_MAYBE_HARD_FLOAT): New macro.
11631 * config/arm/arm-builtins.c (arm_init_builtins): Use
11632 TARGET_MAYBE_HARD_FLOAT.
11633 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11637 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11638 (configargs.h): Include it.
11639 (arm_print_hint_for_fpu_option): New function.
11640 (arm_parse_fpu_option): New function.
11641 (candidate_extension): New class.
11642 (arm_canon_for_multilib): New function.
11643 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11644 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11645 (ARCH_CANONICAL_SPECS): New macro.
11646 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11650 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11651 are set after handling multilib fragments. Set target_cpu_default2
11654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11656 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11658 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11659 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11661 (check_arch): Likewise.
11662 * config/arm/arm.c (arm_configure_build_target): Handle
11663 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11664 options in the default.
11666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11668 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11669 when an option is an alias of another.
11670 * config/arm/parsecpu.awk (optalias): New parser token.
11671 (gen_comm_data): Mark non-alias options as such. Emit entries
11672 for extension aliases.
11673 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11674 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11675 (armv6kz, armv6zk, armv6t2): Likewise.
11676 (armv7): Make vfpv3-d16 an alias.
11677 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11679 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11680 Sort in canonical order.
11681 (armv8-a): Sort in canonical order.
11682 (armv8.1-a, armv8.2-a): Likewise.
11683 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11685 (cortex-a9): Sort in canonical order.
11686 * config/arm/arm.c (selftests.h): Include it.
11687 (arm_test_cpu_arch_data): New function.
11688 (arm_run_self_tests): New function.
11689 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11690 (targetm): Move declaration to the end of the file.
11691 * arm-cpu-cdata.h: Regenerated.
11693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11695 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11696 call to target_mode_check describing the type of option passed.
11697 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11698 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11699 arm_parse_cpu_option_name to match parameters against list of
11701 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11702 arm_arch_core_flags data structure.
11703 * config/arm/arm-cpu_cdata.h: Regenerated.
11705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11707 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11709 (arm_print_hint_for_cpu_option): Likewise.
11710 (arm_print_hint_for_arch_option): Likewise.
11711 (arm_parse_cpu_option_name): Likewise.
11712 (arm_parse_arch_option_name): Likewise.
11713 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11714 of entries in the all_fpus list.
11715 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11716 (arm_parse_cpu_option_name): Declare.
11717 (arm_parse_arch_option_name): Declare.
11718 (arm_parse_option_features): Declare.
11719 (arm_intialize_isa): Declare.
11720 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11722 (gen_comm_data): ... here. Make definitions non-static.
11723 * config/arm/arm-cpu-data.h: Regenerated.
11724 * config/arm/arm-cpu-cdata.h: Regenerated.
11726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11728 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11729 (cpu_arch_extension): New structure.
11730 (cpu_arch_option, arch_option, cpu_option): New structures.
11731 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11732 architecture types.
11733 (gen_data): Generate new format data tables.
11734 * config/arm/arm.c (cpu_tune): New structure.
11735 (cpu_option, processors): Delete.
11736 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11737 (arm_print_hint_for_cpu_option): ... this and ...
11738 (arm_print_hint_for_arch_option): ... this.
11739 (arm_parse_arch_cpu_name): Delete. Replace with ...
11740 (arm_parse_cpu_option_name): ... this and ...
11741 (arm_parse_arch_option_name): ... this.
11742 (arm_unrecognized_feature): Change type of target parameter to
11744 (arm_parse_arch_cpu_features): Delete. Replace with ...
11745 (arm_parse_option_features): ... this.
11746 (arm_configure_build_target): Rework to use new configuration data
11748 (arm_print_tune_info): Rework for new configuration data tables.
11749 * config/arm/arm-cpu-data.h: Regenerated.
11750 * config/arm/arm-cpu.h: Regenerated.
11752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11754 * Makefile.in (OBJS): Move sbitmap.o from here ...
11755 (OBJS-libcommon): ... to here.
11757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11759 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11760 (ISA_ALL_CRYPTO): New macro.
11761 (ISA_ALL_SIMD): New macro
11762 (ISA_ALL_FP): New macro.
11763 * config/arm/arm.c (fpu_bitlist): Update initializer.
11764 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11766 (arm9e): Add fpu. Add option for nofp
11767 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11768 (arm926ej-s, arm1026ej-s): Likewise.
11769 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11770 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11771 neon-fp16, neon-vfpv4, nofp and nosimd.
11772 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11773 (cortex-a8): Add fpu. Add option for nofp.
11774 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11775 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11776 (cortex-r4f): Add fpu.
11777 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11778 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11780 (cortex-r8): Likewise.
11781 (cortex-m4): Add fpu. Add option for nofp.
11782 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11783 (cortex-a17.cortex-a7): Likewise.
11784 (cortex-a32): Add fpu. Add options for crypto and nofp.
11785 (cortex-a35, cortex-a53): Likewise.
11786 (cortex-a57): Add fpu. Add option for crypto.
11787 (cortex-a72, cortex-a73): Likewise.
11788 (exynos-m1): Likewise.
11789 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11790 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11791 (cortex-m33): Add fpu. Add option for nofp.
11792 * config/arm/arm-cpu-cdata.h: Regenerated
11793 * config/arm/arm-cpu-data.h: Regenerated.
11795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11797 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11798 (armv5te, armv5tej): Likewise.
11799 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11800 (armv7): Add options fp and vfpv3-d16.
11801 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11802 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11804 (armv7ve): Likewise.
11805 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11806 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11807 (armv8-a): Add nocrypto option.
11808 (armv8.1-a, armv8.2-a): Likewise.
11809 (armv8-m.main): add options fp, fp.dp and nofp.
11811 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11813 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11815 (armv8-a+crc): Delete.
11816 (armv8.1-a): Add options simd, crypto and nofp.
11817 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11818 (armv8.2-a+fp16): Delete.
11819 (armv8-m.main): Add option dsp.
11820 (armv8-m.main+dsp): Delete.
11821 (cortex-a8): Add fpu. Add nofp option.
11822 (cortex-a9): Add fpu. Add nofp and nosimd options.
11823 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11824 link to main cpu and architecture data structures.
11825 (gen_comm_data): Only put isa attributes from the main architecture
11827 (option): New statement for architecture and CPU entries.
11828 * arm.c (struct cpu_option): New structure.
11829 (struct processors): Add entry for options.
11830 (arm_unrecognized_feature): New function.
11831 (arm_parse_arch_cpu_name): Ignore any characters after the first
11833 (arm_parse_arch_cpu_feature): New function.
11834 (arm_configure_build_target): Separate out any CPU and architecture
11835 features and parse separately. Don't error out if -mfpu=auto is
11836 used with only an architecture string.
11837 (arm_print_asm_arch_directives): New function.
11838 (arm_file_start): Call it.
11839 * config/arm/arm-cpu-cdata.h: Regenerated.
11840 * config/arm/arm-cpu-data.h: Likewise.
11841 * config/arm/arm-tables.opt: Likewise.
11843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11845 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11846 assembler when it is not -mfpu=auto.
11848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11850 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11851 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11852 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11853 (ASM_CPU_SPEC): Rewrite.
11854 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11855 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11856 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11857 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11858 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11859 copied string is NUL-terminated. Also strip any characters prefixed
11861 (arm_rewrite_selected_arch): New function.
11862 (arm_rewrite_march): New function.
11864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11866 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11867 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11868 (march, mcpu, mtune): Convert to string-based options.
11869 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11870 (arm_parse_arch_cpu_name): New function.
11871 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11872 identify selected architecture or CPU.
11873 (arm_option_save): New function.
11874 (TARGET_OPTION_SAVE): Redefine.
11875 (arm_option_restore): Restore string options.
11876 (arm_option_print): Print string options.
11878 2017-06-16 Martin Sebor <msebor@redhat.com>
11880 PR tree-optimization/80933
11881 PR tree-optimization/80934
11882 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11883 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11884 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11885 (gimple_fold_builtin): Call them.
11887 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11889 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11890 as unlikely; update profile.
11892 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11894 * predict.c (force_edge_cold): Handle declaring edges impossible
11897 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11899 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11901 (try_unroll_loop_completely): Fix reporting.
11903 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11905 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11907 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11910 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11911 if given a non-constant argument for an intrinsic which requires a
11914 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11916 * profile.c (compare_freqs): New function.
11917 (branch_prob): Sort edge list.
11918 (find_spanning_tree): Assume that the list is priority sorted.
11920 2017-06-16 Richard Biener <rguenther@suse.de>
11922 PR tree-optimization/81090
11923 * passes.def (pass_record_bounds): Remove.
11924 * tree-pass.h (make_pass_record_bounds): Likewise.
11925 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11926 make_pass_record_bounds): Likewise.
11927 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11928 not free niter estimates at the beginning but at the end.
11929 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11931 2017-06-16 Richard Biener <rguenther@suse.de>
11933 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11934 initializer to workaround ICE in host GCC 4.8.
11936 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11938 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11940 (clone_inlined_nodes): Update.
11942 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11944 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11945 prefetch settings, and enable prefetching by default at -O3.
11947 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11949 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11950 Set flag_prefetch_loop_arrays according to tuning data.
11952 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11954 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11955 New tune structure.
11956 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11957 [Unrelated to main purpose of the patch] Place the pointer field last
11958 to enable type checking errors when tune structure are wrongly merged.
11959 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11960 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11961 (thunderx2t99_prefetch_tune): New tune constants.
11962 (tune_params *_tunings): Update all tunings (no functional change).
11963 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11964 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11965 from tunings structures.
11967 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11970 * ubsan.c (instrument_null): Add T argument, use it instead
11971 of computing it based on IS_LHS.
11972 (instrument_object_size): Likewise.
11973 (pass_ubsan::execute): Adjust instrument_null and
11974 instrument_object_size callers to pass gimple_get_lhs or
11975 gimple_assign_rhs1 result to it. Use instrument_null instead of
11976 calling get_base_address and instrument_mem_ref. Handle
11977 aggregate call arguments for object-size sanitization.
11979 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11981 PR tree-optimization/81089
11982 * tree-vrp.c (is_masked_range_test): Validate operands of
11985 2017-06-15 Martin Sebor <msebor@redhat.com>
11988 * dumpfile.c (dump_register): Avoid calling memset to initialize
11989 a class with a default ctor.
11990 * gcc.c (struct compiler): Remove const qualification.
11991 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11992 * hash-table.h: Ditto.
11993 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11995 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11996 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11998 * params.h (struct param_info): Make struct members non-const.
11999 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12000 with default initialization.
12001 * vec.h (vec_copy_construct, vec_default_construct): New helper
12003 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12004 with vec_copy_construct.
12005 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12006 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12007 * doc/invoke.texi (-Wclass-memaccess): Document.
12009 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12011 * emit-rtl.h (is_leaf): Update comment about local
12012 register allocator.
12014 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12017 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12018 for a variable to have a section before checking if the section has a
12020 Set section to.persistent if persistent attribute is set.
12021 Warn if .persistent attribute is used on an automatic variable.
12023 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12025 PR rtl-optimization/80474
12026 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12028 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12030 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12033 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12035 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12036 (casesi): Emit bounds checking as RTL.
12037 (casesi_internal_mips16_<mode>): Remove bounds checking.
12039 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12041 * config/xtensa/xtensa.c (xtensa_option_override): Append
12042 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12043 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12044 xtensa_doloop_hooks): Define unconditionally.
12045 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12047 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12048 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12049 for it in xtensa_option_override.
12050 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12051 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12053 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12055 * doc/cppopts.texi: Document '-' special value to -MF.
12057 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12059 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12060 (cortex_a53_fconst): Likewise.
12061 (cortex_a53_fpmul): Likewise.
12062 (cortex_a53_f_load_64): Likewise.
12063 (cortex_a53_f_load_many): Likewise.
12064 (cortex_a53_advsimd_alu): Likewise.
12065 (cortex_a53_advsimd_alu_q): Likewise.
12066 (cortex_a53_advsimd_mul): Likewise.
12067 (cortex_a53_advsimd_mul_q): Likewise.
12068 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12069 Add missing fmul, r2f_cvt and fconst cases.
12071 2017-06-14 Richard Biener <rguenther@suse.de>
12073 PR middle-end/81088
12074 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12076 (fold_binary_loc): When associating do not treat pre-existing
12077 TREE_OVERFLOW on literal constants as a reason to allow
12078 TREE_OVERFLOW on associated literal constants.
12080 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12082 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12083 (MASK_FEATURES): New macro.
12084 * config/sparc/sparc.c (sparc_option_override): Remove the special
12085 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12087 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12089 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12090 a division of 0 if non-call exceptions are enabled.
12092 2017-06-14 Andrew Pinski <apinski@cavium.com>
12093 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12096 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12097 Improve vector initialization code gen for only variable case.
12099 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12101 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12103 2017-06-14 Richard Biener <rguenther@suse.de>
12105 PR tree-optimization/81083
12106 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12109 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12111 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12112 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12113 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12114 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12115 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12116 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12118 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12120 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12121 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12123 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12125 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12127 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12129 * config/rs6000/t-rtems: Don't handle SPE.
12131 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12133 * config/rs6000/t-linux: Don't handle SPE.
12135 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12137 * config/rs6000/eabispe.h: Delete file.
12139 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12141 * config/rs6000/t-spe: Delete file.
12143 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12145 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12146 (rs6000_legitimate_offset_address_p): Return false for anything in
12147 V2SImode or V2SFmode.
12149 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12151 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12152 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12153 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12155 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12156 (rs6000_legitimate_offset_address_p): Ditto.
12157 (rs6000_emit_move): Ditto.
12158 (rs6000_init_builtins): Remove V4HI_type_node.
12160 2017-06-13 Martin Liska <mliska@suse.cz>
12163 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12164 (gate_asan): Likewise.
12165 * asan.h (asan_no_sanitize_address_p): Remove the function.
12166 (sanitize_flags_p): New function.
12167 * builtins.def: Fix coding style.
12168 * common.opt: Use renamed enum value.
12169 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12170 * doc/extend.texi: Document no_sanitize attribute.
12171 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12172 to SANITIZE_UNDEFINED_NONDEFAULT.
12173 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12174 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12175 Use sanitize_flags_p.
12176 * gimplify.c (gimplify_function_tree): Likewise.
12177 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12178 * opts.c (parse_no_sanitize_attribute): New function.
12179 (common_handle_option): Use renamed enum value.
12180 * opts.h (parse_no_sanitize_attribute): Declare.
12181 * tree.c (sanitize_flags_p): New function.
12182 * tree.h: Declared here.
12183 * tsan.c: Use sanitize_flags_p.
12184 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12185 (instrument_mem_ref): Likewise.
12186 (instrument_bool_enum_load): Likewise.
12187 (do_ubsan_in_current_function): Remove the function.
12188 (pass_ubsan::execute): Use sanitize_flags_p.
12189 * ubsan.h: Remove do_ubsan_in_current_function
12190 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12191 (dump_function_to_file): Use it here.
12193 2017-06-13 Martin Jambor <mjambor@suse.cz>
12195 PR tree-optimization/80803
12196 PR tree-optimization/81063
12197 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12198 (propagate_subaccesses_across_link): Enqueue subtree whenever
12199 necessary instead of relying on the caller.
12201 2017-06-13 Martin Jambor <mjambor@suse.cz>
12203 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12204 that have a first_link.
12205 (sort_and_splice_var_accesses): Do not check first_link before
12207 (subtree_mark_written_and_enqueue): Likewise.
12208 (propagate_all_subaccesses): Likewise and do not stop at first
12209 parent with a first_link.
12211 2017-06-13 Martin Jambor <mjambor@suse.cz>
12213 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12216 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12218 * match.pd: New pattern.
12220 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12222 * tree-vrp.c (is_masked_range_test): New function.
12223 (register_edge_assert_for): Determine ranges for
12226 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12228 PR tree-optimization/67328
12229 * fold-const.c (maskable_range_p): New function.
12230 (build_range_check): Generate bittests if possible.
12232 2017-06-13 Martin Liska <mliska@suse.cz>
12234 * gimple-pretty-print.c (dump_probability): Add new argument.
12235 (dump_edge_probability): Dump both probability and count.
12236 (dump_gimple_label): Likewise.
12237 (dump_gimple_bb_header): Likewise.
12239 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12242 * config/avr/avr-devices.c: Fix indentation.
12243 * config/avr/gen-avr-mmcu-specs.c: Dito.
12245 2017-06-13 Richard Biener <rguenther@suse.de>
12247 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12248 instead get vector type from stmt_info.
12249 (vectorizable_reduction): Adjust. Remove dead code.
12251 2017-06-13 Richard Biener <rguenther@suse.de>
12253 PR middle-end/81065
12254 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12255 case of C * (x * C2 + C3).
12256 (fold_addr_of_array_ref_difference): Properly fold index difference.
12258 2017-06-12 David S. Miller <davem@davemloft.net>
12261 * config/sparc/sparc.md (return expander): Emit frame blockage if
12262 function uses alloca.
12264 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12266 * combine.c (make_field_assignment): Check len rather than the mode
12267 precision when calling force_to_mode.
12269 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12271 Support multilibs and devices that see flash in RAM address range.
12274 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12275 (avr_mcu_t) <flash_pm_offset>: New field.
12276 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12277 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12278 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12279 (AVR_TINY_PM_OFFSET): Remove macro.
12280 * config/avr/avr.opt (-mshort-calls): New option.
12281 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12282 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12283 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12284 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12285 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12286 instead of avr_arch->have_jmp_call.
12287 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12288 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12289 avr_arch->flash_pm_offset to define.
12290 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12291 new field flash_pm_offset. Add entry for avrxmega3.
12292 (avr_texinfo): Add entry for avrxmega3.
12293 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12294 attiny212, attiny214,
12295 attiny412, attiny414, attiny416, attiny417,
12296 attiny814, attiny816, attiny817,
12297 attiny1614, attiny1616, attiny1617,
12298 attiny3214, attiny3216, attiny3217.
12299 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12300 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12301 (avr_print_operand_address) [AVR_TINY]: Same.
12302 (avr_asm_init_sections) <readonly_data_section>: Only patch
12303 callback if avr_arch->flash_pm_offset = 0.
12304 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12305 for rodata if avr_arch->flash_pm_offset != 0.
12306 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12307 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12308 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12309 (m_options): Append opt_rcall.
12310 (m_dirnames): Append dir_rcall.
12311 * config/avr/t-multilib: Regenerate.
12313 * configure.ac [target=avr]: Check whether avrxmega3 default
12314 linker description file works as needed.
12315 * configure: Regenerate.
12316 * doc/avr-mmcu.texi: Regenerate.
12317 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12318 <__AVR_ARCH__>: Document avrxmega3 and 103.
12319 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12320 <__AVR_SHORT_CALLS__>: Document it.
12321 <__AVR_PM_BASE_ADDRESS__>: Document it.
12322 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12323 (AVR Variable Attributes) <progmem>: Document this is
12324 not needed for avrxmega3.
12325 (AVR Named Address Spaces) <__flash>: Dito.
12327 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12329 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12331 2017-06-12 Doug Rupp <rupp@adacore.com>
12333 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12334 Append vxworks-stdint.h to the tm_file list.
12335 * config/vxworks-stdint.h: New file.
12337 2017-06-12 Martin Liska <mliska@suse.cz>
12339 PR tree-optimization/81041
12340 * tree-profile.c (gimple_gen_ic_func_profiler):
12341 Create an extra BB in profile-generate
12342 (gimple_gen_time_profiler): Likewise.
12344 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12346 PR tree-optimization/81003
12347 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12348 (update_range_test): Use it instead of force_gimple_operand_gsi.
12350 2017-06-12 Richard Biener <rguenther@suse.de>
12352 PR tree-optimization/81053
12353 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12354 with backedge value not defined in loop. Simplify def stmt
12357 2017-06-11 Tom de Vries <tom@codesourcery.com>
12360 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12362 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12363 nvptx_cannot_force_const_mem.
12365 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12367 * opts.c (finish_options): Move test for flag_split_stack after
12368 it has been initialized.
12370 2017-06-11 Jason Merrill <jason@redhat.com>
12372 * tree.h (id_equal): New.
12373 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12374 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12375 instead of strcmp of IDENTIFIER_POINTER.
12377 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12379 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12380 (mark_all_inlined_calls_cdtor): Fix formating.
12381 (inline_transform): Rescale profile before inlining.
12383 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12385 * cgraph.h (cgraph_edge::clone): Update prototype.
12386 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12387 (cgraph_node::create_clone): Update.
12388 (cgraph_node::create_version_clone): Update.
12389 * tree-inline.c (copy_bb): Update.
12390 (expand_call_inline): Update.
12392 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12394 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12395 factored out from ...
12396 (rs6000_emit_prologue): ... here.
12398 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12400 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12401 factored out from ...
12402 (rs6000_emit_prologue): ... here.
12404 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12406 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12408 (handle_missing_profiles): Fix computation of tp_first_run.
12409 (counts_to_freqs): Do not touch freqs when count is 0.
12411 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12413 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12416 2017-06-10 Tom de Vries <tom@codesourcery.com>
12418 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12419 attributes): Document signal effective target.
12421 2017-06-10 Tom de Vries <tom@codesourcery.com>
12423 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12424 Document effective target stack_size.
12426 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12428 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12429 to the edit_context if they can be auto-applied.
12431 2017-06-9 Ian Lance Taylor <iant@golang.org>
12433 * opts.c (finish_options): If -fsplit-stack, disable implicit
12434 -forder-blocks-and-partition.
12435 * doc/invoke.texi (Optimize Options): Document that when using
12436 -fsplit-stack -forder-blocks-and-partition is not implicitly
12439 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12441 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12442 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12443 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12444 * builtins.def (abort, trap, unreachable): Declare cold.
12445 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12446 * tree-core.h (ECF_COLD): New.
12447 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12448 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12450 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12452 * predict.c (unlikely_executed_stmt_p): Cleanup.
12454 2017-06-09 Richard Biener <rguenther@suse.de>
12456 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12457 model if the ref is always written to.
12459 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12461 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12463 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12465 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12468 2017-06-09 Tom de Vries <tom@codesourcery.com>
12471 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12472 "target cannot support label values" when encountering LABEL_REF.
12474 2017-06-09 Martin Liska <mliska@suse.cz>
12476 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12477 (gimple_gen_ic_func_profiler): Emit direct comparison
12478 of __gcov_indirect_call_callee with NULL.
12479 (gimple_gen_time_profiler): Change probability from
12480 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12482 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12484 * profile.c (edge_gcov_counts): Turn to pointer.
12485 (compute_branch_probabilities, compute_branch_probabilities): Update.
12486 (branch_prob): Do not clear edge_gcov_count.
12487 * profile.h (edge_gcov_counts): Turn to pointer.
12488 (edge_gcov_count): Update.
12490 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12492 * gimple.h (gimple_check_failed): Mark cold.
12494 2017-06-09 Richard Biener <rguenther@suse.de>
12496 PR tree-optimization/66623
12497 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12498 refactor check_reduction into two parts, properly computing
12499 whether we have to check reduction validity for outer loop
12502 2017-06-09 Richard Biener <rguenther@suse.de>
12504 PR tree-optimization/79483
12505 * graphite-scop-detection.c (order): New global.
12506 (get_order): Compute bb to order mapping that satisfies code
12507 generation constraints.
12508 (cmp_pbbs): New helper.
12509 (build_scops): Start domwalk at entry block, sort generated
12512 2017-06-09 Richard Biener <rguenther@suse.de>
12514 PR middle-end/81007
12515 * ipa-polymorphic-call.c
12516 (ipa_polymorphic_call_context::restrict_to_inner_class):
12517 Skip FIELD_DECLs with error_mark_node type.
12518 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12521 2017-06-09 Martin Liska <mliska@suse.cz>
12523 * predict.c (struct branch_predictor): New struct.
12524 (test_prediction_value_range): New test.
12525 (predict_c_tests): New function.
12526 * selftest-run-tests.c (selftest::run_tests): Run the function.
12527 * selftest.h: Declare new tests.
12529 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12532 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12533 gen_add3_insn did not fail.
12534 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12535 r0, construct that number in a temporary reg and add that reg to r0.
12536 If asked to put the result in r0 as well, fail.
12538 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12540 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12541 for early expansion of vec_eqv.
12543 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12545 PR middle-end/81005
12546 * ubsan.c (instrument_null): Avoid pointless code temporary.
12547 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12549 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12553 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12556 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12559 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12561 * predict.c (unlikely_executed_edge_p): Move ahead.
12562 (probably_never_executed_edge_p): Use it.
12564 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12566 PR middle-end/79988
12567 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12568 gimple_call_builtin_p call.
12570 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12572 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12573 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12574 rtl_check_failed_type2, rtl_check_failed_code1,
12575 rtl_check_failed_code2, rtl_check_failed_code_mode,
12576 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12577 rtvec_check_failed_bounds, rtl_check_failed_flag,
12578 _fatal_insn_not_found, _fatal_insn): Likewise.
12579 * tree.h (tree_contains_struct_check_failed,
12580 tree_check_failed, tree_not_check_failed,
12581 tree_class_check_failed, tree_range_check_failed,
12582 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12583 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12584 tree_operand_check_failed, omp_clause_check_failed,
12585 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12588 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12590 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12591 flag_branch_probabilities.
12592 * ipa-inline.c (edge_badness): Likewise.
12593 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12594 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12595 * predict.c (maybe_hot_frequency_p): Likewise.
12596 (probably_never_executed): Likewise.
12597 * sched-ebb.c (schedule_ebbs): Likewise.
12598 * sched-rgn.c (find_single_block_region): Likewise.
12599 * tracer.c (tail_duplicate): Likewise.
12601 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12603 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12604 longer requires x_flag_profile_use.
12606 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12608 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12609 instead of flag_reorder_blocks_and_partition.
12610 * dbxout.c (dbxout_function_end): Likewise.
12611 * dwarf2out.c (gen_subprogram_die): Likewise.
12612 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12613 * hw-doloop.c (reorg_loops): Likewise.
12614 * varasm.c (assemble_start_function,
12615 assemble_end_function): Likewise.
12616 (decide_function_section): Do not check for
12617 flag_reorder_blocks_and_partition.
12619 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12621 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12623 (chkp_get_hard_register_fake_addr_expr): Ditto.
12624 (chkp_build_addr_expr): Add check for hard reg case.
12625 (chkp_parse_array_and_component_ref): Ditto.
12626 (chkp_find_bounds_1): Ditto.
12627 (chkp_process_stmt): Don't generate bounds store for
12630 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12632 * predict.c (maybe_hot_bb_p): Do not check profile status.
12633 (maybe_hot_edge_p): Likewise.
12634 (probably_never_executed): Check for zero counts even if profile
12636 (unlikely_executed_edge_p): New function.
12637 (unlikely_executed_stmt_p): New function.
12638 (unlikely_executed_bb_p): New function.
12639 (set_even_probabilities): Use unlikely predicates.
12640 (combine_predictions_for_bb): Likewise.
12641 (predict_paths_for_bb): Likewise.
12642 (predict_paths_leading_to_edge): Likewise.
12643 (determine_unlikely_bbs): New function.
12644 (estimate_bb_frequencies): Use it.
12645 (compute_function_frequency): Use zero counts even if profile is
12647 * profile-count.h: Fix typo.
12649 2017-08-08 Julia Koval <julia.koval@intel.com>
12651 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12652 _mm512_mask_cvtsepi16_storeu_epi8,
12653 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12654 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12655 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12656 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12657 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12658 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12659 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12660 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12661 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12662 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12663 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12664 __builtin_ia32_pmovuswb256mem_mask,
12665 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12666 __builtin_ia32_pmovwb512mem_mask): New builtins.
12668 2017-08-08 Julia Koval <julia.koval@intel.com>
12670 PR target/73350,80862
12671 * config/i386/subst.md (round): Fix round pattern.
12672 * config/i386/i386.c (ix86_erase_embedded_rounding):
12673 Fix erasing rounding for the fixed pattern.
12675 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12677 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12679 2017-06-08 Martin Liska <mliska@suse.cz>
12681 PR gcov-profile/80911
12682 * gcov.c (block_info::block_info): New constructor.
12684 2017-06-07 Carl Love <cel@us.ibm.com>
12686 * config/rs6000/rs6000-c: The return type of the following
12687 built-in functions was implemented as int not long long. Fix sign
12688 of return value for the unsigned version of vec_mulo and vec_mule.
12689 vector unsigned long long vec_bperm (vector unsigned long long,
12690 vector unsigned char)
12691 vector signed long long vec_mule (vector signed int,
12693 vector unsigned long long vec_mule (vector unsigned int,
12694 vector unsigned int)
12695 vector signed long long vec_mulo (vector signed int,
12697 vector unsigned long long vec_mulo (vector unsigned int,
12698 vector unsigned int)
12699 * doc/extend.texi: Fix the documentation for the built-in
12702 2017-06-07 Carl Love <cel@us.ibm.com>
12705 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12708 2017-06-07 Carl Love <cel@us.ibm.com>
12710 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12711 support, Generate doublehv for signed int/float for BE case only.
12713 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12715 * doc/invoke.texi (mcx16): Rewrite.
12717 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12719 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12720 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12721 *mov<mode>_softfloat, and an anonymous splitter): Use
12722 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12724 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12726 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12728 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12729 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12730 (rs6000_debug_reg_global): Adjust.
12731 (rs6000_init_hard_regno_mode_ok): Adjust.
12732 (rs6000_dbx_register_number): Adjust.
12733 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12734 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12735 Remove SPE_ACC and SPEFSCR.
12736 (REG_ALLOC_ORDER): Ditto.
12737 (FRAME_POINTER_REGNUM): Change to 111.
12738 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12739 (REG_CLASS_NAMES): Ditto.
12740 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12741 (REGISTER_NAMES): Ditto.
12742 (ADDITIONAL_REG_NAMES): Ditto.
12743 (rs6000_reg_names): Ditto.
12744 * config/rs6000/rs6000.md: Renumber some register number
12747 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12749 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12751 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12752 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12754 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12755 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12756 Delete the SPE high registers.
12757 (REG_ALLOC_ORDER): Ditto.
12758 (enum reg_class): Remove SPE_HIGH_REGS.
12759 (REG_CLASS_NAMES): Ditto.
12760 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12761 (REGISTER_NAMES): Ditto.
12762 (rs6000_reg_names): Ditto.
12763 * doc/tm.texi.in: Remove SPE as example.
12764 * doc/tm.texi: Regenerate.
12766 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12768 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12769 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12770 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12771 * config/rs6000/rs6000.md (type): Remove "brinc".
12773 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12775 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12776 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12777 * config/rs6000/linuxspe.h: Delete file.
12778 * config/rs6000/rs6000.md: Don't include spe.md.
12779 * config/rs6000/spe.h: Delete file.
12780 * config/rs6000/spe.md: Delete file.
12781 * config/rs6000/t-rs6000: Remove spe.md.
12783 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12785 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12786 (reg_or_none500mem_operand): Delete.
12787 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12788 instead of reg_or_none500mem_operand.
12790 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12792 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12793 handling of SPE flags.
12794 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12798 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12800 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12801 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12802 paired_divv2sf3): Similar.
12803 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12804 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12805 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12807 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12808 Rename the paired_* instruction patterns.
12809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12811 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12812 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12813 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12814 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12815 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12816 PAIRED_VECTOR_MODE.
12817 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12818 (spe_func_has_64bit_regs_p): Delete.
12819 (spe_expand_predicate_builtin): Delete.
12820 (spe_expand_evsel_builtin): Delete.
12821 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12822 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12823 (invalid_e500_subreg): Delete.
12824 (rs6000_legitimize_address): Always force_reg op2 as well, for
12825 paired single memory accesses.
12826 (rs6000_member_type_forces_blk): Delete.
12827 (rs6000_spe_function_arg): Delete.
12828 (rs6000_expand_unop_builtin): Delete SPE handling.
12829 (rs6000_expand_binop_builtin): Ditto.
12830 (spe_expand_stv_builtin): Delete.
12831 (bdesc_2arg_spe): Delete.
12832 (spe_expand_builtin): Delete.
12833 (spe_expand_predicate_builtin): Delete.
12834 (spe_expand_evsel_builtin): Delete.
12835 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12836 (spe_init_builtins): Delete.
12837 (spe_func_has_64bit_regs_p): Delete.
12838 (savres_routine_name): Delete "info" parameter. Adjust callers.
12839 (rs6000_emit_stack_reset): Ditto.
12840 (rs6000_dwarf_register_span): Delete.
12841 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12842 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12843 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12845 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12847 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12848 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12849 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12850 mulv2sf3, divv2sf3): Delete expanders.
12852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12854 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12856 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12858 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12859 * config/rs6000/rs6000.c: Ditto.
12861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12863 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12864 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12865 comparison_operator.
12867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12869 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12870 * config/rs6000/rs6000.opt: Ditto.
12871 * config/rs6000/t-rtems: Ditto.
12873 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12875 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12876 TARGET_E500_SINGLE by 0, simplify.
12877 * config/rs6000/rs6000.c: Ditto.
12878 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12879 (spe_build_register_parallel): Delete.
12880 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12881 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12882 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12883 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12884 (E500_CONVERT): Delete.
12885 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12889 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12890 * config/rs6000/dfp.md: Ditto.
12891 (negdd2, *negdd2_fpr): Merge.
12892 (absdd2, *absdd2_fpr): Merge.
12893 (negtd2, *negtd2_fpr): Merge.
12894 (abstd2, *abstd2_fpr): Merge.
12895 * config/rs6000/e500.h: Delete file.
12896 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12897 TARGET_FPRS by 1 and simplify.
12898 * config/rs6000/rs6000-c.c: Ditto.
12899 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12900 TARGET_DF_SPE by 0.
12901 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12903 * config/rs6000/rs6000.md: Ditto.
12904 (floatdidf2, *floatdidf2_fpr): Merge.
12905 (move_from_CR_gt_bit): Delete.
12906 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12907 (E500_CR_IOR_COMPARE): Delete.
12908 (All patterns that require !TARGET_FPRS): Delete.
12909 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12911 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12913 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12915 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12917 * graphds.c (add_edge): Intitialize edge's attached data.
12918 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12919 pointer parameter. Call pointed function on each edge during
12920 graph traversing. Skip traversing the edge when the function
12922 (graphds_dfs, graphds_scc): Ditto.
12923 (for_each_edge): New parameter. Pass the new parameter to callback
12925 * graphds.h (skip_edge_callback): New function pointer type.
12926 (graphds_dfs, graphds_scc): New function pointer parameter.
12927 (graphds_edge_callback, for_each_edge): New parameter.
12929 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12931 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12932 out code checking if runtime alias check is possible to below ...
12933 Call the new function.
12934 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12935 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12937 2017-06-07 Marek Polacek <polacek@redhat.com>
12940 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12941 TYPE_OVERFLOW_WRAPS check.
12943 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12945 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12946 if versioning is required.
12947 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12948 peeling with the check for versioning.
12950 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12952 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12953 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12954 Set true to new parameter if new ssa variable is defined.
12955 (vect_gen_vector_loop_niters): Refactor. Set range information
12956 for the new vector loop bound variable.
12957 (vect_do_peeling): Ditto.
12959 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12961 * tree-affine.c (ssa.h): Include header file.
12962 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12963 has wrapping overflow behavior.
12965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12967 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12969 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12971 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12972 (tree_to_aff_combination): ... here.
12974 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12976 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12977 reg_pressure model function.
12978 (ivopts_global_cost_for_size): Delete.
12979 (determine_set_costs, iv_ca_recount_cost): Call new model function
12980 ivopts_estimate_reg_pressure.
12982 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12984 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12985 expensive than udiv. Remove floating point cases from mod.
12987 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12989 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12990 Increase idiv cost.
12992 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12994 * config/aarch64/aarch64.md
12995 (copysignsf3): Fix mask generation.
12997 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12999 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13001 (class dump_manager): Add register_dumps method.
13002 * dumpfile.c: Include langhooks.h.
13003 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13004 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13005 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13006 (dump_manager::dump_register): Start with 512 entries instead of 32.
13007 (dump_manager::register_dumps): New method.
13008 * toplev.c (general_init): Instead of invoking register_dumps
13009 langhook, invoke register_dumps method on the dump manager.
13010 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13013 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13015 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13016 Say that # requires an associated define_split to exist, and that
13017 the define_split must be suitable for use after register allocation.
13019 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13021 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13022 (compute_outgoing_frequencies): Also initialize zero counts.
13023 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13024 around loops; preserve more of profile when nothing changes.
13026 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13028 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13030 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13031 * config/arm/arm-cpu-cdata.h: Regenerate.
13032 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13033 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13034 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13035 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13036 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13038 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13040 * config/arm/t-rmprofile: Likewise.
13041 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13043 2017-06-06 David S. Miller <davem@davemloft.net>
13046 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13047 blockage if function uses alloca.
13049 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13051 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13052 New "uid" fields to hold pretty-print IDs of group and ref.
13053 Memory references are now identified as <group_id>:<ref_id>
13054 instead of using [random] addresses.
13055 (dump_mem_details): Simplify, no functional change.
13056 (dump_mem_ref): Simplify and make output more concise.
13057 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13058 (find_or_create_group): Initialize group uid.
13059 (record_ref): Initialize ref uid. Improve debug output.
13060 (prune_group_by_reuse, should_issue_prefetch_p,)
13061 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13062 (mark_nontemporal_store, determine_loop_nest_reuse):
13063 Improve debug output.
13065 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13067 * dbgcnt.def (prefetch): New debug counter.
13068 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13069 (schedule_prefetches): Stop issueing prefetches if debug counter
13072 2017-06-06 Tom de Vries <tom@codesourcery.com>
13074 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13075 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13077 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13079 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13080 Use aarch64_reg_or_zero predicate for operand 4.
13081 (aarch64_compare_and_swap<mode> define_insn_and_split):
13082 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13083 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13085 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13087 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13088 (arm_compute_save_core_reg_mask): This.
13089 (thumb1_compute_save_reg_mask): Rename into ...
13090 (thumb1_compute_save_core_reg_mask): This.
13091 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13092 (arm_compute_frame_layout): Likewise.
13094 2017-06-06 Richard Biener <rguenther@suse.de>
13096 PR tree-optimization/80974
13097 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13098 keep or clear leaders SSA info.
13100 2017-06-06 Tom de Vries <tom@codesourcery.com>
13102 * config/nvptx/nvptx.c (split_mode_p): New function.
13103 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13105 2017-06-06 Tom de Vries <tom@codesourcery.com>
13107 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13109 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13112 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13115 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13117 * shrink-wrap.c (handle_simple_exit): Update profile.
13118 (try_shrink_wrapping): Upate profile.
13120 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13122 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13123 (tree_guess_outgoing_edge_probabilities): New.
13124 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13125 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13127 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13129 * ipa-split.c (split_function): Initialize return bb profile.
13131 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13133 * profile.c (compute_branch_probabilities): Also initialize
13134 EXIT_BLOCK profile.
13136 2017-06-06 Richard Biener <rguenther@suse.de>
13138 PR tree-optimization/80928
13139 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13140 (vect_analyze_loop_operations): Properly guard analysis for
13142 (vect_transform_loop): Likewise.
13143 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13144 (vect_model_induction_cost): Do not cost for pure SLP.
13145 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13146 of induction in inner loop vectorization.
13147 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13148 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13149 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13151 (vect_analyze_slp_cost_1): Cost induction.
13152 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13153 (vect_get_slp_vect_defs): Likewise.
13154 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13155 (vect_transform_stmt): Handle SLP reductions.
13156 * tree-vectorizer.h (vectorizable_induction): Adjust.
13158 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13160 * config/rs6000/rs6000.c (make_resolver_func): Update
13161 init_lowered_empty_function call.
13163 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13165 * doc/invoke.texi: Document the -fprofile-abs-path option.
13166 * common.opt (fprofile-abs-path): New option.
13167 * gcov-io.h (gcov_write_filename): Declare.
13168 * gcov-io.c (gcov_write_filename): New function.
13169 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13170 * profile.c (output_location): Likewise.
13172 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13174 * shring-wrap.c: Revert accidental commit.
13176 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13178 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13180 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13182 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13184 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13185 profile in callgraph edge.
13186 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13187 (apply_scale): Likewise.
13188 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13189 Also scale profile when inlining function with zero profile.
13190 (initialize_cfun): Update exit block profile even when it is zero.
13191 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13192 when profile is read.
13194 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13196 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13197 (CLONE_*): New constants to define the processors we can generate
13198 code for with the target_clone attribute.
13199 (rs6000_clone_map): New array to identify which clone processors
13200 the current program is running on.
13201 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13202 target_clone attribute.
13203 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13204 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13205 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13206 (cpu_expand_builtin): Add support for target_clone attribute.
13207 (rs6000_valid_attribute_p): Allow "default" attribute.
13208 (get_decl_name): New debug function to simplify printing the
13209 current function name in debugging statements.
13210 (rs6000_clone_priority): New functions to support the target_clone
13211 attribute, and be able to generate code to switch between ISA 2.05
13212 through ISA 3.0 (power6 through power9).
13213 (rs6000_compare_version_priority): Likewise.
13214 (rs6000_get_function_versions_dispatcher): Likewise.
13215 (make_resolver_func): Likewise.
13216 (add_condition_to_bb): Likewise.
13217 (dispatch_function_versions): Likewise.
13218 (rs6000_generate_version_dispatcher_body): Likewise.
13219 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13220 (fusion_gpr_load_p): Fix a spacing issue.
13221 * doc/extend.texi (Common Function Attributes): Document that the
13222 PowerPC supports the target_clone attribute.
13224 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13226 * config/arm/arm.h: explain F symbol found in description of ARM
13227 register allocation in its legend.
13229 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13231 * config/mips/frame-header-opt.c: Include profile-count.h.
13232 * config/riscv/riscv.c: Include profile-count.h
13234 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13236 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13238 (sm_set_flag_if_changed): Add bbs field.
13239 (execute_sm_if_changed_flag_set): Pass BBS.
13240 (execute_sm): Update.
13242 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13244 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13247 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13249 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13250 (peephole2): New peephole2 to emit the above.
13251 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13253 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13255 * config/aarch64/aarch64.c (define_peephole2 above
13256 *sub_<shift>_<mode>): New peephole.
13258 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13260 * config/i386/i386.c (make_resolver_func): Update.
13261 * Makefile.in: Add profile-count.h and profile-count.o
13262 * auto-profile.c (afdo_indirect_call): Update to new API.
13263 (afdo_set_bb_count): Update.
13264 (afdo_propagate_edge): Update.
13265 (afdo_propagate_circuit): Update.
13266 (afdo_calculate_branch_prob): Update.
13267 (afdo_annotate_cfg): Update.
13268 * basic-block.h: Include profile-count.h
13269 (struct edge_def): Turn count to profile_count.
13270 (struct basic_block_def): Likewie.
13271 (REG_BR_PROB_BASE): Move to profile-count.h
13272 (RDIV): Move to profile-count.h
13273 * bb-reorder.c (max_entry_count): Turn to profile_count.
13274 (find_traces): Update.
13275 (rotate_loop):Update.
13276 (connect_traces):Update.
13277 (sanitize_hot_paths):Update.
13278 * bt-load.c (migrate_btr_defs): Update.
13279 * cfg.c (RDIV): Remove.
13280 (init_flow): Use alloc_block.
13281 (alloc_block): Uninitialize count.
13282 (unchecked_make_edge): Uninitialize count.
13283 (check_bb_profile): Update.
13284 (dump_edge_info): Update.
13285 (dump_bb_info): Update.
13286 (update_bb_profile_for_threading): Update.
13287 (scale_bbs_frequencies_int): Update.
13288 (scale_bbs_frequencies_gcov_type): Update.
13289 (scale_bbs_frequencies_profile_count): New.
13290 * cfg.h (update_bb_profile_for_threading): Update.
13291 (scale_bbs_frequencies_profile_count): Declare.
13292 * cfgbuild.c (compute_outgoing_frequencies): Update.
13293 (find_many_sub_basic_blocks): Update.
13294 * cfgcleanup.c (try_forward_edges): Update.
13295 (try_crossjump_to_edge): Update.
13296 * cfgexpand.c (expand_gimple_tailcall): Update.
13297 (construct_exit_block): Update.
13298 * cfghooks.c (verify_flow_info): Update.
13299 (dump_bb_for_graph): Update.
13300 (split_edge): Update.
13301 (make_forwarder_block): Update.
13302 (duplicate_block): Update.
13303 (account_profile_record): Update.
13304 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13305 (get_estimated_loop_iterations): Update.
13306 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13307 (single_likely_exit): Update.
13308 * cfgloopmanip.c (scale_loop_profile): Update.
13310 (set_zero_probability): Update.
13311 (lv_adjust_loop_entry_edge): Update.
13312 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13313 (purge_dead_edges): Update.
13314 (rtl_account_profile_record): Update.
13315 * cgraph.c (cgraph_node::create): Uninitialize count.
13316 (symbol_table::create_edge): Uninitialize count.
13317 (cgraph_update_edges_for_call_stmt_node): Update.
13318 (cgraph_edge::dump_edge_flags): Update.
13319 (cgraph_node::dump): Update.
13320 (cgraph_edge::maybe_hot_p): Update.
13321 * cgraph.h: Include profile-count.h
13322 (create_clone), create_edge, create_indirect_edge): Update.
13323 (cgraph_node): Turn count to profile_count.
13324 (cgraph_edge0: Likewise.
13325 (make_speculative, clone): Update.
13326 (create_edge): Update.
13327 (init_lowered_empty_function): Update.
13328 * cgraphclones.c (cgraph_edge::clone): Update.
13329 (duplicate_thunk_for_node): Update.
13330 (cgraph_node::create_clone): Update.
13331 * cgraphunit.c (cgraph_node::analyze): Update.
13332 (cgraph_node::expand_thunk): Update.
13333 * final.c (dump_basic_block_info): Update.
13334 * gimple-streamer-in.c (input_bb): Update.
13335 * gimple-streamer-out.c (output_bb): Update.
13336 * graphite.c (print_global_statistics): Update.
13337 (print_graphite_scop_statistics): Update.
13338 * hsa-brig.c: Include basic-block.h.
13339 * hsa-dump.c: Include basic-block.h.
13340 * hsa-gen.c (T sum_slice): Update.
13341 (convert_switch_statements):Update.
13342 * hsa-regalloc.c: Include basic-block.h.
13343 * ipa-chkp.c (chkp_produce_thunks): Update.
13344 * ipa-cp.c (struct caller_statistics): Update.
13345 (init_caller_stats): Update.
13346 (gather_caller_stats): Update.
13347 (ipcp_cloning_candidate_p): Update.
13348 (good_cloning_opportunity_p): Update.
13349 (get_info_about_necessary_edges): Update.
13350 (dump_profile_updates): Update.
13351 (update_profiling_info): Update.
13352 (update_specialized_profile): Update.
13353 (perhaps_add_new_callers): Update.
13354 (decide_about_value): Update.
13355 (ipa_cp_c_finalize): Update.
13356 * ipa-devirt.c (struct odr_type_warn_count): Update.
13357 (struct decl_warn_count): Update.
13358 (struct final_warning_record): Update.
13359 (possible_polymorphic_call_targets): Update.
13360 (ipa_devirt): Update.
13361 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13362 * ipa-icf.c (sem_function::merge): Update.
13363 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13364 * ipa-inline.c (compute_uninlined_call_time): Update.
13365 (compute_inlined_call_time): Update.
13366 (want_inline_small_function_p): Update.
13367 (want_inline_self_recursive_call_p): Update.
13368 (edge_badness): Update.
13369 (lookup_recursive_calls): Update.
13370 (recursive_inlining): Update.
13371 (inline_small_functions): Update.
13372 (dump_overall_stats): Update.
13373 (dump_inline_stats): Update.
13374 * ipa-profile.c (ipa_profile_generate_summary): Update.
13375 (ipa_propagate_frequency): Update.
13376 (ipa_profile): Update.
13377 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13378 * ipa-utils.c (ipa_merge_profiles): Update.
13379 * loop-doloop.c (doloop_modify): Update.
13380 * loop-unroll.c (report_unroll): Update.
13381 (unroll_loop_runtime_iterations): Update.
13382 * lto-cgraph.c (lto_output_edge): Update.
13383 (lto_output_node): Update.
13384 (input_node): Update.
13385 (input_edge): Update.
13386 (merge_profile_summaries): Update.
13387 * lto-streamer-in.c (input_cfg): Update.
13388 * lto-streamer-out.c (output_cfg): Update.
13389 * mcf.c (create_fixup_graph): Update.
13390 (adjust_cfg_counts): Update.
13391 (sum_edge_counts): Update.
13392 * modulo-sched.c (sms_schedule): Update.
13393 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13394 * predict.c (maybe_hot_count_p): Update.
13395 (probably_never_executed): Update.
13396 (dump_prediction): Update.
13397 (combine_predictions_for_bb): Update.
13398 (propagate_freq): Update.
13399 (handle_missing_profiles): Update.
13400 (counts_to_freqs): Update.
13401 (rebuild_frequencies): Update.
13402 (force_edge_cold): Update.
13403 * predict.h: Include profile-count.h
13404 (maybe_hot_count_p, counts_to_freqs): UPdate.
13405 * print-rtl-function.c: Do not include cfg.h
13406 * print-rtl.c: Include basic-block.h
13407 * profile-count.c: New file.
13408 * profile-count.h: New file.
13409 * profile.c (is_edge_inconsistent): Update.
13410 (correct_negative_edge_counts): Update.
13411 (is_inconsistent): Update.
13412 (set_bb_counts): Update.
13413 (read_profile_edge_counts): Update.
13414 (compute_frequency_overlap): Update.
13415 (compute_branch_probabilities): Update; Initialize and deinitialize
13417 (branch_prob): Update.
13418 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13419 (edge_gcov_count): New.
13420 (bb_gcov_count): New.
13421 * shrink-wrap.c (try_shrink_wrapping): Update.
13422 * tracer.c (better_p): Update.
13423 * trans-mem.c (expand_transaction): Update.
13424 (ipa_tm_insert_irr_call): Update.
13425 (ipa_tm_insert_gettmclone_call): Update.
13426 * tree-call-cdce.c: Update.
13427 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13428 (gimple_duplicate_sese_tail): Update.
13429 (gimple_account_profile_record): Update.
13430 (execute_fixup_cfg): Update.
13431 * tree-inline.c (copy_bb): Update.
13432 (copy_edges_for_bb): Update.
13433 (initialize_cfun): Update.
13434 (freqs_to_counts): Update.
13435 (copy_cfg_body): Update.
13436 (expand_call_inline): Update.
13437 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13438 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13439 (try_unroll_loop_completely): Update.
13440 (try_peel_loop): Update.
13441 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13442 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13443 * tree-ssa-loop-split.c (connect_loops): Update.
13444 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13445 * tree-ssa-reassoc.c (branch_fixup): Update.
13446 * tree-ssa-tail-merge.c (replace_block_by): Update.
13447 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13448 (compute_path_counts): Update.
13449 (update_profile): Update.
13450 (recompute_probabilities): Update.
13451 (update_joiner_offpath_counts): Update.
13452 (estimated_freqs_path): Update.
13453 (freqs_to_counts_path): Update.
13454 (clear_counts_path): Update.
13455 (ssa_fix_duplicate_block_edges): Update.
13456 (duplicate_thread_path): Update.
13457 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13458 (struct switch_conv_info): Update.
13459 * tree-tailcall.c (decrease_profile): Update.
13460 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13461 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13462 * value-prof.c (check_counter): Update.
13463 (gimple_divmod_fixed_value): Update.
13464 (gimple_mod_pow2): Update.
13465 (gimple_mod_subtract): Update.
13466 (gimple_ic_transform): Update.
13467 (gimple_stringop_fixed_value): Update.
13468 * value-prof.h (gimple_ic): Update.
13470 2017-06-02 Carl Love <cel@us.ibm.com>
13472 * config/rs6000/rs6000-c: Add support for built-in functions
13473 vector double vec_doublee (vector signed int);
13474 vector double vec_doublee (vector unsigned int);
13475 vector double vec_doublee (vector float);
13476 vector double vec_doubleh (vector signed int);
13477 vector double vec_doubleh (vector unsigned int);
13478 vector double vec_doubleh (vector float);
13479 vector double vec_doublel (vector signed int);
13480 vector double vec_doublel (vector unsigned int);
13481 vector double vec_doublel (vector float);
13482 vector double vec_doubleo (vector signed int);
13483 vector double vec_doubleo (vector unsigned int);
13484 vector double vec_doubleo (vector float);.
13485 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13486 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13488 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13489 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13490 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13492 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13493 vec_doublel, vec_doubleh.
13494 * doc/extend.texi: Update the built-in documentation file for the
13495 new built-in functions.
13497 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13500 * ipa-inline-analysis.c (free_growth_caches): Set
13501 edge_removal_hook_holder to NULL after removing it.
13503 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13505 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13506 comparision with zero.
13508 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13510 for early expansion of vec_min and vec_max builtins.
13511 (builtin_function_type): Add min/max unsigned variants to those
13512 identified as having unsigned arguments.
13514 2017-06-02 Olivier Hainque <hainque@adacore.com>
13516 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13518 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13520 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13521 Use VALL_F16 iterator rather than VALL.
13523 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13525 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13526 Emit CBNZ inside loop when doing a strong exchange and comparing
13527 against zero. Generate the CC flags after the loop.
13529 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13531 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13532 (dl_section_ref): New.
13533 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13534 On AIX, append an expression to subtract the size of the
13535 section length to dl_section_ref.
13537 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13539 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13540 for early expansion of vector absolute builtins.
13542 2017-06-02 Richard Biener <rguenther@suse.de>
13544 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13545 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13547 2017-06-02 Richard Biener <rguenther@suse.de>
13549 PR tree-optimization/80948
13550 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13551 stmt order as well.
13553 2017-06-02 Richard Biener <rguenther@suse.de>
13555 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13557 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13558 for inductions as relevant.
13560 2017-06-02 Richard Biener <rguenther@suse.de>
13562 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13563 (vectorizable_induction): ... this. Remove dead code.
13565 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13567 * builtins. (expand_builtin_alloca): Remove second parameter and
13568 infer its value from the first parameter instead.
13569 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13571 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13573 PR rtl-optimization/80903
13574 * loop-doloop.c (add_test): Unshare sequence.
13576 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13578 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13580 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13582 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13584 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13585 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13586 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13587 (xlogue_layout::get_stub_rtx): Make static.
13588 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13589 (xlogue_layout::compute_stub_managed_regs): Rename to...
13590 (xlogue_layout::count_stub_managed_regs): ...this.
13591 (xlogue_layout::is_stub_managed_reg): New function.
13592 (xlogue_layout::m_stub_names): Rename to...
13593 (xlogue_layout::s_stub_names): ...this, make static.
13594 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13595 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13596 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13597 xlogue_layout::s_stub_names): Instantiate statics.
13598 (stub_managed_regs): Remove.
13599 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13600 (disable_call_ms2sysv_xlogues): Rename to...
13601 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13602 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13604 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13605 change after reload_completed.
13606 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13608 (ix86_expand_prologue): Likewise.
13609 (ix86_expand_epilogue): Likewise.
13610 (ix86_expand_split_stack_prologue): Likewise.
13611 (ix86_compute_frame_layout): Remove frame parameter ...
13612 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13613 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13615 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13616 (ix86_frame): Move from here ...
13617 * config/i386/i386.h (ix86_frame): ... to here.
13618 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13619 complete ix86_frame data structure instead. Remove some_ld_name.
13621 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13623 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13624 symbols that hold a DECL_VALUE_EXPR.
13626 2017-06-01 Martin Jambor <mjambor@suse.cz>
13628 PR tree-optimization/80898
13629 * tree-sra.c (process_subtree_disqualification): Removed.
13630 (disqualify_candidate): Do not acll
13631 process_subtree_disqualification.
13632 (subtree_mark_written_and_enqueue): New function.
13633 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13634 RHS has been disqualified and re-queue LHS if necessary. Apart
13635 from that, ignore disqualified RHS.
13637 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13639 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13640 address fetch for z10 or later.
13642 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13644 * config/arc/arc.md (tst_movb): Add guard when splitting.
13646 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13648 * config/arc/arc.c (arc_can_eliminate): Test against
13649 arc_frame_pointer_needed.
13651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13653 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13654 to prevent store reordering.
13655 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13656 (type): Add block type.
13657 (stack_tie): Define special instruction to be used in
13660 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13662 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13663 constraint. It is not valid for the pattern.
13664 (noncommutative_binary_comparison): Likewise.
13666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13668 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13671 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13673 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13674 be used by the reg-alloc.
13676 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13678 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13679 reg-alloc when having mul64 or mul32x16 instructions.
13680 (mulsidi3): Likewise.
13681 (umulsidi3): Likewise.
13682 (mulsi32x16): New pattern.
13683 (mulsi64): Likewise.
13684 (mulsidi64): Likewise.
13685 (umulsidi64): Likewise.
13686 (MUL32x16_REG): Define.
13687 (mul64_600): Use MUL32x16_REG.
13688 (mac64_600): Likewise.
13689 (umul64_600): Likewise.
13690 (umac64_600): Likewise.
13692 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13694 * config/arc/arc.md (mulsi3_700): Make it commutative.
13696 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13698 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13700 (*sign_extendsidi2_insn): Likewise for movstosw.
13702 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13704 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13705 the type of the input discriminant value. Convert the
13706 discriminant value of signedness vary.
13708 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13710 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13711 Add to -Wall section.
13713 2017-06-01 Richard Biener <rguenther@suse.de>
13715 PR middle-end/66313
13716 * fold-const.c (fold_plusminus_mult_expr): If the factored
13717 factor may be zero use a wrapping type for the inner operation.
13718 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13719 and handle moved defs.
13720 (process_assignment): Properly guard the unary op case. Return a
13721 tri-state indicating that moving the stmt before the call may allow
13722 to continue. Pass through to_move.
13723 (find_tail_calls): Handle moving unrelated defs before
13726 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13729 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13730 splitter result in the canonical way.
13732 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13734 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13735 also for 32bit target. Update insn attributes.
13736 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13738 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13740 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13741 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13742 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13743 (_mm_maskz_min_ss): New intrinsics.
13745 2017-05-31 Martin Liska <mliska@suse.cz>
13747 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13748 Change comment style to one we normally use.
13749 (vectorizable_reduction): Likewise.
13750 (vectorizable_induction): Likewise.
13751 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13752 (vectorizable_call): Likewise.
13753 (vectorizable_simd_clone_call): Likewise.
13754 (vectorizable_conversion): Likewise.
13755 (vectorizable_assignment): Likewise.
13756 (vectorizable_shift): Likewise.
13757 (vectorizable_operation): Likewise.
13758 (vectorizable_store): Likewise.
13759 (vectorizable_load): Likewise.
13760 * tree-vectorizer.h: Likewise.
13762 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13764 * passes.c (emergency_dump_function): New.
13765 * tree-pass.h (emergency_dump_function): Declare.
13766 * plugin.c (plugins_internal_error_function): Remove.
13767 * plugin.h (plugins_internal_error_function): Remove declaration.
13768 * toplev.c (internal_error_function): New static function. Use it...
13769 (general_init): ...here.
13771 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13773 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13774 (arc_rtx_costs): Add costs for new patterns.
13775 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13776 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13778 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13780 * tree-ssa-strlen.c (get_next_strinfo): New function.
13781 (get_stridx_plus_constant): Use it.
13782 (zero_length_string): Likewise.
13783 (adjust_related_strinfos): Likewise.
13784 (adjust_last_stmt): Likewise.
13786 2017-05-31 Richard Biener <rguenther@suse.de>
13789 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13790 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13792 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13794 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13795 loop_vinfo argument and use of dependence distance vectors.
13796 Check instead whether the two references differ only in their
13797 initial value and assume that they have the same alignment if the
13798 difference is a multiple of the vector alignment.
13799 (vect_analyze_data_refs_alignment): Update call accordingly.
13801 2017-05-31 Martin Liska <mliska@suse.cz>
13804 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13806 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13808 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13809 (create_intersect_range_checks): Move from ...
13810 * tree-data-ref.c (create_intersect_range_checks_index)
13811 (create_intersect_range_checks): ... to here.
13812 (create_runtime_alias_checks): New function factored from ...
13813 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13814 here. Call above function.
13815 * tree-data-ref.h (create_runtime_alias_checks): New function.
13817 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13819 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13820 segment length for dr_b and compute it in wide_int.
13822 2017-05-31 Richard Biener <rguenther@suse.de>
13824 PR tree-optimization/80906
13825 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13826 and pass through iv_map.
13827 (copy_bb_and_scalar_dependences): Adjust.
13828 (translate_pending_phi_nodes): Likewise.
13829 (copy_loop_close_phi_args): Handle code-generating IVs instead
13832 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13834 * diagnostic-color.c (color_dict): Add "type-diff".
13835 (parse_gcc_colors): Update comment.
13836 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13837 -fdiagnostics-show-template-tree and -fno-elide-type.
13838 (GCC_COLORS): Add type-diff to example.
13840 (-fdiagnostics-show-template-tree): New.
13841 (-fno-elide-type): New.
13842 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13843 the pp_format_decoder callback. Call any m_format_postprocessor's
13845 (pretty_printer::pretty_printer): Initialize
13846 m_format_postprocessor.
13847 (pretty_printer::~pretty_printer): Delete any
13848 m_format_postprocessor.
13849 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13850 (class format_postprocessor): New class.
13851 (struct pretty_printer::format_decoder): Document the new parameters.
13852 (struct pretty_printer::m_format_postprocessor): New field.
13853 * tree-diagnostic.c (default_tree_printer): Update for new
13854 bool and const char ** params.
13855 * tree-diagnostic.h (default_tree_printer): Likewise.
13857 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13859 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13860 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13861 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13862 rs6000_gen_cell_microcode code.
13863 (rs6000_final_prescan_insn): Delete.
13864 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13865 "warn-cell-microcode" entries.
13866 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13867 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13868 throughout. Change cc_reg_not_micro_cr0_operand to
13869 cc_reg_not_cr0_operand throughout.
13870 (*extendhi<mode>2_noload): Delete.
13871 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13872 (mwarn-cell-microcode): Delete.
13873 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13874 -mgen-cell-microcode and -mwarn-cell-microcode.
13876 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13879 * config/i386/constraints.md (Yd): New constraint.
13881 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13882 and (?Yd, r) alternatives. Update insn attributes.
13883 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13884 and (?*Yd, r) alternatives. Update insn attributes.
13885 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13887 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13889 * gimplify.c (gimplify_modify_expr): Don't create a
13890 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13893 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13895 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13897 2017-05-30 Richard Biener <rguenther@suse.de>
13899 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13900 and reduc_def fields.
13901 (STMT_VINFO_REDUC_TYPE): New define.
13902 (STMT_VINFO_REDUC_DEF): Likewise.
13903 (vect_force_simple_reduction): Adjust prototype.
13904 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13905 (vect_is_simple_reduction): Remove check_reduction argument.
13906 (vect_force_simple_reduction): Adjust and set
13907 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13908 (vectorizable_reduction): Do not re-do reduction analysis
13909 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13910 * tree-parloops.c (gather_scalar_reductions): Adjust.
13912 2017-05-30 Richard Biener <rguenther@suse.de>
13914 PR middle-end/80901
13915 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13918 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13920 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13921 Introduce unknown_misalignment parameter and remove vf.
13922 (vect_peeling_hash_get_lowest_cost):
13923 Pass unknown_misalignment parameter.
13924 (vect_enhance_data_refs_alignment):
13925 Fix unsupportable data ref treatment.
13927 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13929 * tree-vect-data-refs.c (vect_get_data_access_cost):
13930 Workaround for SLP handling.
13931 (vect_enhance_data_refs_alignment):
13932 Compute costs for doing no peeling at all, compare to the best
13933 peeling costs so far and avoid peeling if cheaper.
13935 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13937 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13938 Return peeling info and set costs to zero for unlimited cost
13940 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13941 with unknown misalignment. Compute and costs for unknown
13942 misalignment, compare them to the costs for known misalignment
13943 and choose the cheapest for peeling.
13945 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13947 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13948 (vect_get_peeling_costs_all_drs): Create function.
13949 (vect_peeling_hash_get_lowest_cost):
13950 Use vect_get_peeling_costs_all_drs.
13951 (vect_peeling_supportable): Create function.
13952 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13954 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13956 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13957 DR_HAS_NEGATIVE_STEP.
13958 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13959 (vect_enhance_data_refs_alignment): Use.
13960 (vect_duplicate_ssa_name_ptr_info): Use.
13961 * tree-vectorizer.h (dr_misalignment): Use.
13962 (known_alignment_for_access_p): Use.
13964 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13967 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13969 (has_section_name): New function.
13971 2017-05-30 Martin Liska <mliska@suse.cz>
13974 * auto-profile.c (get_function_decl_from_block): Fix
13977 2017-05-30 Richard Biener <rguenther@suse.de>
13979 PR middle-end/80876
13980 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13982 2017-05-30 Martin Liska <mliska@suse.cz>
13984 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13985 * dumpfile.h (struct dump_file_info): Remove ctors.
13987 2017-05-30 Martin Liska <mliska@suse.cz>
13989 * predict.def: Fix GNU coding style.
13991 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13993 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13994 Mark 'to' argument with ATTRIBUTE_UNUSED.
13996 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13998 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13999 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14000 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14003 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14005 * doc/install.texi (Options specification): Restore entry of
14006 --enable-sjlj-exceptions.
14008 2017-05-27 Michael Eager <eager@eagercon.com>
14011 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14013 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14015 * config/microblaze/microblaze.h
14016 (FIXED_REGISTERS): Update in macro.
14017 (CALL_USED_REGISTERS): Update in macro.
14019 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14021 * doc/install.texi: Add links to macOS binary distributions.
14023 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14027 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14029 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14031 2017-05-26 Martin Liska <mliska@suse.cz>
14033 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14035 2017-05-26 Martin Liska <mliska@suse.cz>
14037 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14038 always leading ';; '.
14039 (dump_bb_info): Likewise.
14040 (brief_dump_cfg): Likewise.
14041 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14042 * dumpfile.c: Remove usage of TDF_VERBOSE.
14043 * dumpfile.h (enum dump_kind): Likewise.
14044 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14045 * print-tree.c (debug_verbose): Remove.
14046 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14047 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14048 * tree-diagnostic.c (default_tree_printer): Replace
14049 TDF_DIAGNOSTIC with TDF_SLIM.
14051 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14053 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14054 in parameter loop, rather than loop_vinfo.
14055 (create_intersect_range_checks): Ditto.
14056 (vect_create_cond_for_alias_checks): Update call to above functions.
14058 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14060 PR tree-optimization/80815
14061 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14062 for merging runtime alias checks. Handle negative DR_STEPs.
14064 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14066 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14068 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14069 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14070 out code pruning runtime alias checks.
14071 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14072 factored out from above.
14073 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14075 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14077 (prune_runtime_alias_test_list): New decalaration.
14079 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14081 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14082 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14083 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14084 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14085 (operator==, comp_dr_with_seg_len_pair): Ditto.
14086 (vect_prune_runtime_alias_test_list): Ditto.
14088 2017-05-26 Martin Liska <mliska@suse.cz>
14091 * params.def: Bound partial-inlining-entry-probability param.
14093 2017-05-26 Marek Polacek <polacek@redhat.com>
14096 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14099 2017-05-26 Richard Biener <rguenther@suse.de>
14101 PR tree-optimization/80842
14102 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14105 2017-05-26 Richard Biener <rguenther@suse.de>
14107 PR tree-optimization/80844
14108 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14110 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14112 * doc/md.texi (Machine Constraints): Update x86 family
14113 machine constraints section to match 'config/i386/constraints.md'.
14115 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14117 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14119 2017-05-25 Nathan Sidwell <nathan@acm.org>
14121 * doc/invoke.texi (--enable-languages): Update documentation.
14123 2017-05-25 Martin Liska <mliska@suse.cz>
14125 * dumpfile.c: Add TDF_FOLDING.
14126 * dumpfile.h (enum dump_kind): Likewise.
14127 * genmatch.c (dt_simplify::gen_1): Use it.
14129 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14131 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14133 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14135 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14136 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14138 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14140 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14141 * match.pd (X == C): Rewrite it here.
14142 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14143 with_certain_nonzero_bits2): New predicates.
14144 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14146 2017-05-24 Nathan Sidwell <nathan@acm.org>
14148 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14151 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14154 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14156 * config/powerpcspe: New port. Files are copied from the rs6000
14157 port, with "rs6000" in filenames replaced by "powerpcspe".
14159 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14161 PR rtl-optimization/80754
14162 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14164 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14166 * config/sparc/sparc.md (length): Return the correct value for -mflat
14167 sibcalls to match output_sibcall.
14169 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14173 * config/rs6000/rs6000.c (struct machine_function): Add new field
14175 (rs6000_get_separate_components): Init that field, use it.
14176 (rs6000_components_for_bb): Use the field.
14178 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14180 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14182 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14184 PR middle-end/80823
14185 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14187 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14190 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14191 address against address_operand predicate.
14192 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14194 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14196 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14197 parameters passed indirectly.
14199 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14201 * config/i386/i386.md (*movdi_internal): Remove SSE4
14202 alternative 18 (?r, *v). Update insn attributes.
14203 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14204 Update insn attributes.
14205 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14206 Update insn attributes.
14207 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14208 alternative 1 (r, v). Remove isa attribute.
14209 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14210 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14211 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14213 2017-05-23 Tom de Vries <tom@codesourcery.com>
14215 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14218 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14220 * cgraphunit.c (symbol_table::process_new_functions): Update.
14221 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14222 (inline_generate_summary): Rename to ...
14223 (ipa_fn_summary_generate): ... this one.
14224 (inline_read_summary): Rename to ...
14225 (ipa_fn_summary_read): ... this one.
14226 (inline_write_summary): Rename to ...
14227 (ipa_fn_summary_write): ... this one.
14228 (inline_free_summary): Rename to ...
14229 (ipa_free_fn_summary): ... this one.
14230 (pass_data_local_fn_summary, pass_local_fn_summary,
14231 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14232 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14233 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14234 make_pass_ipa_fn_summary): New.
14235 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14236 inline_write_summary, inline_free_summary): Remove.
14237 (ipa_free_fn_summary) : New.
14238 * ipa-inline.c (ipa_inline): Update.
14239 (pass_ipa_inline): Do not generate summaries.
14240 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14242 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14243 and add pass_ipa_fn_summary.
14244 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14246 (make_pass_inline_parameters): Remove.
14248 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14250 * omp-low.c (struct omp_context): Remove "default_kind" member.
14253 * omp-offload.c (execute_oacc_device_lower): Remove the
14254 parallelism dimensions function attributes for unparallelized
14255 OpenACC kernels constructs.
14257 2017-05-23 Martin Liska <mliska@suse.cz>
14259 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14261 (cgraph_edge::make_speculative): Likewise.
14262 (cgraph_edge::resolve_speculation): Likewise.
14263 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14264 (cgraph_node::dump): Likewise.
14265 * cgraph.h: Likewise.
14266 * cgraphunit.c (analyze_functions): Likewise.
14267 (symbol_table::compile): Likewise.
14268 * ipa-cp.c (print_all_lattices): Likewise.
14269 (determine_versionability): Likewise.
14270 (initialize_node_lattices): Likewise.
14271 (ipcp_verify_propagated_values): Likewise.
14272 (estimate_local_effects): Likewise.
14273 (update_profiling_info): Likewise.
14274 (create_specialized_node): Likewise.
14275 (perhaps_add_new_callers): Likewise.
14276 (decide_about_value): Likewise.
14277 (decide_whether_version_node): Likewise.
14278 (identify_dead_nodes): Likewise.
14279 (ipcp_store_bits_results): Likewise.
14280 * ipa-devirt.c (dump_targets): Likewise.
14281 (ipa_devirt): Likewise.
14282 * ipa-icf.c (sem_item::dump): Likewise.
14283 (sem_function::equals): Likewise.
14284 (sem_variable::equals): Likewise.
14285 (sem_item_optimizer::read_section): Likewise.
14286 (sem_item_optimizer::execute): Likewise.
14287 (congruence_class::dump): Likewise.
14288 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14289 (dump_inline_summary): Likewise.
14290 (estimate_node_size_and_time): Likewise.
14291 (inline_analyze_function): Likewise.
14292 * ipa-inline-transform.c (inline_call): Likewise.
14293 * ipa-inline.c (report_inline_failed_reason): Likewise.
14294 (want_early_inline_function_p): Likewise.
14295 (edge_badness): Likewise.
14296 (update_edge_key): Likewise.
14297 (inline_small_functions): Likewise.
14298 * ipa-profile.c (ipa_profile): Likewise.
14299 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14300 (ipa_make_edge_direct_to_target): Likewise.
14301 (remove_described_reference): Likewise.
14302 (ipa_impossible_devirt_target): Likewise.
14303 (propagate_controlled_uses): Likewise.
14304 (ipa_print_node_params): Likewise.
14305 (ipcp_transform_function): Likewise.
14306 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14307 (propagate_pure_const): Likewise.
14308 * ipa-reference.c (generate_summary): Likewise.
14309 (read_write_all_from_decl): Likewise.
14310 (propagate): Likewise.
14311 (ipa_reference_read_optimization_summary): Likewise.
14312 * ipa-utils.c (ipa_merge_profiles): Likewise.
14313 * ipa.c (walk_polymorphic_call_targets): Likewise.
14314 (symbol_table::remove_unreachable_nodes): Likewise.
14315 (ipa_single_use): Likewise.
14316 * passes.c (execute_todo): Likewise.
14317 * predict.c (drop_profile): Likewise.
14318 * symtab.c (symtab_node::get_dump_name): New function.
14319 (symtab_node::dump_name): Likewise.
14320 (symtab_node::dump_asm_name): Likewise.
14321 (symtab_node::dump_references): Likewise.
14322 (symtab_node::dump_referring): Likewise.
14323 (symtab_node::dump_base): Likewise.
14324 (symtab_node::debug_symtab): Likewise.
14325 * tree-sra.c (convert_callers_for_node): Likewise.
14326 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14327 * value-prof.c (init_node_map): Likewise.
14329 2017-05-23 Martin Liska <mliska@suse.cz>
14331 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14332 and symtab_node::debug_symtab to symbol_table::debug.
14333 * cgraphunit.c (analyze_functions): Use the renamed function.
14334 (symbol_table::compile): Likewise.
14335 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14336 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14337 * passes.c (execute_todo): Likewise.
14338 * symtab.c (symbol_table::dump): New function.
14339 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14341 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14343 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14344 that nonconst implies exec.
14346 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14348 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14349 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14350 (inline_edge_summary_vec): Turn into ...
14351 (ipa_call_summaries): ... this one.
14352 (redirect_to_unreachable, edge_set_predicate,
14353 evaluate_properties_for_edge, inline_summary_alloc,
14354 reset_ipa_call_summary, reset_inline_summary,
14355 inline_summary_t::duplicate): Update.
14356 (inline_edge_duplication_hook): Turn to ...
14357 (ipa_call_summary_t::duplicate): ... this one.
14358 (inline_edge_removal_hook): Turn to ...
14359 (ipa_call_summary_t::remove): ... this one.
14360 (dump_inline_edge_summary): Turn to ...
14361 (dump_ipa_call_summary): ... this one.
14362 (estimate_function_body_sizes): Update.
14363 (inline_update_callee_summaries): Update.
14364 (remap_edge_change_prob): Update.
14365 (remap_edge_summaries): Update.
14366 (inline_merge_summary): Update.
14367 (do_estimate_edge_time): Update.
14368 (inline_generate_summary): Update.
14369 (inline_read_section): Update.
14370 (inline_read_summary): Update.
14371 (inline_free_summary): Update.
14372 * ipa-inline.c (can_inline_edge_p): Update.
14373 (compute_inlined_call_time): Update.
14374 (want_inline_small_function_p): Update.
14375 (edge_badness): Update.
14376 (early_inliner): Update.
14377 * ipa-inline.h (inline_edge_summary): Turn to ...
14378 (ipa_call_summary): ... this one.
14379 (ipa_call_summary_t): New class.
14380 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14381 (ipa_call_summaries): New.
14382 (inline_edge_summary): Remove.
14383 (estimate_edge_growth): Update.
14384 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14385 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14386 * ipa-split.c (execute_split_functions): Update.
14387 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14389 2017-05-23 Tom de Vries <tom@codesourcery.com>
14391 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14392 attributes): Document rdrand effective target.
14394 2017-05-23 Tom de Vries <tom@codesourcery.com>
14396 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14397 attributes): Sort alphabetically.
14399 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14401 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14403 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14406 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14407 V2DF/V2DI splat into two separate patterns, one that handles
14408 registers, and the other that only handles memory. Drop support
14409 for splatting from a GPR on ISA 2.07 and then splitting the
14410 splat into direct move and splat.
14411 (vsx_splat_<mode>_reg): Likewise.
14412 (vsx_splat_<mode>_mem): Likewise.
14414 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14416 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14418 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14420 PR middle-end/80809
14421 * omp-low.c (finish_taskreg_remap): New function.
14422 (finish_taskreg_scan): If unit size of ctx->record_type
14423 is non-constant, unshare the size expression and replace
14424 decls in it with possible outer var refs.
14426 PR middle-end/80809
14427 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14428 GOVD_SHARED rather than GOVD_PRIVATE with it.
14429 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14430 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14432 PR middle-end/80853
14433 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14434 as last argument to build_outer_var_ref for pointer bases of array
14435 section reductions.
14437 2017-05-19 Martin Sebor <msebor@redhat.com>
14439 * print-tree.c (print_node): Print DECL_READ_P flag.
14441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14443 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14444 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14445 * cgraph.c: Likewise.
14446 * cgraphunit.c: Likewise.
14447 * gengtype.c: Likewise.
14448 * ipa-cp.c: Likewise.
14449 * ipa-devirt.c: Likewise.
14450 * ipa-icf.c: Likewise.
14451 * ipa-predicate.c: Likewise.
14452 * ipa-profile.c: Likewise.
14453 * ipa-prop.c: Likewise.
14454 * ipa-split.c: Likewise.
14456 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14457 edge_predicate_pool, dump_inline_hints,
14458 inline_summary::account_size_time, redirect_to_unreachable,
14459 edge_set_predicate, set_hint_predicate,
14460 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14461 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14462 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14463 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14464 ipa_call_summary_t::remove, initialize_growth_caches,
14465 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14466 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14467 mark_modified, unmodified_parm_1, unmodified_parm,
14468 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14469 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14470 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14471 will_be_nonconstant_predicate, record_modified_bb_info,
14472 get_minimal_bb, record_modified, param_change_prob,
14473 phi_result_unknown_predicate, predicate_for_phi_result,
14474 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14475 estimate_function_body_sizes, compute_inline_parameters,
14476 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14477 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14478 inline_update_callee_summaries, remap_edge_change_prob,
14479 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14480 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14481 inline_analyze_function, inline_summary_t::insert,
14482 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14483 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14484 inline_free_summary): Move to ipa-fnsummary.h
14485 (predicate_t): Remove.
14486 * ipa-fnsummary.c: New file.
14487 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14488 (enum inline_hints_vals, inline_hints, agg_position_info,
14489 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14490 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14491 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14492 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14493 inline_read_summary, inline_write_summary, inline_free_summary,
14494 inline_analyze_function, initialize_inline_failed,
14495 inline_merge_summary, inline_update_overall_summary,
14496 compute_inline_parameters): Move to ipa-fnsummary.h
14497 * ipa-fnsummary.h: New file.
14498 * ipa-inline-transform.h: Include ipa-inline.h.
14499 * ipa-inline.c: LIkewise.
14501 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14503 * ipa-inline.c (edge_badness): Use inlined_time instead of
14504 inline_summaries->get.
14506 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14508 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14510 2017-05-22 Nathan Sidwell <nathan@acm.org>
14512 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14513 (fdump-lang): Document 'raw' option.
14514 * dumpfile.h (TDI_tu): Delete.
14515 * dumpfile.c (dump_files): Remove translation-unit.
14516 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14518 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14520 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14521 command option from $(AWK) call.
14522 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14523 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14524 [FORMAT]: Remove handling of variable.
14525 * config/avr/t-multilib: Regenerate.
14527 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14529 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14531 (dump_inline_summary): Do not print self_time.
14532 (estimate_function_body_sizes): Do not set self_time.
14533 (compute_inline_parameters): Likewise.
14534 (inline_read_section, inline_write_summary): Do not stream self_time.
14535 * ipa-inline.h (inline_summary): Drop self_time.
14537 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14539 * ipa-inline-analysis.c (account_size_time): Rename to ...
14540 (inline_summary::account_size_time): ... this one.
14541 (reset_ipa_call_summary): Turn to ...
14542 (ipa_call_summary::reset): ... this one.
14543 (reset_inline_summary): Turn to ...
14544 (inline_summary::reset): ... this one.
14545 (inline_summary_t::remove): Update.
14546 (inline_summary_t::duplicate): Update.
14547 (ipa_call_summary_t::remove): Update.
14548 (dump_inline_summary): Update.
14549 (estimate_function_body_sizes): Update.
14550 (compute_inline_parameters): Update.
14551 (estimate_node_size_and_time): Update.
14552 (inline_merge_summary): Update.
14553 (inline_update_overall_summary): Update.
14554 (inline_read_section): Update.
14555 (inline_write_summary): Update.
14556 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14557 add account_size_time and reset member functions.
14558 (ipa_call_summary): Add reset function.
14559 * ipa-predicate.h (predicate::operator &): Constify.
14561 2017-05-22 Richard Biener <rguenther@suse.de>
14563 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14565 2017-05-19 Jason Merrill <jason@redhat.com>
14567 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14569 2017-05-19 Marek Polacek <polacek@redhat.com>
14572 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14573 TYPE_OVERFLOW_WRAPS checks.
14575 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14577 * tree-core.h (enum omp_clause_default_kind): Add
14578 "OMP_CLAUSE_DEFAULT_PRESENT".
14579 * tree-pretty-print.c (dump_omp_clause): Handle it.
14580 * gimplify.c (enum gimplify_omp_var_data): Add
14581 "GOVD_MAP_FORCE_PRESENT".
14582 (gimplify_adjust_omp_clauses_1): Map it to
14583 "GOMP_MAP_FORCE_PRESENT".
14584 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14586 * gimplify.c (oacc_default_clause): Clarify.
14588 2017-05-19 Nathan Sidwell <nathan@acm.org>
14590 LANG_HOOK_REGISTER_DUMPS
14591 * toplev.c (general_init): Call register dump lang hook.
14592 * doc/invoke.texi: Document -fdump-lang option family.
14593 * dumpfile.c (dump_files): Remove class dump here.
14594 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14595 * dumpfile.h (tree_dump_index): Remove TDI_class.
14596 * langhooks-def.h (lhd_register_dumps): Declare.
14597 (LANG_HOOKS_REGISTER_DUMPS): Define.
14598 (LANG_HOOKS_INITIALIZER): Add it.
14599 * langhooks.c (lhd_register_dumps): Define.
14600 * langhooks.h (struct lang_hooks): Add register_dumps.
14602 2017-05-19 Nathan Sidwell <nathan@acm.org>
14604 * context.h (context::set_passes): New.
14605 * context.c (context::context): Do not create pass manager.
14606 * toplev.c (general_init): Create pass manager here.
14608 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14610 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14611 use this splitter if two add or or instructions would also work for
14612 the constant we want to generate.
14614 2017-05-19 Richard Biener <rguenther@suse.de>
14617 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14618 predicate evaluation.
14620 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14622 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14624 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14627 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14629 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14630 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14631 (inline_edge_summary_vec): Turn into ...
14632 (ipa_call_summaries): ... this one.
14633 (redirect_to_unreachable, edge_set_predicate,
14634 evaluate_properties_for_edge, inline_summary_alloc,
14635 reset_ipa_call_summary, reset_inline_summary,
14636 inline_summary_t::duplicate): Update.
14637 (inline_edge_duplication_hook): Turn to ...
14638 (ipa_call_summary_t::duplicate): ... this one.
14639 (inline_edge_removal_hook): Turn to ...
14640 (ipa_call_summary_t::remove): ... this one.
14641 (dump_inline_edge_summary): Turn to ...
14642 (dump_ipa_call_summary): ... this one.
14643 (estimate_function_body_sizes): Update.
14644 (inline_update_callee_summaries): Update.
14645 (remap_edge_change_prob): Update.
14646 (remap_edge_summaries): Update.
14647 (inline_merge_summary): Update.
14648 (do_estimate_edge_time): Update.
14649 (inline_generate_summary): Update.
14650 (inline_read_section): Update.
14651 (inline_read_summary): Update.
14652 (inline_free_summary): Update.
14653 * ipa-inline.c (can_inline_edge_p): Update.
14654 (compute_inlined_call_time): Update.
14655 (want_inline_small_function_p): Update.
14656 (edge_badness): Update.
14657 (early_inliner): Update.
14658 * ipa-inline.h (inline_edge_summary): Turn to ...
14659 (ipa_call_summary): ... this one.
14660 (ipa_call_summary_t): New class.
14661 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14662 (ipa_call_summaries): New.
14663 (inline_edge_summary): Remove.
14664 (estimate_edge_growth): Update.
14665 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14666 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14667 * ipa-split.c (execute_split_functions): Update.
14668 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14670 2017-05-19 Richard Biener <rguenther@suse.de>
14672 PR middle-end/80764
14673 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14675 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14677 * config/rs6000/rs6000.c (struct machine_function): Add field
14678 fpr_is_wrapped_separately.
14679 (rs6000_get_separate_components): Use 64 components. Handle the
14680 new FPR components.
14681 (rs6000_components_for_bb): Handle the FPR components.
14682 (rs6000_emit_prologue_components): Handle the FPR components.
14683 (rs6000_emit_epilogue_components): Handle the FPR components.
14684 (rs6000_set_handled_components): Handle the FPR components.
14685 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14686 that are already separately shrink-wrapped.
14687 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14688 that are already separately shrink-wrapped.
14690 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14693 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14696 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14697 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14698 cases where the register allocator uses a move and an offsettable
14699 memory operation to/from a FPR register on ISA 2.06/2.07.
14700 (define_peephole2 for Altivec d-form store): Likewise.
14702 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14705 * config/i386/mmx.md (*mov<mode>_internal): Enable
14706 alternatives 11, 12, 13 and 14 also for 32bit targets.
14707 Remove alternatives 15, 16, 17 and 18.
14708 * config/i386/sse.md (vec_concatv2di): Change
14709 alternative (!x, *y) to (x, ?!*Yn).
14711 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14713 * dumpfile.h (enum dump_kind): Remove stray comma.
14715 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14717 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14718 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14719 predicate::num_conditions
14720 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14721 (CHANGED): turn into predicate::changed.
14722 (agg_position_info): Move to ipa-predicate.h
14723 (add_condition, predicate::add_clause, predicate::operator &=,
14724 predicate::or_with, predicate::evaluate, predicate::probability,
14725 dump_condition, dump_clause, predicate::dump,
14726 predicate::remap_after_duplication, predicate::remap_after_inlining,
14727 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14728 (evaluate_conditions_for_known_args): Update.
14729 (set_cond_stmt_execution_predicate): Update.
14730 * ipa-inline.h: Include ipa-predicate.h
14731 (condition, inline_param_summary, conditions, agg_position_info,
14732 predicate): Move to ipa-predicate.h
14733 * ipa-predicate.c: New file.
14734 * ipa-predicate.h: New file.
14736 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14738 * final.c (leaf_function_p): Check we are not in a sequence.
14740 2017-05-18 Martin Liska <mliska@suse.cz>
14742 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14743 * dumpfile.c (dump_register): Use new enum dump_kind.
14744 (get_dump_file_name): Likewise.
14745 (dump_enable_all): Likewise.
14746 (dump_switch_p_1): Likewise.
14747 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14748 * dumpfile.h (enum dump_kind): New enum type.
14749 (struct dump_file_info): Create constructor and
14750 format fields and comments.
14751 * passes.c (pass_manager::register_one_dump_file):
14753 * statistics.c (statistics_early_init): Likewise.
14754 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14755 TDF_TREE with TDF_SLIM.
14756 (gather_memory_references_ref): Likewise.
14758 2017-05-18 Martin Liska <mliska@suse.cz>
14760 * vec.h (struct vnull): Use it.
14762 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14764 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14765 (true_predicate, false_predicate, true_predicate_p,
14766 false_predicate_p): Remove.
14767 (single_cond_predicate, not_inlined_predicate): Turn to member function
14769 (add_condition): Update.
14770 (add_clause): Turn to...
14771 (predicate::add_clause): ... this one; update; allow passing NULL
14773 (and_predicates): Turn to ...
14774 (predicate::operator &=): ... this one.
14775 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14776 (or_predicates): Turn to ...
14777 (predicate::or_with): ... this one.
14778 (evaluate_predicate): Turn to ...
14779 (predicate::evaluate): ... this one.
14780 (predicate_probability): Turn to ...
14781 (predicate::probability): ... this one.
14782 (dump_condition): Update.
14783 (dump_predicate): Turn to ...
14784 (predicate::dump): ... this one.
14785 (account_size_time): Update.
14786 (edge_set_predicate): Update.
14787 (set_hint_predicate): UPdate.
14788 (evaluate_conditions_for_known_args): Update.
14789 (evaluate_properties_for_edge): Update.
14790 (remap_predicate_after_duplication): Turn to...
14791 (predicate::remap_after_duplication): ... this one.
14792 (remap_hint_predicate_after_duplication): Update.
14793 (inline_summary_t::duplicate): UPdate.
14794 (dump_inline_edge_summary): Update.
14795 (dump_inline_summary): Update.
14796 (set_cond_stmt_execution_predicate): Update.
14797 (set_switch_stmt_execution_predicate): Update.
14798 (compute_bb_predicates): Update.
14799 (will_be_nonconstant_expr_predicate): Update.
14800 (will_be_nonconstant_predicate): Update.
14801 (phi_result_unknown_predicate): Update.
14802 (predicate_for_phi_result): Update.
14803 (array_index_predicate): Update.
14804 (estimate_function_body_sizes): Update.
14805 (estimate_node_size_and_time): Update.
14806 (estimate_ipcp_clone_size_and_time): Update.
14807 (remap_predicate): Rename to ...
14808 (predicate::remap_after_inlining): ... this one.
14809 (remap_hint_predicate): Update.
14810 (inline_merge_summary): Update.
14811 (inline_update_overall_summary): Update.
14812 (estimate_size_after_inlining): Update.
14813 (read_predicate): Rename to ...
14814 (predicate::stream_in): ... this one.
14815 (read_inline_edge_summary): Update.
14816 (write_predicate): Rename to ...
14817 (predicate::stream_out): ... this one.
14818 (write_inline_edge_summary): Update.
14819 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14820 (clause_t): Turn to uint32_t
14821 (predicate): Turn to class; implement constructor and operators
14823 (size_time_entry): Update.
14824 (inline_summary): Update.
14825 (inline_edge_summary): Update.
14827 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14829 * fold-const.c (fold_binary_loc): Move transformation...
14830 * match.pd (C - X CMP X): ... here.
14832 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14834 * config/sparc/sparc.c (sparc_option_override): Set function
14835 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14836 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14838 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14840 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14842 2017-05-18 Marek Polacek <polacek@redhat.com>
14845 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14846 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14847 gimple_assign_load_p.
14849 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14851 PR middle-end/80692
14852 * real.c (do_compare): Give decimal_do_compare preference over
14853 comparing just the signs.
14855 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14857 * doc/md.texi (Canonicalization of Instructions): Describe the
14858 canonical form of instructions that inherently set a condition
14861 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14863 PR middle-end/80775
14864 * tree-cfg.c: Move deletion of unreachable case statements to after
14865 the merging of consecutive case labels.
14867 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14869 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14870 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14871 restoring of callee-saved registers.
14873 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14875 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14876 * config/visium/visium.c (single_set_and_flags): Likewise.
14877 * config/visium/visium.md (Substitutions): Likewise.
14879 2017-05-17 Martin Liska <mliska@suse.cz>
14881 * cfg.c: Introduce dump_flags_t type and
14882 use it instead of int type.
14884 * cfghooks.c: Likewise.
14885 * cfghooks.h (struct cfg_hooks): Likewise.
14886 * cfgrtl.c: Likewise.
14887 * cfgrtl.h: Likewise.
14888 * cgraph.c (cgraph_node::get_body): Likewise.
14889 * coretypes.h: Likewise.
14890 * domwalk.c: Likewise.
14891 * domwalk.h: Likewise.
14892 * dumpfile.c (struct dump_option_value_info): Likewise.
14893 (dump_enable_all): Likewise.
14894 (dump_switch_p_1): Likewise.
14895 (opt_info_switch_p): Likewise.
14896 * dumpfile.h (enum tree_dump_index): Likewise.
14897 (struct dump_file_info): Likewise.
14898 * genemit.c: Likewise.
14899 * generic-match-head.c: Likewise.
14900 * gengtype.c (open_base_files): Likewise.
14901 * gimple-pretty-print.c: Likewise.
14902 * gimple-pretty-print.h: Likewise.
14903 * graph.c (print_graph_cfg): Likewise.
14904 * graphite-scop-detection.c (dot_all_sese): Likewise.
14905 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14906 * loop-unroll.c (report_unroll): Likewise.
14907 * passes.c (pass_manager::register_one_dump_file): Likewise.
14908 * print-tree.c: Likewise.
14909 * statistics.c: Likewise.
14910 * tree-cfg.c: Likewise.
14911 * tree-cfg.h: Likewise.
14912 * tree-dfa.c: Likewise.
14913 * tree-dfa.h: Likewise.
14914 * tree-dump.c (dump_function): Likewise.
14915 * tree-dump.h (struct dump_info): Likewise.
14916 * tree-pretty-print.c: Likewise.
14917 * tree-pretty-print.h: Likewise.
14918 * tree-ssa-live.c: Likewise.
14919 * tree-ssa-live.h: Likewise.
14920 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14921 * tree-vect-loop.c: Likewise.
14922 * tree-vect-slp.c: Likewise.
14924 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14925 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14927 PR tree-optimization/80457
14928 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14929 of all arguments to a statement as scalar_to_vec operations.
14930 (vectorizable_call): Adjust call to vect_model_simple_cost for
14932 (vectorizable_conversion): Likewise.
14933 (vectorizable_assignment): Likewise.
14934 (vectorizable_shift): Likewise.
14935 (vectorizable_operation): Likewise.
14936 (vectorizable_comparison): Likewise.
14937 (vect_is_simple_cond): Record the def types for operands.
14938 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14939 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14940 for statement argument count.
14942 2017-05-16 Carl Love <cel@us.ibm.com>
14944 * config/rs6000/rs6000-c: Add support for built-in functions
14945 vector unsigned long long vec_bperm (vector unsigned long long,
14946 vector unsigned char)
14947 vector signed long long vec_mule (vector signed int,
14949 vector unsigned long long vec_mule (vector unsigned int,
14950 vector unsigned int)
14951 vector signed long long vec_mulo (vector signed int,
14953 vector unsigned long long vec_mulo (vector unsigned int,
14954 vector unsigned int)
14955 vector signed char vec_sldw (vector signed char,
14956 vector signed char,
14958 vector unsigned char vec_sldw (vector unsigned char,
14959 vector unsigned char,
14961 vector signed short vec_sldw (vector signed short,
14962 vector signed short,
14964 vector unsigned short vec_sldw (vector unsigned short,
14965 vector unsigned short,
14967 vector signed int vec_sldw (vector signed int,
14970 vector unsigned int vec_sldw (vector unsigned int,
14971 vector unsigned int,
14973 vector signed long long vec_sldw (vector signed long long,
14974 vector signed long long,
14976 vector unsigned long long vec_sldw (vector unsigned long long,
14977 vector unsigned long long,
14979 * config/rs6000/rs6000-c: Add support for built-in functions
14980 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14981 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14982 * doc/extend.texi: Update the built-in documentation for the
14983 new built-in functions.
14985 2017-05-16 Marek Polacek <polacek@redhat.com>
14989 * tree.c (save_expr): Don't fold the expression.
14991 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14993 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14994 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14995 and (?*y,m). Update insn attributes.
14997 2017-05-16 Martin Liska <mliska@suse.cz>
14999 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15000 flags argument of print_gimple_stmt, print_gimple_expr,
15001 print_generic_stmt and print_generic_expr.
15002 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15003 * coretypes.h: Likewise.
15004 * except.c (dump_eh_tree): Likewise.
15005 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15006 * gimple-pretty-print.h: Likewise.
15007 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15008 (backprop::push_to_worklist): Likewise.
15009 (backprop::pop_from_worklist): Likewise.
15010 (backprop::process_use): Likewise.
15011 (backprop::intersect_uses): Likewise.
15012 (note_replacement): Likewise.
15013 * gimple-ssa-store-merging.c
15014 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15015 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15016 (pass_store_merging::execute): Likewise.
15017 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15018 (ssa_base_cand_dump_callback): Likewise.
15019 (dump_incr_vec): Likewise.
15020 (replace_refs): Likewise.
15021 (replace_mult_candidate): Likewise.
15022 (create_add_on_incoming_edge): Likewise.
15023 (create_phi_basis): Likewise.
15024 (insert_initializers): Likewise.
15025 (all_phi_incrs_profitable): Likewise.
15026 (introduce_cast_before_cand): Likewise.
15027 (replace_one_candidate): Likewise.
15028 * gimplify.c (gimplify_expr): Likewise.
15029 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15030 (set_rename): Likewise.
15031 (rename_uses): Likewise.
15032 (copy_loop_phi_nodes): Likewise.
15033 (add_close_phis_to_merge_points): Likewise.
15034 (copy_loop_close_phi_args): Likewise.
15035 (copy_cond_phi_args): Likewise.
15036 (graphite_copy_stmts_from_block): Likewise.
15037 (translate_pending_phi_nodes): Likewise.
15038 * graphite-poly.c (print_pdr): Likewise.
15039 (dump_gbb_cases): Likewise.
15040 (dump_gbb_conditions): Likewise.
15041 (print_scop_params): Likewise.
15042 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15043 (build_cross_bb_scalars_use): Likewise.
15044 (gather_bbs::before_dom_children): Likewise.
15045 * hsa-dump.c (dump_hsa_immed): Likewise.
15046 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15047 (get_replacement_map): Likewise.
15048 * ipa-inline-analysis.c (dump_condition): Likewise.
15049 (estimate_function_body_sizes): Likewise.
15050 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15051 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15052 * ipa-prop.c (ipa_dump_param): Likewise.
15053 (ipa_print_node_jump_functions_for_edge): Likewise.
15054 (ipa_modify_call_arguments): Likewise.
15055 (ipa_modify_expr): Likewise.
15056 (ipa_dump_param_adjustments): Likewise.
15057 (ipa_dump_agg_replacement_values): Likewise.
15058 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15059 * ipa-pure-const.c (check_stmt): Likewise.
15060 (pass_nothrow::execute): Likewise.
15061 * ipa-split.c (execute_split_functions): Likewise.
15062 * omp-offload.c (dump_oacc_loop_part): Likewise.
15063 (dump_oacc_loop): Likewise.
15064 * trans-mem.c (tm_log_emit): Likewise.
15065 (tm_memopt_accumulate_memops): Likewise.
15066 (dump_tm_memopt_set): Likewise.
15067 (dump_tm_memopt_transform): Likewise.
15068 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15069 (print_loop): Likewise.
15070 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15071 (chkp_gather_checks_info): Likewise.
15072 (chkp_get_check_result): Likewise.
15073 (chkp_remove_check_if_pass): Likewise.
15074 (chkp_use_outer_bounds_if_possible): Likewise.
15075 (chkp_reduce_bounds_lifetime): Likewise.
15076 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15077 (chkp_mark_completed_bounds): Likewise.
15078 (chkp_register_incomplete_bounds): Likewise.
15079 (chkp_mark_invalid_bounds): Likewise.
15080 (chkp_maybe_copy_and_register_bounds): Likewise.
15081 (chkp_build_returned_bound): Likewise.
15082 (chkp_get_bound_for_parm): Likewise.
15083 (chkp_build_bndldx): Likewise.
15084 (chkp_get_bounds_by_definition): Likewise.
15085 (chkp_generate_extern_var_bounds): Likewise.
15086 (chkp_get_bounds_for_decl_addr): Likewise.
15087 * tree-chrec.c (chrec_apply): Likewise.
15088 * tree-data-ref.c (dump_data_reference): Likewise.
15089 (dump_subscript): Likewise.
15090 (dump_data_dependence_relation): Likewise.
15091 (analyze_overlapping_iterations): Likewise.
15092 * tree-inline.c (expand_call_inline): Likewise.
15093 (tree_function_versioning): Likewise.
15094 * tree-into-ssa.c (dump_defs_stack): Likewise.
15095 (dump_currdefs): Likewise.
15096 (dump_names_replaced_by): Likewise.
15097 (dump_update_ssa): Likewise.
15098 (update_ssa): Likewise.
15099 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15100 * tree-parloops.c (build_new_reduction): Likewise.
15101 (try_create_reduction_list): Likewise.
15102 (ref_conflicts_with_region): Likewise.
15103 (oacc_entry_exit_ok_1): Likewise.
15104 (oacc_entry_exit_single_gang): Likewise.
15105 * tree-pretty-print.h: Likewise.
15106 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15107 (get_scalar_evolution): Likewise.
15108 (add_to_evolution): Likewise.
15109 (get_loop_exit_condition): Likewise.
15110 (analyze_evolution_in_loop): Likewise.
15111 (analyze_initial_condition): Likewise.
15112 (analyze_scalar_evolution): Likewise.
15113 (instantiate_scev): Likewise.
15114 (number_of_latch_executions): Likewise.
15115 (gather_chrec_stats): Likewise.
15116 (final_value_replacement_loop): Likewise.
15117 (scev_const_prop): Likewise.
15118 * tree-sra.c (dump_access): Likewise.
15119 (disqualify_candidate): Likewise.
15120 (create_access): Likewise.
15121 (reject): Likewise.
15122 (maybe_add_sra_candidate): Likewise.
15123 (create_access_replacement): Likewise.
15124 (analyze_access_subtree): Likewise.
15125 (analyze_all_variable_accesses): Likewise.
15126 (sra_modify_assign): Likewise.
15127 (initialize_constant_pool_replacements): Likewise.
15128 (find_param_candidates): Likewise.
15129 (decide_one_param_reduction): Likewise.
15130 (replace_removed_params_ssa_names): Likewise.
15131 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15132 * tree-ssa-copy.c (dump_copy_of): Likewise.
15133 (copy_prop_visit_cond_stmt): Likewise.
15134 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15135 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15136 (record_equivalences_from_stmt): Likewise.
15137 * tree-ssa-dse.c (compute_trims): Likewise.
15138 (delete_dead_call): Likewise.
15139 (delete_dead_assignment): Likewise.
15140 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15141 (forward_propagate_into_cond): Likewise.
15142 (pass_forwprop::execute): Likewise.
15143 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15144 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15146 (move_computations_worker): Likewise.
15147 (execute_sm): Likewise.
15148 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15149 (remove_exits_and_undefined_stmts): Likewise.
15150 (remove_redundant_iv_tests): Likewise.
15151 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15152 (adjust_iv_update_pos): Likewise.
15153 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15154 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15155 (value_replacement): Likewise.
15156 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15157 * tree-ssa-pre.c (print_pre_expr): Likewise.
15158 (get_representative_for): Likewise.
15159 (create_expression_by_pieces): Likewise.
15160 (insert_into_preds_of_block): Likewise.
15161 (eliminate_insert): Likewise.
15162 (eliminate_dom_walker::before_dom_children): Likewise.
15163 (eliminate): Likewise.
15164 (remove_dead_inserted_code): Likewise.
15165 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15166 * tree-ssa-reassoc.c (get_rank): Likewise.
15167 (eliminate_duplicate_pair): Likewise.
15168 (eliminate_plus_minus_pair): Likewise.
15169 (eliminate_not_pairs): Likewise.
15170 (undistribute_ops_list): Likewise.
15171 (eliminate_redundant_comparison): Likewise.
15172 (update_range_test): Likewise.
15173 (optimize_range_tests_var_bound): Likewise.
15174 (optimize_vec_cond_expr): Likewise.
15175 (rewrite_expr_tree): Likewise.
15176 (rewrite_expr_tree_parallel): Likewise.
15177 (linearize_expr): Likewise.
15178 (break_up_subtract): Likewise.
15179 (linearize_expr_tree): Likewise.
15180 (attempt_builtin_powi): Likewise.
15181 (attempt_builtin_copysign): Likewise.
15182 (transform_stmt_to_copy): Likewise.
15183 (transform_stmt_to_multiply): Likewise.
15184 (dump_ops_vector): Likewise.
15185 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15186 (print_scc): Likewise.
15187 (set_ssa_val_to): Likewise.
15188 (visit_reference_op_store): Likewise.
15189 (visit_use): Likewise.
15190 (sccvn_dom_walker::before_dom_children): Likewise.
15191 (run_scc_vn): Likewise.
15192 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15194 (expr_hash_elt::print): Likewise.
15195 (const_and_copies::pop_to_marker): Likewise.
15196 (const_and_copies::record_const_or_copy_raw): Likewise.
15197 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15198 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15199 (dump_predicates): Likewise.
15200 (find_uninit_use): Likewise.
15201 (warn_uninitialized_phi): Likewise.
15202 (pass_late_warn_uninitialized::execute): Likewise.
15203 * tree-ssa.c (verify_vssa): Likewise.
15204 (verify_ssa): Likewise.
15205 (maybe_optimize_var): Likewise.
15206 * tree-vrp.c (dump_value_range): Likewise.
15207 (dump_all_value_ranges): Likewise.
15208 (dump_asserts_for): Likewise.
15209 (register_edge_assert_for_2): Likewise.
15210 (vrp_visit_cond_stmt): Likewise.
15211 (vrp_visit_switch_stmt): Likewise.
15212 (vrp_visit_stmt): Likewise.
15213 (vrp_visit_phi_node): Likewise.
15214 (simplify_cond_using_ranges_1): Likewise.
15215 (fold_predicate_in): Likewise.
15216 (evrp_dom_walker::before_dom_children): Likewise.
15217 (evrp_dom_walker::push_value_range): Likewise.
15218 (evrp_dom_walker::pop_value_range): Likewise.
15219 (execute_early_vrp): Likewise.
15221 2017-05-16 Richard Biener <rguenther@suse.de>
15223 * dwarf2out.c (loc_list_from_tree_1): Do not create
15224 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15226 2017-05-16 Richard Biener <rguenther@suse.de>
15228 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15230 (note_variable_value_in_expr): If we resolved the decl ref
15231 do not push to the stack.
15233 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15235 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15236 operations in fast-math mode.
15237 (vaddq_f16): Likewise.
15238 (vmul_f16): Likewise.
15239 (vmulq_f16): Likewise.
15240 (vsub_f16): Likewise.
15241 (vsubq_f16): Likewise.
15242 * config/arm/neon.md (add<mode>3): New.
15244 (fma:<VH:mode>3): New. Also remove outdated comment.
15247 2017-05-16 Martin Liska <mliska@suse.cz>
15251 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15252 (odr_types_equivalent_p): Likewise.
15254 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15256 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15258 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15261 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15262 non-interunit SSE move alternatives with '?'.
15263 (zero-extendsidi peephole2): New peephole to skip intermediate
15264 general register in SSE zero-extend sequence.
15266 2017-05-15 Jeff Law <law@redhat.com>
15268 * reorg.c (relax_delay_slots): Create a new variable to hold
15269 the temporary target rather than clobbering TARGET_LABEL.
15271 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15272 missing argument to extract_bit_field call.
15273 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15275 2017-05-15 Martin Liska <mliska@suse.cz>
15278 * gcc.c (process_command): Do not allow empty argument of -o option.
15280 2017-05-15 Renlin Li <renlin.li@arm.com>
15282 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15283 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15284 * config/aarch64/constraints.md (Usf): Add long call check.
15285 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15286 (call_value): Likewise.
15287 (sibcall): Likewise.
15288 (sibcall_value): Likewise.
15290 (call_value_insn): New.
15291 (sibcall_insn): Update rtx pattern.
15292 (sibcall_value_insn): Likewise.
15293 (call_internal): Remove.
15294 (call_value_internal): Likewise.
15295 (sibcall_internal): Likewise.
15296 (sibcall_value_internal): Likewise.
15297 (call_reg): Likewise.
15298 (call_symbol): Likewise.
15299 (call_value_reg): Likewise.
15300 (call_value_symbol): Likewise.
15302 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15305 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15307 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15309 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15310 compatible with CCGOCmode and with CCZmode.
15312 2017-05-14 Martin Sebor <msebor@redhat.com>
15314 PR middle-end/77671
15315 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15316 (gimple_fold_builtin_snprintf): Same.
15317 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15318 (gimple_fold_builtin_snprintf): Same.
15319 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15320 of character types.
15321 (is_call_safe): New function.
15322 (try_substitute_return_value): Call it.
15323 (try_simplify_call): New function.
15324 (pass_sprintf_length::handle_gimple_call): Call it.
15326 2017-05-14 Martin Sebor <msebor@redhat.com>
15328 PR middle-end/80669
15329 * builtins.c (expand_builtin_stpncpy): Simplify.
15331 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15333 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15334 * config/i386/i386.h
15335 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15336 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15337 (struct machine_function): Add new members call_ms2sysv,
15338 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15339 (struct machine_frame_state): New fields sp_realigned and
15340 sp_realigned_offset.
15341 * config/i386/i386.c
15342 (enum xlogue_stub): New enum.
15343 (enum xlogue_stub_sets): New enum.
15344 (class xlogue_layout): New class.
15345 (struct ix86_frame): New fields stack_realign_allocate_offset,
15346 stack_realign_offset and outlined_save_offset. Modify comments to
15347 detail stack layout when using out-of-line stubs.
15348 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15349 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15350 -mcall-ms2sysv-xlogues.
15351 (stub_managed_regs): New static variable.
15352 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15353 registers managed by out-of-line stub.
15354 (disable_call_ms2sysv_xlogues): New function.
15355 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15356 m->call_ms2sysv when appropriate and compute frame layout for
15358 (sp_valid_at, fp_valid_at): New inline functions.
15359 (choose_basereg): New function.
15360 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15362 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15363 Use align parameter of choose_baseaddr to generated aligned SSE movs
15365 (pro_epilogue_adjust_stack): Modify to track
15366 machine_frame_state::sp_realigned.
15367 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15368 (ix86_nsaved_sseregs): Likewise.
15369 (ix86_emit_save_regs): Likewise.
15370 (ix86_emit_save_regs_using_mov): Likewise.
15371 (ix86_emit_save_sse_regs_using_mov): Likewise.
15372 (get_scratch_register_on_entry): Likewise.
15373 (gen_frame_set): New function.
15374 (gen_frame_load): Likewise.
15375 (gen_frame_store): Likewise.
15376 (emit_outlined_ms2sysv_save): Likewise.
15377 (emit_outlined_ms2sysv_restore): Likewise.
15378 (ix86_expand_prologue): Modify stack re-alignment code and call
15379 emit_outlined_ms2sysv_save when appropriate.
15380 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15381 parameter rtx_insn *insn, which allows the function to be used to only
15382 generate the notes.
15383 (ix86_expand_epilogue): Modify validity checks of frame and stack
15384 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15385 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15386 * config/i386/predicates.md
15387 (save_multiple): New predicate.
15388 (restore_multiple): Likewise.
15389 * config/i386/sse.md
15390 (save_multiple<mode>): New pattern.
15391 (save_multiple_realign<mode>): Likewise.
15392 (restore_multiple<mode>): Likewise.
15393 (restore_multiple_and_return<mode>): Likewise.
15394 (restore_multiple_leave_return<mode>): Likewise.
15395 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15397 2017-05-14 Julia Koval <julia.koval@intel.com>
15399 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15400 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15401 (__builtin_ia32_xsetbv): New builtins.
15402 * config/i386/i386.c (ix86_expand_special_args_builtin):
15404 (ix86_expand_builtin): Special expand for new intrinsics.
15405 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15406 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15407 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15409 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15411 * cfganal.c (inverted_post_order_compute): Change argument type
15413 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15414 * df-core.c (rest_of_handle_df_initialize): Adjust.
15415 (rest_of_handle_df_finish): Likewise.
15416 (df_analyze_1): Likewise.
15417 (df_analyze): Likewise.
15418 (loop_inverted_post_order_compute): Change argument to be a vec *.
15419 (df_analyze_loop): Adjust.
15420 (df_get_n_blocks): Likewise.
15421 (df_get_postorder): Likewise.
15422 * df.h (struct df_d): Change field to be a vec.
15423 * lcm.c (compute_laterin): Adjust.
15424 (compute_available): Likewise.
15425 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15426 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15427 * tree-ssa-pre.c (compute_antic): Likewise.
15429 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15431 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15432 (depth_first_search::depth_first_search): Change structure init
15433 function to this constructor.
15434 (depth_first_search::add_bb): Rename function to this member.
15435 (depth_first_search::execute): Likewise.
15436 (flow_dfs_compute_reverse_finish): Adjust.
15438 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15440 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15441 (longest_simple_path): Likewise.
15442 * shrink-wrap.c (spread_components): Likewise.
15443 (disqualify_problematic_components): Likewise.
15444 (emit_common_heads_for_components): Likewise.
15445 (emit_common_tails_for_components): Likewise.
15446 (insert_prologue_epilogue_for_components): Likewise.
15448 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15450 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15453 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15455 * df-core.c (df_set_blocks): Start using auto_bitmap.
15456 (df_compact_blocks): Likewise.
15457 * df-problems.c (df_rd_confluence_n): Likewise.
15458 * df-scan.c (df_insn_rescan_all): Likewise.
15459 (df_process_deferred_rescans): Likewise.
15460 (df_update_entry_block_defs): Likewise.
15461 (df_update_exit_block_uses): Likewise.
15462 (df_entry_block_bitmap_verify): Likewise.
15463 (df_exit_block_bitmap_verify): Likewise.
15464 (df_scan_verify): Likewise.
15465 * lra-constraints.c (lra_constraints): Likewise.
15466 (undo_optional_reloads): Likewise.
15467 (lra_undo_inheritance): Likewise.
15468 * lra-remat.c (calculate_gen_cands): Likewise.
15469 (do_remat): Likewise.
15470 * lra-spills.c (assign_spill_hard_regs): Likewise.
15471 (spill_pseudos): Likewise.
15472 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15473 (bitmap_set_subtract_values): Likewise.
15475 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15477 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15478 management with auto_bitmap.
15479 (fix_inter_tick): Likewise.
15480 (fix_recovery_deps): Likewise.
15481 * ira.c (add_store_equivs): Likewise.
15482 (find_moveable_pseudos): Likewise.
15483 (split_live_ranges_for_shrink_wrap): Likewise.
15484 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15485 (rtx_reuse_manager::seen_def_p): Likewise.
15486 (rtx_reuse_manager::set_seen_def): Likewise.
15487 * print-rtl.h (class rtx_reuse_manager): Likewise.
15489 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15491 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15493 (migrate_btr_def): Likewise.
15494 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15495 * df-core.c (loop_post_order_compute): Likewise.
15496 (loop_inverted_post_order_compute): Likewise.
15497 * hsa-common.h: Likewise.
15498 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15499 * init-regs.c (initialize_uninitialized_regs): Likewise.
15500 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15501 (inline_small_functions): Likewise.
15502 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15503 * ira.c (combine_and_move_insns): Likewise.
15504 (build_insn_chain): Likewise.
15505 * loop-invariant.c (find_invariants): Likewise.
15506 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15507 * predict.c (tree_predict_by_opcode): Likewise.
15508 (predict_paths_leading_to): Likewise.
15509 (predict_paths_leading_to_edge): Likewise.
15510 (estimate_loops_at_level): Likewise.
15511 (estimate_loops): Likewise.
15512 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15513 (spread_components): Likewise.
15514 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15515 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15516 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15517 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15518 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15519 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15520 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15521 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15522 (mark_threaded_blocks): Likewise.
15523 (thread_through_all_blocks): Likewise.
15524 * tree-ssa.c (verify_ssa): Likewise.
15525 (execute_update_addresses_taken): Likewise.
15526 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15528 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15530 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15532 (post_order_compute): Likewise.
15533 (inverted_post_order_compute): Likewise.
15534 (pre_and_rev_post_order_compute_fn): Likewise.
15536 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15538 * genrecog.c (int_set::int_set): Explicitly construct our
15539 auto_vec base class.
15540 * vec.h (auto_vec::auto_vec): New constructor.
15542 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15544 * bitmap.h (class auto_bitmap): New constructor taking
15545 bitmap_obstack * argument.
15547 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15549 * bitmap.h (class auto_bitmap): Change type of m_bits to
15550 bitmap_head, and adjust ctor / dtor and member operators.
15552 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15554 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15555 when returned register mode doesn't match original mode.
15557 2017-05-12 Jeff Law <law@redhat.com>
15558 Jakub Jelinek <jakub@redhat.com>
15560 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15561 we look for cc setter after the compare-elim changes.
15562 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15563 within the vector to match what compare-elim now expects.
15564 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15565 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15567 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15568 after the compare-elim changes.
15569 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15570 the vector to match what compare-elim now expects.
15571 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15572 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15573 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15574 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15575 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15577 * config/visium/visium.c (single_set_and_flags): Fix where
15578 we look for cc setter after the compare-elim changes.
15579 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15580 with the vector to match what compare-elim now expects.
15581 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15582 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15583 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15584 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15585 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15586 (neg<mode>2_insn_set_overflow): Likewise.
15588 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15590 PR middle-end/79794
15591 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15592 maybe_expand_insn call, set ops[0].target. If still set after call,
15593 set alt_rtl. Add extra arg to recursive calls.
15594 (extract_bit_field): Add alt_rtl argument. Pass to
15596 * expmed.h (extract_bit_field): Fix prototype.
15597 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15598 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15599 to extract_bit_field_calls.
15600 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15601 Pass alt_rtl to extract_bit_field calls.
15602 * calls.c (store_unaligned_arguments_into_psuedos)
15603 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15604 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15606 * optabs.h (struct expand_operand): Add target bitfield.
15608 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15610 * compare-elim.c (try_eliminate_compare): Canonicalize
15611 operation with embedded compare to
15612 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15613 (set (reg) (operation)].
15615 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15617 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15620 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15621 cost of adding a carry flag for ADC instruction.
15622 [case MINUS]: Ignore the cost of subtracting a carry flag
15623 for SBB instruction.
15625 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15627 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15629 * config/rs6000/bmiintrin.h: New file.
15630 * config/rs6000/bmi2intrin.h: New file.
15631 * config/rs6000/x86intrin.h: New file.
15633 2017-05-12 Jeff Law <law@redhat.com>
15635 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15638 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15640 PR middle-end/80707
15641 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15643 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15645 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15646 early expansion of vector divide builtins.
15647 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15648 builtins identified as having unsigned arguments.
15650 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15652 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15653 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15654 expansion of vector logical operations (and, andc, or, xor,
15657 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15659 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15660 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15662 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15664 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15665 early GIMPLE expansion of vector multiplies.
15667 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15669 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15670 TARGET_HAVE_MOVT conditional.
15671 (movt splitter): Likewise.
15673 2017-05-12 Richard Biener <rguenther@suse.de>
15675 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15676 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15677 Fold all stmts not inplace.
15679 2017-05-12 Richard Biener <rguenther@suse.de>
15681 PR tree-optimization/80713
15682 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15683 inserted_exprs bit for not removed stmts.
15685 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15687 PR middle-end/69921
15688 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15689 parallelized" attribute for parallelized OpenACC kernels.
15690 * omp-offload.c (execute_oacc_device_lower): Use it.
15692 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15693 Set "oacc kernels" attribute.
15694 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15695 parameter. Adjust all users.
15696 (oacc_fn_attrib_kernels_p): Remove function.
15697 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15698 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15699 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15700 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15701 assert "oacc kernels" attribute is set.
15703 2017-05-11 Carl Love <cel@us.ibm.com>
15705 * config/rs6000/rs6000-c: Add support for built-in functions
15706 vector unsigned char vec_popcnt (vector signed char)
15707 vector unsigned char vec_popcnt (vector unsigned char)
15708 vector unsigned short vec_popcnt (vector signed short)
15709 vector unsigned short vec_popcnt (vector unsigned short)
15710 vector unsigned int vec_popcnt (vector signed int)
15711 vector unsigned int vec_popcnt (vector unsigned int)
15712 vector unsigned long long vec_popcnt (vector signed long long)
15713 vector unsigned long long vec_popcnt (vector unsigned long long)
15714 vector signed long long vec_slo (vector signed long long,
15715 vector signed char)
15716 vector signed long long vec_slo (vector signed long long,
15717 vector unsigned char)
15718 vector unsigned long long vec_slo (vector unsigned long long,
15719 vector signed char)
15720 vector unsigned long long vec_slo (vector unsigned long long,
15721 vector unsigned char)
15722 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15723 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15724 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15725 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15726 * doc/extend.texi: Update the built-in documentation file for the
15727 new built-in functions.
15729 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15731 * attribs.h (sorted_attr_string): Move machine independent
15732 functions for target clone support from the i386 port to common
15733 code. Rename ix86_function_versions to common_function_versions.
15734 Rename make_name to make_unique_name.
15735 (common_function_versions): Likewise.
15736 (make_unique_name): Likewise.
15737 (make_dispatcher_decl): Likewise.
15738 (is_function_default_version): Likewise.
15739 * attribs.c (attr_strcmp): Likewise.
15740 (sorted_attr_string): Likewise.
15741 (common_function_versions): Likewise.
15742 (make_unique_name): Likewise.
15743 (make_dispatcher_decl): Likewise.
15744 (is_function_default_version): Likewise.
15745 * config/i386/i386.c (attr_strcmp): Likewise.
15746 (sorted_attr_string): Likewise.
15747 (ix86_function_versions): Likewise.
15748 (make_name): Likewise.
15749 (make_dispatcher_decl): Likewise.
15750 (is_function_default_version): Likewise.
15751 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15753 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15756 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15757 Account for direct move costs for vec_construct of integer
15760 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15763 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15764 (UNSPEC_STX_ATOMIC): Ditto.
15765 (loaddi_via_sse): New insn.
15766 (storedi_via_sse): Ditto.
15767 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15768 Update corresponding peephole2 patterns.
15769 (atomic_storedi_fpu): Ditto.
15771 2017-05-11 Julia Koval <julia.koval@intel.com>
15773 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15774 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15776 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15777 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15778 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15780 2017-05-11 Nathan Sidwell <nathan@acm.org>
15782 * graphite-poly.c: Include dumpfile.h.
15784 * dumpfle.h (dump_function): Declare here ...
15785 * tree-dump.h (dump_function): ... not here.
15786 * dumpfile.c: #include tree-cfg.h.
15787 (dump_function): Move here from ...
15788 * tree-dump.c (dump_function): ... here.
15789 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15790 * graphite-poly.c: Don't include tree-dump.h.
15791 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15792 * print-tree.c: Likewise.
15793 * stor-layout.c: Likewise.
15794 * tree-nested.c: Likewise.
15796 * dumpfile.c (dump_start): Use TDF_FLAGS.
15797 (dump_enable_all): Fix TDF_KIND check thinko.
15799 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15801 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15802 array entries to represent two legal parameterizations of the
15803 overloaded __builtin_cmpb function, as represented by the
15804 P6_OV_BUILTIN_CMPB constant.
15805 (altivec_resolve_overloaded_builtin): Add special case handling
15806 for the __builtin_cmpb function, as represented by the
15807 P6_OV_BUILTIN_CMPB constant.
15808 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15809 (BU_P6_64BIT_2): New macro.
15810 (BU_P6_OVERLOAD_2): New macro
15811 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15812 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15813 (CMPB): Add overload support to represent both 32-bit and 64-bit
15814 compare-bytes function.
15815 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15816 support for TARGET_CMPB.
15817 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15818 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15819 documentation of the __builtin_cmpb overloaded built-in function.
15821 2017-05-11 Richard Biener <rguenther@suse.de>
15823 PR tree-optimization/80705
15824 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15825 bases are not vectorizable.
15827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15829 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15830 when counting register pressure.
15832 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15834 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15835 (iv_ca_more_deps): Renamed to ...
15836 (iv_ca_compare_deps): ... this.
15837 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15841 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15843 (determine_group_iv_costs): ... here.
15844 (find_inv_vars_cb): Record inv var if it's not recorded before.
15846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15848 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15849 (get_shiftadd_cost): Ditto.
15851 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15853 * tree-ssa-address.c: Include header file.
15854 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15856 (add_to_parts): Refactor.
15857 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15858 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15863 PR tree-optimization/53090
15864 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15866 (extract_cond_operands): Detect condition with IV on both sides
15867 and return COMP_IV_EXPR_2.
15868 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15869 (rewrite_use_compare): Simplify by removing call to function
15870 extract_cond_operands.
15872 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15874 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15875 (extract_cond_operands): Detect condition comparing against non-
15876 invariant bound and return appropriate enum value.
15877 (find_interesting_uses_cond): Update use of extract_cond_operands.
15878 Handle its return value accordingly.
15879 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15883 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15884 nonlinear iv_use computation in loop invariant sensitive way.
15886 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15888 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15889 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15891 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15893 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15894 (dump_cand): Support iv_cand.inv_exprs.
15895 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15897 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15900 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15902 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15904 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15905 as local function. Include necessary header files.
15906 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15910 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15912 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15914 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15915 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15916 RSHIFT_EXPR and BIT_NOT_EXPR.
15918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15920 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15921 (adjust_setup_cost): New parameter supporting round up adjustment.
15922 (struct address_cost_data): Delete.
15923 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15924 (split_address_cost, ptr_difference_cost): Delete.
15925 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15926 (struct ainc_cost_data): New struct.
15927 (get_address_cost_ainc): New function.
15928 (get_address_cost, get_computation_cost): Reimplement.
15929 (determine_group_iv_cost_address): Record inv_expr for all uses of
15931 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15932 (iv_ca_has_deps): Reimplemented to ...
15933 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15935 (iv_ca_extend): Call iv_ca_more_deps.
15937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15939 * tree-ssa-address.c (struct mem_address): Move to header file.
15940 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15941 * tree-ssa-address.h (struct mem_address): Move from C file.
15942 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15944 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15946 * tree-affine.h (aff_combination_type): New interface.
15947 (aff_combination_zero_p): Remove static.
15948 (aff_combination_const_p): New interface.
15949 (aff_combination_singleton_var_p): New interfaces.
15951 2017-05-11 Richard Biener <rguenther@suse.de>
15953 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15954 Skip unreachable blocks and destinations.
15955 (eliminate): Move stmt removal and fixup ...
15956 (fini_eliminate): ... here. Skip inserted exprs.
15957 (pass_pre::execute): Move fini_pre after fini_eliminate.
15958 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15959 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15960 PRE to get rid of dead code that has invalid SSA form and
15961 split critical edges again.
15963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15965 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15967 2017-05-11 Richard Biener <rguenther@suse.de>
15969 * passes.c (execute_function_todo): Verify loops if they are
15970 said to be up-to-date.
15971 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15972 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15974 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15977 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15978 handle calling assemble_external ourself.
15981 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15982 modes with zero size. Enhance comment.
15984 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15986 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15987 built-ins for vec_xl and vec_xst with short and char pointer
15990 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15992 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15993 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15994 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15995 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15996 (_mm_maskz_min_round_ss): New intrinsics.
15997 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15998 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15999 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16000 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16001 (__builtin_ia32_minss_mask_round): New builtins.
16002 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16003 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16004 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16006 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16007 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16009 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16012 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16014 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16015 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16016 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16017 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16018 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16019 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16020 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16021 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16022 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16023 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16024 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16025 (__builtin_ia32_mulss_mask_round): New builtins.
16026 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16027 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16028 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16030 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16031 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16033 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16036 2017-05-10 Julia Koval <julia.koval@intel.com>
16038 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16039 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16040 (_mm256_setr_m128i): New intrinsics.
16042 2017-05-10 Julia Koval <julia.koval@intel.com>
16044 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16045 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16046 (_mm_maskz_rcp14_ss): New intrinsics.
16047 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16048 (__builtin_ia32_rcp14ss_mask): New builtins.
16049 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16051 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16053 PR tree-optimization/51513
16054 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16055 (assert_unreachable_fallthru_edge_p): Use it.
16056 (group_case_labels_stmt): Likewise.
16057 * tree-cfg.h: Prototype it.
16058 * stmt.c: Include cfghooks.h and tree-cfg.h.
16059 (emit_case_dispatch_table) <gap_label>: New local variable.
16060 Use it to fill dispatch table gaps.
16061 Test for default_label before updating probabilities.
16062 (expand_case) <default_label>: Remove unneeded initialization.
16063 Test for unreachable default case statement and remove its edge.
16064 Set default_label accordingly.
16065 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16067 2017-05-10 Carl Love <cel@us.ibm.com>
16069 * config/rs6000/rs6000-c: Add support for built-in functions
16070 vector signed char vec_neg (vector signed char)
16071 vector signed short int vec_neg (vector short int)
16072 vector signed int vec_neg (vector signed int)
16073 vector signed long long vec_neg (vector signed long long)
16074 vector float vec_neg (vector float)
16075 vector double vec_neg (vector double)
16076 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16078 * config/rs6000/altivec.h: Add define for vec_neg
16079 * doc/extend.texi: Update the built-in documentation for the
16080 new built-in functions.
16082 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16084 PR tree-optimization/77644
16085 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16087 2017-05-10 Nathan Sidwell <nathan@acm.org>
16089 * dumpfile.h (TDI_lang_all): New.
16090 (TDF_KIND): New. Renumber others
16091 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16093 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16095 (get_dump_file_name): Adjust suffix generation.
16096 (dump_enable_all): Use TDF_KIND.
16097 * doc/invoke.texi (-fdump-lang-all): Document.
16099 * dumpfile.h: Tabify.
16101 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16104 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16105 Move member access before delete.
16107 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16109 * tree-inline.c (expand_call_inline): Split block at stmt
16112 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16115 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16116 are now unused after splitting mov{sf,sd}_hardfloat.
16117 (f32_lr2): Likewise.
16118 (f32_lm): Likewise.
16119 (f32_lm2): Likewise.
16120 (f32_li): Likewise.
16121 (f32_li2): Likewise.
16122 (f32_lv): Likewise.
16123 (f32_sr): Likewise.
16124 (f32_sr2): Likewise.
16125 (f32_sm): Likewise.
16126 (f32_sm2): Likewise.
16127 (f32_si): Likewise.
16128 (f32_si2): Likewise.
16129 (f32_sv): Likewise.
16130 (f32_dm): Likewise.
16131 (f32_vsx): Likewise.
16132 (f32_av): Likewise.
16133 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16134 For movsf, order stores so the VSX stores occur before the GPR
16135 store which encourages the register allocator to use a traditional
16136 FPR instead of a GPR. For movsd, order the stores so that the GPR
16137 store comes before the VSX stores to allow the power6 to work.
16138 This is due to the power6 not having a 32-bit integer store
16139 instruction from a FPR.
16140 (movsf_hardfloat): Likewise.
16141 (movsd_hardfloat): Likewise.
16143 2017-05-09 Martin Sebor <msebor@redhat.com>
16145 PR translation/80280
16146 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16149 PR translation/80280
16150 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16151 data member added in r247778.
16152 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16154 2017-05-09 Nathan Sidwell <nathan@acm.org>
16156 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16158 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16161 2017-05-09 Marek Polacek <polacek@redhat.com>
16163 * doc/invoke.texi: Fix typo.
16165 2017-05-09 Richard Biener <rguenther@suse.de>
16167 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16168 (vrp_val_is_min): Likewise.
16169 (set_value_range_to_value): Likewise.
16170 (set_value_range_to_nonnegative): Likewise.
16171 (gimple_assign_nonzero_p): Likewise.
16172 (gimple_stmt_nonzero_p): Likewise.
16173 (vrp_int_const_binop): Likewise. Remove unreachable case.
16174 (adjust_range_with_scev): Adjust comments.
16175 (compare_range_with_value): Likewise.
16176 (extract_range_from_phi_node): Likewise.
16177 (test_for_singularity): Likewise.
16179 2017-05-09 Richard Biener <rguenther@suse.de>
16181 * tree-vrp.c (get_single_symbol): Add assert that we don't
16182 get overflowed constants as invariant part.
16183 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16184 checks. Use wi::cmp instead of recursing for integer constants.
16185 (compare_values): Just ignore whether we assumed undefined
16186 overflow instead of failing the compare.
16187 (extract_range_for_var_from_comparison_expr): Add comment before the
16188 TREE_NO_WARNING sets.
16189 (test_for_singularity): Likewise.
16190 (extract_range_from_comparison): Do not disable optimization
16191 when we assumed undefined overflow.
16192 (extract_range_basic): Remove init of unused var.
16194 2017-05-09 Richard Biener <rguenther@suse.de>
16196 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16197 (extract_range_from_multiplicative_op_1): Adjust.
16198 (extract_range_from_binary_expr_1): Use int_const_binop.
16200 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16203 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16204 rs6000_store_data_bypass_p in seven define_bypass directives and
16205 in several comments.
16206 * config/rs6000/rs6000-protos.h: Add prototype for
16207 rs6000_store_data_bypass_p function.
16208 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16209 function implements slightly different (rs6000-specific) semantics
16210 than store_data_bypass_p, returning false rather than aborting
16211 with assertion error when arguments do not satisfy the
16212 requirements of store data bypass.
16213 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16214 rs6000_store_data_bypass_p.
16216 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16218 * config/xtensa/xtensa-protos.h
16219 (xtensa_initial_elimination_offset): New declaration.
16220 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16221 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16222 macro definition, add case for FRAME_POINTER_REGNUM when
16223 FRAME_GROWS_DOWNWARD.
16224 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16225 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16226 xtensa_initial_elimination_offset.
16228 2017-05-08 Nathan Sidwell <nathan@acm.org>
16230 * doc/invoke.texi: Alphabetize -fdump options.
16232 2017-05-08 Martin Sebor <msebor@redhat.com>
16234 PR translation/80280
16235 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16237 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16239 * target.def (compute_frame_layout): New optional target hook.
16240 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16241 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16242 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16244 * reload1.c (verify_initial_elim_offsets): Likewise.
16245 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16246 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16247 (arm_get_frame_offsets): Split up into this ...
16248 (arm_compute_frame_layout): ... and this function.
16250 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16252 * config/aarch64/constraints.md (Usa): New constraint.
16253 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16255 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16257 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16258 with_multilib_list after it has been checked.
16260 2017-05-08 Richard Biener <rguenther@suse.de>
16262 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16263 (bitmap_set_subtract_values): Likewise.
16265 2017-05-08 Richard Biener <rguenther@suse.de>
16267 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16268 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16270 (gimple_stmt_nonzero_warnv_p): Rename to ...
16271 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16273 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16274 (extract_range_basic): Adjust, do not disable propagation on
16275 strict overflow sensitive simplification.
16276 (vrp_visit_cond_stmt): Likewise.
16278 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16280 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16281 body size unconditionally.
16283 2017-05-07 Jeff Law <law@redhat.com>
16286 2017-05-06 Jeff Law <law@redhat.com>
16287 PR tree-optimization/78496
16288 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16291 PR tree-optimization/78496
16292 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16293 (simplify_stmt_using_ranges): Call it.
16294 (vrp_dom_walker::before_dom_children): Extract equivalences
16295 from an ASSERT_EXPR with an equality comparison against a
16298 2017-05-06 Jeff Law <law@redhat.com>
16300 PR tree-optimization/78496
16301 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16304 PR tree-optimization/78496
16305 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16306 (simplify_stmt_using_ranges): Call it.
16307 (vrp_dom_walker::before_dom_children): Extract equivalences
16308 from an ASSERT_EXPR with an equality comparison against a
16311 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16313 * lra-constraints.c (lra_copy_reg_equiv): New function.
16314 (split_reg): Use it to copy equivalence information from the
16315 original register to the spill register.
16317 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16319 PR rtl-optimization/75964
16320 * simplify-rtx.c (simplify_const_relational_operation): Remove
16321 invalid handling of comparisons of integer ABS.
16323 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16325 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16326 initialize to zero.
16327 (init_regs): Remove declaration.
16328 (function_arg_advance_32): Initialize error_p as boolean variable.
16330 2017-05-05 Nathan Sidwell <nathan@acm.org>
16332 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16333 lines. Use for (;;).
16335 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16337 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16338 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16339 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16340 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16341 VF=2 that require versioning.
16343 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16345 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16348 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16350 * diagnostic.h (diagnostic_override_option_index): Convert from
16351 macro to inline function.
16353 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16355 * diagnostic.c (last_module_changed_p): New function.
16356 (set_last_module): New function.
16357 (diagnostic_report_current_module): Convert macro usage to
16358 the above functions.
16359 * diagnostic.h (diagnostic_context::last_module): Strengthen
16360 from const line_map * to const line_map_ordinary *.
16361 (diagnostic_last_module_changed): Delete macro.
16362 (diagnostic_set_last_module): Delete macro.
16364 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16366 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16367 with diagnostic_report_diagnostic.
16368 (diagnostic_n_impl_richloc): Likewise.
16369 * diagnostic.h (report_diagnostic): Delete macro.
16370 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16371 with diagnostic_report_diagnostic.
16372 * substring-locations.c (format_warning_va): Likewise.
16374 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16376 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16377 save/restor of format_spec. Move option-printing code to...
16378 (print_option_information): ...this new function, and
16379 reimplement by simply printing to the pretty_printer,
16380 rather than appending to the format string.
16382 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16384 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16385 handling logic into...
16386 (update_effective_level_from_pragmas): ...this new function.
16388 2017-05-04 Andrew Waterman <andrew@sifive.com>
16390 * config/riscv/riscv.opt (mstrict-align): New option.
16391 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16392 (SLOW_UNALIGNED_ACCESS): Define.
16393 (riscv_slow_unaligned_access): Declare.
16394 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16396 (riscv_slow_unaligned_access): New variable.
16397 (rocket_tune_info): Set slow_unaligned_access to true.
16398 (optimize_size_tune_info): Set slow_unaligned_access to false.
16399 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16400 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16401 (riscv_option_override): Set riscv_slow_unaligned_access.
16402 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16404 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16406 * config/riscv/riscv.md: Unify indentation.
16408 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16413 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16415 (extendsi<mode>2): Add support for doing sign extension via
16416 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16417 don't have ISA 3.0 instructions.
16418 (extendsi<mode>2 splitter): Likewise.
16419 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16420 generate the normal insns since SImode can now go in vector
16421 registers. Disallow the special UNSPECs needed for previous
16422 machines to hide SImode being used. Add new insns
16423 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16424 (fix_trunc<mode>si2_stfiwx): Likewise.
16425 (fix_trunc<mode>si2_internal): Likewise.
16426 (fixuns_trunc<mode>si2): Likewise.
16427 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16428 (fctiw<u>z_<mode>_smallint): Likewise.
16429 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16430 of floating point to 32-bit integer from doing a direct move to
16431 the GPR registers to do a store.
16432 (fctiwz_<mode>): Break long line.
16434 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16436 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16437 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16438 (addr_list, addr_offset_valid_p): New.
16439 (split_address_groups): Check offset validity with above function.
16440 (gt-tree-ssa-loop-ivopts.h): Include header file.
16442 2017-05-05 Nathan Sidwell <nathan@acm.org>
16444 * config.gcc (arm*-*-*): Add missing 'fi'.
16446 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16448 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16449 included in -fopt-info does not matter.
16450 * doc/optinfo.texi (-fopt-info): Fix description of default
16451 behavour. Explicitly say order of options included in -fopt-info
16454 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16456 * config.gcc: Allow combinations of aprofile and rmprofile values for
16457 --with-multilib-list.
16458 * config/arm/t-multilib: New file.
16459 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16460 variables. Remove setting of ISA and floating-point ABI in
16461 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16462 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16463 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16464 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16466 * config/arm/t-rmprofile: Likewise except for the matches changes.
16467 * doc/install.texi (--with-multilib-list): Document the combination of
16468 aprofile and rmprofile values and warn about pitfalls in doing that.
16470 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16472 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16473 (movdi_aarch64): Likewise.
16475 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16477 PR tree-optimization/80632
16478 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16480 (build_arrays): Initialize it for virtual phis.
16481 (fix_phi_nodes): Use it for virtual phis.
16483 PR tree-optimization/80558
16484 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16485 [x, y] op z into [x op, y op z] for op & or | if conditions
16488 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16489 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16492 * config/arm/arm.md (use_literal_pool): Remove.
16493 (64-bit immediate split): No longer takes cost into consideration
16494 if arm_disable_literal_pool is enabled.
16495 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16496 used when arm_disable_literal_pool is enabled.
16497 (arm_max_const_double_inline_cost): Remove use of
16498 arm_disable_literal_pool.
16499 (push_minipool_fix): Add assert.
16500 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16501 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16502 (no_literal_pool_sf_immediate): New.
16504 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16506 PR tree-optimization/80613
16507 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16508 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16510 2017-05-05 Richard Biener <rguenther@suse.de>
16512 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16514 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16516 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16517 of this flag from insn conditions due to removal from r247495.
16519 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16521 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16523 (arm_early_store_addr_dep_ptr): Likewise.
16524 * config/arm/aarch-common-protos.h
16525 (arm_early_load_addr_dep_ptr): Add prototype.
16526 (arm_early_store_addr_dep_ptr): Likewise.
16527 * config/arm/cortex-a53.md: Add new bypasses.
16529 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16531 * tree.c (next_type_uid): Change type to unsigned.
16532 (type_hash_canon): Decrement back next_type_uid if
16533 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16534 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16537 2017-05-04 Martin Sebor <msebor@redhat.com>
16539 * builtins.c: Fix a trivial typo in a comment.
16541 PR middle-end/79234
16542 * builtins.c (check_sizes): Adjust to handle reading past the end.
16543 Avoid printing excessive upper bound of ranges. Use %E to print
16544 tree nodes instead of converting them to %wu.
16545 (expand_builtin_memchr): New function.
16546 (compute_dest_size): Rename...
16547 (compute_objsize): ...to this.
16548 (expand_builtin_memcpy): Adjust.
16549 (expand_builtin_mempcpy): Adjust.
16550 (expand_builtin_strcat): Adjust.
16551 (expand_builtin_strcpy): Adjust.
16552 (check_strncat_sizes): Adjust.
16553 (expand_builtin_strncat): Adjust.
16554 (expand_builtin_strncpy): Adjust and simplify.
16555 (expand_builtin_memset): Adjust.
16556 (expand_builtin_bzero): Adjust.
16557 (expand_builtin_memcmp): Adjust.
16558 (expand_builtin): Handle memcmp.
16559 (maybe_emit_chk_warning): Check strncat just once.
16561 2017-05-04 Martin Sebor <msebor@redhat.com>
16563 PR preprocessor/79214
16564 PR middle-end/79222
16565 PR middle-end/79223
16566 * builtins.c (check_sizes): Add inlining context and issue
16567 warnings even when -Wno-system-headers is set.
16568 (check_strncat_sizes): Same.
16569 (expand_builtin_strncat): Same.
16570 (expand_builtin_memmove): New function.
16571 (expand_builtin_stpncpy): Same.
16572 (expand_builtin): Handle memmove and stpncpy.
16574 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16576 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16577 which is not used any more.
16579 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16581 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16583 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16585 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16586 (cortexa53_tunings): Likewise.
16587 (cortexa57_tunings): Likewise.
16588 (cortexa72_tunings): Likewise.
16589 (cortexa73_tunings): Likewise.
16591 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16593 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16594 Set loop alignment to 8.
16596 2017-05-04 Martin Sebor <msebor@redhat.com>
16598 PR translation/80280
16599 * builtins.c (expand_builtin_object_size): Add missing quoting to
16600 %D and like directives.
16601 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16602 (hsa_type_for_tree_type): Same.
16603 (verify_function_arguments): Same.
16604 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16605 * varasm.c (get_section): Same.
16608 2017-05-04 Martin Sebor <msebor@redhat.com>
16610 PR translation/80280
16611 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16613 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16615 * config/aarch64/aarch64.c (generic_addrcost_table):
16616 Change HI/TI mode setting.
16618 2017-05-04 Martin Jambor <mjambor@suse.cz>
16620 PR tree-optimization/80622
16621 * tree-sra.c (comes_initialized_p): New function.
16622 (build_accesses_from_assign): Only set write lazily when
16623 comes_initialized_p is false.
16624 (analyze_access_subtree): Use comes_initialized_p.
16625 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16626 instead of testing for PARM_DECL.
16628 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16630 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16631 constraint on operand 0 to allow more general addressing modes.
16632 Adjust output template.
16633 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16635 * config/aarch64/aarch64-protos.h
16636 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16637 * config/aarch64/constraints.md (Dp): New address constraint.
16638 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16641 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16643 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16644 update use of estimate_ipcp_clone_size_and_time.
16645 (estimate_local_effects): Update use of
16646 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16647 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16648 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16649 Return nonspecialized time.
16651 2017-05-04 Richard Biener <rguenther@suse.de>
16653 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16654 for the last VUSE which def dominates the PHI. Directly call
16656 (get_continuation_for_phi_1): Remove.
16658 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16660 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16661 to explain the use of truncating division. Cap the number of
16662 iterations to the maximum given by nb_iterations_upper_bound,
16665 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16667 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16668 * configure: Regenerate.
16669 * config.in: Regenerate.
16670 * config/i386/driver-mingw32.c: new file.
16671 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16672 * config.host: Link driver-mingw32.o on MinGW host.
16673 * doc/install.texi: Document new --enable-mingw-wildcard configure
16676 2017-05-04 Marek Polacek <polacek@redhat.com>
16678 PR tree-optimization/80612
16679 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16681 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16682 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16684 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16685 (movt splitter): Likewise.
16686 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16687 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16688 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16689 block for Thumb-1 with MOVT.
16690 (thumb2_legitimate_address_p): Move code block ...
16691 (can_avoid_literal_pool_for_label_p): ... into this new function.
16692 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16694 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16695 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16696 "M-profile targets with the MOVT instruction".
16698 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16700 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16701 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16702 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16704 2017-05-04 Martin Liska <mliska@suse.cz>
16706 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16707 variable cond_code.
16709 2017-05-04 Richard Biener <rguenther@suse.de>
16711 * tree.c (array_at_struct_end_p): Handle arrays at struct
16712 end with flexarrays more conservatively. Refactor and treat
16713 arrays of arrays or aggregates more strict. Fix
16714 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16715 * tree.c (array_at_struct_end_p): Adjust prototype.
16716 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16717 * gimple-fold.c (get_range_strlen): Likewise.
16718 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16720 2017-05-04 Richard Biener <rguenther@suse.de>
16722 PR tree-optimization/31130
16723 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16725 (supports_overflow_infinity): Likewise.
16726 (is_negative_overflow_infinity): Likewise.
16727 (is_positive_overflow_infinity): Likewise.
16728 (is_overflow_infinity): Likewise.
16729 (stmt_overflow_infinity): Likewise.
16730 (overflow_infinity_range_p): Likewise.
16731 (usable_range_p): Remove as always returning true.
16732 (make_overflow_infinity): Remove.
16733 (negative_overflow_infinity): Likewise.
16734 (positive_overflow_infinity): Likewise.
16735 (avoid_overflow_infinity): Likewise.
16736 (set_value_range): Adjust accordingly.
16737 (set_value_range_to_nonnegative): Likewise, remove now unused
16738 overflow_infinity arg.
16739 (vrp_operand_equal_p): Adjust.
16740 (update_value_range): Likewise.
16741 (range_int_cst_singleton_p): Likewise.
16742 (operand_less_p): Likewise.
16743 (compare_values_warnv): Likewise.
16744 (extract_range_for_var_from_comparison_expr): Likewise.
16745 (vrp_int_const_binop): Likewise.
16746 (zero_nonzero_bits_from_vr): Likewise.
16747 (extract_range_from_multiplicative_op_1): Likewise.
16748 (extract_range_from_binary_expr_1): Likewise.
16749 (extract_range_from_unary_expr): Likewise.
16750 (extract_range_from_comparison): Likewise.
16751 (extract_range_basic): Likewise.
16752 (adjust_range_with_scev): Likewise.
16753 (compare_ranges): Likewise.
16754 (compare_range_with_value): Likewise.
16755 (dump_value_range): Likewise.
16756 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16758 (simplify_cond_using_ranges): Adjust.
16760 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16762 * brig-builtins.def: Added a builtin for class_f64.
16763 * builtin-types.def: Added a builtin type needed by class_f64.
16765 2017-05-03 Jason Merrill <jason@redhat.com>
16767 * timevar.def: Add TV_CONSTEXPR.
16769 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16771 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16773 2017-05-03 Martin Jambor <mjambor@suse.cz>
16775 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16776 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16777 * ipa-cp.c (ipcp_propagate_stage): Do not call
16778 ipa_update_after_lto_read.
16779 * ipa-inline.c (ipa_inline): Likewise.
16781 2017-05-03 Martin Jambor <mjambor@suse.cz>
16783 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16784 tag. Added a default constructor and a destructor.
16785 (ipa_edge_args_sum_t): New class;
16786 (ipa_edge_args_sum): Declare.
16787 (ipa_edge_args_vector): Remove declaration.
16788 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16789 (ipa_free_edge_args_substructures): Remove declaration.
16790 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16791 (ipa_edge_args_info_available_for_edge_p): Likewise.
16792 * ipa-prop.c (ipa_edge_args_vector): Removed.
16793 (edge_removal_hook_holder): Likewise.
16794 (edge_duplication_hook_holder): Likewise.
16795 (ipa_edge_args_sum): New variable.
16796 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16797 ipa_edge_args_vector.
16798 (ipa_free_edge_args_substructures): Likewise.
16799 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16800 ipa_edge_args_vector.
16801 (ipa_edge_removal_hook): Turned into method
16802 ipa_edge_args_sum_t::remove.
16803 (ipa_edge_duplication_hook): Turned into method
16804 ipa_edge_args_sum_t::duplicate.
16805 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16806 registering edge hooks.
16807 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16808 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16809 ipa_edge_args_sum instead of ipa_edge_args_vector.
16810 * ipa-profile.c (ipa_profile): Likewise.
16812 2017-05-03 Martin Jambor <mjambor@suse.cz>
16814 * symbol-summary.h (function_summary): New method exists.
16815 (function_summary::symtab_removal): Deallocate through release.
16816 (call_summary): New class.
16817 (gt_ggc_mx): New overload.
16818 (gt_pch_nx): Likewise.
16819 (gt_pch_nx): Likewise.
16821 2017-05-03 Jeff Law <law@redhat.com>
16823 PR tree-optimization/78496
16824 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16825 from simplify_cond_using_ranges. Split off code to walk
16826 backwards through casts into ...
16827 (simplify_cond_using_ranges_2): New function.
16828 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16829 (execute_vrp): After identifying jump threads, call
16830 simplify_cond_using_ranges_2.
16832 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16835 * ipa-inline.h (inline_summary): Add ctor.
16836 (create_ggc): Do not use ggc_cleared_alloc.
16838 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16839 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16841 * gcc.c (handle_braces): Support escaping in switch matching
16843 * doc/invoke.texi (Spec Files): Document it.
16844 Remove superfluous @code markup in items.
16846 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16848 * diagnostic-show-locus.c (struct column_range): New struct.
16849 (get_affected_columns): New function.
16850 (get_printed_columns): New function.
16851 (struct correction): New struct.
16852 (correction::ensure_capacity): New function.
16853 (correction::ensure_terminated): New function.
16854 (struct line_corrections): New struct.
16855 (line_corrections::~line_corrections): New dtor.
16856 (line_corrections::add_hint): New function.
16857 (layout::print_trailing_fixits): Reimplement in terms of the new
16859 (selftest::test_overlapped_fixit_printing): New function.
16860 (selftest::diagnostic_show_locus_c_tests): Call it.
16862 2017-05-03 Nathan Sidwell <nathan@acm.org>
16864 Canonicalize canonical type hashing
16865 * tree.h (type_hash_canon_hash): Declare.
16866 * tree.c (type_hash_list, attribute_hash_list): Move into
16867 type_hash_canon_hash.
16868 (build_type_attribute_qual_variant): Break out hash code calc into
16869 type_hash_canon_hash.
16870 (type_hash_canon_hash): New. Generic type hash computation.
16871 (build_range_type_1, build_array_type_1, build_function_type,
16872 build_method_type_directly, build_offset_type, build_complex_type,
16873 make_vector_type): Call it.
16875 2017-05-03 Richard Biener <rguenther@suse.de>
16877 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16878 When all DRs have unknown misaligned do not always peel
16879 when there is a store but apply the same costing model as if
16880 there were only loads.
16882 2017-05-03 Richard Biener <rguenther@suse.de>
16885 PR tree-optimization/80492
16886 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16887 compare_base_decls returning dont-know properly.
16889 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16891 * config/arm/iterators.md (CCSI): New mode iterator.
16892 (arch): New mode attribute.
16893 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16894 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16895 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16896 code iterator for success result mode.
16897 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16898 the corresponding new insn generators.
16900 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16903 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16904 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16906 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16908 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16909 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16910 (DDR_A): Wrap DDR argument in brackets.
16911 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16912 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16913 (DDR_REVERSED_P): Likewise.
16915 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16917 PR tree-optimization/79472
16918 * tree-switch-conversion.c (struct switch_conv_info): Add
16919 contiguous_range and default_case_nonstandard fields.
16920 (collect_switch_conv_info): Compute contiguous_range and
16921 default_case_nonstandard fields, don't clear final_bb if
16922 contiguous_range and only the default case doesn't have the required
16924 (check_all_empty_except_final): Set default_case_nonstandard instead
16925 of failing if contiguous_range and the default case doesn't have empty
16927 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16928 and only the default case doesn't have the required constants. Skip
16930 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16931 if default_case_nonstandard.
16932 (build_constructors): Build constant 1 just once. Assert that default
16933 values aren't inserted in between cases if contiguous_range. Skip
16935 (build_arrays): Skip virtual phis.
16936 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16937 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16938 Handle virtual phis.
16939 (gen_inbound_check): Handle default_case_nonstandard case.
16940 (process_switch): Adjust check_final_bb caller. Call
16941 gather_default_values with the first non-default case instead of
16942 default case if default_case_nonstandard.
16944 2017-05-02 Nathan Sidwell <nathan@acm.org>
16946 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16947 check. Fix formatting.
16949 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16951 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16952 errors when comparing specialized and unspecialized times.
16954 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16956 * diagnostic-show-locus.c
16957 (layout::should_print_annotation_line_p): Make private.
16958 (layout::print_annotation_line): Make private.
16959 (layout::annotation_line_showed_range_p): Make private.
16960 (layout::show_ruler): Make private.
16961 (layout::print_source_line): Make private. Pass in line and
16962 line_width, rather than calling location_get_source_line. Drop
16964 (layout::print_leading_fixits): New method.
16965 (layout::print_any_fixits): Rename to...
16966 (layout::print_trailing_fixits): ...this, and make private.
16967 Don't print newline fixits.
16968 (diagnostic_show_locus): Move logic for printing one row into...
16969 (layout::print_line): ...this new function. Move the
16970 location_get_source_line call and error-handling from
16971 print_source_line to here. Call print_leading_fixits, and rename
16972 print_any_fixits to print_trailing_fixits.
16973 (selftest::test_fixit_insert_containing_newline): Update now that
16974 newlines are partially supported.
16975 (selftest::test_fixit_insert_containing_newline_2): New test.
16976 (selftest::test_fixit_replace_containing_newline): Update comments.
16977 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16978 * edit-context.c (class added_line): New class.
16979 (class edited_line): Describe newline handling in comment.
16980 (edited_line::actually_edited_p): New method.
16981 (edited_line::print_content): Delete redundant decl.
16982 (edited_line::m_predecessors): New field.
16983 (edited_file::print_content): Call edited_line::print_content.
16984 (edited_file::print_diff): Update to support newlines.
16985 (edited_file::print_diff_hunk): Likewise.
16986 (edited_file::print_run_of_changed_lines): New function.
16987 (edited_file::print_diff_line): Convert to...
16988 (print_diff_line): ...this.
16989 (edited_file::get_effective_line_count): New function.
16990 (edited_line::edited_line): Initialize new field m_predecessors.
16991 (edited_line::~edited_line): Clean up m_predecessors.
16992 (edited_line::apply_fixit): Handle newlines.
16993 (edited_line::get_effective_line_count): New function.
16994 (edited_line::print_content): New function.
16995 (edited_line::print_diff_lines): New function.
16996 (selftest::test_applying_fixits_insert_containing_newline): New
16998 (selftest::test_applying_fixits_replace_containing_newline): New
17000 (selftest::insert_line): New function.
17001 (selftest::test_applying_fixits_multiple_lines): Add example of
17003 (selftest::edit_context_c_tests): Call the new tests.
17005 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17007 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17008 parameter cand. Update dump information.
17009 (get_computation_cost): Update uses.
17011 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17013 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17014 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17015 (get_computation_at, rewrite_use_address): Update use of
17016 get_computation_aff.
17018 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17020 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17021 (get_computation): Delete.
17022 (get_computation_cost): Implement like get_computation_cost_at.
17023 Use get_computation_at.
17024 (get_computation_cost_at): Delete.
17025 (rewrite_use_nonlinear_expr): Use get_computation_at.
17026 (rewrite_use_compare, remove_unused_ivs): Ditto.
17028 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17030 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17034 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17035 (ivopts_global_cost_for_size): Rename parameter and update uses.
17036 (iv_ca_recount_cost): Update uses.
17037 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17038 candidates seperately in n_invs and n_cands.
17039 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17041 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17043 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17044 (find_inv_vars_cb): New.
17045 (find_depends): Renamed to ...
17046 (find_inv_vars): ... this.
17047 (add_candidate_1, force_var_cost): Call find_inv_vars.
17048 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17050 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17052 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17053 inv_vars. Add inv_exprs.
17054 (struct iv_cand): Rename depends_on to inv_vars.
17055 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17056 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17057 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17058 (dump_cand): Dump inv_vars.
17059 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17060 (record_invariant, find_depends, add_candidate_1): Ditto.
17061 (set_group_iv_cost, force_var_cost): Ditto.
17062 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17063 (get_computation_cost_at, get_computation_cost): Ditto.
17064 (determine_group_iv_cost_generic): Ditto.
17065 (determine_group_iv_cost_address): Ditto.
17066 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17067 (determine_group_iv_costs): Ditto.
17068 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17069 (iv_ca_set_remove_invariants): Renamed to ...
17070 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17071 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17072 (iv_ca_set_add_invariants): Renamed to ...
17073 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17074 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17075 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17076 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17077 (create_new_ivs): Remove useless dump.
17079 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17081 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17083 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17084 (iv_ca_set_no_cp, create_new_iv): Ditto.
17086 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17088 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17090 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17092 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17093 function tree_check2.
17095 2017-05-02 Martin Liska <mliska@suse.cz>
17097 * doc/gcov.texi: Add missing preposition.
17098 * gcov.c (function_info::function_info): Properly fill up
17099 all member variables.
17101 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17103 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17105 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17107 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17109 2017-05-02 Martin Liska <mliska@suse.cz>
17112 * lto-streamer-in.c (lto_read_tree_1): Remove
17113 LTO_STREAMER_DEBUG.
17114 * lto-streamer.c (struct tree_hash_entry): Likewise.
17115 (struct tree_entry_hasher): Likewise.
17116 (tree_entry_hasher::hash): Likewise.
17117 (tree_entry_hasher::equal): Likewise.
17118 (lto_streamer_init): Likewise.
17119 (lto_orig_address_map): Likewise.
17120 (lto_orig_address_get): Likewise.
17121 (lto_orig_address_remove): Likewise.
17122 * lto-streamer.h: Likewise.
17123 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17124 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17126 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17128 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17129 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17130 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17131 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17132 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17133 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17134 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17135 (mm_maskz_sub_ss): New intrinsics.
17136 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17137 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17138 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17139 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17140 (__builtin_ia32_subss_mask_round): New builtins.
17141 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17142 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17143 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17145 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17146 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17148 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17151 2017-05-02 Martin Jambor <mjambor@suse.cz>
17153 PR tree-optimization/78687
17154 * tree-sra.c (access): New field parent.
17155 (process_subtree_disqualification): New function.
17156 (disqualify_candidate): Call it.
17157 (build_accesses_from_assign): Reset write flag if creating an
17159 (build_access_subtree): Fill in parent field and also prpagate
17160 down grp_write flag.
17161 (create_artificial_child_access): New parameter set_grp_write, set
17162 grp_write to its value.
17163 (propagate_subaccesses_across_link): Also propagate grp_write flag
17165 (propagate_all_subaccesses): Push the closest parent back to work
17166 queue if add_access_to_work_queue returned true.
17168 2017-05-02 Richard Biener <rguenther@suse.de>
17170 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17171 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17172 -fstrict-overflow documentation.
17173 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17174 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17175 flag_strict_overflow.
17176 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17177 * lto-opts.c (lto_write_options): Do not stream it.
17178 * lto-wrapper.c (merge_and_complain): Do not handle it.
17179 * opts.c (default_options_table): Do not set -fstrict-overflow.
17180 (finish_options): Likewise do not clear it when sanitizing.
17181 * simplify-rtx.c (simplify_const_relational_operation): Do not
17182 test flag_strict_overflow.
17184 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17186 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17187 using enabled attribute.
17188 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17189 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17190 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17191 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17192 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17193 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17194 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17195 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17196 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17197 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17199 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17201 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17203 2017-05-02 Richard Biener <rguenther@suse.de>
17205 PR tree-optimization/80591
17207 2017-04-10 Richard Biener <rguenther@suse.de>
17209 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17212 2017-05-02 Richard Biener <rguenther@suse.de>
17214 PR tree-optimization/80549
17215 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17216 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17217 headers if they do not have a preheader.
17219 2017-05-02 Martin Liska <mliska@suse.cz>
17222 * common.opt: Fix typo.
17223 * doc/invoke.texi: Likewise.
17225 2017-05-01 Jan Beulich <jbeulich@suse.com>
17227 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17228 swapping, add (x,x,m,x,n) alternative.
17230 2017-05-01 Nathan Sidwell <nathan@acm.org>
17232 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17233 unnecessary unadjusted_alignment check.
17235 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17238 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17240 * gimplify.c (gimplify_cilk_detach): New function.
17241 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17242 * tree-core.h: Document EXPR_CILK_SPAWN.
17243 * tree.h (EXPR_CILK_SPAWN): Define.
17245 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17247 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17248 to use new fixit_hint representation, using the "replace" logic.
17249 (get_line_span_for_fixit_hint): Likewise.
17250 (layout::print_any_fixits): Likewise.
17251 (selftest::test_one_liner_many_fixits): Rename to...
17252 (selftest::test_one_liner_many_fixits_1): ...this, and update
17253 comment and expected output to reflect that the multiple fix-it
17254 hints are now consolidated into one insertion.
17255 (selftest::test_one_liner_many_fixits_2): New test.
17256 (selftest::test_diagnostic_show_locus_one_liner): Update for
17258 (selftest::test_fixit_consolidation): Update for fix-it API
17260 * diagnostic.c (print_parseable_fixits): Likewise.
17261 * edit-context.c (edited_line::m_line_events): Convert from
17262 auto_vec <line_event *> to auto_vec <line_event>.
17263 (class line_event): Convert from abstract base class to a concrete
17264 class, taking over the role of replace_event.
17265 (class insert_event): Delete.
17266 (class replace_event): Rename to class line_event. Convert to
17268 (edit_context::add_fixits): Reimplement.
17269 (edit_context::apply_insert): Delete.
17270 (edit_context::apply_replace): Rename to...
17271 (edit_context::apply_fixit): ...this. Convert to half-open range.
17272 (edited_file::apply_insert): Delete.
17273 (edited_file::apply_replace): Rename to...
17274 (edited_file::apply_fixit): ...this.
17275 (edited_line::~edited_line): Drop deletion of events.
17276 (edited_line::apply_insert): Delete.
17277 (edited_line::apply_replace): Rename to...
17278 (edited_line::apply_fixit): ...this. Convert to half-open range.
17279 Update for change to type of m_line_events.
17280 * edit-context.h (edit_context::apply_insert): Delete.
17281 (edit_context::apply_replace): Rename to...
17282 (edit_context::apply_fixit): ...this.
17284 2017-05-01 Martin Sebor <msebor@redhat.com>
17286 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17289 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17292 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17293 __get_cpuid_max returns 0.
17294 (__get_cpuid_count): Ditto.
17296 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17298 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17299 replacement expression is another instance of one of its arguments.
17301 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17304 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17305 check for stack push/pop autoinc.
17306 * config/i386/i386.c (ix86_agi_dependent): Return false
17307 if the only reason why modified_in_p returned true is that
17308 addr is SP based and set_insn is a push or pop.
17310 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17312 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17315 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17318 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17319 (account_size_time): Use two predicates - exec_pred and
17321 (evaluate_conditions_for_known_args): Compute both clause and
17323 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17324 (inline_summary_t::duplicate): Update.
17325 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17327 (compute_inline_parameters): Likewise.
17328 (estimate_edge_size_and_time): Update caluclation of time.
17329 (estimate_node_size_and_time): Compute both time and nonspecialized
17331 (estimate_ipcp_clone_size_and_time): Update.
17332 (inline_merge_summary): Update.
17333 (do_estimate_edge_time): Update.
17334 (do_estimate_edge_size): Update.
17335 (do_estimate_edge_hints): Update.
17336 (inline_read_section, inline_write_summary): Stream both new predicates.
17337 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17339 (compute_inlined_call_time): Cleanup.
17340 (big_speedup_p): Update.
17341 (edge_badness): Update.
17342 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17343 (size_time_entry): Replace predicate by exec_predicate and
17344 nonconst_predicate.
17345 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17346 (estimate_edge_time): Return also nonspec_time.
17347 (reset_edge_growth_cache): Update.
17349 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17351 PR rtl-optimization/80491
17352 * ifcvt.c (noce_process_if_block): When looking for x setter
17353 with missing else_bb, don't check only the insn right before
17354 cond_earliest, but look for the last insn that x is modified in
17355 within the same bb.
17357 PR rtl-optimization/80491
17358 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17360 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17362 PR tree-optimization/80487
17363 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17365 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17367 PR tree-optimization/79697
17368 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17369 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17370 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17372 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17373 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17375 2017-04-28 Martin Sebor <msebor@redhat.com>
17377 PR tree-optimization/80523
17378 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17379 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17381 (maybe_warn, format_directive, parse_directive): Use new functions.
17382 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17384 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17386 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17388 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17390 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17391 target_header_dir): Set correctly.
17392 * configure: Regenerated.
17393 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17394 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17395 instead of SYSTEM_HEADER_DIR.
17397 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17399 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17400 (estimate_local_effects): Likewise.
17401 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17402 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17403 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17404 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17405 do_estimate_edge_time, estimate_edge_time): Likewise.
17406 * ipa-inline-analysis.c (estimate_node_size_and_time,
17407 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17408 (estimate_time_after_inlining): Remove.
17410 2017-04-28 Martin Liska <mliska@suse.cz>
17412 * doc/gcov.texi: Enhance documentation of gcov.
17414 2017-04-28 Martin Liska <mliska@suse.cz>
17416 * doc/gcov.texi: Sort options in alphabetic order.
17417 * doc/gcov-dump.texi: Likewise.
17418 * doc/gcov-tool.texi: Likewise.
17419 * gcov.c (print_usage): Likewise.
17420 * gcov-dump.c (print_usage): Likewise.
17421 * gcov-tool.c (print_merge_usage_message): Likewise.
17422 (print_rewrite_usage_message): Likewise.
17423 (print_overlap_usage_message): Likewise.
17425 2017-04-28 Martin Liska <mliska@suse.cz>
17427 PR gcov-profile/53915
17428 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17430 2017-04-28 Martin Liska <mliska@suse.cz>
17432 PR gcov-profile/79891
17433 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17434 is marked by compiler as living on a line.
17435 (get_cycles_count): Remove usage of the union.
17436 (output_intermediate_file): Likewise.
17437 (find_source): Fix GNU coding style.
17438 (accumulate_line_counts): Remove old non-all block mode.
17439 (output_lines): Remove usage of the union.
17440 * profile.c (output_location): Include all BBs, even if
17441 belonging to a same line (and file) as a previous BB.
17443 2017-04-28 Martin Liska <mliska@suse.cz>
17445 * gcov.c (process_args): Handle new argument 'w'.
17446 (read_graph_file): Assign ID to BBs.
17447 (output_branch_count): Display BB # if verbose flag is set.
17448 (output_lines): Likewise for arcs.
17449 (print_usage): Add '--verbose' option help.
17450 * doc/gcov.texi: Document --verbose (-w) option.
17452 2017-04-28 Martin Liska <mliska@suse.cz>
17454 * gcov.c (struct block_location_info): New struct.
17455 (process_file): Fill up the new structure.
17456 (read_graph_file): Replace usage of encoding by the newly added
17458 (add_line_counts): Likewise.
17459 (accumulate_line_counts): Remove usage of the union.
17460 (function_info::function_info): New function.
17461 (function_info::~function_info): Likewise.
17462 (process_file): Call delete instead of release_function.
17463 (release_function): Release the function.
17464 (release_structures): Call delete instead of release_function.
17465 (solve_flow_graph): Replace usage of num_blocks.
17466 (find_exception_blocks): Likewise.
17467 (output_lines): Fix GNU coding style.
17469 2017-04-28 Martin Liska <mliska@suse.cz>
17472 * coverage.c (coverage_remove_note_file): New function.
17473 * coverage.h: Declare the function.
17474 * toplev.c (finalize): Clean if an error has been seen.
17476 2017-04-28 Martin Liska <mliska@suse.cz>
17478 PR gcov-profile/80031
17479 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17480 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17481 * gcov.c (read_graph_file): Read just number of blocks.
17482 * profile.c (branch_prob): Do not stream 0 flags per a basic
17485 2017-04-28 Martin Liska <mliska@suse.cz>
17487 * gcov-dump.c (tag_*): Add new argument to declarations.
17488 (dump_gcov_file): Likewise.
17489 (tag_blocks): Add and use new argument depth.
17490 (tag_arcs): Likewise.
17491 (tag_lines): Likewise.
17492 (tag_counters): Likewise.
17493 (tag_summary): Likewise.
17494 (dump_working_sets): Use depth to do a proper indentation.
17496 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17499 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17500 * symtab.c (symtab_node::debug_symtab): Move definition here.
17502 2017-04-28 Richard Biener <rguenther@suse.de>
17504 * lto-streamer.h (LTO_major_version): Bump to 7.
17506 2017-04-28 Richard Biener <rguenther@suse.de>
17508 * tree-vrp.c (assert_info): New struct.
17509 (add_assert_info): New helper.
17510 (register_edge_assert_for_2): Refactor to add asserts to a vector
17512 (register_edge_assert_for_1): Likewise.
17513 (register_edge_assert_for): Likewise.
17514 (finish_register_edge_assert_for): New helper actually registering
17515 asserts where live on edge.
17516 (find_conditional_asserts): Adjust.
17517 (find_switch_asserts): Likewise.
17518 (evrp_dom_walker::try_find_new_range): Generalize.
17519 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17521 2017-04-27 Marek Polacek <polacek@redhat.com>
17524 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17525 arg10 and arg11 to itype.
17527 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17529 * doc/extend.texi (Object Size Checking): Improve grammar.
17531 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17534 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17535 that the logic for permitting reciprocal estimates matches that
17538 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17541 * tree.c (type_cache_hasher::equal): Only compare
17542 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17543 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17544 non-aggregate element types.
17545 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17546 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17548 2017-04-27 Richard Biener <rguenther@suse.de>
17550 PR middle-end/80533
17551 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17552 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17553 keeping a reference to a trailing array.
17555 2017-04-27 Richard Biener <rguenther@suse.de>
17557 PR middle-end/80539
17558 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17559 being in loop-closed SSA form conservatively.
17560 (chrec_fold_multiply_poly_poly): Likewise.
17562 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17564 PR middle-end/79665
17565 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17566 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17568 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17571 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17572 (aarch64_function_arg_alignment): Return unsigned int again, but still
17573 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17574 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17575 Don't emit -Wpsabi note.
17576 (aarch64_function_arg_boundary): Likewise.
17577 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17580 2017-04-26 Nathan Sidwell <nathan@acm.org>
17582 * tree.h (crc32_unsigned_n): Declare.
17583 (crc32_unsigned, crc32_unsigned): Make inline.
17584 * tree.c (crc32_unsigned_bits): Replace with ...
17585 (crc32_unsigned_n): ... this.
17586 (crc32_unsigned, crc32_byte): Remove.
17587 (crc32_string): Remove unnecessary braces.
17589 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17591 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17592 * ipa-inline-analysis.c (MAX_TIME): Remove.
17593 (account_size_time): Use sreal for time.
17594 (dump_inline_summary): Update.
17595 (estimate_function_body_sizes): Update.
17596 (estimate_edge_size_and_time): Update.
17597 (estimate_calls_size_and_time): Update.
17598 (estimate_node_size_and_time): Update.
17599 (inline_merge_summary): Update.
17600 (inline_update_overall_summary): Update.
17601 (estimate_time_after_inlining): Update.
17602 (inline_read_section): Update.
17603 (inline_write_summary): Update.
17604 * ipa-inline.c (compute_uninlined_call_time): Update.
17605 (compute_inlined_call_time): Update.
17606 (recursive_inlining): Update.
17607 (inline_small_functions): Update.
17608 (dump_overall_stats): Update.
17609 * ipa-inline.h: Include sreal.h.
17610 (size_time_entry): Turn time to sreal.
17611 (inline_summary): Turn self_time nad time to sreal.
17613 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17615 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17617 (sreal::stream_out, sreal::stream_in): New.
17618 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17620 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17622 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17625 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17628 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17630 (dimode_scalar_chain::compute_convert_gain): Ditto.
17631 (dimode_scalar_chain::make_vector_copies): Ditto.
17632 (dimode_scalar_chain::convert_reg): Ditto.
17633 (dimode_scalar_chain::convert_insn): Ditto.
17634 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17635 (VI248_AVX512BW_1): New mode iterator.
17636 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17637 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17640 2017-04-25 Martin Sebor <msebor@redhat.com>
17642 PR tree-optimization/80497
17643 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17644 constants are representable in HOST_WIDE_INT.
17645 (parse_directive): Ditto.
17647 2017-04-25 Martin Sebor <msebor@redhat.com>
17650 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17651 (new_zero_array): Adjust signature.
17652 (dom_info::dom_init): Used unsigned rather that size_t.
17653 (dom_info::dom_info): Same.
17655 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17656 Jakub Jelinek <jakub@redhat.com>
17659 * config/arm/arm.c: Include gimple.h.
17660 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17661 returns negative, increment ncrn only if it returned positive.
17662 (arm_needs_doubleword_align): Return int instead of bool,
17663 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17664 members, but if there is any such non-FIELD_DECL
17665 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17666 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17667 returns negative, increment nregs only if it returned positive.
17668 (arm_setup_incoming_varargs): Likewise.
17669 (arm_function_arg_boundary): Emit -Wpsabi note if
17670 arm_needs_doubleword_align returns negative, return
17671 DOUBLEWORD_ALIGNMENT only if it returned positive.
17673 2017-04-25 Marek Polacek <polacek@redhat.com>
17676 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17677 first argument to type.
17679 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17682 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17683 type checks to test for compatibility instead of equality.
17685 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17686 Jakub Jelinek <jakub@redhat.com>
17689 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17691 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17692 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17693 the alignment computation, but return their maximum in warn_alignment.
17694 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17695 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17697 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17698 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17701 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17703 * config/arc/simdext.md (dmpyh): Fix typo.
17705 2017-04-25 Richard Biener <rguenther@suse.de>
17707 PR tree-optimization/80492
17708 * alias.c (compare_base_decls): Handle registers with asm
17709 specification conservatively.
17710 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17711 compare_base_decls returning dont-know properly.
17713 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17715 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17716 (legitimate_offset_address_p): New function.
17717 (arc_legitimate_address_p): Use above function.
17719 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17721 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17723 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17725 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17726 ACCH registers whenever they are available.
17728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17730 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17731 double regs fix when not used.
17733 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17735 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17737 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17738 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17740 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17742 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17743 when emitting short ADD instructions.
17745 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17747 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17749 (cmpsi_cc_c_insn): Likewise.
17750 (cbranchsi4_scratch): Compute proper instruction length using
17751 compact_hreg_operand.
17752 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17754 2017-04-25 Richard Biener <rguenther@suse.de>
17756 PR middle-end/80509
17757 * passes.c (pass_manager::pass_manager): Initialize
17758 m_name_to_pass_map.
17760 2017-04-25 Richard Biener <rguenther@suse.de>
17762 PR tree-optimization/79201
17763 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17765 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17768 * config/s390/vector.md: Split MEM->GPR vector moves for
17769 non-s_operand addresses.
17771 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17774 * config/s390/predicates.md (reload_const_wide_int_operand): New
17776 * config/s390/s390.md ("movti"): Remove d/P alternative.
17777 ("movti_bigconst"): New pattern definition.
17779 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17782 * s390-protos.h (s390_expand_cs_hqi): Removed.
17783 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17784 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17785 modes as well as CCZ1mode and CCZmode.
17786 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17787 signature of s390_emit_compare_and_swap.
17788 (s390_expand_cs_hqi): Likewise, make static.
17789 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17790 compare-and-swap, in some cases.
17791 (s390_expand_cs): Wrapper function.
17792 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17794 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17795 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17796 patterns for small and large integers. Forbid symref memory operands.
17797 Move expander to s390.c. Require cc register.
17798 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17799 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17800 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17801 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17802 symref memory operands. Remove CC mode and call s390_match_ccmode
17804 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17806 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17808 * config/s390/s390.md (define_peephole2): New peephole to help
17809 combining the load-and-test pattern with volatile memory.
17811 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17813 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17814 with CCZmode for TARGET_Z196.
17816 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17818 PR rtl-optimization/80501
17819 * combine.c (make_compound_operation_int): Set subreg_code to SET
17820 even for AND with mask of the sign bit of mode.
17822 PR rtl-optimization/80500
17823 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17824 sum's initial value.
17826 2017-04-25 Julian Brown <julian@codesourcery.com>
17827 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17829 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17831 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17833 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17835 2017-04-25 Julian Brown <julian@codesourcery.com>
17836 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17838 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17839 (thunderx2t99_sha): New Reservation.
17841 2017-04-25 Julian Brown <julian@codesourcery.com>
17842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17844 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17845 type for 1-element load.
17847 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17849 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17851 2017-04-24 Martin Jambor <mjambor@suse.cz>
17853 PR tree-optimization/80293
17854 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17855 char arrays not totally scalarizable if it is false.
17856 (analyze_all_variable_accesses): Pass correct value in the new
17857 parameter. Add a statistics counter.
17859 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17861 PR middle-end/79931
17862 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17864 2017-04-24 Richard Biener <rguenther@suse.de>
17866 PR tree-optimization/80494
17867 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17868 out for complex types.
17870 2017-04-24 Richard Biener <rguenther@suse.de>
17872 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17873 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17874 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17875 (DFS): Adjust and never fail.
17876 (sccvn_dom_walker::fail): Remove.
17877 (sccvn_dom_walker::before_dom_children): Adjust.
17878 (run_scc_vn): Likewise and never fail.
17879 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17880 (pass_fre::execute): Likewise.
17882 2017-04-24 Richard Biener <rguenther@suse.de>
17884 PR tree-optimization/79725
17885 * tree-ssa-sink.c (statement_sink_location): Return whether
17886 failure reason was zero uses. Move that check later.
17887 (sink_code_in_bb): Deal with zero uses by removing the stmt
17890 2017-04-24 Richard Biener <rguenther@suse.de>
17893 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17894 pointer-based references.
17896 2017-04-24 Richard Biener <rguenther@suse.de>
17899 * pass_manager.h (pass_manager::operator new): Remove.
17900 (pass_manager::operator delete): Likewise.
17901 * passes.c (pass_manager::operator new): Remove.
17902 (pass_manager::operator delete): Likewise.
17903 (pass_manager::pass_manager): Zero individual pass members.
17905 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17908 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17909 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17910 Check "XEXP (src, 1)" operand here.
17911 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17912 Check "XEXP (src, 1)" operand here.
17913 (dimode_scalar_chain::make_vector_copies): Detect count register
17914 of a shift instruction. Zero extend count register from QImode
17915 to DImode to satisfy vector shift pattern count operand predicate.
17916 Substitute vector shift count operand with a DImode copy.
17917 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17920 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17922 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17923 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17924 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17925 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17926 (UNSPEC_NOREX_MEM): Remove definition.
17928 2017-04-21 Richard Biener <rguenther@suse.de>
17930 PR tree-optimization/79547
17931 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17932 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17933 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17934 without any constraints.
17936 2017-04-21 Richard Biener <rguenther@suse.de>
17938 PR tree-optimization/78847
17939 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17941 2017-04-21 Richard Biener <rguenther@suse.de>
17943 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17944 (build_distinct_type_copy): Likewise.
17945 (build_variant_type_copy): Likewise.
17946 * tree.c (build_qualified_type): Pass down mem-stat info.
17947 (build_distinct_type_copy): Likewise.
17948 (build_variant_type_copy): Likewise.
17950 2017-04-21 Richard Biener <rguenther@suse.de>
17952 PR tree-optimization/80237
17953 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17955 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17956 for a simplified result.
17958 2016-04-21 Richard Biener <rguenther@suse.de>
17960 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17961 sth as strict as a simple_iv but a chrec without symbols and an
17962 operand defined in the loop we are peeling (and not some subloop).
17963 (propagate_constants_for_unrolling): Propagate all constants.
17965 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17968 * config/i386/i386.c (print_reg): Remove assert for disalowed
17969 regno values, call output_operand_lossage instead.
17971 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17974 * config/i386/constraints.md (Yc): New register constraint.
17975 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17976 Use Yc constraint for alternative 2 of operand 0. Remove
17977 preferred_for_speed attribute.
17979 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17981 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17982 lastprivate clauses in SIMT case.
17984 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17986 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17988 2017-04-20 Richard Biener <rguenther@suse.de>
17990 PR tree-optimization/57796
17991 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17992 as N scalar stores.
17993 (vect_model_load_cost): Cost gathers as N scalar loads.
17995 2017-04-20 Richard Biener <rguenther@suse.de>
17997 * ggc-page.c (ggc_allocated_p): Rename to ...
17998 (safe_lookup_page_table_entry): ... this and return the lookup
18000 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18002 2017-04-20 Richard Biener <rguenther@suse.de>
18004 PR tree-optimization/80453
18005 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18006 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18007 from the conditions.
18008 (vn_phi_eq): Pass them down.
18009 (vn_phi_lookup): Record them.
18010 (vn_phi_insert): Likewise.
18012 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18014 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18015 uninitialized variable warning to avoid buffer overrun.
18017 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18020 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18021 is suppressed for '{ 0 }' in C.
18023 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18025 * BASE-VER: Set to 8.0.0.
18027 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18029 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18030 priority .init_array and .fini_array section with SECTION_NOTYPE
18033 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18035 PR middle-end/80423
18036 * tree.h (build_array_type): Add typeless_storage default argument.
18037 * tree.c (type_cache_hasher::equal): Also compare
18038 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18039 (build_array_type): Add typeless_storage argument, set
18040 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18042 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18043 (build_array_type): Likewise. Add typeless_storage argument.
18045 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18046 Jakub Jelinek <jakub@redhat.com>
18048 PR tree-optimization/80426
18049 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18050 operation on symbolic operands, also compute the overflow for the
18051 invariant part when the operation degenerates into a negation.
18053 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18056 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18057 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18060 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18062 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18065 * config/avr/avr.c (tree.h): Include it.
18066 (cgraph.h): Include it.
18067 (avr_encode_section_info): Don't warn for uninitialized progmem
18068 variable if it's just an alias.
18070 2017-04-19 Richard Biener <rguenther@suse.de>
18073 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18074 when needed by AutoPGO.
18076 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18079 * doc/lto.texi: Remove an extra 'that'.
18081 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18083 PR rtl-optimization/80429
18084 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18085 are only used in debug insns.
18087 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18088 Vladimir Makarov <vmakarov@redhat.com>
18090 * config/sparc/predicates.md (input_operand): Add comment. Return
18091 true for any memory operand when LRA is in progress.
18092 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18094 2017-04-18 Jeff Law <law@redhat.com>
18097 * mips.md ({return,simple_return}_internal): Do not overwrite
18100 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18102 PR tree-optimization/80443
18103 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18104 instead of adding 1, subtract -1 and similarly instead of subtracting
18107 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18109 PR rtl-optimization/80357
18110 * haifa-sched.c (tmp_bitmap): New variable.
18111 (model_recompute): Handle duplicate use records.
18112 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18113 (free_global_sched_pressure_data): Free it.
18115 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18118 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18119 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18120 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18121 instead of SYSTEM_HEADER_DIR.
18123 2017-04-18 Jeff Law <law@redhat.com>
18125 PR middle-end/80422
18126 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18127 predecessors after walking up the insn chain.
18129 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18132 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18133 sizetype type into debug info.
18135 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18138 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18139 unneeded test for TARGET_UPPER_REGS_SF.
18140 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18142 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18145 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18146 instead of gsi_after_labels.
18148 2017-04-18 Jeff Law <law@redhat.com>
18150 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18154 2017-04-13 Jeff Law <law@redhat.com>
18155 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18156 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18158 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18161 * config/avr/avr.c (intl.h): Include it.
18162 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18164 2017-04-18 Martin Liska <mliska@suse.cz>
18166 PR gcov-profile/78783
18167 * gcov-tool.c (gcov_output_files): Validate that destination
18168 file is either removed by the tool or by a user.
18170 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18171 Guy Benyei <guybe@mellanox.com>
18173 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18174 block, and do not negate it, the stored id is already negative.
18176 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18178 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18180 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18183 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18184 masks of options that should be turned off if the VSX vector
18185 options are turned off.
18186 (OTHER_P8_VECTOR_MASKS): Likewise.
18187 (OTHER_VSX_VECTOR_MASKS): Likewise.
18188 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18189 rs6000_disable_incompatible_switches to validate no type switches
18191 (rs6000_incompatible_switch): New function to disallow turning on
18192 other vector options if -mno-vsx, -mno-power8-vector, or
18193 -mno-power9-vector are specified.
18195 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18197 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18199 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18201 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18202 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18203 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18204 (ARG_POINTER_CFA_OFFSET): Likewise.
18206 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18208 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18209 conditions to take advantage of various optimizations.
18211 2017-04-13 Jeff Law <law@redhat.com>
18213 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18214 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18215 (zero_extendsidi2_dext): Likewise.
18217 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18220 * fold-const.c (fold_ternary_loc): Revert
18221 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18224 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18226 PR rtl-optimization/80343
18227 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18228 new scratch pseudo.
18230 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18233 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18234 to ubsan_encode_value.
18236 2017-04-13 Jeff Law <law@redhat.com>
18238 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18239 appearing in DEBUG_INSNs.
18241 2017-04-13 Martin Liska <mliska@suse.cz>
18243 PR gcov-profile/80413
18244 * gcov-io.c (gcov_write_string): Copy to buffer just when
18245 allocated size is greater than zero.
18247 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18250 * dwarf2out.c (decls_for_scope): Ignore declarations of
18251 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18253 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18256 * ipa-visibility.c (non_local_p): Fix typos.
18257 (localize_node): When localizing symbol in same comdat group,
18258 dissolve the group only when we know external symbols are going
18260 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18262 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18264 PR tree-optimization/79390
18265 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18266 order does not result in usable sequence, retry with reversed operand
18272 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18273 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18274 op0 instead of fold_convert_loc (loc, type, arg0).
18276 2017-04-12 Jeff Law <law@redhat.com>
18278 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18279 has a delay slot in the generated code.
18281 * config/cris/cris.md (cris_preferred_reload_class): Return
18282 GENNONACR_REGS rather than GENERAL_REGS.
18284 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18287 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18288 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18289 signedness of the result type.
18291 2017-04-12 Richard Biener <rguenther@suse.de>
18292 Jeff Law <law@redhat.com>
18294 PR tree-optimization/80359
18295 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18296 trim stores to TARGET_MEM_REFs.
18298 2017-04-12 Richard Biener <rguenther@suse.de>
18300 PR tree-optimization/79390
18301 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18302 threading case even more.
18304 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18307 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18308 for quad_address_p for TImode, instead of just not indexed_address.
18310 2017-04-12 Richard Biener <rguenther@suse.de>
18311 Bernd Edlinger <bernd.edlinger@hotmail.de>
18313 PR middle-end/79671
18314 * alias.c (component_uses_parent_alias_set_from): Handle
18315 TYPE_TYPELESS_STORAGE.
18316 (get_alias_set): Likewise.
18317 * tree-core.h (tree_type_common): Add typeless_storage flag.
18318 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18319 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18320 for types containing members with TYPE_TYPELESS_STORAGE.
18321 (place_field): Likewise.
18322 (layout_type): Likewise for ARRAY_TYPE.
18323 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18324 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18325 TYPE_TYPELESS_STORAGE.
18326 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18328 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18331 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18332 first argument to type.
18334 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18338 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18339 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18340 (rs6000_expand_binop_builtin): Likewise.
18341 (rs6000_expand_ternop_builtin): Likewise; also add missing
18342 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18343 vshasigma built-ins.
18344 * doc/extend.texi: Document that vec_xxpermdi's third argument
18345 must be a constant.
18347 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18349 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18350 Use shift_const cost parameter when calculating gain of STV shifts.
18352 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18354 PR rtl-optimization/70478
18355 * lra-constraints.c (process_alt_operands): Check memory for
18356 disfavoring memory insn operand.
18358 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18360 PR middle-end/80100
18361 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18362 left shift in unsigned HOST_WIDE_INT type.
18364 PR rtl-optimization/80385
18365 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18366 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18369 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18370 if they have any depend clauses.
18372 2017-04-11 Martin Liska <mliska@suse.cz>
18375 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18376 * ipa-split.c (split_function): Create a local comdat symbol
18377 if caller is in a comdat group.
18379 2017-04-11 Martin Liska <mliska@suse.cz>
18382 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18385 2017-04-11 Martin Sebor <msebor@redhat.com>
18387 PR middle-end/80364
18388 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18389 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18391 (directive::set_width, directive::set_precision, format_character):
18393 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18396 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18399 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18400 conflict, set target->arch_name instead of target->cpu_name.
18402 2017-04-11 Richard Biener <rguenther@suse.de>
18404 PR tree-optimization/80374
18405 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18406 build_zero_cst, remove fold_convertible_p check again.
18408 2017-04-11 Martin Liska <mliska@suse.cz>
18411 * ubsan.c (instrument_object_size): Do not instrument register
18414 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18417 * config/i386/i386-builtin-types.def
18418 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18419 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18420 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18421 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18422 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18423 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18424 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18425 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18426 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18427 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18428 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18429 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18430 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18431 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18432 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18433 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18434 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18435 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18436 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18437 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18438 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18439 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18440 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18441 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18442 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18443 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18444 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18445 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18446 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18447 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18448 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18449 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18450 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18451 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18452 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18453 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18454 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18455 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18456 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18457 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18458 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18459 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18460 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18461 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18462 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18463 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18465 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18466 flag to second_arg_count, handle 4 argument function type _COUNT
18467 aliases, handle second_arg_count on second argument rather than last.
18469 2017-04-10 Jeff Law <law@redhat.com>
18471 PR tree-optimization/80374
18472 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18473 record anything if we can not convert integer_zero_node to the
18476 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18479 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18480 Enhance special handling given to the TARGET_P9_MINMAX option in
18481 relation to certain other options.
18483 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18485 PR tree-optimization/80153
18486 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18487 remove POINTER_PLUS_EXPR's base part directly, rather than through
18490 2017-04-10 Richard Biener <rguenther@suse.de>
18491 Bin Cheng <bin.cheng@arm.com>
18493 PR tree-optimization/80153
18494 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18495 the first element of pointer type aff_tree. Build result expr in
18497 (add_elt_to_tree): Convert to type unconditionally. Remove other
18498 fold_convert calls.
18499 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18500 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18502 2017-04-10 Richard Biener <rguenther@suse.de>
18504 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18507 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18509 PR rtl-optimization/70478
18510 * lra-constraints.c (curr_small_class_check): New.
18511 (update_and_check_small_class_inputs): New.
18512 (process_alt_operands): Update curr_small_class_check. Disfavor
18513 alternative insn memory operands. Check available regs for small
18516 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18519 * config/mips/mips.opt (-mvirt): Update description.
18520 * doc/invoke.texi (-mvirt): Likewise.
18522 2017-04-10 Richard Biener <rguenther@suse.de>
18524 PR middle-end/80362
18525 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18526 looking for NEGATE_EXPR in -A / -B to A / B folding.
18528 2017-04-10 Martin Liska <mliska@suse.cz>
18530 PR gcov-profile/80224
18531 * gcov.c (print_usage): Fix usage string.
18532 (get_gcov_intermediate_filename): Remove.
18533 (output_gcov_file): Use both for normal and intermediate format.
18534 (generate_results): Do not initialize special file for
18535 intermediate format.
18537 2017-04-10 Richard Biener <rguenther@suse.de>
18539 PR tree-optimization/80304
18540 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18543 2017-04-10 Nathan Sidwell <nathan@acm.org>
18546 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18547 (rs6000_init_builtins): Use it.
18549 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18551 * config/arm/arm.md (<mrc>): Add mode to SET source.
18552 (<mrrc>): Likewise.
18554 2017-04-10 Richard Biener <rguenther@suse.de>
18556 PR middle-end/80344
18557 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18559 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18562 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18563 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18564 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18565 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18566 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18567 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18568 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18569 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18570 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18571 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18572 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18573 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18574 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18575 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18576 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18577 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18578 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18579 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18580 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18581 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18582 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18583 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18584 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18586 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18588 PR rtl-optimization/70478
18589 * lra-constraints.c: Reverse the last patch.
18591 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18593 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18594 Add comment for WCHAR_T.
18596 2017-04-08 Martin Liska <mliska@suse.cz>
18599 2017-04-07 Martin Liska <mliska@suse.cz>
18602 * ipa-split.c (split_function): Add function part to a same comdat
18605 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18608 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18610 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18612 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18613 (vec_pairing): ...to this.
18614 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18615 (rs6000_sched_init): Adjust for name change.
18616 (struct rs6000_sched_context): Likewise.
18617 (rs6000_init_sched_context): Likewise.
18618 (rs6000_set_sched_context): Likewise.
18620 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18626 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18628 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18629 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18630 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18632 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18634 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18636 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18638 PR rtl-optimization/70703
18639 * ira-color.c (update_conflict_hard_regno_costs): Use
18640 int64_t instead of HOST_WIDE_INT.
18642 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18644 PR rtl-optimization/70478
18645 * lra-constraints.c (process_alt_operands): Disfavor alternative
18646 insn memory operands.
18648 2017-04-07 Jeff Law <law@redhat.com>
18650 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18651 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18653 2017-04-07 Martin Liska <mliska@suse.cz>
18656 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18657 Show error message instead of an ICE.
18659 2017-04-07 Martin Liska <mliska@suse.cz>
18662 * ipa-split.c (split_function): Add function part to a same comdat
18665 2017-04-07 Richard Biener <rguenther@suse.de>
18667 PR middle-end/80341
18668 * tree.c (get_unwidened): Also handle ! for_type case for
18670 * convert.c (do_narrow): Split out from ...
18671 (convert_to_integer_1): ... here. Do not pass final truncation
18672 type to get_unwidened for TRUNC_DIV_EXPR.
18674 2017-04-07 Richard Biener <rguenther@suse.de>
18676 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18678 (aff_combination_const): Adjust.
18679 (aff_combination_scale): Likewise.
18680 (aff_combination_add_elt): Likewise.
18681 (aff_combination_add_cst): Likewise.
18682 (aff_combination_convert): Likewise.
18683 (add_elt_to_tree): Likewise. Remove unused argument.
18684 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18686 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18688 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18690 * config/arm/arm.c (arm_default_short_enums): Use
18691 ARM_DEFAULT_SHORT_ENUMS.
18692 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18694 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18697 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18698 members with redundant out-of-class redeclaration.
18700 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18703 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18704 * config/i386/i386.md (*zero_extendsidi2):
18705 Add (?*x,*x) and (?*v,*v) alternatives.
18707 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18710 * config/i386/i386.c (ix86_expand_builtin)
18711 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18712 mode from insn data. Convert operands to insn operand mode.
18713 Copy operands that don't satisfy insn predicate to a register.
18715 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18717 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18720 2017-04-06 Richard Biener <rguenther@suse.de>
18722 PR tree-optimization/80334
18723 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18724 preserve alignment of accesses.
18726 2017-04-06 Richard Biener <rguenther@suse.de>
18728 PR tree-optimization/80262
18729 * tree-sra.c (build_ref_for_offset): Preserve address-space
18731 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18732 Drop useless address-space information on MEM_REF offsets.
18734 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18736 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18738 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18740 PR rtl-optimization/70703
18741 * ira-color.c (update_conflict_hard_regno_costs): Use
18742 HOST_WIDE_INT instead of long.
18744 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18747 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18748 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18750 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18751 for x86_64 target. Handle -m3dnowa option.
18753 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18755 PR rtl-optimization/70703
18756 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18757 (update_conflict_hard_regno_costs): Use long instead of unsigned
18758 arithmetic for cost calculation.
18760 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18761 Bernd Edlinger <bernd.edlinger@hotmail.de>
18764 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18767 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18770 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18771 ptr_mode with Pmode throughout.
18772 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18773 into probe_stack_range and use DImode.
18775 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18778 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18779 call_eh_return is true.
18781 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18783 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18784 Initialize last_match_fntype_index.
18786 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18789 * tree-nvr.c: Include internal-fn.h.
18790 (pass_return_slot::execute): Ignore internal calls without
18793 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18794 Richard Biener <rguenther@suse.de>
18797 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18798 captures used multiple times, except for the last use.
18799 * generic-match-head.c: Include gimplify.h.
18801 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18803 PR tree-optimization/79390
18804 * target.h (struct noce_if_info): Declare.
18805 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18806 * target.def (noce_conversion_profitable_p): New target hook.
18807 * ifcvt.h (struct noce_if_info): New type, moved from ...
18808 * ifcvt.c (struct noce_if_info): ... here.
18809 (noce_conversion_profitable_p): Renamed to ...
18810 (default_noce_conversion_profitable_p): ... this. No longer
18812 (noce_try_store_flag_constants, noce_try_addcc,
18813 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18814 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18815 instead of noce_conversion_profitable_p.
18816 * config/i386/i386.c: Include ifcvt.h.
18817 (ix86_option_override_internal): Don't override
18818 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18819 (ix86_noce_conversion_profitable_p): New function.
18820 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18821 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18822 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18823 * doc/tm.texi: Regenerated.
18825 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18827 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18830 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18833 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18834 instructions for small multiply cores.
18836 2017-04-04 Jeff Law <law@redhat.com>
18838 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18840 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18841 that are not set by the loop over the elements.
18843 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18846 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18847 int mode, convert_modes it to mode as unsigned, otherwise use
18848 lowpart_subreg to mode rather than SImode.
18849 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18850 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18851 Use DImode instead of SImode for the shift count operand.
18852 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18855 2017-04-04 Richard Biener <rguenther@suse.de>
18857 PR middle-end/80281
18858 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18859 arithmetic done for the negate or the plus. Simplify.
18860 (A - (-B) -> A + B): Likewise.
18861 * fold-const.c (split_tree): Make sure to not negate pointers.
18863 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18865 PR rtl-optimization/60818
18866 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18867 a compare of comparisons with the thing compared if this results
18868 in a different machine mode.
18870 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18872 * alias.c (base_alias_check): Fix typo in comment.
18873 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18874 * cgraphunit.c (symbol_table::compile): Likewise.
18875 * collect2.c (maybe_run_lto_and_relink): Likewise.
18876 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18877 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18878 * config/avr/avr.c (avr_map_op_t): Likewise.
18879 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18880 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18881 * config/epiphany/epiphany.md (movcc): Likewise.
18882 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18883 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18885 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18886 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18887 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18888 * config/sh/sh.c (sh_rtx_costs): Likewise.
18889 * fold-const.c (fold_truth_andor): Likewise.
18890 * genautomata.c (collapse_flag): Likewise.
18891 * gengtype.h (struct type::u::s): Likewise.
18892 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18893 * input.c (FORMAT_AMOUNT): Likewise.
18894 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18895 (known_aggs_to_agg_replacement_list): Likewise.
18896 * ipa-inline-analysis.c: Likewise.
18897 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18898 * ipa-polymorphic-call.c
18899 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18900 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18901 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18903 * modulo-sched.c (apply_reg_moves): Likewise.
18904 * omp-expand.c (build_omp_regions_1): Likewise.
18905 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18906 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18907 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18908 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18909 * value-prof.c: Likewise.
18910 * var-tracking.c (val_reset): Likewise.
18912 2017-04-03 Richard Biener <rguenther@suse.de>
18914 PR tree-optimization/80275
18915 * fold-const.c (split_address_to_core_and_offset): Handle
18918 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18920 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18921 descriptors is at least equal to that of functions.
18923 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18925 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18927 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18930 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18931 (mov<IMOD4:mode>): New expander.
18932 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18934 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18936 PR rtl-optimization/79405
18937 * fwprop.c (propagations_left): New variable.
18938 (forward_propagate_into): Decrement it.
18939 (fwprop_init): Initialize it.
18940 (fw_prop): If the variable has reached zero, stop propagating.
18941 (fwprop_addr): Ditto.
18943 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18946 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18947 a FUNCTION_DECL, pass it as decl instead of origin to
18950 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18952 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18955 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18958 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18959 TARGET_VSX_SMALL_INTEGER.
18961 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18963 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18964 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18966 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18968 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18969 extraction from odd-numbered MSA register.
18971 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18973 PR middle-end/80173
18974 * expmed.c (store_bit_field_1): Don't attempt to create
18975 a word subreg out of hard registers wider than word if they
18976 have HARD_REGNO_NREGS of 1 for their mode.
18978 PR middle-end/80163
18979 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18980 conversions to integer types wider than word and pointer.
18983 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18984 (rtx_equal_for_cselib_p): Pass 0 to it.
18985 * cselib.c (cselib_hasher::equal): Likewise.
18986 (rtx_equal_for_cselib_1): Add depth argument. If depth
18987 is 128, don't look up VALUE locs and punt. Increment
18988 depth in recursive calls when walking VALUE locs.
18990 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18992 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18993 (make_gcov_file_name): Use the canonical path name for generating
18995 (read_line): Fix handling of files with ascii null bytes.
18997 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18999 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19000 to initialise a vector register instead
19001 of using a const_int.
19003 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19005 PR translation/80189
19006 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19007 diagnostic messages.
19009 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19012 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19013 (dfp_diex_<mode>): Update mode of operand 1.
19014 * doc/extend.texi (dxex, dxexq): Document change to return type.
19015 (diex, diexq): Document change to argument type.
19017 2017-03-30 Martin Jambor <mjambor@suse.cz>
19020 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19021 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19022 it reflects the signature changes performed at the callee side.
19023 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19024 to cgraph_build_function_type_skip_args.
19025 (build_function_decl_skip_args): Adjust call to the above function.
19027 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19030 * config/i386/sse.md
19031 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19032 register as dest whenever it is a MEM not rtx_equal_p to the
19033 corresponding dup operand, and when forcing into reg move the
19034 reg into the memory afterwards.
19035 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19036 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19037 for the force_reg mode.
19038 (avx512vl_vextractf128<mode>): Use register as dest either
19039 always when a MEM, or when it is a MEM not rtx_equal_p to the
19040 corresponding dup operand, or even not when it is a CONST_VECTOR
19041 depending on the mode and lo vs. hi.
19042 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19044 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19045 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19046 Likewise. Require that operands[2] is even.
19047 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19048 Remove extraneous parens. Require that operands[2] is a multiple
19050 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19051 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19052 disallow memory then.
19054 2017-03-30 Richard Biener <rguenther@suse.de>
19056 PR tree-optimization/77498
19057 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19058 to non-constants over backedges.
19060 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19062 PR rtl-optimization/80233
19063 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19064 as last_combined_insn. Do not test for BARRIER_P separately.
19066 2017-03-29 Andreas Schwab <schwab@suse.de>
19069 * calls.c (prepare_call_address): Convert funexp to Pmode before
19070 copying to temp reg.
19072 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19074 PR tree-optimization/80158
19075 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19076 Handle possible future case of more than one alternate
19078 (replace_rhs_if_not_dup): Likewise.
19079 (replace_one_candidate): Likewise.
19081 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19083 PR rtl-optimization/80193
19084 * ira.c (ira): Do not check allocation for LRA.
19086 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19088 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19089 (nvptx_output_simt_exit): Declare.
19090 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19091 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19092 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19093 (nvptx_declare_function_name): ...here. Emit declaration of local
19094 memory space buffer for omp_simt_enter insn.
19095 (nvptx_output_unisimt_switch): New.
19096 (nvptx_output_softstack_switch): New.
19097 (nvptx_output_simt_enter): New.
19098 (nvptx_output_simt_exit): New.
19099 * config/nvptx/nvptx.h (struct machine_function): New fields
19100 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19101 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19102 (UNSPECV_SIMT_EXIT): Ditto.
19103 (omp_simt_enter_insn): New insn.
19104 (omp_simt_enter): New expansion.
19105 (omp_simt_exit): New insn.
19106 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19108 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19109 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19110 (expand_GOMP_SIMT_EXIT): New.
19111 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19112 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19113 (GOMP_SIMT_EXIT): Ditto.
19114 * target-insns.def (omp_simt_enter): New insn.
19115 (omp_simt_exit): Ditto.
19116 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19118 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19119 (lower_rec_input_clauses): Likewise.
19120 (lower_lastprivate_clauses): Handle SIMT privatization.
19122 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19123 (ompdevlow_adjust_simt_enter): New.
19124 (find_simtpriv_var_op): New.
19125 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19126 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19128 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19129 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19130 (copy_decl_for_dup_finish): Ditto.
19132 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19134 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19137 * config/i386/i386.c (ix86_option_override_internal): Always
19138 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19140 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19142 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19144 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19146 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19147 mark new edge's irreducible flag accordign to it.
19148 (vect_do_peeling): Check loop preheader edge's irreducible flag
19149 and pass it to function slpeel_add_loop_guard.
19151 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19153 PR tree-optimization/80218
19154 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19155 Update block frequencies and counts.
19157 2017-03-28 Richard Biener <rguenther@suse.de>
19159 PR tree-optimization/78644
19160 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19161 of a simplification result we may not use it at all.
19163 2017-03-28 Richard Biener <rguenther@suse.de>
19166 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19167 without arguments, generate default definition of a SSA name.
19169 2017-03-28 Richard Biener <rguenther@suse.de>
19171 PR middle-end/80222
19172 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19173 TYPE_REF_CAN_ALIAS_ALL references.
19174 * fold-const.c (fold_indirect_ref_1): Likewise.
19176 2017-03-28 Martin Liska <mliska@suse.cz>
19179 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19180 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19182 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19183 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19185 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19186 (EXTRA_SPECS): Define.
19187 (SUBTARGET_EXTRA_SPECS): Likewise.
19188 (SUBTARGET_CPP_SPEC): Likewise.
19189 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19190 SUBTARGET_EXTRA_SPECS.
19191 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19193 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19195 * config/arc/simdext.md (vst64_insn): Update pattern.
19196 (vld32wh_insn): Likewise.
19197 (vld32wl_insn): Likewise.
19198 (vld64_insn): Likewise.
19199 (vld32_insn): Likewise.
19201 2017-03-28 Marek Polacek <polacek@redhat.com>
19204 * fold-const.c (fold_comparison): Use protected_set_expr_location
19205 instead of SET_EXPR_LOCATION.
19207 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19209 * tree.c (add_expr): Avoid name lookup warning.
19211 2017-03-27 Jeff Law <law@redhat.com>
19213 PR tree-optimization/80216
19214 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19215 function name. Limit recursion depth.
19216 (record_temporary_equivalences): Corresponding changes.
19218 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19220 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19223 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19226 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19228 * cfgcleanup.c (reg_note_cfa_p): New array.
19229 (insns_have_identical_cfa_notes): New function.
19230 (old_insns_match_p): Don't cross-jump in between /f
19231 and non-/f instructions. If both i1 and i2 are frame related,
19232 verify all CFA notes, their order and content.
19234 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19237 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19238 HImode and SImode with zero extend to DImode to one insn.
19239 (bswap<mode>2_extenddi): Likewise.
19240 (bswapsi2_extenddi): Likewise.
19241 (bswaphi2_extendsi): Likewise.
19242 (bswaphi2): Combine bswap HImode and SImode into one insn.
19243 Separate memory insns from swapping register.
19244 (bswapsi2): Likewise.
19245 (bswap<mode>2): Likewise.
19246 (bswaphi2_internal): Delete, no longer used.
19247 (bswapsi2_internal): Likewise.
19248 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19249 store, and gpr<-gpr swap insns.
19250 (bswap<mode>2_store): Likewise.
19251 (bswaphi2_reg): Register only splitter, combine with the splitter.
19252 (bswaphi2 splitter): Likewise.
19253 (bswapsi2_reg): Likewise.
19254 (bswapsi2 splitter): Likewise.
19255 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19256 the insns into load, store, and register/register insns.
19257 (bswapdi2_ldbrx): Likewise.
19258 (bswapdi2_load): Likewise.
19259 (bswapdi2_store): Likewise.
19260 (bswapdi2_reg): Likewise.
19262 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19264 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19265 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19267 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19270 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19273 special handling for target option conflicts between dform
19274 options (-mpower9-dform, -mpower9-dform-vector,
19275 -mpower9-dform-scalar) and -mno-direct-move.
19277 2017-03-27 Richard Biener <rguenther@suse.de>
19279 PR tree-optimization/80181
19280 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19282 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19284 * config/arc/predicates.md (move_double_src_operand): Replace the
19285 call to move_double_src_operand with a call to address_operand.
19287 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19289 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19290 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19291 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19293 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19295 * config/arc/predicates.md (long_immediate_loadstore_operand):
19296 Consider scaled addresses cases.
19298 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19300 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19301 restored when in interrupt.
19302 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19303 doesn't have delay slot.
19305 2017-03-27 Richard Biener <rguenther@suse.de>
19308 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19309 inlined thunk clones.
19311 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19314 * asan.c (instrument_derefs): Copy over last operand from
19315 original COMPONENT_REF to the new COMPONENT_REF with
19316 DECL_BIT_FIELD_REPRESENTATIVE.
19317 * ubsan.c (instrument_object_size): Likewise.
19319 2017-03-27 Richard Biener <rguenther@suse.de>
19321 PR tree-optimization/80170
19322 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19323 sure DR/SCEV didnt fold in constants we do not see when looking
19324 at the reference base alignment.
19326 2017-03-27 Richard Biener <rguenther@suse.de>
19328 PR middle-end/80171
19329 * gimple-fold.c (fold_ctor_reference): Properly guard against
19330 NULL return value from canonicalize_constructor_val.
19332 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19335 * config/i386/i386.c (ix86_expand_builtin)
19336 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19337 flags reg setting and flags reg using instructions.
19338 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19339 clobbering instructions to zero extend op2.
19341 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19343 * doc/install.texi (Configuration) <--with-aix-soname>:
19344 Update link to AIX ld.
19346 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19348 PR rtl-optimization/80160
19349 PR rtl-optimization/80159
19350 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19351 reg_alternate_class into account.
19353 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19356 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19357 to consider in curr_insn_transform.
19359 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19361 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19362 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19363 and emit_mode_inner.
19365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19367 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19368 argument to the overloaded builtin variants. Use the new flag to
19369 deprecate certain builtin variants.
19370 * config/s390/s390-builtin-types.def: Add new builtin types.
19371 * config/s390/s390-builtins.h: Support new flags field for
19372 overloaded builtins.
19373 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19374 (s390_macro_to_expand): Enable vector float data type.
19375 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19376 builtins by incrementing the __VEC__ version number.
19377 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19379 (s390_resolve_overloaded_builtin): Emit error messages depending
19380 on the builtin flags.
19381 * config/s390/s390.c (s390_expand_builtin): Support additional
19382 flags argument. Change error message to match the messages
19383 emitted in s390-c.c.
19384 * config/s390/s390.md: New UNSPEC_* constants.
19385 (op_type): Add new instruction types.
19386 * config/s390/vecintrin.h: Add new builtins and test data class
19388 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19389 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19390 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19391 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19392 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19393 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19395 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19396 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19397 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19398 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19400 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19401 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19402 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19403 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19405 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19406 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19407 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19408 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19409 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19410 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19411 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19413 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19414 ("vec_scatter_element<V_HW_4:mode>_DI")
19415 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19416 ("vec_fpint<mode>", "vflls")
19417 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19418 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19419 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19420 ("*vec_cmphe<mode>_cc"): ... these.
19422 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19423 mode constant instead of magic value.
19425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19427 * config/s390/s390.c (s390_expand_vec_compare): Support other
19428 vector floating point modes than just V2DF.
19429 (s390_expand_vcond): Likewise.
19430 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19431 (s390_cannot_change_mode_class): Prevent mode changes between TF
19432 and V1TF in vector registers.
19433 * config/s390/s390.md (DF, SF): New mode attributes.
19434 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19435 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19436 SFmode support for VRs.
19437 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19439 (VFT, VF_HW): New mode iterators.
19440 (vw, sdx): New mode attributes.
19441 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19442 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19443 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19444 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19445 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19446 also the new vector floating point modes. Renaming to ...
19448 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19449 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19450 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19451 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19452 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19453 ("vec_unordered<mode>"): ... these.
19455 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19456 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19457 ("*vec_extendv2df"): New insn definitions.
19459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19461 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19462 ("mulditi3_2", "*muldi3_sign"): New patterns.
19463 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19464 rename the pattern definition.
19466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19468 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19470 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19474 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19475 instruction if possible.
19476 * config/s390/vector.md (vec_halfnumelts): New mode
19478 ("*vec_vllezlf<mode>"): New pattern.
19480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19482 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19483 ("popcountv4si2", "popcountv2di2"): Rename to ...
19484 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19485 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19487 ("popcount<mode>2_vxe"): New pattern.
19489 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19491 * common/config/s390/s390-common.c (processor_flags_table): Add
19493 * config.gcc: Add arch12.
19494 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19495 Default to arch12 for unknown CPU model numbers.
19496 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19497 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19498 PROCESSOR_max sanity check.
19499 * config/s390/s390-opts.h (enum processor_type): Add
19501 * config/s390/s390.c (processor_table): Add arch12.
19502 (s390_expand_builtin): Add check for B_VXE flag.
19503 (s390_issue_rate): Add PROCESSOR_ARCH12.
19504 (s390_get_sched_attrmask): Likewise.
19505 (s390_get_unit_mask): Likewise.
19506 (s390_sched_score): Enable z13 scheduling for arch12.
19507 (s390_sched_reorder): Likewise.
19508 (s390_sched_variable_issue): Likewise.
19509 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19511 (s390_tune_attr): Use z13 scheduling also for arch12.
19512 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19513 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19514 (TARGET_VXE_P): New macros.
19515 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19516 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19517 * config/s390/s390.opt: Add arch12 as processor_type.
19519 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19521 * config/s390/s390.md
19522 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19523 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19524 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19526 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19527 Rename expanders to ...
19529 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19530 ("fixuns_truncdddi2_emu"): ... these.
19532 ("fixuns_trunc<mode>si2_emu"): New expander.
19534 ("*fixuns_truncdfdi2_z13"): Rename to ...
19535 ("*fixuns_truncdfdi2_vx"): ... this.
19537 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19539 * config/s390/2964.md: Remove the single element vector compare
19540 instructions which are no longer used.
19541 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19543 (s390_canonicalize_comparison): Remove handling of DFmode
19545 (s390_expand_vec_compare_scalar): Remove function.
19546 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19547 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19549 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19551 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19553 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19555 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19556 will anyway by matched by mov<mode>_64dfp.
19558 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19560 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19561 vlef/vstef. Add missing operand to vleif.
19563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19565 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19566 pair for all vector types with 64 bit elements.
19567 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19568 * config/s390/vector.md (V_HW_64): ... here.
19569 (V_128_NOSINGLE): New mode iterator.
19570 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19571 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19572 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19573 ("*vec_load_pairv2di"): Change to ...
19574 ("*vec_load_pair<mode>"): ... this one.
19576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19578 * config/s390/constraints.md: Add comments.
19579 (jKK): Reject element sizes > 8 bytes.
19580 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19582 * config/s390/s390.md: Add the s_operand checks formerly in
19583 s390_split_ok_p to various splitters where they are still
19585 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19586 for 128 bit vectors. Plus two splitters.
19588 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19590 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19596 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19597 error if the boundary argument is not constant.
19599 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19601 PR rtl-optimization/80112
19602 * loop-doloop.c (doloop_condition_get): Don't check condition
19603 if cmp isn't SET with IF_THEN_ELSE src.
19605 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19607 PR tree-optimization/80158
19608 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19609 replacing a candidate statement, also replace it for the
19610 candidate's alternate interpretation.
19611 (replace_rhs_if_not_dup): Likewise.
19612 (replace_one_candidate): Likewise.
19614 2017-03-24 Richard Biener <rguenther@suse.de>
19616 PR tree-optimization/80167
19617 * graphite-isl-ast-to-gimple.c
19618 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19620 (translate_isl_ast_to_gimple::get_rename): Likewise.
19622 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19625 handling of certain combinations of target options, including the
19626 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19627 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19629 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19632 * config/arm/arm.md (*load_multiple): Add reload_completed to
19633 matching condition.
19635 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19636 Richard Biener <rguenth@suse.de>
19638 PR tree-optimization/79908
19639 PR tree-optimization/80136
19640 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19641 been cast away, gimplify_and_add suffices.
19643 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19645 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19647 2017-03-23 Richard Biener <rguenther@suse.de>
19649 PR tree-optimization/80032
19650 * gimplify.c (gimple_push_cleanup): Forced unconditional
19651 cleanups still have to go to the conditional_cleanups
19654 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19656 PR tree-optimization/80072
19657 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19659 (next_operand_entry_id): Change type to unsigned int.
19660 (sort_by_operand_rank): Make sure to return the right return value
19661 even if unsigned fields are bigger than INT_MAX.
19662 (struct oecount): Change cnt and id type to unsigned int.
19663 (oecount_hasher::equal): Formatting fix.
19664 (oecount_cmp): Make sure to return the right return value
19665 even if unsigned fields are bigger than INT_MAX.
19666 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19669 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19670 TREE_READONLY on result if writing it more than once.
19673 * doc/invoke.texi (-fsanitize=thread): Document that with
19674 -fnon-call-exceptions atomics are not able to throw
19678 * tsan.c: Include tree-eh.h.
19679 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19680 maybe_clean_or_replace_eh_stmt where needed.
19681 (instrument_memory_accesses): Add cfg_changed argument.
19682 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19683 if it returned true.
19684 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19686 PR rtl-optimization/63191
19687 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19688 wrapper function, moved the whole old content into ...
19689 (ix86_delegitimize_address_1): ... this. New inline function.
19690 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19691 true as last argument instead of ix86_delegitimize_address.
19693 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19695 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19696 cortexa57_branch_cost.
19698 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19700 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19702 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19705 * doc/md.texi (Constraints): Document wA constraint.
19706 * config/rs6000/constraints.md (wA): New.
19707 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19708 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19709 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19710 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19712 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19715 * gimplify.c (is_oacc_declared): New function.
19716 (oacc_default_clause): Use it to set default flags for acc declared
19717 variables inside parallel regions.
19718 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19719 declared variables.
19720 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19721 declare attribute to any decl as necessary.
19723 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19726 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19727 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19728 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19729 (arm_arch_lpae): This.
19730 * config/arm/arm.c (arm_arch7ve): Rename into ...
19731 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19732 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19735 2017-03-22 Martin Liska <mliska@suse.cz>
19738 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19739 error message instead of an ICE.
19741 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19743 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19745 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19747 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19749 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19752 2017-03-21 Martin Sebor <msebor@redhat.com>
19754 * doc/extend.texi: Use "cannot" instead of "can't."
19755 * doc/hostconfig.texi: Same.
19756 * doc/install.texi: Same.
19757 * doc/invoke.texi: Same.
19758 * doc/loop.texi: Same.
19759 * doc/md.texi: Same.
19760 * doc/objc.texi: Same.
19761 * doc/rtl.texi: Same.
19762 * doc/tm.texi: Same.
19763 * doc/tm.texi.in: Same.
19764 * doc/trouble.texi: Same.
19766 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19769 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19770 (collect_checksum_attributes): Set it.
19771 (die_checksum_ordered): Use it.
19773 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19775 PR tree-optimization/79908
19776 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19777 change: For a VA_ARG whose LHS has been cast away, use
19778 force_gimple_operand to construct the side effects.
19780 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19782 PR translation/80001
19783 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19784 more amenable to translation.
19785 (oacc_loop_auto_partitions): Likewise.
19787 2017-03-21 Marek Polacek <polacek@redhat.com>
19788 Martin Sebor <msebor@redhat.com>
19790 PR tree-optimization/80109
19791 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19792 on INTEGRAL_TYPE_P.
19794 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19795 Segher Boessenkool <segher@kernel.crashing.org>
19798 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19799 check reg_used_between_p between insn and one of succ or succ2
19800 depending on if succ is artificial insn not inserted into insn
19803 2017-03-21 Martin Liska <mliska@suse.cz>
19805 PR gcov-profile/80081
19806 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19807 * doc/gcc.texi: Include gcov-dump stuff.
19808 * doc/gcov-dump.texi: New file.
19810 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19812 PR rtl-optimization/79150
19813 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19814 conditional jump, if the jump is the last insn of the loop.
19816 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19817 Richard Biener <rguenth@suse.de>
19819 PR tree-optimization/79908
19820 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19821 been cast away, use force_gimple_operand to construct the side
19824 2017-03-21 Martin Liska <mliska@suse.cz>
19826 PR libfortran/79956
19827 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19830 2017-03-21 Brad Spengler <spender@grsecurity.net>
19833 * plugin.c (htab_hash_plugin): New function.
19834 (add_new_plugin): Use it and adjust.
19835 (parse_plugin_arg_opt): Adjust.
19836 (init_one_plugin): Likewise.
19838 2017-03-21 Richard Biener <rguenther@suse.de>
19840 PR tree-optimization/80032
19841 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19842 if set force the cleanup to happen unconditionally.
19843 (gimplify_target_expr): Push inserted clobbers with force_uncond
19844 to avoid them being removed by control-dependent DCE.
19846 2017-03-21 Richard Biener <rguenther@suse.de>
19848 PR tree-optimization/80122
19849 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19850 va_arg_pack_len when the inlined call stmt requires pack
19852 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19854 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19857 * tsan.c (instrument_builtin_call): If the memory model argument
19858 is not a constant, assume it is valid.
19861 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19864 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19866 PR rtl-optimization/79910
19867 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19868 if its dest is used by an insn before I2 (other than the combined
19869 insns themselves, which are properly handled already).
19871 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19874 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19876 * combine.c (record_used_regs): New static function.
19877 (try_combine): Handle situations where there is an additional
19878 instruction between I2 and I3 which needs to have a LOG_LINK
19882 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19884 * combine.c (try_combine): Delete redundant i1 test. Call
19885 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19887 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19890 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19891 alternatives 13/14.
19893 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19895 PR tree-optimization/80054
19896 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19897 the optimization if a PHI or any of its arguments is not dominated
19898 by the candidate's basis. Use gphi* rather than gimple* as
19900 (replace_profitable_candidates): Clean up a gimple* variable that
19901 should be a gphi* variable.
19903 2017-03-20 Martin Sebor <msebor@redhat.com>
19906 * doc/extend.texi (attribute constructor): Document present limitation.
19908 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19911 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19912 __POWER9_VECTOR__ #ifdef control, change template definition to
19913 use Power9-specific built-in function.
19914 (vec_any_eq): Likewise.
19915 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19916 to control outcomes from this test.
19917 (vector_ae_<mode>p): For VEC_F modes, likewise.
19919 2017-03-20 Ian Lance Taylor <iant@google.com>
19921 * config/i386/i386.c (ix86_function_regparm): Save an extra
19922 register for -fsplit-stack with DECL_STATIC_CHAIN.
19924 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19927 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19928 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19930 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19932 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19934 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19937 2017-03-20 Marek Polacek <polacek@redhat.com>
19940 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19942 2017-03-20 Richard Biener <rguenther@suse.de>
19944 PR tree-optimization/80113
19945 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19946 allocate extra SSA name for PHI def.
19947 (add_close_phis_to_outer_loops): Likewise.
19948 (add_close_phis_to_merge_points): Likewise.
19949 (copy_loop_close_phi_args): Likewise.
19950 (copy_cond_phi_nodes): Likewise.
19952 2017-03-20 Martin Liska <mliska@suse.cz>
19954 PR middle-end/79753
19955 * tree-chkp.c (chkp_build_returned_bound): Do not build
19956 returned bounds for a LHS that's not a BOUNDED_P type.
19958 2017-03-20 Martin Liska <mliska@suse.cz>
19962 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19963 COMPLEX_CST and VECTOR_CST.
19965 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19968 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19969 target operand. A new splitter adds the clobber statement in case
19970 the target operand is dead anyway.
19972 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19974 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19975 to age-old versions of binutils and glibc.
19977 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19979 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19981 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19983 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19985 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19987 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19988 requirement for binutils 2.13.
19990 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19992 * combine.c (try_combine): Delete redundant i1 test. Call
19993 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19995 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19997 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19998 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20000 <riscv64-*-elf>: Re-arrange section
20001 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20002 <riscv32-*-linux>: Likewise.
20003 <riscv64-*-elf>: Likewise
20004 <riscv64-*-linux>: Likewise.
20006 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20009 * aarch64.opt(verbose-cost-dump): Fix typo.
20011 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20014 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20015 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20017 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20019 * reload.c (find_reloads): When reloading a nonoffsettable address,
20020 use RELOAD_OTHER for it and its address reloads.
20022 PR rtl-optimization/79910
20023 * combine.c (record_used_regs): New static function.
20024 (try_combine): Handle situations where there is an additional
20025 instruction between I2 and I3 which needs to have a LOG_LINK
20028 2017-03-17 Jeff Law <law@redhat.com>
20030 PR tree-optimization/71437
20031 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20032 conditional in the hash table first.
20033 (vrp_dom_walker::before_dom_children): Extract condition from
20034 ASSERT_EXPR. Record condition, its inverion and any implied
20035 conditions as well.
20037 2017-03-17 Marek Polacek <polacek@redhat.com>
20038 Markus Trippelsdorf <markus@trippelsdorf.de>
20040 PR tree-optimization/80079
20041 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20044 2017-03-17 Richard Biener <rguenther@suse.de>
20046 PR middle-end/80075
20047 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20048 Properly verify the LHS before the RHS possibly claims to be
20050 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20053 2017-03-17 Martin Jambor <mjambor@suse.cz>
20055 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20056 (List of -O2 options): Likewise.
20057 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20060 2017-03-17 Tom de Vries <tom@codesourcery.com>
20062 * gcov-dump.c (print_usage): Print bug_report_url.
20064 2017-03-17 Richard Biener <rguenther@suse.de>
20066 PR middle-end/80050
20067 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20068 (parser::peek): Likewise.
20070 2017-03-17 Richard Biener <rguenther@suse.de>
20072 PR tree-optimization/80048
20073 * sese.c (free_sese_info): Properly release rename_map and
20074 copied_bb_map elements.
20076 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20078 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20079 Add linked-list forward and backlinks. Insert on
20080 construction, remove on destruction.
20081 (class pass_store_merging): Add m_stores_head field.
20082 (pass_store_merging::terminate_and_process_all_chains):
20083 Iterate over m_stores_head list.
20084 (pass_store_merging::terminate_all_aliasing_chains):
20086 (pass_store_merging::execute): Check for debug stmts first.
20087 Push new chains onto the m_stores_head stack.
20089 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20092 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20093 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20094 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20096 2017-03-16 Jeff Law <law@redhat.com>
20098 PR tree-optimization/71437
20099 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20100 member function. Implementation moved into after_dom_children
20101 member function and into the threader's thread_outgoing_edges
20103 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20104 some code into new thread_outgoing_edges.
20105 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20106 definition. Simplify marker handling (do it here). Assume we always
20107 have the available expression and the const/copies tables.
20108 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20110 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20111 * tree-vrp.c (equiv_stack): No longer file scoped.
20112 (vrp_dom_walker): New class.
20113 (vrp_dom_walker::before_dom_children): New member function.
20114 (vrp_dom_walker::after_dom_children): Likewise.
20115 (identify_jump_threads): Setup domwalker. Use it rather than
20116 walking edges in a random order by hand. Simplify setup/finalization.
20117 (finalize_jump_threads): Remove.
20118 (vrp_finalize): Do not call identify_jump_threads here.
20119 (execute_vrp): Do it here instead and call thread_through_all_blocks
20122 PR tree-optimization/71437
20123 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20125 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20127 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20128 (dom_opt_dom_walker::thread_across_edge): Remove
20129 handle_dominating_asserts argument. All callers changed.
20130 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20131 changes. Remove calls to lhs_of_dominating_assert. Other
20132 uses of handle_dominating_asserts turn into unconditional code
20133 (simplify_control_stmt_condition_1): Likewise.
20134 (simplify_control_stmt_condition): Likewise.
20135 (thread_through_normal_block, thread_across_edge): Likewise.
20136 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20137 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20138 object if it is not an SSA_NAME.
20139 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20140 before calling into the VRP specific simplifiers.
20141 (identify_jump_threads): Remove handle_dominating_asserts
20144 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20147 * tree-diagnostic.c (default_tree_printer): No longer static.
20148 * tree-diagnostic.h (default_tree_printer): New prototype.
20150 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20152 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20153 Change ins into fmov.
20155 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20157 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20158 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20159 Use h_con constraint for operand 1.
20160 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20161 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20163 2017-03-15 Jeff Law <law@redhat.com>
20165 PR tree-optimization/71437
20166 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20167 (record_temporary_equivalences): Use it.
20169 PR tree-optimization/71437
20170 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20171 tree-ssa-scopedtables.
20172 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20173 (record_conditions, record_cond, vuse_eq): Likewise.
20174 (record_edge_info): Adjust to API tweak of record_conditions.
20175 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20176 (record_temporary_equivalences, optimize_stmt): Likewise.
20177 (eliminate_redundant_computations): Likewise.
20178 (record_equivalences_from_stmt): Likewise.
20179 * tree-ssa-scopedtables.c: Include options.h and params.h.
20180 (vuse_eq): New function, moved from tree-ssa-dom.c
20181 (build_and_record_new_cond): Likewise.
20182 (record_conditions): Likewise. Accept vector of conditions rather
20183 than edge_equivalence structure for first argument.
20184 for the first argument.
20185 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20186 from tree-ssa-dom.c.
20187 (avail_exprs_stack::record_cond): Likewise.
20188 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20189 from tree-ssa-dom.c.
20190 (avail_exprs_stack): Add new member functions lookup_avail_expr
20192 (record_conditions): Declare.
20194 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20197 * lra-constraints.c (process_alt_operands): Increase reject for
20198 reloading an input/output operand.
20200 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20203 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20204 insns to convert from signed/unsigned char/short to IEEE 128-bit
20206 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20208 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20211 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20212 subreg of inner mode for values already in registers.
20214 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20216 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20217 iteration reg is used after the loop.
20219 2017-03-14 Martin Sebor <msebor@redhat.com>
20221 PR tree-optimization/79800
20222 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20223 precision in negative-positive range.
20224 (format_floating): Call non-const overload with adjusted precision.
20226 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20229 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20232 2017-03-14 Martin Sebor <msebor@redhat.com>
20234 PR middle-end/80020
20235 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20236 * builtins.def (aligned_alloc): Use it.
20239 * Makefile.in (GTFILES): Add calls.c.
20240 * calls.c: Include "gt-calls.h".
20242 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20244 PR rtl-optimization/79728
20245 * regs.h (struct target_regs): New field
20246 x_contains_allocatable_regs_of_mode.
20247 (contains_allocatable_regs_of_mode): New macro.
20248 * reginfo.c (init_reg_sets_1): Initialize it, and change
20249 contains_reg_of_mode so it includes global regs as well.
20250 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20251 rather than contains_regs_of_mode.
20253 2017-03-14 Martin Liska <mliska@suse.cz>
20255 * doc/invoke.texi: Document options that can't be combined with
20256 -fcheck-pointer-bounds.
20258 2017-03-14 Martin Liska <mliska@suse.cz>
20260 PR middle-end/79831
20261 * doc/invoke.texi (-Wchkp): Document the option.
20263 2017-03-14 Martin Liska <mliska@suse.cz>
20265 * Makefile.in: Install gcov-dump.
20267 2017-03-14 Martin Liska <mliska@suse.cz>
20269 * multiple_target.c (expand_target_clones): Bail out for
20270 an invalid attribute.
20272 2017-03-14 Richard Biener <rguenther@suse.de>
20274 * alias.c (struct alias_set_entry): Pack properly.
20275 * cfgloop.h (struct loop): Likewise.
20276 * cse.c (struct set): Likewise.
20277 * ipa-utils.c (struct searchc_env): Likewise.
20278 * loop-invariant.c (struct invariant): Likewise.
20279 * lra-remat.c (struct cand): Likewise.
20280 * recog.c (struct change_t): Likewise.
20281 * rtl.h (struct address_info): Likewise.
20282 * symbol-summary.h (function_summary): Likewise.
20283 * tree-loop-distribution.c (struct partition): Likewise.
20284 * tree-object-size.c (struct object_size_info): Likewise.
20285 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20286 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20287 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20288 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20289 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20290 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20291 (struct _stmt_vec_info): Likewise.
20293 2017-03-14 Martin Liska <mliska@suse.cz>
20296 * multiple_target.c (create_dispatcher_calls): Check that
20297 a target can create a function dispatcher.
20299 2017-03-14 Martin Liska <mliska@suse.cz>
20302 * multiple_target.c (expand_target_clones): Drop local.local
20303 flag for default implementation.
20305 2017-03-14 Richard Biener <rguenther@suse.de>
20307 PR tree-optimization/80030
20308 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20310 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20312 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20313 gcc_fallthrough() instead of __attribute__((fallthrough));
20315 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20317 * doc/gcc.texi: Remove "up" link to (DIR).
20318 * doc/gccint.texi: Ditto.
20320 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20322 * doc/install.texi (Specific) <avr>: Remove reference to
20325 2017-03-13 Jeff Law <law@redhat.com>
20327 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20328 attribute rather than comments.
20330 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20331 match_scratch operand is highest.
20333 2017-03-13 Martin Liska <mliska@suse.cz>
20335 PR middle-end/78339
20336 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20337 is a CHKP clone, use original declaration.
20339 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20341 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20342 (arc_conditional_register_usage): Use a different allocation order
20343 when optimizing for size.
20344 * common/config/arc/arc-common.c (arc_option_optimization_table):
20345 Section anchors default on when optimizing for size.
20347 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20349 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20351 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20353 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20354 * config/arc/arc.md (cpu_facility): Add cd variant.
20355 (*movqi_insn): Add code density variant.
20356 (*movhi_insn): Likewise.
20357 (*movqi_insn): Likewise.
20358 (*addsi3_mixed): Likewise.
20359 (subsi3_insn): Likewise.
20361 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20363 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20365 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20367 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20368 expressions with MINUS and UNARY ops.
20370 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20373 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20375 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20376 between vec_select and vector argument.
20377 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20378 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20379 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20380 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20381 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20382 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20384 2017-03-13 Richard Biener <rguenther@suse.de>
20387 * params.def (vect-max-peeling-for-alignment): Fix typo.
20389 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20391 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20392 issue that only occurred with binutils below 2.18.
20394 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20396 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20397 refer to binutils 2.11/2.12 minimum.
20399 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20401 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20402 ftp.kernel.org and simplify binutils requirement.
20404 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20406 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20408 (Optimize Options): Ditto. Also remove redundancy.
20410 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20412 PR translation/79848
20413 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20415 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20416 to G_ to avoid double translation.
20418 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20420 PR translation/79923
20421 * auto-profile.c (get_combined_location): Convert leading
20422 character of diagnostics to lower case and remove trailing period.
20423 (read_profile): Likewise for various diagnostics.
20424 * config/arm/arm.c (arm_option_override): Remove trailing period
20425 from various diagnostics.
20426 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20427 (msp430_expand_delay_cycles): Likewise.
20429 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20432 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20433 full command-line argument, rather than just "str".
20434 (aarch64_validate_march): Likewise.
20435 (aarch64_validate_mtune): Likewise.
20437 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20439 PR rtl-optimization/78911
20440 * lra-assigns.c (must_not_spill_p): New function.
20441 (spill_for): Use it.
20443 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20445 PR tree-optimization/79981
20446 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20447 ATOMIC_COMPARE_EXCHANGE ifn result.
20448 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20449 IFN_ATOMIC_COMPARE_EXCHANGE.
20451 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20454 * opts.c (parse_sanitizer_options): Add missing question mark to
20455 "did you mean" message.
20457 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20459 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20461 (VMULEUH_UNS): Likewise.
20462 (VMULOUB_UNS): Likewise.
20463 (VMULOUH_UNS): Likewise.
20464 * config/rs6000/rs6000.c (builtin_function_type): Remove
20465 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20467 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20470 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20471 x with result of extra_parsing_for_operand_code_0.
20472 (function_reader::extra_parsing_for_operand_code_0): Convert
20473 return type from void to rtx, returning x. When reading
20474 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20475 larger size containing struct block_symbol.
20477 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20480 -mfloat128-hardware without -m64.
20482 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20485 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20486 entries to the case statement that marks unsigned arguments to
20487 overloaded functions.
20489 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20491 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20492 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20494 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20497 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20498 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20500 2017-03-10 Martin Liska <mliska@suse.cz>
20504 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20505 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20508 2017-03-10 Olivier Hainque <hainque@adacore.com>
20510 * tree-switch-conversion (array_value_type): Start by resetting
20511 candidate type to it's main variant.
20513 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20515 PR rtl-optimization/79909
20516 * combine.c (try_combine): Use simplify_replace_rtx on individual
20517 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20518 of the whole CALL_INSN_FUNCTION_USAGE.
20520 PR tree-optimization/79972
20521 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20522 get_range_info on SSA_NAMEs. Formatting fixes.
20524 2017-03-10 Richard Biener <rguenther@suse.de>
20525 Jakub Jelinek <jakub@redhat.com>
20527 PR tree-optimization/77975
20528 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20529 edge to be constant.
20530 (get_val_for): For constant x return it. Formatting fix.
20531 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20532 would use the same bases as the current one.
20534 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20536 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20537 instead of vec_select for V1TImode.
20538 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20540 (VSX_LE_128): Add V1TI to this mode iterator.
20541 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20542 (*vsx_le_perm_store_<mode>): Likewise.
20543 (pre-reload splitter for VSX stores): Likewise.
20544 (post-reload splitter for VSX stores): Likewise.
20545 (*vsx_xxpermdi2_le_<mode>): Likewise.
20546 (*vsx_lxvd2x2_le_<mode>): Likewise.
20547 (*vsx_stxvd2x2_le_<mode>): Likewise.
20549 2017-03-09 Michael Eager <eager@eagercon.com>
20551 Correct failures with --enable-checking=yes,rtl.
20553 * config/microblaze/microblaze.c (microblaze_expand_shift):
20554 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20555 test for const0_rtx.
20556 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20557 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20559 2017-03-09 Richard Biener <rguenther@suse.de>
20561 PR tree-optimization/79977
20562 * graphite-scop-detection.c (scop_detection::merge_sese):
20563 Handle the case of extra exits to blocks dominating the entry.
20565 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20567 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20570 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20572 PR rtl-optimization/79949
20573 * lra-constraints.c (process_alt_operands): Check memory when
20574 trying to predict a cycle. Print about the overall increase.
20576 2017-03-09 Richard Biener <rguenther@suse.de>
20578 PR middle-end/79971
20579 * gimple-expr.c (useless_type_conversion_p): Preserve
20580 TYPE_SATURATING for fixed-point types.
20582 2017-03-09 Richard Biener <rguenther@suse.de>
20585 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20586 alignment of BLKmode params.
20588 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20591 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20592 (VALL_NO_V2Q): Likewise.
20594 * config/aarch64/aarch64-simd.md
20595 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20597 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20598 VALL_NO_V2Q mode iterator.
20599 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20601 2017-03-09 Martin Liska <mliska@suse.cz>
20603 PR tree-optimization/79631
20604 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20605 tree_int_cst_sign_bit just for INTEGER constants.
20607 2017-03-09 Martin Liska <mliska@suse.cz>
20611 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20614 2017-03-09 Marek Polacek <polacek@redhat.com>
20617 * tree.c (inchash::add_expr): Handle TREE_VEC.
20619 2017-03-09 Martin Liska <mliska@suse.cz>
20622 (chkp_narrow_size_and_offset): New function.
20623 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20624 (void chkp_parse_bit_field_ref): New function.
20625 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20626 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20628 2017-03-09 Martin Liska <mliska@suse.cz>
20631 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20632 (chkp_find_bounds_1): Remove gcc_unreachable.
20634 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20637 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20638 BUILT_IN_SYNC*, determine the access type from the size suffix and
20639 always build a MEM_REF with that type. Handle forgotten
20640 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20643 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20644 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20645 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20646 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20647 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20648 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20649 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20650 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20651 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20652 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20653 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20654 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20655 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20656 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20657 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20658 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20659 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20660 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20661 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20662 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20663 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20664 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20665 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20666 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20667 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20668 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20669 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20670 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20671 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20672 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20673 definitions outside of __OPTIMIZE__ guarded section.
20676 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20677 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20678 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20679 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20682 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20684 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20685 ("vfenez<mode>"): Add missing constraints.
20687 2017-03-08 Martin Sebor <msebor@redhat.com>
20690 * config/nds32/nds32.c (nds32_option_override):
20691 Fix misspelled diagnostic.
20693 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20696 * gimplify.c (gimplify_omp_for): Replace index var in outer
20697 taskloop statement with an artificial variable and add
20698 OMP_CLAUSE_PRIVATE clause for it.
20700 2017-03-08 Richard Biener <rguenther@suse.de>
20702 PR tree-optimization/79955
20703 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20704 for accesses that are completely outside of the variable.
20706 2017-03-08 Andrew Haley <aph@redhat.com>
20708 PR tree-optimization/79943
20709 * tree-ssa-loop-split.c (compute_new_first_bound): When
20710 calculating the new upper bound, (END-BEG) should be added, not
20713 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20715 * config/avr/avr.md (setmemhi): Make sure match_dup
20716 operand number comes before match_scratch.
20718 2017-03-08 Richard Biener <rguenther@suse.de>
20720 PR tree-optimization/79920
20721 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20722 with ncopies == 1 to ...
20723 (vect_transform_slp_perm_load): ... here. Properly compute
20724 all element loads by iterating VF times over the group. Do
20725 not handle ncopies (computed in a broken way) in
20726 vect_create_mask_and_perm.
20728 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20731 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20732 is a uniform vector, use uniform_vector_p return value instead of
20733 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20735 2017-03-07 Marek Polacek <polacek@redhat.com>
20737 PR middle-end/79809
20738 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20739 (alloca_call_type): Likewise.
20741 2017-03-07 Martin Liska <mliska@suse.cz>
20743 * gcov.c (process_args): Put comment to correct location.
20745 2017-03-07 Martin Liska <mliska@suse.cz>
20747 PR middle-end/68270
20748 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20749 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20750 (chkp_narrow_bounds_for_field): Likewise.
20751 (chkp_parse_array_and_component_ref): Pass one more argument to
20754 2017-03-07 Richard Biener <rguenther@suse.de>
20756 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20759 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20761 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20762 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20764 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20767 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20768 to end of description.
20769 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20771 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20773 PR rtl-optimization/79901
20774 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20776 (*avx512f_<code><mode>3<mask_name>): ... this.
20777 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20778 iterator instead of VI8_AVX2_AVX512BW.
20780 PR rtl-optimization/79901
20781 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20782 min/max expander, expand it using expand_vec_cond_expr.
20785 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20788 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20791 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20792 to void * for PCH reasons.
20793 * dwarf2out.c (output_loc_operands, output_die): Cast
20794 v.val_vec.array to unsigned char *.
20796 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20799 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20802 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20804 PR rtl-optimization/79571
20805 * lra-constraints.c (process_alt_operands): Calculate static
20806 reject and subtract it from overall when only addresses will be
20809 2017-03-06 Julia Koval <julia.koval@intel.com>
20812 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20813 incoming stack boundary to 128 for 64-bit targets.
20815 2017-03-06 Richard Biener <rguenther@suse.de>
20817 PR tree-optimization/79894
20818 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20819 to NULL after folding it.
20821 2017-03-06 Richard Biener <rguenther@suse.de>
20823 PR tree-optimization/79824
20824 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20825 check disabling peeling for gaps.
20827 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20829 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20830 attributes): Document gettimeofday.
20832 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20834 * config/s390/s390.c (s390_option_override_internal): Set
20835 PARAM_MIN_VECT_LOOP_BOUND
20837 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20839 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20840 * config/s390/s390.md: Likewise.
20842 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20845 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20846 (<avx2_avx512>_perm<mode>): Rename to ...
20847 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20849 (<avx512>_perm<mode>_mask): Rename to ...
20850 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20852 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20853 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20854 instead of VI8F_256_512.
20855 (avx512f_perm<mode>): New define_expand.
20856 (avx512f_perm<mode>_mask): Likewise.
20857 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20858 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20860 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20862 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20863 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20865 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20867 2017-03-06 Martin Liska <mliska@suse.cz>
20870 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20871 when having a SSA NAME w/o VAR_DECL assigned to it.
20873 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20875 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20876 msa_dpsub_<su>_d): Fix MODE for vec_select.
20878 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20880 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20882 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20884 2017-03-06 Richard Biener <rguenther@suse.de>
20886 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20887 * plugin.c (register_plugin_info): Likewise.
20888 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20890 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20892 * config/i386/sse.md (sse_storehps, sse_storelps,
20893 avx_<castmode><avxsizesuffix>_<castmode>,
20894 avx512f_<castmode><avxsizesuffix>_<castmode>,
20895 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20896 in condition that at least one operand is not a MEM.
20898 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20900 PR middle-end/79805
20901 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20902 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20904 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20905 gimple_call_nothrow_p flag based on whether original builtin can throw.
20906 If it can, emit following stmts on the fallthrough edge.
20907 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20908 don't create new bb if inserting just debug stmts on the edge, try to
20909 insert them on the fallthru bb or just reset debug stmts.
20911 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20914 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20915 restore recog_data (including the operand rtxes inside it) around
20916 the call to get_insn_template.
20918 2017-03-03 Martin Sebor <msebor@redhat.com>
20920 PR tree-optimization/79699
20921 * context.c (context::~context): Free MPFR caches to avoid
20922 a memory leak on program exit.
20924 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20926 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20927 Use wide_int::ulow () instead of .elt (0).
20929 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20931 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20932 (*pushxf): Limit oF constraint to 32bit targets and add oC
20933 constraint for 64bit targets.
20934 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20935 (*pushdf): Change rmF constraint to rmC.
20937 2017-03-03 Martin Liska <mliska@suse.cz>
20939 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20940 Remove unused variable.
20942 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20945 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20946 is a memory operand, increase num_memory.
20947 (ix86_expand_args_builtin): Likewise.
20949 2017-03-03 Jan Hubicka <jh@suse.cz>
20952 * ipa-devirt.c (maybe_record_node): Properly handle
20953 __cxa_pure_virtual visibility.
20955 2017-03-03 Martin Liska <mliska@suse.cz>
20957 PR tree-optimization/79803
20958 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20960 (pass_loop_prefetch::execute): Disabled optimization if an
20961 assumption about L1 cache size is not met.
20963 2017-03-03 Martin Liska <mliska@suse.cz>
20965 PR rtl-optimization/79574
20966 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20967 (hash_scan_set): Likewise.
20968 (dump_hash_table): Likewise.
20969 (hoist_code): Likewise.
20971 2017-03-03 Richard Biener <rguenther@suse.de>
20973 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20974 in place of uhigh ().
20975 (fixed_convert_from_real): Likewise.
20977 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20980 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20982 2017-03-03 Richard Biener <rguenther@suse.de>
20984 PR middle-end/79818
20985 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20986 TYPE_OVERFLOW_UNDEFINED check.
20988 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20990 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20992 (vector_ae_<mode>_p): Likewise.
20993 (vector_nez_<mode>_p): Likewise.
20994 (vector_ne_v2di_p): Likewise.
20995 (vector_ae_v2di_p): Likewise.
20996 (vector_ne_<mode>_p): Likewise.
20997 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20999 (vsx_tsqrt<mode>2_fe): Likewise.
21001 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21004 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21006 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21008 PR rtl-optimization/79780
21009 * cprop.c (one_cprop_pass): When second and further conditional trap
21010 in a single basic block is turned into an unconditional trap, turn it
21011 into a deleted note to avoid RTL verification failures.
21013 2017-03-02 Richard Biener <rguenther@suse.de>
21015 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21017 2017-03-02 Richard Biener <rguenther@suse.de>
21019 PR tree-optimization/79345
21021 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21022 param and abort the walk, returning -1 if it is hit.
21023 (walk_aliased_vdefs): Take a limit param and pass it on.
21024 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21025 defaulting to 0 and return a signed int.
21026 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21027 (check_defs): New helper.
21028 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21029 about uninitialized memory.
21030 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21031 bogus uninitialized warning.
21032 (fixed_convert_from_real): Likewise.
21034 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21036 PR tree-optimization/66768
21037 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21038 iv_use if base object can't be determined.
21040 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21042 PR tree-optimization/79345
21043 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21044 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21045 (get_pattern_stats): Initialize it.
21046 * genemit.c (gen_expand): Verify match_scratch numbers come after
21047 match_operand/match_dup numbers.
21048 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21049 match_scratch numbers.
21050 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21052 * config/s390/s390.md (trunctdsd2): Likewise.
21054 2017-03-02 Richard Biener <rguenther@suse.de>
21056 * wide-int.h (wide_int_storage::operator=): Implement in terms
21059 2017-03-02 Richard Biener <rguenther@suse.de>
21061 PR tree-optimization/79777
21062 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21063 the to insert expression to sth existing.
21065 2017-03-01 Martin Sebor <msebor@redhat.com>
21067 PR middle-end/79692
21068 * gimple-ssa-sprintf.c
21069 (directive::known_width_and_precision): New function.
21070 (format_integer): Use it.
21071 (get_mpfr_format_length): Consider the full range of precision
21072 when computing %g output with the # flag. Set the likely byte
21073 count to 3 rather than 1 when precision is indeterminate.
21074 (format_floating): Correct the lower bound of precision.
21076 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21078 * doc/invoke.texi: Document default code model for 64-bit Linux.
21080 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21083 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21084 udiv rather than div since input pattern is unsigned.
21086 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21088 * config/i386/i386.c (print_reg): Warn for values of
21089 unsupported size in integer register.
21091 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21094 * config/rs6000/predicates.md (current_file_function_operand): Do
21095 not allow self calls to be local if the function is replaceable.
21097 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21100 * config/rs6000/altivec.h (vec_ctz and others): Change the
21101 preprocessor macro that controls conditional compilation from
21102 _ARCH_PWR9 to __POWER9_VECTOR__.
21103 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21104 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21105 control (instead of _ARCH_PWR9 control) so that template
21106 definition uses power9-specific function.
21107 (vec_any_eq): Likewise.
21108 (vec_all_ne): Change macro definition to use a power9-specific
21109 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21110 _ARCH_PWR9 control).
21111 (vec_any_eq) Likewise.
21112 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21113 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21114 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21115 support for xvcmpnedp instruction.
21116 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21117 macro expansion so that Power9 implementation of vec_all_ne does
21118 not use the AltiVec predicate framework.
21119 (VCMPNEH_P): Likewise.
21120 (VCMPNEW_P): Likewise.
21121 (VCMPNED_P): Likewise.
21122 (VCMPNEFP_P): Likewise.
21123 (VCMPNEDP_P): Likewise.
21124 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21125 implementation of vec_any_eq to not use AltiVec predicate
21127 (VCMPAEH_P): Likewise.
21128 (VCMPAEW_P): Likewise.
21129 (VCMPAED_P): Likewise.
21130 (VCMPAEFP_P): Likewise.
21131 (VCMPAEDP_P): Likewise.
21132 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21133 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21134 not use the AltiVec predicate framework.
21135 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21136 of vec_any_eq to not use AltiVec predicate framework.
21137 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21138 support for predefined __POWER9_VECTOR__ macro to indicate that
21139 Power9 instruction selection is enabled.
21140 (altivec_overloaded_builtins): Remove extraneous
21141 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21142 function argument types RS6000_BTI_bool_V16QI and
21143 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21144 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21145 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21146 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21147 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21148 Power9 for implementations of vec_cmpne. Change the signature for
21149 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21150 (representing vec_all_ne) to remove the previously described first
21151 argument of type RS6000_BTI_INTSI, as this was an artifact of
21152 reliance on the AltiVec predicate framework, which is no longer
21153 used in the implementation of these functions. Add
21154 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21155 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21156 since, unlike the AltiVec predicate framework implementation, we
21157 do not share function descriptors between vec_alle and vec_anyeq.
21158 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21159 set of modes that receive special treatment even when
21160 TARGET_P9_VECTOR is true. The special treatment emits code that
21161 does not depend on Power9 instructions.
21162 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21163 define_expand to not rely on AltiVec predicate framework.
21164 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21166 (vector_ne_v2di_p): Change this define_expand to not rely on
21167 AltiVec predicate framework.
21168 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21170 (vector_ne_<mode>_p): Change this define_expand to not rely on
21171 AltiVec predicate framework.
21172 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21174 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21175 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21176 define_insn pattern.
21177 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21178 define_insn pattern because the xvcmpne<VSs>. instruction is not
21180 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21181 instruction is not supported.
21183 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21185 * config/nvptx/nvptx.c: Include intl.h.
21187 2017-03-01 Martin Jambor <mjambor@suse.cz>
21190 * ipa-prop.h (ipa_bits): Removed field known.
21191 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21192 to pointers. Adjusted their comments to warn about their sharing.
21193 (ipcp_transformation_summary): Change bits to a vector of pointers.
21194 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21195 (ipa_get_ipa_bits_for_value): Declare.
21196 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21197 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21198 (ipa_bits_hash_table): Likewise.
21199 (ipa_vr_ggc_hash_traits): Likewise.
21200 (ipa_vr_hash_table): Likewise.
21201 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21202 being pointers and vr_known being removed.
21203 (ipa_set_jf_unknown): Likewise.
21204 (ipa_get_ipa_bits_for_value): New function.
21205 (ipa_set_jfunc_bits): Likewise.
21206 (ipa_get_value_range): New overloaded functions.
21207 (ipa_set_jfunc_vr): Likewise.
21208 (ipa_compute_jump_functions_for_edge): Use the above functions to
21209 construct bits and vr parts of jump functions.
21210 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21211 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21213 (ipcp_grow_transformations_if_necessary): Also allocate
21214 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21216 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21217 them. Fix too long lines.
21218 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21219 vr_known being removed.
21220 (ipa_read_jump_function): Use new setter functions to construct bits
21221 and vr parts of jump functions or set them to NULL.
21222 (write_ipcp_transformation_info): Adjust for bits being pointers.
21223 (read_ipcp_transformation_info): Likewise.
21224 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21226 Include gt-ipa-prop.h.
21227 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21229 (ipcp_store_bits_results): Likewise.
21230 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21231 Do not write to existing jump functions but use a temporary instead.
21233 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21236 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21237 attempt to use its first operand as BIT_FIELD_REF base.
21239 2017-03-01 Richard Biener <rguenther@suse.de>
21241 PR middle-end/79721
21242 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21243 interpolating formula in wrapping arithmetic.
21244 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21246 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21248 PR tree-optimization/79734
21249 * tree-vect-generic.c (expand_vector_condition): Optimize
21250 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21251 Handle VEC_COND_EXPR where comparison has different inner width from
21252 type's inner width.
21254 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21256 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21257 markup, and similar issues. Remove @opindex entries for things
21258 that aren't options. Add missing -mmpy-option entries.
21260 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21262 PR tree-optimization/79737
21263 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21264 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21265 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21266 instead of byte_size. Formatting fix.
21267 (shift_bytes_in_array_right): Formatting fix.
21269 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21272 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21273 condition on optimize for the leaf function test.
21275 2017-02-28 Martin Liska <mliska@suse.cz>
21278 * read-rtl-function.c (function_reader::handle_unknown_directive):
21279 Bail out when one uses -flto.
21281 2017-02-28 Martin Liska <mliska@suse.cz>
21283 * common.opt: Replace space with tabular for options of <number>
21285 * config/i386/i386.opt: Show <number> value for
21286 -mlarge-data-threshold.
21287 * opts.c (print_filtered_help): Do not display number in hexadecimal
21290 2017-02-28 Martin Liska <mliska@suse.cz>
21292 * common.opt: Fix --help=option -Q for options which are of
21295 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21297 * config/i386/i386.c (print_reg): Error out for values
21298 of 8-bit size in invalid integer register.
21300 2017-02-28 Martin Sebor <msebor@redhat.com>
21302 PR tree-optimization/79691
21303 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21305 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21308 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21309 gcc_unreachable with output_operand_lossage.
21311 2017-02-28 Richard Biener <rguenther@suse.de>
21313 PR tree-optimization/79740
21314 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21316 (visit_nary_op): Insert the nary into the hashtable if we
21317 pattern-matched sth.
21318 * tree-ssa-pre.c (eliminate_insert): Robustify.
21320 2017-02-28 Richard Biener <rguenther@suse.de>
21322 PR middle-end/79731
21323 * fold-const.c (decode_field_reference): Reject out-of-bound
21326 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21328 * config/i386/i386.c: Include intl.h.
21329 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21330 instead of just cond ? "..." : "...".
21331 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21332 * coverage.c (read_counts_file): Likewise.
21333 * omp-offload.c: Include intl.h.
21334 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21335 of just cond ? "..." : "...".
21336 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21337 of just cond ? "..." : "...".
21339 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21342 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21344 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21345 'tune for' CPU name.
21346 * config/arm/arm-cpu-data.h: Regenerated.
21348 2017-02-28 Richard Biener <rguenther@suse.de>
21350 PR tree-optimization/79732
21351 * tree-inline.c (expand_call_inline): Do not shadow var.
21353 2017-02-28 Richard Biener <rguenther@suse.de>
21355 PR tree-optimization/79723
21356 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21357 address-space properly.
21359 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21361 * doc/optinfo.texi (Optimization groups): Fix option used for
21363 * doc/invoke.texi (-fopt-info): Document "omp".
21364 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21365 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21366 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21367 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21368 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21370 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21372 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21373 OPTGROUP_OMP with "omp".
21375 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21378 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21379 for arithmetic shift of unsigned V2DI.
21381 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21383 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21384 arc/linux.h headers.
21385 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21386 (LINK_SPEC): Likewise.
21387 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21388 (EXTRA_SPECS): Likewise.
21389 (STARTFILE_SPEC): Likewise.
21390 (ENDFILE_SPEC): Likewise.
21391 (LIB_SPEC): Likewise.
21392 (TARGET_SDATA_DEFAULT): Likewise.
21393 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21394 (MULTILIB_DEFAULTS): Likewise.
21395 (DWARF2_UNWIND_INFO): Likewise.
21396 * config/arc/big.h: New file.
21397 * config/arc/elf.h: Likewise.
21398 * config/arc/linux.h: Likewise.
21399 * config/arc/t-uClibc: Remove.
21401 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21403 PR tree-optimization/77536
21404 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21405 (tree_transform_and_unroll_loop): Use above function to compute the
21406 estimated niter of unrolled loop and use it when scaling profile.
21407 Also use count info rather than frequency if it's non-zero.
21408 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21409 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21410 (vect_transform_loop): Call above function.
21412 2017-02-27 Richard Biener <rguenther@suse.de>
21414 PR tree-optimization/45397
21415 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21416 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21417 (visit_nary_op): Add pattern matching for CSEing sign-changed
21418 or truncated operations with wider ones.
21420 2017-02-27 Richard Biener <rguenther@suse.de>
21422 PR tree-optimization/79690
21423 * tree-vect-stmts.c (vectorizable_store): Use vector type
21424 built from the DR with address-space.
21426 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21428 * doc/invoke.texi (Optimize Options): Refine the description
21429 of asan-use-after-return.
21431 2017-02-25 Alan Modra <amodra@gmail.com>
21433 PR rtl-optimization/79584
21434 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21435 base, not ad->base_term, the reg within base. Remove assertion
21436 that ad->base == ad->base_term. Replace gen_int_mode using
21437 bogus mode with const0_rtx.
21439 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21441 PR middle-end/79396
21442 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21443 FMA_EXPR like tcc_binary or tcc_unary.
21445 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21448 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21450 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21451 (output_loc_operands): Handle DW_OP_call_ref and
21452 DW_OP_GNU_variable_value.
21453 (struct variable_value_struct): New type.
21454 (struct variable_value_hasher): Likewise.
21455 (variable_value_hash): New variable.
21456 (string_types): Remove.
21457 (copy_loc_descr): New function.
21458 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21459 (prepend_loc_descr_to_each): New function.
21460 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21461 instead of add_loc_descr_to_each if the first argument is single
21462 location list and the second has multiple.
21463 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21464 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21465 when looking for variable value which doesn't have other location info.
21466 (loc_list_from_tree): Formatting fix.
21467 (gen_array_type_die): Simplify DW_AT_string_length handling.
21468 (adjust_string_types): Remove.
21469 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21470 string_types. Call resolve_variable_values.
21471 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21472 (resolve_addr_in_expr): Likewise. Add A argument.
21473 (copy_deref_exprloc): Remove deref argument. Adjust for the
21474 original expression being DW_OP_GNU_variable_value with optionally
21475 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21476 optionally after it.
21477 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21478 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21479 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21480 (variable_value_hasher::hash, variable_value_hasher::equal): New
21482 (resolve_variable_value_in_expr, resolve_variable_value,
21483 resolve_variable_values, note_variable_value_in_expr,
21484 note_variable_value): New functions.
21485 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21488 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21491 * opts.h (handle_generated_option): Add GENERATED_P argument.
21492 * opts-common.c (handle_option): Adjust function comment.
21493 (handle_generated_option): Add GENERATED_P argument, pass it to
21495 (control_warning_option): Pass false to handle_generated_option
21497 * opts.c (maybe_default_option): Pass true to handle_generated_option
21499 * optc-gen.awk: Likewise.
21501 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21503 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21504 a REG, look at the REG it is a SUBREG of.
21505 (splitter for cmpeqsi_t): Ditto.
21507 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21509 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21510 the special USEs with the pattern of the insn, not the insn itself.
21512 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21515 * doc/invoke.texi: Document -mload-store-pairs.
21517 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21518 Sandra Loosemore <sandra@codesourcery.com>
21520 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21521 argument isn't a CONST_INT.
21522 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21523 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21524 (nios2_validate_compare): Bypass alternate compare logic if *op2
21525 is not a CONST_INT.
21526 (ldstwm_operation_p): Return false if first_base is not a REG or
21527 if first_offset is not a CONST_INT.
21529 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21531 * config/cris/cris.md: Use correct operand in a define_peephole2.
21533 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21535 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21537 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21539 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21540 this_insn if it is an INSN or JUMP_INSN.
21541 (force_offsettable): Look at base, not at addr.
21542 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21543 on things that aren't necessarily CONST_INTs.
21545 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21547 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21548 -mfpmath=sse is the default also for x86-32 targets with SSE2
21549 instruction set when @option{-ffast-math} is enabled
21551 2017-02-24 Jeff Law <law@redhat.com>
21553 PR rtl-optimizatoin/79286
21554 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21557 2017-02-24 Richard Biener <rguenther@suse.de>
21559 PR tree-optimization/79389
21560 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21563 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21565 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21566 function comment to reflect reality.
21567 (loop_exits_before_overflow): Fix typo in function description.
21569 2017-02-24 Richard Biener <rguenther@suse.de>
21571 PR tree-optimization/79389
21572 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21573 properly that a threading opportunity exists. Detect conditional
21574 copy/constant propagation opportunities.
21576 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21578 * config/visium/visium.md (type): Add trap.
21579 (b): New mode attribute.
21580 (*btst): Rename into...
21581 (*btst<mode>): ...this and adjust.
21582 (*cbranchsi4_btst_insn): Rename into...
21583 (*cbranch<mode>4_btst_insn): ...this and adjust.
21584 (trap): New define_insn.
21586 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21588 PR tree-optimization/79389
21589 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21590 (noce_reversed_cond_code): New function.
21591 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21592 reversed_comparison_code. Formatting fix.
21593 (noce_try_store_flag): Test rev_cond != NULL in addition to
21594 reversed_comparison_code.
21595 (noce_try_store_flag_constants): Likewise.
21596 (noce_try_store_flag_mask): Likewise.
21597 (noce_try_addcc): Use rev_cond if non-NULL instead of
21598 reversed_comparison_code.
21599 (noce_try_cmove_arith): Likewise. Formatting fixes.
21600 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21601 (noce_find_if_block): Initialize rev_cond.
21602 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21603 instead of false as last argument never attempt to reverse it
21606 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21608 PR tree-optimization/79663
21609 * tree-predcom.c (combine_chains): Process refs in reverse order
21610 only for ZERO length chains, and add explaining comment.
21612 2017-02-23 Jeff Law <law@redhat.com>
21614 PR tree-optimization/79578
21615 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21616 in call to operand_equal_p.
21618 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21621 * config/i386/cpuid.h: Fix another undefined behavior.
21623 2017-02-23 Richard Biener <rguenther@suse.de>
21625 PR tree-optimization/79683
21626 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21627 vector types for data-refs.
21629 2017-02-23 Martin Liska <mliska@suse.cz>
21631 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21633 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21635 PR middle-end/79665
21636 * internal-fn.c (get_range_pos_neg): Moved to ...
21637 * tree.c (get_range_pos_neg): ... here. No longer static.
21638 * tree.h (get_range_pos_neg): New prototype.
21639 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21640 are known to be in between 0 and signed maximum inclusive, try to
21641 expand both unsigned and signed divmod and use the cheaper one from
21644 2017-02-22 Jeff Law <law@redhat.com>
21646 PR tree-optimization/79578
21647 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21648 to compare base operands.
21650 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21653 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21654 gpc_reg_operand instead of fpr_reg_operand.
21656 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21658 * config/mips/mips.c (mips_return_in_memory): Force FP
21659 vector types to be returned in memory for o32 ABI.
21661 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21663 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21664 instead of DW_TAG_member for static data member declarations and don't
21665 set no_linkage_name for static inline data members.
21666 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21669 2017-02-22 Martin Liska <mliska@suse.cz>
21671 * doc/invoke.texi: Replace inequality signs with square brackets
21674 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21676 PR tree-optimization/68644
21677 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21679 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21682 * lra-constraints.c (simplify_operand_subreg): Handle
21683 WORD_REGISTER_OPERATIONS targets.
21685 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21688 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21689 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21690 elimination by swapping fld*.
21692 2017-02-22 Richard Biener <rguenther@suse.de>
21694 PR tree-optimization/79673
21695 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21696 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21697 irrelevant address-space qualifiers and avoiding a
21698 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21700 2017-02-22 Richard Biener <rguenther@suse.de>
21702 PR tree-optimization/79666
21703 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21704 to not symbolically negate if that may introduce undefined
21707 2017-02-22 Martin Liska <mliska@suse.cz>
21710 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21711 * data-streamer-out.c (streamer_write_gcov_count_stream):
21713 * value-prof.c (stream_out_histogram_value): Make assert more
21714 precise based on type of counter.
21716 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21719 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21720 Use nonimmediate_operand instead of memory_operand for operand 1.
21721 (float-extend standard_x87sse_constant_load splitter): Ditto.
21723 2017-02-21 Jeff Law <law@redhat.com>
21725 PR tree-optimization/79621
21726 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21727 blocks with edges to themselves.
21729 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21732 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21733 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21734 Use gimple_call_builtin_p.
21737 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21738 on temporarily removed DEBUG_INSNs.
21740 PR tree-optimization/79649
21741 * tree-loop-distribution.c (classify_partition): Give up on
21742 non-generic address space loads/stores.
21744 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21746 * doc/loop.texi (Loop manipulation): Remove nonexistent
21747 tree_ssa_loop_version from the documentation.
21748 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21750 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21753 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21754 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21755 * config/rs6000/rs6000.c: Include except.h.
21756 (rs6000_expand_split_stack_prologue): Call
21757 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21759 2017-02-21 Martin Jambor <mjambor@suse.cz>
21762 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21763 have been analyzed.
21765 2017-02-21 Martin Jambor <mjambor@suse.cz>
21767 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21768 for backward compatibility only.
21769 * doc/invoke.texi (Option Summary): Remove all references to
21770 -fipa-cp-alignment.
21772 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21776 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21778 * lra-constraints.c (curr_insn_transform): Handle
21779 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21781 2017-02-21 Martin Liska <mliska@suse.cz>
21783 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21785 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21787 PR translation/79638
21788 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21790 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21793 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21794 (arm_function_ok_for_sibcall): Return false for an indirect call by
21795 descriptor if all the argument registers are used.
21796 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21797 alignment of the function.
21799 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21801 PR tree-optimization/61441
21802 * simplify-rtx.c (simplify_const_unary_operation): For
21803 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21804 the sNaN unmodified.
21806 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21808 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21809 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21810 instead of SYSTEM_HEADER_DIR.
21812 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21813 Martin Liška <mliska@suse.cz>
21815 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21816 Fix typos and grammar, use active voice, and clarify.
21818 2017-02-20 Marek Polacek <polacek@redhat.com>
21820 PR middle-end/79537
21821 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21824 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21826 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21829 * config/i386/i386.c (ix86_expand_builtin): Handle
21830 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21831 ix86_builtins_isa[fcode].isa as a requirement of those
21832 flags and any other flag in the bitmask.
21833 (ix86_init_mmx_sse_builtins): Use 0 instead of
21834 ~OPTION_MASK_ISA_64BIT as mask.
21835 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21836 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21837 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21838 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21840 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21843 * lra-constraints.c (split_reg): Check requested split mode
21844 is supported by the register.
21846 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21848 * lra-constraints.c (simplify_operand_subreg): Remove early
21851 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21854 * lra-constraints.c (curr_insn_transform): Tighten condition
21855 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21857 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21860 * lra-constraints.c (curr_insn_transform): Handle
21861 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21863 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21866 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21868 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21870 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21873 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21876 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21878 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21881 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21883 * final.c (last_columnnum, override_columnnum): New variables.
21884 (final_start_function): Set last_columnnum, pass it to begin_prologue
21885 hook and pass 0 to dwarf2out_begin_prologue.
21886 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21887 to source_line debug hook.
21888 (notice_source_line): Compute last_columnnum and for debug_column_info
21889 return true on column changes.
21890 * debug.h (struct gcc_debug_hooks): Add column argument to
21891 source_line and begin_prologue hooks.
21892 (debug_nothing_int_charstar_int_bool): Remove prototype.
21893 (debug_nothing_int_int_charstar,
21894 debug_nothing_int_int_charstar_int_bool): New prototypes.
21895 (dwarf2out_begin_prologue): Add column argument.
21896 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21897 begin_prologue hooks.
21898 (debug_nothing_int_charstar_int_bool): Remove.
21899 (debug_nothing_int_int_charstar,
21900 debug_nothing_int_int_charstar_int_bool): New functions.
21901 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21902 through to dwarf2out_source_line.
21903 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21904 (dwarf2out_source_line): Add column argument, emit it if requested.
21905 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21907 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21908 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21909 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21910 through to dwarf2out_begin_prologue.
21911 (vmsdbgout_source_line): Add column argument, pass it through to
21912 dwarf2out_source_line.
21913 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21914 dbxout_source_line caller.
21915 (dbxout_source_line): Add column argument.
21917 * common.opt (gno-column-info, gcolumn-info): New options.
21918 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21919 (check_die): Also test for multiple DW_AT_decl_column attributes.
21920 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21921 DW_AT_decl_column if requested.
21922 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21924 (gen_variable_die): Likewise.
21925 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21926 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21929 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21930 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21931 (ix86_handle_option): Handle OPT_m3dnowa.
21932 * doc/invoke.texi (-m3dnowa): Document.
21933 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21934 -m3dnowa instead of -m3dnow -march=athlon.
21937 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21938 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21940 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21943 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21944 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21945 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21946 generator for vsx_xxpermdi_<mode>_be.
21947 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21948 force big-endian semantics.
21949 (vsx_xxpermdi_<mode>_be): New define_expand with same
21950 implementation as previous version of vsx_xxpermdi_<mode>.
21952 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21954 PR tree-optimization/79327
21955 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21956 variable, its initialization and use.
21958 2017-02-17 Julia Koval <julia.koval@intel.com>
21960 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21961 (OPTION_MASK_ISA_PKU_UNSET): New.
21962 (ix86_handle_option): Handle -mrdpid.
21963 * config/i386/cpuid.h (bit_RDPID): New.
21964 * config/i386/driver-i386.c (host_detect_local_cpu):
21965 Detect RDPID feature.
21966 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21967 * config/i386/i386-c.c (ix86_target_macros_internal):
21969 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21970 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21971 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21972 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21973 * config/i386/i386.md (define_insn "rdpid"): New.
21974 * config/i386/i386.opt Add -mrdpid.
21975 * config/i386/immintrin.h (_rdpid_u32): New.
21977 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21979 PR rtl-optimization/79541
21980 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21981 instead of transforming it into USE.
21983 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21985 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21986 If HONOR_SNANS (SFmode) force the input to a register.
21987 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21988 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21989 an frsp or similar insn.
21991 2017-02-17 Martin Liska <mliska@suse.cz>
21993 PR rtl-optimization/79577
21994 * params.def (selsched-max-sched-times): Increase minimum to 1.
21996 2017-02-17 Martin Liska <mliska@suse.cz>
21998 PR rtl-optimization/79574
21999 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22001 2017-02-17 Martin Liska <mliska@suse.cz>
22003 PR tree-optimization/79529
22004 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22005 ssa_defined_default_def_p to handle cases which are implicitly
22007 * tree-ssa.c (ssa_defined_default_def_p): New function.
22008 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22009 which are implicitly defined.
22010 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22012 2017-02-17 Richard Biener <rguenther@suse.de>
22014 PR middle-end/79576
22015 * params.def (max-ssa-name-query-depth): Limit to 10.
22017 2017-02-17 Richard Biener <rguenther@suse.de>
22019 PR tree-optimization/79552
22020 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22023 2017-02-17 Richard Biener <rguenther@suse.de>
22026 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22028 2017-02-17 Marek Polacek <polacek@redhat.com>
22030 PR middle-end/79536
22031 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22032 (fold_negate_expr): New wrapper.
22034 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22036 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22037 Correct terminology and de-emphasize pre-standard behavior.
22039 2017-02-16 Alan Modra <amodra@gmail.com>
22041 PR rtl-optimization/79286
22042 * ira.c (def_dominates_uses): New function.
22043 (update_equiv_regs): Don't create an equivalence for insns that
22044 may trap where the register def does not dominate the use.
22046 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22048 PR rtl-optimization/78127
22049 * lra.c (lra): Call lra_eliminate before finish the loop after
22052 2017-02-16 Richard Biener <rguenther@suse.de>
22054 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22056 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22057 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22058 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22059 (isl_val_int_from_wi): New function.
22060 (extract_affine_gmp): Rename to ...
22061 (extract_affine_wi): ... this, take a widest_int.
22062 (extract_affine_int): Just wrap extract_affine_wi.
22063 (add_param_constraints): Use isl_val_int_from_wi.
22064 (add_loop_constraints): Likewise, and extract_affine_wi.
22066 2017-02-15 Jeff Law <law@redhat.com>
22068 PR middle-end/79521
22069 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22070 ira_init_register_move_cost_if_necessary.
22072 2017-02-15 Martin Sebor <msebor@redhat.com>
22074 PR middle-end/32003
22075 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22076 removed in a prior commit.
22078 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22080 PR tree-optimization/79347
22081 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22082 counters during peeling.
22084 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22086 * Makefile.in (site.exp): Remove "set ISLVER".
22088 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22091 * real.c (real_from_integer): Call real_convert even for decimal.
22093 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22096 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22098 2017-02-14 Andrew Pinski <apinski@cavium.com>
22100 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22101 cores and change the partno/implementer to be correct.
22102 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22103 the 'B" as the implementer.
22104 * config/aarch64/aarch64-tune.md: Regenerate.
22106 2017-02-14 Carl Love <cel@us.ibm.com>
22108 * config/rs6000/rs6000.c: Add case statement entry to make the
22109 xvcvuxdsp built-in argument unsigned.
22110 * config/rs6000/vsx.md: Fix the source and return operand types so they
22111 match the instruction definitions from the ISA document. Fix typo
22112 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22115 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22118 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22119 member early_clobber_alts.
22120 * lra-lives.c (reg_early_clobber_p): New.
22121 (process_bb_lives): Use it.
22122 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22123 (debug_operand_data): Initialize early_clobber_alts.
22124 (setup_operand_alternative): Set up early_clobber_alts.
22125 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22126 alternatives to new_insn_reg.
22127 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22129 (lra_update_insn_regno_info): Pass the new arg.
22131 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22133 PR middle-end/79505
22134 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22135 (new_oacc_loop_raw): Don't clear already cleared fields.
22138 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22139 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22140 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22142 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22145 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22147 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22150 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22151 the extra instruction to the right place to store 128-bit constant
22154 2017-02-14 Martin Sebor <msebor@redhat.com>
22156 PR middle-end/79448
22157 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22158 warning for strings of unknown length.
22160 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22162 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22164 2017-02-14 Jeff Law <law@redhat.com>
22167 * ira-costs.c (scan_one_insn): Initialize register move costs
22168 for pseudos seen in USE/CLOBBER insns.
22170 PR tree-optimization/79095
22171 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22172 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22173 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22174 if the operands are known to be not equal, then the resulting range
22176 (intersect_ranges): If the new range is ~[0,0] and the old range is
22177 wide, then prefer ~[0,0].
22178 * tree-vrp.c (overflow_comparison_p_1): New function.
22179 (overflow_comparison_p): New function.
22180 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22181 if NAME is used in an overflow test.
22182 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22183 overflow check that can be expressed as an equality test, then adjust
22184 ops to be that equality test.
22186 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22188 * config/s390/s390-builtin-types.def: Remove flags argument.
22189 * config/s390/s390.c (s390_init_builtins): Likewise.
22191 2017-02-14 Martin Liska <mliska@suse.cz>
22193 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22194 vector. Fix trailing white spaces.
22196 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22198 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22201 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22203 PR rtl-optimization/68664
22204 * config/arm/arm.c (arm_sched_can_speculate_insn):
22205 New function. Declare prototype.
22206 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22208 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22210 PR rtl-optimization/68664
22211 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22213 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22215 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22217 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22218 max skip bytes for function, loop and jump.
22220 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22222 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22223 ABS_EXPR for gimple dump.
22225 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22228 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22230 PR tree-optimization/79408
22231 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22232 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22233 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22234 also if rhs1 is INTEGER_CST.
22236 2017-02-14 Richard Biener <rguenther@suse.de>
22238 PR middle-end/79432
22239 * tree-into-ssa.c (insert_phi_nodes): When the function can
22240 have abnormal edges rewrite SSA names with broken use-def
22241 dominance out of SSA and register them for PHI insertion.
22243 2017-02-13 Martin Sebor <msebor@redhat.com>
22245 PR middle-end/79496
22246 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22247 clearing info.nowrite flag when snprintf size argument is a range.
22249 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22251 * cprop.c (cprop_jump): Add missing space in string literal.
22252 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22253 (get_constraint_for_component_ref): Likewise.
22254 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22255 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22256 * lra-constraints.c (process_alt_operands): Likewise.
22257 * ipa-inline.c (inline_small_functions): Likewise.
22258 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22259 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22260 * trans-mem.c (diagnose_tm_1_op): Likewise.
22261 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22262 (grid_parallel_clauses_gridifiable): Likewise.
22264 * config/nvptx/mkoffload.c (process): Add space in between
22267 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22268 "MOD4_SSE_REGS" and "ALL_REGS".
22270 * spellcheck.c (test_data): Add , in between "foo" and "food".
22272 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22275 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22276 boundary crossing check and subsequent code generation agree.
22278 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22280 * config/aarch64/aarch64.c (has_memory_op): Delete.
22281 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22284 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22286 PR rtl-optimization/79388
22287 PR rtl-optimization/79450
22288 * combine.c (distribute_notes): When removing TEM_INSN for which
22289 corresponding dest has last value recorded, invalidate that last
22292 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22294 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22295 of explicit '@'. Add missing assembly comment marker on branch costs
22298 2017-02-13 Nathan Sidwell <nathan@acm.org>
22300 * gengtype-lex.l (<in_struct>): Add '/'.
22302 2017-02-13 Martin Liska <mliska@suse.cz>
22305 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22307 2017-02-13 Richard Biener <rguenther@suse.de>
22309 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22311 * configure: Re-generate.
22312 * config.in: Likewise.
22313 * graphite-dependences.c: Simplify as if
22314 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22315 * graphite-isl-ast-to-gimple.c: Likewise.
22316 * graphite-optimize-isl.c: Likewise.
22317 * graphite-poly.c: Likewise.
22318 * graphite-sese-to-poly.c: Likewise.
22319 * graphite.h: Likewise.
22320 * toplev.c: Include isl/version.h and use isl_version () for
22321 printing the ISL version.
22322 * doc/install.texi: Update ISL requirement.
22324 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22326 * doc/standards.texi (Standards): Update reference to
22329 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22331 * doc/extend.texi (Named Address Spaces): sourceware.org now
22333 * doc/install.texi (Binaries): Ditto.
22336 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22338 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22339 standard terminology "stringize"/"stringizing" throughout.
22340 * doc/cppinternals.texi: Likewise.
22342 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22344 * doc/extend.texi: Fix some spelling mistakes and typos.
22345 * doc/invoke.texi: Likewise.
22347 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22350 * params.def (inline-min-speedup) Change from 10 to 8.
22352 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22354 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22357 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22360 * ipa-inline-analysis.c (get_minimal_bb): New function.
22361 (record_modified): Use it.
22362 (remap_edge_change_prob): Handle also ancestor functions.
22364 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22366 * doc/contrib.texi (Contributors): Remove broken link into
22367 the Mauve CVS repository.
22369 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22371 PR middle-end/79454
22372 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22373 result computation whenever lhs doesn't have vector mode, not
22374 just when it has BLKmode.
22376 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22378 * doc/makefile.texi (profiledbootstrap): Refer to the
22379 installation instructions only in textual form.
22381 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22384 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22386 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22388 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22389 (Specific): Update mingw-w64 reference.
22391 (Specific): Remove broken link to Renesas RX processor.
22393 2017-02-10 Richard Biener <rguenther@suse.de>
22395 * toplev.c (process_options): Do not mention obsolete graphite
22396 options when printing sorry message about missing graphite support.
22397 Mention -floop-nest-optimize.
22399 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22401 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22402 (vtst_p16): Likewise.
22403 (vtstq_p8): Likewise.
22404 (vtstq_p16): Likewise.
22406 (vtstq_p64): Likewise.
22407 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22408 (vset_lane_p64): New.
22409 (vsetq_lane_p64): New.
22411 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22413 PR tree-optimization/79411
22414 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22415 stmt operands are SSA_NAMEs used in abnormal phis.
22416 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22419 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22422 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22425 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22427 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22429 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22431 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22435 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22436 not arbitrary TREE_CONSTANT.
22439 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22440 "omp declare target link" attribute unless is_global_var.
22441 * omp-offload.c (find_link_var_op): Likewise.
22443 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22444 Chung-Lin Tang <cltang@codesourcery.com>
22446 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22448 (gimplify_adjust_omp_clauses): Don't delete TILE.
22449 (gimplify_omp_for): Deal with TILE.
22450 * internal-fn.c (expand_GOACC_TILE): New function.
22451 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22453 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22454 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22456 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22457 avoid DIV for outermost collapse var.
22458 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22459 Remove out of date comments, fix whitespace.
22460 * omp-general.c (omp_extract_for_data): Deal with tiling.
22461 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22462 adjust OLF_DIM_BASE value.
22463 (struct omp_for_data): Add tiling field.
22464 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22465 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22466 for auto loops. Remove default auto determining, moved to
22467 oacc_loop_fixed_partitions.
22468 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22469 stmts, add e_mask field.
22470 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22471 (oacc_thread_numbers): Use oacc_dim_call.
22472 (oacc_xform_tile): New.
22473 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22474 (finish_oacc_loop): Adjust for ifns vector.
22475 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22476 add case for GOACC_TILE fns.
22477 (oacc_loop_xform_loop): Delete.
22478 (oacc_loop_process): Iterate over call list directly, and add
22479 handling for GOACC_TILE fns.
22480 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22482 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22483 vector partitioning to outer loops. Assign 2 partitions to loops
22484 when available. Add TILE handling.
22485 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22486 (execite_oacc_device_lower): Process GOACC_TILE fns,
22487 ignore unknown specs.
22488 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22489 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22490 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22492 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22494 * configure.ac (ACX_BUGURL): Update.
22495 * configure: Regenerate.
22497 2017-02-09 Richard Biener <rguenther@suse.de>
22499 PR tree-optimization/69823
22500 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22501 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22503 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22505 * config/arc/arc-c.def: Add __NPS400__ definition.
22506 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22507 (TARGET_NPS400): Define.
22509 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22511 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22513 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22514 pointer, arch_info.
22515 (arc_cpu_types): Fill the arch_info field with a pointer into the
22516 arc_arch_types table.
22517 (arc_selected_cpu): Declare.
22518 * config/arc/arc.c (arc_selected_cpu): Make global.
22519 (arc_selected_arch): Delete.
22520 (arc_base_cpu): Delete.
22521 (arc_override_options): Remove references to deleted variables,
22522 update access to arch information.
22523 (ARC_OPT): Update access to arch information.
22524 (ARC_OPTX): Likewise.
22525 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22526 (TARGET_ARC600): Update access to arch information.
22527 (TARGET_ARC601): Likewise.
22528 (TARGET_ARC700): Likewise.
22529 (TARGET_EM): Likewise.
22530 (TARGET_HS): Likewise.
22531 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22534 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22537 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22538 condition/operands for integer GE/LE/GEU/LEU operations.
22540 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22542 PR translation/79397
22543 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22546 2017-02-08 Martin Jambor <mjambor@suse.cz>
22549 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22550 whether allocation happened.
22551 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22552 nothing was allocated.
22554 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22556 PR tree-optimization/79408
22557 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22558 constant, but SSA_NAME with a known integer range, use the minimum
22559 of that range instead of op1 to determine if modulo can be replaced
22560 with its first operand.
22562 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22564 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22566 2017-02-08 Richard Biener <rguenther@suse.de>
22568 PR tree-optimization/71824
22569 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22570 Check all loops contained in the merged region.
22572 2017-02-07 Andrew Pinski <apinski@cavium.com>
22574 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22576 2017-02-07 Andrew Pinski <apinski@cavium.com>
22578 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22579 (thunderxt88): Likewise.
22580 (thunderxt81): Disable LSE and change v8.1 to v8.
22581 (thunderxt83): Likewise.
22583 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22584 Richard Biener <rguenther@suse.de>
22586 PR middle-end/79399
22587 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22588 type from int to size_t.
22589 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22591 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22593 PR rtl-optimization/79386
22594 * cprop.c (bypass_conditional_jumps): Initialize
22595 bypass_last_basic_block already before splitting bbs after
22596 unconditional traps...
22597 (bypass_conditional_jumps): ... rather than here.
22600 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22601 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22602 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22603 fix -masm=intel patterns.
22605 2017-02-07 Richard Biener <rguenther@suse.de>
22607 PR tree-optimization/79256
22608 PR middle-end/79278
22609 * builtins.c (get_object_alignment_2): Use min_align_of_type
22610 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22611 and ADJUST_FIELD_ALIGN.
22613 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22615 * doc/tm.texi: Regenerate.
22616 * stor-layout.c (layout_decl): Adjust.
22617 (update_alignment_for_field): Likewise.
22618 (place_field): Likewise.
22619 (min_align_of_type): Likewise.
22620 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22621 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22622 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22623 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22624 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22625 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22626 * config/i386/i386.c (x86_field_alignment): Likewise.
22627 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22628 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22629 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22630 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22631 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22632 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22636 2017-01-30 Richard Biener <rguenther@suse.de>
22638 PR tree-optimization/79256
22639 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22640 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22643 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22645 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22646 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22647 builtins to SImode and emit a zero-extend, if necessary.
22649 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22651 * docs/invoke.texi (RISC-V Options): Alphabetize.
22653 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22655 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22658 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22660 * config/riscv/riscv.c: New file.
22661 * common/config/riscv/riscv-common.c: Likewise.
22662 * config.gcc: Likewise.
22663 * config/riscv/constraints.md: Likewise.
22664 * config/riscv/elf.h: Likewise.
22665 * config/riscv/generic.md: Likewise.
22666 * config/riscv/linux.h: Likewise.
22667 * config/riscv/multilib-generator: Likewise.
22668 * config/riscv/peephole.md: Likewise.
22669 * config/riscv/pic.md: Likewise.
22670 * config/riscv/predicates.md: Likewise.
22671 * config/riscv/riscv-builtins.c: Likewise.
22672 * config/riscv/riscv-c.c: Likewise.
22673 * config/riscv/riscv-ftypes.def: Likewise.
22674 * config/riscv/riscv-modes.def: Likewise.
22675 * config/riscv/riscv-opts.h: Likewise.
22676 * config/riscv/riscv-protos.h: Likewise.
22677 * config/riscv/riscv.h: Likewise.
22678 * config/riscv/riscv.md: Likewise.
22679 * config/riscv/riscv.opt: Likewise.
22680 * config/riscv/sync.md: Likewise.
22681 * config/riscv/t-elf-multilib: Likewise.
22682 * config/riscv/t-linux: Likewise.
22683 * config/riscv/t-linux-multilib: Likewise.
22684 * config/riscv/t-riscv: Likewise.
22685 * configure.ac: Likewise.
22686 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22687 Waterman as RISC-V maintainers.
22688 * doc/install.texi: Add RISC-V entries.
22689 * doc/invoke.texi: Add RISC-V options section.
22690 * doc/md.texi: Add RISC-V constraints section.
22691 * configure: Regenerated.
22693 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22696 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22697 false values to be constant vectors with all 0 or all 1 bits set.
22698 (vcondu<mode><mode>): Likewise.
22699 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22701 (fpmask_comparison_operator): Update comment.
22702 (vecint_comparison_operator): New predicate.
22703 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22704 vector conditionals when the true and false values are constant
22705 vectors with all 0 bits or all 1 bits set.
22707 2017-02-06 Martin Sebor <msebor@redhat.com>
22709 PR tree-optimization/79376
22710 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22712 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22714 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22715 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22716 to simplify split condition.
22718 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22720 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22721 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22724 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22726 PR rtl-optimization/68664
22727 * target.def (can_speculate_insn): New hook.
22728 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22729 * doc/tm.texi: Regenerate.
22730 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22731 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22732 (rs6000_sched_can_speculate_insn): New function.
22734 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22736 PR tree-optimization/79284
22737 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22738 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22739 vectorizable_mask_load_store, vectorizable_operation,
22740 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22741 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22742 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22743 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22744 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22745 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22746 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22747 is_gimple_assign (stmt). Replace another such test with
22748 is_gimple_assign (stmt).
22750 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22753 * config/avr/avr.c (rtl-iter.h): Include it.
22754 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22755 (avr_legitimate_combined_insn): ...and implementation.
22757 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22759 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22760 * config/s390/s390.c (s390_const_operand_ok)
22761 (s390_canonicalize_comparison, s390_extract_part)
22762 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22763 (s390_contiguous_bitmask_p, s390_rtx_costs)
22764 (legitimize_pic_address): Likewise.
22765 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22766 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22767 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22768 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22769 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22771 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22773 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22774 REGNO($0) == REGNO($1).
22776 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22778 * config/s390/linux.h(SIZE_TYPE): Add comment.
22780 2017-02-06 Julian Brown <julian@codesourcery.com>
22781 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22782 Virendra Pathak <virendra.pathak@broadcom.com>
22784 * config/aarch64/aarch64-cores.def: Change the scheduler
22786 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22787 * config/aarch64/thunderx2t99.md: New file.
22789 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22791 * doc/standards.texi (Go Language): Update link to language
22794 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22796 * tree-eh.c (lower_resx): Sanitize profile.
22797 (cleanup_empty_eh_move_lp): Likewise.
22799 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22802 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22804 * cfgloopmanip.h (loop_version): Update prototype.
22805 * modulo-sched.c (sms_schedule): Update call of loop_version.
22806 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22807 * tree-parloops.c (gen_parallel_loop): Likewise.
22808 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22809 * tree-ssa-loop-split.c (split_loop): Likewise.
22810 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22811 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22813 2017-02-05 Martin Liska <mliska@suse.cz>
22816 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22818 (print_operand_address): Initialize a struct to zero.
22820 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22822 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22823 garbage collector only in textual form.
22825 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22827 * doc/extend.texi (x86 specific memory model extensions for
22828 transactional memory): Simplify a phrase.
22830 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22833 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22834 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22835 (atomic_storedi_1): Likewise.
22837 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22839 PR tree-optimization/79338
22840 * tree-parloops.c (gather_scalar_reductions): Don't call
22841 vect_analyze_loop_form for loop->inner before destroying loop's
22844 2017-02-03 Martin Sebor <msebor@redhat.com>
22846 PR tree-optimization/79327
22847 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22848 when precision has resulted in leading zeros.
22849 (format_integer): Adjust the likely counter to assume an unknown
22850 argument that may be zero is non-zero.
22852 2017-02-03 Jason Merrill <jason@redhat.com>
22855 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22856 avoid copying non-taken branch.
22858 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22860 PR tree-optimization/79340
22861 * tree-vect-loop.c (vectorizable_reduction): Release
22862 vec_defs elements after safe_splicing them into other vectors.
22865 PR tree-optimization/79327
22866 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22867 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22869 (format_integer): Use wide_int_to_tree instead of build_int_cst
22870 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22871 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22872 of shortest and longest sequence.
22874 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22876 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22877 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22879 2017-02-03 Walter Lee <walt@tilera.com>
22882 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22883 after initial stackframe link reg save.
22884 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22886 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22889 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22890 wu for stxssp alternative.
22892 2017-02-03 Martin Sebor <msebor@redhat.com>
22894 PR tree-optimization/79352
22895 * gimple-fold.c (get_range_strlen): Add argument.
22896 (get_range_strlen): Change return type to bool.
22897 (get_maxval_strlen): Pass in a dummy argument.
22898 * gimple-fold.h (get_range_strlen): Change return type to bool.
22899 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22900 * tree.h (array_at_struct_end_p): Add argument.
22901 * tree.c (array_at_struct_end_p): Handle it.
22903 2017-02-03 Martin Liska <mliska@suse.cz>
22906 * multiple_target.c (create_dispatcher_calls): Redirect edge
22907 from a caller of a dispatcher.
22908 (expand_target_clones): Make the clones local.
22909 (ipa_target_clone): Do both target clones and resolvers.
22910 (ipa_dispatcher_calls): Remove the pass.
22911 (pass_dispatcher_calls::gate): Likewise.
22912 (make_pass_dispatcher_calls): Likewise.
22913 * passes.def (pass_target_clone): Put as very first IPA early
22916 2017-02-03 Martin Liska <mliska@suse.cz>
22918 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22919 in case of a function with ifunc attribute.
22921 2017-02-03 Martin Liska <mliska@suse.cz>
22923 * cgraph.c (cgraph_node::dump): Dump function version info.
22924 * symtab.c (symtab_node::dump_base): Add missing new line.
22926 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22928 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22929 (ifcombine_ifandif): Use it.
22931 2017-02-03 Martin Liska <mliska@suse.cz>
22933 * doc/invoke.texi: Document default value for
22934 use-after-scope-direct-emission-threshold.
22936 2017-02-03 Martin Liska <mliska@suse.cz>
22938 PR tree-optimization/79339
22939 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22940 (format_floating): Likewise.
22942 2017-02-03 Martin Liska <mliska@suse.cz>
22945 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22947 (ipa_node_params_t::remove): Likewise.
22948 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22949 initialization from removed ipa_node_params_t::insert.
22950 (ipa_node_params::~ipa_node_params): Move from removed
22951 ipa_node_params_t::release.
22952 * symbol-summary.h (symbol_summary::m_released): New member.
22953 Do not release a summary twice. Do not allow to call finalizer
22954 for types of a summary that live in GGC memory.
22956 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22958 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22961 2017-02-02 Martin Sebor <msebor@redhat.com>
22963 PR middle-end/79275
22964 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22965 (format_string): Tighten up the range of output for non-constant
22966 strings and correct the expected range for wide non-constant strings.
22968 2017-02-02 Martin Sebor <msebor@redhat.com>
22970 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22972 PR middle-end/32003
22973 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22975 (-fdump-tree-@var): Add to index and document how to come up
22976 with pass-specific option and dump file names.
22977 (-fdump-passes): Clarify where to look for output.
22979 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22981 PR middle-end/77445
22982 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22983 statistics of the analyzed path; allow threading for speed when
22984 any of BBs along the path are optimized for speed.
22986 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22988 PR middle-end/78468
22989 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22990 settings of the virtual registers.
22993 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22995 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22996 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22999 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23001 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23002 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23004 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23006 * config/s390/s390.md: Add missing comments with the expanded
23008 * config/s390/vector.md: Likewise.
23009 * config/s390/vx-builtins.md: Likewise.
23011 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23014 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23015 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23016 conditions on a single line.
23018 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23020 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23021 __S390_VX__ to __VX__.
23023 2017-02-01 Andrew Pinski <apinski@cavium.com>
23025 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23026 stmt_info to record_stmt_cost.
23027 (vect_get_known_peeling_cost): Pass stmt_info if known to
23029 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23030 cpu_vector_cost field into
23031 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23032 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23033 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23034 splitting of scalar_stmt_cost and vec_stmt_cost.
23035 (thunderx_vector_cost): Likewise.
23036 (cortexa57_vector_cost): LIkewise.
23037 (exynosm1_vector_cost): Likewise.
23038 (xgene1_vector_cost): Likewise.
23039 (thunderx2t99_vector_cost): Improve after the splitting of the two
23041 (aarch64_builtin_vectorization_cost): Update for the splitting of
23042 scalar_stmt_cost and vec_stmt_cost.
23044 2017-02-01 Torvald Riegel <triegel@redhat.com>
23045 Richard Henderson <rth@redhat.com>
23047 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23048 conditional on existance of a fast atomic load.
23049 * optabs-query.c (can_atomic_load_p): New function.
23050 * optabs-query.h (can_atomic_load_p): Declare it.
23051 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23052 no fast atomic load is available for the particular size of access.
23053 (expand_atomic_compare_and_swap): Likewise.
23054 (expand_atomic_load): Likewise.
23055 (expand_atomic_store): Likewise.
23056 (expand_atomic_fetch_op): Likewise.
23057 * testsuite/lib/target-supports.exp
23058 (check_effective_target_sync_int_128): Remove x86 because it provides
23059 no fast atomic load.
23060 (check_effective_target_sync_int_128_runtime): Likewise.
23062 2017-02-01 Richard Biener <rguenther@suse.de>
23064 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23065 (graphite_transform_loops): Provide opt-info for optimized nests.
23066 * tree-parloop.c (parallelize_loops): Provide opt-info for
23067 parallelized loops.
23069 2017-02-01 Richard Biener <rguenther@suse.de>
23071 PR middle-end/79315
23072 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23073 was not set before.
23075 2017-02-01 Richard Biener <rguenther@suse.de>
23077 PR tree-optimization/71824
23078 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23079 Verify the loops are valid in the merged SESE region.
23080 (scop_detection::can_represent_loop_1): Check analyzing the
23081 evolution of the number of iterations in the region succeeds.
23083 2017-01-31 Ian Lance Taylor <iant@golang.org>
23085 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23086 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23088 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23090 PR preprocessor/79210
23091 * input.c (get_substring_ranges_for_loc): Replace line_width
23092 assertion with error-handling.
23094 2017-01-31 Richard Biener <rguenther@suse.de>
23096 PR tree-optimization/77318
23097 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23098 (create_pw_aff_from_tree): Take loop parameter.
23099 (add_condition_to_pbb): Pass loop of the condition to
23100 create_pw_aff_from_tree.
23102 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23104 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23105 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23107 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23111 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23113 (convert_int_to_float128): Likewise.
23114 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23115 (convert_int_to_float128): Likewise.
23116 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23117 (UNSPEC_IEEE128_CONVERT): Likewise.
23118 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23119 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23120 Use local variables for IBM extended format.
23121 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23122 (fix_trunc<mode>si2_fprs): Likewise.
23123 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23124 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23125 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23126 to know that we can now have integers of all sizes in vector
23128 (fix<uns>_<mode>di2_hw): Likewise.
23129 (float<uns>_<mode>si2_hw): Likewise.
23130 (fix_<mode>si2_hw): Likewise.
23131 (fixuns_<mode>si2_hw): Likewise.
23132 (float<uns>_<mode>di2_hw): Likewise.
23133 (float_<mode>di2_hw): Likewise.
23134 (float_<mode>si2_hw): Likewise.
23135 (floatuns_<mode>di2_hw): Likewise.
23136 (floatuns_<mode>si2_hw): Likewise.
23137 (xscvqp<su>wz_<mode>): Delete, no longer used.
23138 (xscvqp<su>dz_<mode>): Likewise.
23139 (xscv<su>dqp_<mode>): Likewise.
23140 (ieee128_mfvsrd_64bit): Likewise.
23141 (ieee128_mfvsrd_32bit): Likewise.
23142 (ieee128_mfvsrwz): Likewise.
23143 (ieee128_mtvsrw): Likewise.
23144 (ieee128_mtvsrd_64bit): Likewise.
23145 (ieee128_mtvsrd_32bit): Likewise.
23147 2017-01-31 Martin Liska <mliska@suse.cz>
23150 * ipa-prop.c (ipa_free_all_node_params): Call release method
23151 instead of ~sumbol_summary to not to trigger double times
23154 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23156 PR tree-optimization/71691
23157 * bitmap.h (class auto_bitmap): New.
23158 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23159 is_maybe_undefined instead of ssa_undefined_value_p.
23161 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23163 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23164 __S390_ARCH_LEVEL__ to __ARCH__.
23166 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23168 PR tree-optimization/79267
23169 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23170 if should_remove_lhs_p is true.
23172 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23175 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23176 (add_alignment_attribute): New.
23177 (base_type_die): Add alignment attribute.
23178 (subrange_type_die): Likewise.
23179 (modified_type_die): Likewise.
23180 (gen_array_type_die): Likewise.
23181 (gen_descr_array_type_die: Likewise.
23182 (gen_enumeration_type_die): Likewise.
23183 (gen_subprogram_die): Likewise.
23184 (gen_variable_die): Likewise.
23185 (gen_field_die): Likewise.
23186 (gen_ptr_to_mbr_type_die): Likewise.
23187 (gen_struct_or_union_type_die): Likewise.
23188 (gen_subroutine_type_die): Likewise.
23189 (gen_typedef_die): Likewise.
23190 (base_type_cmp): Compare alignment attribute.
23192 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23195 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23196 (setb_unsigned) New pattern for setb with CCUNS.
23197 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23198 subfc./subfe sequence to avoid overflow problems. Generate a
23199 shorter sequence with cmpld/setb for power9.
23200 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23201 for generating subfc. instruction.
23202 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23203 now uses this instruction.
23205 2017-01-30 Ian Lance Taylor <iant@google.com>
23208 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23209 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23211 2017-01-30 Martin Sebor <msebor@redhat.com>
23213 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23214 Move constant to the right of a relational operator.
23215 (get_mpfr_format_length, format_character, format_string): Ditto.
23216 (should_warn_p, maybe_warn): Same.
23218 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23220 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23223 * asan.c (get_translation_unit_decl): Remove function.
23224 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23226 2017-01-30 Martin Liska <mliska@suse.cz>
23228 PR gcov-profile/79259
23229 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23230 -fprofile-generate.
23232 2017-01-30 Martin Liska <mliska@suse.cz>
23235 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23236 Initialize variables with NULL value.
23238 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23241 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23243 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23245 2017-01-30 Richard Biener <rguenther@suse.de>
23247 PR tree-optimization/79276
23248 * tree-vrp.c (process_assert_insertions): Properly adjust common
23249 when removing a duplicate.
23251 * gcc.dg/torture/pr79276.c: New testcase.
23253 2017-01-30 Richard Biener <rguenther@suse.de>
23255 PR tree-optimization/79256
23256 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23257 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23259 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23261 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23264 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23265 ("*r<noxa>sbg_<mode>_sll_bitmask")
23266 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23267 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23268 Use contiguous_bitmask_nowrap_operand.
23270 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23273 * config/rs6000/altivec.h (vec_xl): Revise #define.
23274 (vec_xst): Likewise.
23276 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23278 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23280 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23282 PR rtl-optimization/79194
23283 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23284 traps before call to bypass_conditional_jumps.
23286 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23288 PR tree-optimization/71374
23289 * lra-constraints.c (check_conflict_input_operands): New.
23290 (match_reload): Use it.
23292 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23295 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23296 account to calculate conflict_set.
23298 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23300 PR rtl-optimization/78559
23301 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23302 other_insn in combine.
23304 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23306 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23307 uint16_type_node for BT_UINT16.
23309 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23311 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23312 "RTL Tests" to menu.
23313 (GIMPLE Tests): New node.
23314 (RTL Tests): New node.
23316 2017-01-27 Richard Biener <rguenther@suse.de>
23318 PR tree-optimization/79245
23319 * tree-loop-distribution.c (distribute_loop): Apply cost
23320 modeling also to detected patterns.
23322 2017-01-27 Richard Biener <rguenther@suse.de>
23324 PR tree-optimization/71433
23325 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23326 (compare_assert_loc): New function.
23327 (process_assert_insertions): Sort and optimize assert locations
23328 to remove duplicates and push down identical assertions on
23329 edges to their destination block.
23331 2017-01-27 Richard Biener <rguenther@suse.de>
23333 PR tree-optimization/79244
23334 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23335 out SSA names even if abnormal.
23337 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23339 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23340 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23341 instead of MPFR_RNDN.
23343 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23346 * arm.c (arm_option_override): Don't call build_target_option_node
23347 until after doing all option overrides.
23348 (arm_valid_target_attribute_tree): Likewise.
23350 2017-01-27 Martin Liska <mliska@suse.cz>
23352 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23353 for {cd}tors and C++ {cd}tors.
23355 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23357 * config/s390/s390.md ("*setmem_long_and")
23358 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23360 2017-01-26 Martin Sebor <msebor@redhat.com>
23362 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23365 2017-01-26 Martin Sebor <msebor@redhat.com>
23367 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23368 HAVE_DFmode before using XFmode or DFmode.
23369 (parse_directive): Avoid using the z length modifier to avoid
23370 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23372 PR middle-end/78703
23373 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23374 to accept adjustment as an array.
23375 (get_int_range): New function.
23376 (struct directive): Make width and prec arrays.
23377 (directive::set_width, directive::set_precision): Call get_int_range.
23378 (format_integer, format_floating): Handle width and precision ranges.
23379 (format_string, parse_directive): Same.
23381 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23384 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23385 comdat_type_p set, just clone them, but keep the children in the
23389 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23390 which have direct callers with -fvar-tracking-assignments enabled
23392 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23393 inside of type units.
23395 2017-01-26 Martin Sebor <msebor@redhat.com>
23397 PR middle-end/78703
23398 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23400 (struct format_result): Replace number_chars, number_chars_min,
23401 and number_chars_max with a single member of struct result_range.
23403 (format_result::operator+=): Adjust.
23404 (struct fmtresult): Remove bounded. Handle likely and unlikely
23406 (fmtresult::adjust_for_width_or_precision): New function.
23407 (fmtresult:type_max_digits): New function.
23408 (bytes_remaining): Handle likely and unlikely counters.
23409 (min_bytes_remaining): Remove.
23410 (format_percent): Simplify.
23411 (format_integer, format_floating): Set likely and unlikely counters.
23412 (get_string_length, format_character, format_string): Same.
23413 (format_plain, should_warn_p): New function.
23414 (maybe_warn): Call should_warn_p. Update diagnostic messages
23415 and handle those for all directives, including plain strings.
23416 (format_directive): Handle likely and unlikely counters.
23417 Remove unnecessary quoting from diagnostics. Add an informational
23419 (add_bytes): Remove.
23420 (pass_sprintf_length::compute_format_length): Simplify.
23421 (try_substitute_return_value): Handle likely and unlikely counters.
23423 2017-01-26 Carl Love <cel@us.ibm.com>
23425 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23426 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23428 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23431 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23432 endianess for subregs into account.
23433 * lra-constraints.c (lra_constraints): Do risky transformations
23434 always on the first iteration.
23435 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23436 last_call_used_reg_set.
23437 (process_bb_lives): Define and use last_call_used_reg_set.
23438 * lra.c (lra): Always continue after lra_constraints on the first
23441 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23443 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23445 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23447 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23449 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23450 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23451 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23452 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23453 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23454 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23455 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23456 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23457 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23459 2017-01-26 Marek Polacek <polacek@redhat.com>
23462 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23463 for the third operand.
23465 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23467 PR middle-end/79236
23468 * omp-low.c (struct omp_context): Add simt_stmt field.
23469 (scan_omp_for): Return omp_context *.
23470 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23471 context to the _simt_ SIMD stmt.
23472 (lower_omp_for): For combined SIMD with sibling _simt_
23473 SIMD, make sure to use the same decls in _looptemp_
23474 clauses as in the sibling.
23476 2017-01-26 David Sherwood <david.sherwood@arm.com>
23478 PR middle-end/79212
23479 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23482 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23485 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23486 emit fld b; fld a; if possible.
23488 * brig-builtins.def: Update copyright years.
23489 * config/arm/arm_acle_builtins.def: Update copyright years.
23491 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23494 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23495 constraint instead of o for the stxsd instruction.
23497 2017-01-25 Carl Love <cel@us.ibm.com>
23499 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23500 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23502 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23504 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23506 2017-01-25 Richard Biener <rguenther@suse.de>
23508 PR tree-optimization/69264
23509 * target.def (vector_alignment_reachable): Improve documentation.
23510 * doc/tm.texi: Regenerate.
23511 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23513 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23514 earlier changes with respect to TYPE_USER_ALIGN.
23515 (vector_alignment_reachable_p): Likewise. Improve dumping.
23517 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23520 * config/arm/arm.md (xordi3): Force constant operand into a register
23523 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23525 * doc/invoke.texi (-fstore-merging): Correct default optimization
23526 levels at which it is enabled.
23527 (-O): Move -fstore-merging from list to...
23530 2017-01-25 Richard Biener <rguenther@suse.de>
23533 * omp-expand.c: Include debug.h.
23534 (expand_omp_taskreg): Make sure to generate early debug before
23535 outlining anything from a function.
23536 (expand_omp_target): Likewise.
23537 (grid_expand_target_grid_body): Likewise.
23539 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23542 * asan.c (get_translation_unit_decl): New function.
23543 (asan_add_global): Extract modules file name from globals
23544 TRANSLATION_UNIT_DECL name.
23546 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23549 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23550 for long calls with APCS frame and VFP.
23552 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23554 * cfg.c (original_copy_tables_initialized_p): New function.
23555 * cfg.h (original_copy_tables_initialized_p): New decl.
23556 * cfgrtl.c (relink_block_chain): Guard the call to
23557 free_original_copy_tables with a call to
23558 original_copy_tables_initialized_p.
23559 * cgraph.h (symtab_node::native_rtl_p): New decl.
23560 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23561 (symtab_node::needed_p): Don't assert for early assembly output
23562 for __RTL functions.
23563 (cgraph_node::finalize_function): Set "force_output" for __RTL
23565 (cgraph_node::analyze): Bail out early for __RTL functions.
23566 (analyze_functions): Update assertion to support __RTL functions.
23567 (cgraph_node::expand): Bail out early for __RTL functions.
23568 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23570 * function.h (struct function): Update comment for field
23572 * gimple-expr.c: Include "tree-pass.h".
23573 (gimple_has_body_p): Return false for __RTL functions.
23574 * Makefile.in (OBJS): Add run-rtl-passes.o.
23575 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23577 (gcc::pass_manager::get_clean_slate): New accessor.
23578 * passes.c: Include "insn-addr.h".
23579 (should_skip_pass_p): Add logging. Update logic for running
23580 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23581 property-provider override so it is only done for gimple passes.
23583 (skip_pass): New function.
23584 (execute_one_pass): Call skip_pass when skipping passes.
23585 * read-md.c (md_reader::read_char): Support filtering
23586 the input to a subset of line numbers.
23587 (md_reader::md_reader): Initialize fields
23588 m_first_line and m_last_line.
23589 (md_reader::read_file_fragment): New function.
23590 * read-md.h (md_reader::read_file_fragment): New decl.
23591 (md_reader::m_first_line): New field.
23592 (md_reader::m_last_line): New field.
23593 * read-rtl-function.c (function_reader::create_function): Only
23594 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23595 curr_properties. Set DECL_INITIAL to a dummy block.
23596 (read_rtl_function_body_from_file_range): New function.
23597 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23599 * run-rtl-passes.c: New file.
23600 * run-rtl-passes.h: New file.
23602 2017-01-24 Jeff Law <law@redhat.com>
23604 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23607 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23609 PR tree-optimization/79159
23610 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23611 (record_nonwrapping_iv): Improve boundary using above function if no
23612 value range information.
23614 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23615 Martin Jambor <mjambor@suse.cz>
23617 * brig-builtins.def: New file.
23618 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23619 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23620 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23621 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23622 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23623 * builtin-types.def (BT_INT8): New.
23624 (BT_INT16): Likewise.
23625 (BT_UINT8): Likewise.
23626 (BT_UINT16): Likewise.
23627 (BT_FN_ULONG): Likewise.
23628 (BT_FN_UINT_INT): Likewise.
23629 (BT_FN_UINT_ULONG): Likewise.
23630 (BT_FN_UINT_LONG): Likewise.
23631 (BT_FN_UINT_PTR): Likewise.
23632 (BT_FN_ULONG_PTR): Likewise.
23633 (BT_FN_INT8_FLOAT): Likewise.
23634 (BT_FN_INT16_FLOAT): Likewise.
23635 (BT_FN_UINT32_FLOAT): Likewise.
23636 (BT_FN_UINT16_FLOAT): Likewise.
23637 (BT_FN_UINT8_FLOAT): Likewise.
23638 (BT_FN_UINT64_FLOAT): Likewise.
23639 (BT_FN_UINT16_UINT32): Likewise.
23640 (BT_FN_UINT32_UINT16): Likewise.
23641 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23642 (BT_FN_INT_PTR_INT): Likewise.
23643 (BT_FN_UINT_PTR_UINT): Likewise.
23644 (BT_FN_LONG_PTR_LONG): Likewise.
23645 (BT_FN_ULONG_PTR_ULONG): Likewise.
23646 (BT_FN_VOID_UINT64_UINT64): Likewise.
23647 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23648 (BT_FN_INT8_INT8_INT8): Likewise.
23649 (BT_FN_INT16_INT16_INT16): Likewise.
23650 (BT_FN_INT_INT_INT): Likewise.
23651 (BT_FN_UINT_FLOAT_UINT): Likewise.
23652 (BT_FN_FLOAT_UINT_UINT): Likewise.
23653 (BT_FN_ULONG_UINT_UINT): Likewise.
23654 (BT_FN_ULONG_UINT_PTR): Likewise.
23655 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23656 (BT_FN_UINT_UINT_UINT): Likewise.
23657 (BT_FN_VOID_UINT_PTR): Likewise.
23658 (BT_FN_UINT_UINT_PTR: Likewise.
23659 (BT_FN_UINT32_UINT64_PTR): Likewise.
23660 (BT_FN_INT_INT_UINT_UINT): Likewise.
23661 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23662 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23663 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23664 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23665 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23666 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23667 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23668 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23669 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23670 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23671 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23672 * doc/frontends.texi: List BRIG FE.
23673 * doc/install.texi (Testing): Add BRIG tesring requirements.
23674 * doc/invoke.texi (Overall Options): Mention BRIG.
23675 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23677 2017-01-24 Richard Biener <rguenther@suse.de>
23679 PR translation/79208
23680 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23682 2017-01-24 Martin Jambor <mjambor@suse.cz>
23685 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23686 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23687 and known_contexts.
23689 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23691 PR middle-end/79123
23692 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23693 casts from signed to unsigned really don't have a range.
23695 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23697 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23698 GMP_RNDx for compatiblity.
23700 2017-01-24 Martin Liska <mliska@suse.cz>
23703 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23704 that would prevent us to call alloca with -1 as argument.
23706 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23708 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23709 Avoid -Wformat-security warning.
23711 2017-01-23 Andrew Pinski <apinski@cavium.com>
23713 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23716 2017-01-23 Martin Sebor <msebor@redhat.com>
23718 PR middle-end/78703
23719 * gimple-ssa-sprintf.c (warn_level): New global.
23720 (format_integer): Use it here and throughout the rest of the file.
23721 Use the same switch to compute sign as base.
23722 (maybe_warn): New function.
23723 (format_directive): Factor out warnings into maybe_warn.
23724 Add debugging output. Use warn_level.
23725 (add_bytes): Use warn_level.
23726 (pass_sprintf_length::compute_format_length): Add debugging output.
23727 (try_substitute_return_value): Same.
23728 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23730 PR middle-end/78703
23731 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23732 (struct fmtresult, format_integer, format_floating): Adjust.
23733 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23734 (get_string_length, format_string,format_directive): Same.
23735 (pass_sprintf_length::compute_format_length): Same.
23736 (try_substitute_return_value): Simplify slightly.
23738 PR middle-end/78703
23739 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23740 (fmtresult::operator+=): Outlined.
23741 (struct fmtresult): Add ctors.
23742 (struct conversion_spec): Rename...
23743 (struct directive): ...to this. Add and remove data members.
23744 (directive::set_width, directive::set_precision): New functions.
23745 (format_percent): Use fmtresult ctor.
23746 (get_width_and_precision): Remove.
23747 (format_integer): Make naming changes. Avoid computing width and
23749 (format_floating): Same. Adjust indentation.
23750 (format_character, format_none): New functions.
23751 (format_string): Moved character handling to format_character.
23752 (format_directive): Remove arguments, change return type.
23753 (parse_directive): New function.
23754 (pass_sprintf_length::compute_format_length): Move directive
23755 parsing to parse_directive.
23757 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23759 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23760 (assign_assembler_name_if_needed): ... this.
23761 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23762 (assign_assembler_name_if_needed): ... this.
23763 (free_lang_data_in_cgraph): Adjust callers.
23764 * cgraphunit.c (cgraph_node::analyze): Likewise.
23765 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23768 2017-01-23 Richard Biener <rguenther@suse.de>
23770 PR tree-optimization/79088
23771 PR tree-optimization/79188
23772 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23773 resetting loop bounds after last path deletion. Reset loop
23774 bounds of the target loop, make code match the comments.
23775 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23776 Make sure loops need no fixups.
23778 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23780 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23781 exponent support with double type for first argument.
23782 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23783 type returned by __builtin_vec_extract_sig,
23784 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23785 functions from "vector int" to "vector unsigned int" or from
23786 "vector long long int" to "vector unsigned long long int".
23787 Changed type returned by __builtin_vec_extract_exp,
23788 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23789 functions from "vector int" to "vector unsigned int" or from
23790 "vector long long int" to "vector unsigned long long int".
23791 Changed return type of __builtin_vec_test_data_class,
23792 __builtin_vec_test_data_class_sp, and
23793 __builtin_vec_test_data_class_dp from "vector int" to
23794 "vector bool int" or from "vector long long int" to "vector bool
23795 long long int" and changed second argument type from "unsigned
23796 int" to "int". Added new overloaded function forms "vector float
23797 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23798 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23799 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23800 double, vector unsigned long long int)" and "vector double
23801 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23802 long int)". Changed return type of
23803 __builtin_scalar_test_data_class and
23804 __builtin_scalar_test_data_class_sp and
23805 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23806 int" and changed second argument from "unsigned int" to "int".
23807 Changed type returned by __builtin_scalar_test_neg,
23808 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23809 from "int" to "bool int". Added new overloaded function form
23810 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23811 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23812 exponent double-precision with floating point first argument.
23813 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23814 documentation of scalar_test_data_class, scalar_test_neg,
23815 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23816 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23817 vec_test_data_class built-in functions to reflect refinements in
23818 their type signatures.
23820 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23822 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23824 (aarch64_elf_asm_destructor): Likewise.
23826 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23828 PR rtl-optimization/78634
23829 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23830 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23831 * ifcvt.c (noce_try_cmove): Add missing cost check.
23833 PR rtl-optimization/71724
23834 * combine.c (if_then_else_cond): Look for situations where it is
23835 beneficial to undo the work of one of the recursive calls.
23837 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23839 PR tree-optimization/70754
23840 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23841 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23842 combined stmt before it if not NULL.
23843 (combine_chains): Process refs reversely and compute dominance point
23846 2017-01-23 Martin Liska <mliska@suse.cz>
23848 PR tree-optimization/79196
23849 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23850 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23852 (strlen_optimize_stmt): Call the renamed function.
23854 2017-01-23 Michael Matz <matz@suse.de>
23856 PR tree-optimization/78384
23857 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23859 2017-01-23 Richard Biener <rguenther@suse.de>
23861 PR tree-optimization/79186
23862 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23863 both incoming edges before moving an assert.
23865 2017-01-23 Martin Jambor <mjambor@suse.cz>
23867 * ipa-prop.c (load_from_param_1): Removed.
23868 (load_from_unmodified_param): Bits from load_from_param_1 put back
23870 (load_from_param): Removed.
23871 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23872 with stmt. Reverted back to use of load_from_unmodified_param.
23874 2017-01-23 Martin Jambor <mjambor@suse.cz>
23877 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23878 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23879 field a pointer to garbage collected vector, mark lattices and
23880 ipcp_orig_node with GTY((skip)).
23881 (ipa_get_param_count): Adjust to descriptors being a pointer.
23882 (ipa_get_param): Likewise.
23883 (ipa_get_type): Likewise.
23884 (ipa_get_param_move_cost): Likewise.
23885 (ipa_set_param_used): Likewise.
23886 (ipa_get_controlled_uses): Likewise.
23887 (ipa_set_controlled_uses): Likewise.
23888 (ipa_is_param_used): Likewise.
23889 (ipa_node_params_t): Move into garbage collector. New methods insert
23891 (ipa_node_params_sum): Annotate wth GTY(()).
23892 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23894 (ipa_load_from_parm_agg): Adjust declaration.
23895 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23896 * ipa-profile.c (ipa_profile): Likewise.
23897 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23898 (ipa_populate_param_decls): Make descriptors parameter garbage
23900 (ipa_dump_param): Adjust to descriptors being a pointer.
23901 (ipa_alloc_node_params): Likewise.
23902 (ipa_initialize_node_params): Likewise.
23903 (load_from_param_1): Make descriptors parameter garbage collected.
23904 (load_from_unmodified_param): Likewise.
23905 (load_from_param): Likewise.
23906 (ipa_load_from_parm_agg): Likewise.
23907 (ipa_node_params::~ipa_node_params): Removed.
23908 (ipa_free_all_node_params): Remove call to delete operator.
23909 (ipa_node_params_t::insert): New.
23910 (ipa_node_params_t::remove): Likewise.
23911 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23912 copy known_csts and known_contexts vectors.
23913 (ipa_read_node_info): Adjust to descriptors being a pointer.
23914 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23916 (ipcp_transform_function): Make descriptors variable garbage
23919 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23921 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23922 * config/i386/avx512dqintrin.h: Ditto.
23923 * config/i386/avx512fintrin.h: Ditto.
23924 * config/i386/i386.c: Handle new builtins.
23925 * config/i386/i386-builtin.def: Add new builtins.
23926 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23927 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23929 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23930 Martin Liska <mliska@suse.cz>
23932 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23933 * asan.c (asan_expand_poison_ifn): Support stores and use
23934 appropriate ASAN report function.
23935 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23936 * internal-fn.def (ASAN_POISON_USE): Declare.
23937 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23938 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23939 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23940 ASAN_POISON calls w/o LHS.
23941 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23942 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23943 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23944 * gimplify.c (asan_poison_variables): Add attribute
23945 use_after_scope_memory to variables that really needs to live
23947 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23948 having the attribute.
23950 2017-01-23 Martin Liska <mliska@suse.cz>
23952 * asan.c (create_asan_shadow_var): New function.
23953 (asan_expand_poison_ifn): Likewise.
23954 * asan.h (asan_expand_poison_ifn): New declaration.
23955 * internal-fn.c (expand_ASAN_POISON): Likewise.
23956 * internal-fn.def (ASAN_POISON): New builtin.
23957 * sanopt.c (pass_sanopt::execute): Expand
23958 asan_expand_poison_ifn.
23959 * tree-inline.c (copy_decl_for_dup_finish): Make function
23961 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23962 * tree-ssa.c (is_asan_mark_p): New function.
23963 (execute_update_addresses_taken): Rewrite local variables
23964 (identified just by use-after-scope as addressable) into SSA.
23966 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23968 * doc/install.texi (Specific): opensource.apple.com uses https
23969 now. Remove trailing slash.
23971 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23973 * README.Portability: Remove note on an Irix compatibility issue.
23975 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23977 * gcov.c (INCLUDE_ALGORITHM): Define.
23978 (INCLUDE_VECTOR): Define.
23979 No longer include <vector> and <algorithm> directly.
23981 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23983 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23985 * doc/invoke.texi (Code Gen Options): Ditto.
23987 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23990 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23992 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23994 rtl-optimization/79125
23995 * cprop.c (local_cprop_pass): Handle cases where we make an
23996 unconditional trap.
23998 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24002 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24003 read from, for big endian.
24005 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24007 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24008 register pauth builtins for LP64 only.
24010 2017-01-20 Marek Polacek <polacek@redhat.com>
24013 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24016 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24018 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24020 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24021 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24022 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24024 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24027 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24028 in big-endian mode when they are not a single duplicated value.
24030 2017-01-20 Richard Biener <rguenther@suse.de>
24032 * BASE-VER: Bump to 7.0.1.
24034 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24036 * omp-low.c (omplow_simd_context): New struct. Use it...
24037 (lower_rec_simd_input_clauses): ...here and...
24038 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24039 references to idx, lane, max_vf, is_simt.
24041 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24043 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24046 2017-01-20 Martin Jambor <mjambor@suse.cz>
24048 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24049 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24051 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24052 (GTFILES): Rename hsa.c to hsa-common.c.
24053 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24054 * hsa-dump.c: Likewise.
24055 * hsa-gen.c: Likewise.
24056 * hsa-regalloc.c: Likewise.
24057 * ipa-hsa.c: Likewise.
24058 * omp-expand.c: Likewise.
24059 * omp-low.c: Likewise.
24060 * toplev.c: Likewise.
24062 2017-01-20 Marek Polacek <polacek@redhat.com>
24065 * doc/invoke.texi: Document -Wduplicated-branches.
24066 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24067 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24068 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24069 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24070 return 0 only when not OEP_LEXICOGRAPHIC.
24071 (fold_build_cleanup_point_expr): Use the expression
24072 location when building CLEANUP_POINT_EXPR.
24073 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24074 * tree.c (add_expr): Handle error_mark_node.
24076 2017-01-20 Martin Liska <mliska@suse.cz>
24079 * tree-profile.c (init_ic_make_global_vars): Do not call
24081 (gimple_init_gcov_profiler): Likewise.
24083 2017-01-20 Martin Liska <mliska@suse.cz>
24086 * cgraph.h (maybe_create_reference): Remove argument and
24088 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24090 * ipa-cp.c (create_specialized_node): Likewise.
24091 * symtab.c (symtab_node::maybe_create_reference): Handle
24092 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24094 2017-01-20 Martin Liska <mliska@suse.cz>
24096 * read-rtl-function.c (function_reader::create_function): Use
24097 build_decl instread of build_decl_stat.
24099 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24101 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24102 * config/i386/avx512dqintrin.h: Ditto.
24103 * config/i386/avx512fintrin.h: Ditto.
24104 * config/i386/i386-builtin-types.def: Add new types.
24105 * config/i386/i386.c: Handle new types.
24106 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24107 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24108 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24109 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24110 (__builtin_ia32_kshiftridi): New.
24111 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24113 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24117 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24118 define to rs6000_init_stack_protect_guard.
24119 (rs6000_init_stack_protect_guard): New function.
24121 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24122 Yunqiang Su <yunqiang.su@imgtec.com>
24124 * config.gcc (supported_defaults): Add madd4.
24125 (with_madd4): Add validation.
24126 (all_defaults): Add madd4.
24127 * config/mips/mips.opt (mmadd4): New option.
24128 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24130 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24132 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24133 (ISA_HAS_FUSED_MADD4): Likewise.
24134 * doc/invoke.texi (-mmadd4): Document the new option.
24135 * doc/install.texi (--with-madd4): Document the new option.
24137 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24139 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24140 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24141 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24142 (aarch64_init_pauth_hint_builtins): New.
24143 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24144 (aarch64_expand_builtin): Expand new builtins.
24146 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24148 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24149 * combine-stack-adj.c (no_unhandled_cfa): Handle
24150 REG_CFA_TOGGLE_RA_MANGLE.
24151 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24152 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24153 info for return address signing.
24154 (aarch64_expand_epilogue): Likewise.
24156 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24158 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24159 * config/aarch64/aarch64-protos.h
24160 (aarch64_return_address_signing_enabled): New declaration.
24161 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24163 (aarch64_expand_prologue): Sign return address before it's pushed onto
24165 (aarch64_expand_epilogue): Authenticate return address fetched from
24167 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24168 (aarch64_attributes): New function attributes for "sign-return-address".
24169 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24170 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24171 ("*do_return"): Generate combined instructions according to key index.
24172 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24173 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24175 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24176 * config/aarch64/aarch64.opt (msign-return-address=): New.
24177 * doc/extend.texi (AArch64 Function Attributes): Documents
24178 "sign-return-address=".
24179 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24181 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24183 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24184 overall option summary.
24186 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24188 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24189 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24190 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24191 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24193 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24195 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24196 -mpower9-minmax by default for -mcpu=power9.
24197 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24198 128-bit floating point.
24200 2017-01-20 Alan Modra <amodra@gmail.com>
24202 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24203 optimizing for size.
24205 2017-01-20 Alan Modra <amodra@gmail.com>
24208 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24209 for strcmp and strncmp from corresponding builtin decl.
24211 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24213 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24214 instead of i386/rtems-64.h.
24215 * config/i386/rtems-64.h: Remove.
24217 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24221 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24223 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24225 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24227 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24228 Change int to HOST_WIDE_INT.
24229 * config/aarch64/aarch64-protos.h
24230 (aarch64_simd_gen_const_vector_dup): Likewise.
24231 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24233 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24235 * langhooks-def.h (lhd_type_for_size): New decl.
24236 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24237 * langhooks.c (lhd_type_for_size): New function, taken from
24240 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24242 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24243 define_bypass for CR latency.
24244 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24245 (power9-alu2): Add define_bypass for CR latency.
24247 (power9-mul): Update insn latency.
24248 (power9-mul-compare): Update insn latency, bypass latency and remove
24251 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24253 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24255 * config/aarch64/aarch64.md
24256 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24257 aarch64_nopcrelative_literal_loads.
24258 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24260 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24262 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24263 TARGET_LOONGSON_3A.
24264 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24266 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24269 * config.gcc (supported_defaults): Add lxc1-sxc1.
24270 (with_lxc1_sxc1): Add validation.
24271 (all_defaults): Add lxc1-sxc1.
24272 * config/mips/mips.opt (mlxc1-sxc1): New option.
24273 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24275 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24276 __mips_no_lxc1_sxc1.
24277 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24278 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24279 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24281 2017-01-19 Richard Biener <rguenther@suse.de>
24283 PR tree-optimization/72488
24284 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24285 sure to restore SSA info.
24286 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24288 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24290 PR rtl-optimization/79121
24291 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24292 of the inner type when shifting an extended value.
24294 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24297 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24298 interposable aliases.
24300 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24303 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24304 Use the evmergelohi instruction.
24305 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24306 (mov_sitf_e500_subreg8_2_be): Likewise.
24307 (mov_sitf_e500_subreg12_2_le): Likewise.
24308 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24309 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24310 (mov_sitf_e500_subreg8_2_le): Likewise.
24311 (mov_sitf_e500_subreg12_2_be): Likewise.
24313 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24315 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24316 attribute from vecsimple to vecperm.
24317 (altivec_vbpermq2): Likewise.
24319 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24322 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24324 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24325 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24326 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24327 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24328 case where N arg is SIZE_MAX.
24329 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24330 (cmpstrsi): Add pattern.
24332 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24334 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24335 __builtin_vec_revb builtins.
24336 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24337 built-in functions to support generation of the ISA 3.0 XXBR<x>
24338 vector byte reverse instructions.
24339 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24340 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24341 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24342 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24343 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24344 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24345 (P9V_BUILTIN_VEC_REVB): Likewise.
24346 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24347 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24348 (p9_xxbrq_v16qi): Likewise.
24349 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24350 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24351 (p9_xxbrh_v8hi): Likewise.
24352 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24353 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24354 vec_revb built-in functions.
24356 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24358 PR rtl-optimization/78952
24359 * config/i386/i386.md (any_extract): New code iterator.
24360 (*insvqi_2): Use any_extract for source operand.
24361 (*insvqi_3): Use any_shiftrt for source operand.
24363 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24365 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24367 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24369 2017-01-18 Matthias Klose <doko@ubuntu.com>
24371 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24373 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24375 * config/rs6000/altivec.h (vec_bperm): Change #define.
24376 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24377 (altivec_vbpermq2): New define_insn.
24378 (altivec_vbpermd): Likewise.
24379 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24380 function interface.
24381 (VBPERMD): Likewise.
24382 (VBPERM): New polymorphic function interface.
24383 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24384 Add entries for P9V_BUILTIN_VEC_VBPERM.
24385 * doc/extend.texi: Add interfaces for vec_bperm.
24387 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24389 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24390 first letter of error messages.
24391 (s390_resolve_overloaded_builtin): Likewise.
24392 * config/s390/s390.c (s390_expand_builtin): Likewise.
24393 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24394 (s390_valid_target_attribute_inner_p): Likewise.
24395 * config/s390/s390.md ("tabort"): Likewise.
24397 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24399 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24400 (ISA_AVOID_DIV_HILO): New macro.
24401 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24402 (ISA_HAS_DDIV): Likewise.
24404 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24406 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24408 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24410 * doc/invoke.texi (fabi-version): Spelling fix.
24412 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24415 * doc/invoke.texi (fabi-version): Mention mangling fix for
24418 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24421 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24423 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24426 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24427 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24429 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24431 (stack_protect_test): Ditto.
24432 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24433 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24435 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24436 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24437 -mstack-protector-guard-offset=.
24438 (RS/6000 and PowerPC Options): Ditto.
24440 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24442 * config/i386/i386.h (MASK_CLASS_P): New define.
24443 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24444 there are no registers from different register sets also when
24445 mask registers are used. Update function comment.
24446 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24447 to (*k/*r) and (*k/*km) alternatives.
24449 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24451 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24452 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24453 (EH_RETURN_HANDLER_RTX): New define.
24454 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24455 Force frame pointer in EH return functions.
24456 (aarch64_expand_epilogue): Add barrier for eh_return.
24457 (aarch64_final_eh_return_addr): Remove.
24458 (aarch64_eh_return_handler_rtx): New function.
24459 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24461 (aarch64_eh_return_handler_rtx): New prototype.
24463 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24465 * config/rs6000/altivec.h (vec_rlmi): New #define.
24466 (vec_vrlnm): Likewise.
24467 (vec_rlnm): Likewise.
24468 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24469 (UNSPEC_VRLNM): Likewise.
24470 (VIlong): New mode iterator.
24471 (altivec_vrl<VI_char>mi): New define_insn.
24472 (altivec_vrl<VI_char>nm): Likewise.
24473 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24475 (VRLDNM): Likewise.
24476 (RLNM): New polymorphic function entry.
24477 (VRLWMI): New monomorphic function entry.
24478 (VRLDMI): Likewise.
24479 (RLMI): New polymorphic function entry.
24480 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24481 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24482 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24485 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24488 * dwarf2out.c (field_byte_offset): Restore the
24489 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24490 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24491 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24494 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24497 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24499 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24502 * dwarf2out.c (add_data_member_location_attribute): For constant
24503 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24504 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24505 DW_AT_byte_size attributes.
24507 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24509 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24510 after forcing to constant memory when the code model is medium.
24512 2017-01-17 Julia Koval <julia.koval@intel.com>
24515 * config/i386/avx512fintrin.h
24516 (_mm512_i32gather_ps): Change __addr type to void const*.
24517 (_mm512_mask_i32gather_ps): Ditto.
24518 (_mm512_i32gather_pd): Ditto.
24519 (_mm512_mask_i32gather_pd): Ditto.
24520 (_mm512_i64gather_ps): Ditto.
24521 (_mm512_mask_i64gather_ps): Ditto.
24522 (_mm512_i64gather_pd): Ditto.
24523 (_mm512_mask_i64gather_pd): Ditto.
24524 (_mm512_i32gather_epi32): Ditto.
24525 (_mm512_mask_i32gather_epi32): Ditto.
24526 (_mm512_i32gather_epi64): Ditto.
24527 (_mm512_mask_i32gather_epi64): Ditto.
24528 (_mm512_i64gather_epi32): Ditto.
24529 (_mm512_mask_i64gather_epi32): Ditto.
24530 (_mm512_i64gather_epi64): Ditto.
24531 (_mm512_mask_i64gather_epi64): Ditto.
24532 (_mm512_i32scatter_ps): Change __addr type to void*.
24533 (_mm512_mask_i32scatter_ps): Ditto.
24534 (_mm512_i32scatter_pd): Ditto.
24535 (_mm512_mask_i32scatter_pd): Ditto.
24536 (_mm512_i64scatter_ps): Ditto.
24537 (_mm512_mask_i64scatter_ps): Ditto.
24538 (_mm512_i64scatter_pd): Ditto.
24539 (_mm512_mask_i64scatter_pd): Ditto.
24540 (_mm512_i32scatter_epi32): Ditto.
24541 (_mm512_mask_i32scatter_epi32): Ditto.
24542 (_mm512_i32scatter_epi64): Ditto.
24543 (_mm512_mask_i32scatter_epi64): Ditto.
24544 (_mm512_i64scatter_epi32): Ditto.
24545 (_mm512_mask_i64scatter_epi32): Ditto.
24546 (_mm512_i64scatter_epi64): Ditto.
24547 (_mm512_mask_i64scatter_epi64): Ditto.
24548 * config/i386/avx512pfintrin.h
24549 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24550 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24551 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24552 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24553 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24554 (_mm512_prefetch_i32scatter_ps): Ditto.
24555 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24556 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24557 (_mm512_prefetch_i64scatter_pd): Ditto.
24558 (_mm512_prefetch_i64scatter_ps): Ditto.
24559 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24560 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24561 * config/i386/avx512vlintrin.h
24562 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24563 (_mm_mmask_i32gather_ps): Ditto.
24564 (_mm256_mmask_i32gather_pd): Ditto.
24565 (_mm_mmask_i32gather_pd): Ditto.
24566 (_mm256_mmask_i64gather_ps): Ditto.
24567 (_mm_mmask_i64gather_ps): Ditto.
24568 (_mm256_mmask_i64gather_pd): Ditto.
24569 (_mm_mmask_i64gather_pd): Ditto.
24570 (_mm256_mmask_i32gather_epi32): Ditto.
24571 (_mm_mmask_i32gather_epi32): Ditto.
24572 (_mm256_mmask_i32gather_epi64): Ditto.
24573 (_mm_mmask_i32gather_epi64): Ditto.
24574 (_mm256_mmask_i64gather_epi32): Ditto.
24575 (_mm_mmask_i64gather_epi32): Ditto.
24576 (_mm256_mmask_i64gather_epi64): Ditto.
24577 (_mm_mmask_i64gather_epi64): Ditto.
24578 (_mm256_i32scatter_ps): Change __addr type to void*.
24579 (_mm256_mask_i32scatter_ps): Ditto.
24580 (_mm_i32scatter_ps): Ditto.
24581 (_mm_mask_i32scatter_ps): Ditto.
24582 (_mm256_i32scatter_pd): Ditto.
24583 (_mm256_mask_i32scatter_pd): Ditto.
24584 (_mm_i32scatter_pd): Ditto.
24585 (_mm_mask_i32scatter_pd): Ditto.
24586 (_mm256_i64scatter_ps): Ditto.
24587 (_mm256_mask_i64scatter_ps): Ditto.
24588 (_mm_i64scatter_ps): Ditto.
24589 (_mm_mask_i64scatter_ps): Ditto.
24590 (_mm256_i64scatter_pd): Ditto.
24591 (_mm256_mask_i64scatter_pd): Ditto.
24592 (_mm_i64scatter_pd): Ditto.
24593 (_mm_mask_i64scatter_pd): Ditto.
24594 (_mm256_i32scatter_epi32): Ditto.
24595 (_mm256_mask_i32scatter_epi32): Ditto.
24596 (_mm_i32scatter_epi32): Ditto.
24597 (_mm_mask_i32scatter_epi32): Ditto.
24598 (_mm256_i32scatter_epi64): Ditto.
24599 (_mm256_mask_i32scatter_epi64): Ditto.
24600 (_mm_i32scatter_epi64): Ditto.
24601 (_mm_mask_i32scatter_epi64): Ditto.
24602 (_mm256_i64scatter_epi32): Ditto.
24603 (_mm256_mask_i64scatter_epi32): Ditto.
24604 (_mm_i64scatter_epi32): Ditto.
24605 (_mm_mask_i64scatter_epi32): Ditto.
24606 (_mm256_i64scatter_epi64): Ditto.
24607 (_mm256_mask_i64scatter_epi64): Ditto.
24608 (_mm_i64scatter_epi64): Ditto.
24609 (_mm_mask_i64scatter_epi64): Ditto.
24610 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24611 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24612 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24613 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24614 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24615 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24616 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24617 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24618 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24619 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24620 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24621 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24622 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24623 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24624 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24625 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24626 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24627 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24628 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24629 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24630 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24631 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24632 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24633 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24634 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24635 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24636 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24637 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24638 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24639 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24640 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24641 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24642 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24643 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24644 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24645 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24646 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24647 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24648 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24649 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24650 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24651 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24652 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24653 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24654 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24655 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24656 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24657 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24658 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24659 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24660 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24661 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24662 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24663 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24664 definitions accordingly.
24666 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24667 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24670 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24673 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24676 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24677 hard regs for inner regno.
24679 2017-01-17 Martin Liska <mliska@suse.cz>
24682 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24683 assumption and add comment.
24685 2017-01-17 Nathan Sidwell <nathan@acm.org>
24687 * ipa-visibility.c (localize_node): New function, broken out of ...
24688 (function_and_variable_visibility): ... here. Call it.
24690 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24692 PR middle-end/77445
24693 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24694 correctly set frequency of oudgoing edge.
24695 (duplicate_thread_path): Fix profile updating.
24697 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24700 * configure.ac: Add GCC_BASE_VER.
24701 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24702 version from BASE-VER file.
24703 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24704 (gcc.o): Depend on $(BASEVER).
24705 * common.opt (dumpfullversion): New option.
24706 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24707 * doc/invoke.texi: Document -dumpfullversion.
24708 * doc/install.texi: Document --with-gcc-major-version-only.
24709 * configure: Regenerated.
24711 2017-01-17 Richard Biener <rguenther@suse.de>
24713 PR tree-optimization/71433
24714 * tree-vrp.c (register_new_assert_for): Merge same asserts
24715 on all incoming edges.
24716 (process_assert_insertions_for): Handle insertions at the
24719 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24721 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24722 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24724 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24727 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24730 2017-01-17 Alan Modra <amodra@gmail.com>
24733 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24734 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24735 symbolic stack limit when pic.
24737 2017-01-16 Martin Sebor <msebor@redhat.com>
24739 PR tree-optimization/78608
24740 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24742 2017-01-16 Jeff Law <law@redhat.com>
24745 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24746 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24747 for several include directories that may be relative to sysroot.
24748 * config/i386/x-mingw32 (gplus_includedir): Define.
24749 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24750 (native_system_includedir): Likewise.
24751 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24752 override if TARGET_SYSTEM_ROOT is defined.
24753 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24755 PR tree-optimization/79090
24756 PR tree-optimization/33562
24757 PR tree-optimization/61912
24758 PR tree-optimization/77485
24759 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24760 and computed trims into the dump file.
24762 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24764 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24766 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24769 * gimplify.c (gimplify_init_constructor): If want_value and
24770 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24774 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24775 sequence. Formatting fixes.
24776 (doloop_optimize): Formatting fixes.
24779 * gcc.c (debug_level_greater_than_spec_func): New function.
24780 (static_spec_functions): Add debug-level-gt spec function.
24781 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24783 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24784 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24785 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24786 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24787 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24788 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24790 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24792 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24793 QImode fixups to general and mask registers only.
24795 2017-01-16 Carl Love <cel@us.ibm.com>
24797 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24798 for built-in functions
24799 vector signed char vec_nabs (vector signed char)
24800 vector signed short vec_nabs (vector signed short)
24801 vector signed int vec_nabs (vector signed int)
24802 vector signed long long vec_nabs (vector signed long long)
24803 vector float vec_nabs (vector float)
24804 vector double vec_nabs (vector double)
24805 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24807 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24808 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24809 * doc/extend.texi: Update the documentation file for the new built-in
24812 2017-01-16 Martin Sebor <msebor@redhat.com>
24814 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24817 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24819 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24820 UNSPEC_VSX__XXSPLTD to require special splat handling.
24822 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24825 * system.h: Poison strndup.
24827 2017-01-16 Alan Modra <amodra@gmail.com>
24830 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24833 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24835 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24837 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24839 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24840 call recog here. Assert that INSN_CODE (insn) is non-negative.
24842 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24845 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24847 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24848 in the currently scheduled RTL fragment.
24850 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24852 PR rtl-optimization/78751
24853 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24856 2017-01-14 Jeff Law <law@redhat.com>
24858 PR tree-optimization/79090
24859 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24860 variable length stores.
24861 (compute_trims): Delete dead assignment to *trim_tail.
24862 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24865 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24867 PR rtl-optimization/78626
24868 PR rtl-optimization/78727
24869 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24870 of a block, and split such blocks after everything else is finished.
24872 2017-01-14 Alan Modra <amodra@gmail.com>
24875 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24876 target legitimate_combined_insn.
24877 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24878 (rs6000_legitimate_combined_insn): New function.
24879 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24881 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24882 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24883 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24885 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24887 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24889 2017-01-13 Jeff Law <law@redhat.com>
24891 PR tree-optimization/33562
24892 PR tree-optimization/61912
24893 PR tree-optimization/77485
24894 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24896 (delete_dead_assignment): Likewise.
24897 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24898 statement to delete_dead_call and delete_dead_assignment.
24900 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24903 * substring-locations.c (format_warning_va): Strengthen case 1 so
24904 that both endpoints of the substring must be within the format
24905 range for just the substring to be printed.
24907 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24909 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24910 * config/i386/i386.c (ix86_target_string): Add missing options
24911 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24912 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24913 flags_other and ix86_target_other to flags2_other. Display unknown
24915 (ix86_valid_target_attribute_inner_p): Add missing options and
24916 reorder options by implied ISAs, as in ix86_target_string.
24918 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24920 * hash-table.h (hash_table::too_empty_p): New function.
24921 (hash_table::expand): Use it.
24922 (hash_table::traverse): Likewise.
24923 (hash_table::empty_slot): Use sizeof (value_type) instead of
24924 sizeof (PTR) to convert bytes to elements. Shrink the table
24925 if the current size is excessive for the current number of
24928 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24930 * ira-costs.c (record_reg_classes): Break from the inner loop
24931 early once alt_fail is known to be true. Update outer loop
24932 handling accordingly.
24934 2017-01-13 Jeff Law <law@redhat.com>
24936 * tree-ssa-dse.c (decrement_count): New function.
24937 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24938 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24939 when we know the partially dead statement is a mem* function.
24941 PR tree-optimization/61912
24942 PR tree-optimization/77485
24943 * tree-ssa-dse.c: Include expr.h.
24944 (maybe_trim_constructor_store): New function.
24945 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24947 PR tree-optimization/33562
24948 PR tree-optimization/61912
24949 PR tree-optimization/77485
24950 * doc/invoke.texi: Document new dse-max-object-size param.
24951 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24952 * tree-ssa-dse.c: Include params.h.
24953 (dse_store_status): New enum.
24954 (initialize_ao_ref_for_dse): New, partially extracted from
24956 (valid_ao_ref_for_dse, normalize_ref): New.
24957 (setup_live_bytes_from_ref, compute_trims): Likewise.
24958 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24959 (maybe_trim_partially_dead_store): Likewise.
24960 (maybe_trim_complex_store): Likewise.
24961 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24962 Track what bytes live from the original store. Return tri-state
24963 for dead, partially dead or live.
24964 (dse_dom_walker): Add constructor, destructor and new private members.
24965 (delete_dead_call, delete_dead_assignment): New extracted from
24967 (dse_optimize_stmt): Make a member of dse_dom_walker.
24968 Use initialize_ao_ref_for_dse.
24970 PR tree-optimization/33562
24971 PR tree-optimization/61912
24972 PR tree-optimization/77485
24973 * sbitmap.h (bitmap_count_bits): Prototype.
24974 (bitmap_clear_range, bitmap_set_range): Likewise.
24975 * sbitmap.c (bitmap_clear_range): New function.
24976 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24978 2017-01-13 Martin Liska <mliska@suse.cz>
24981 * function.c (set_cfun): Add new argument force.
24982 * function.h (set_cfun): Likewise.
24983 * ipa-inline-transform.c (inline_call): Use the function when
24984 strict alising from is dropped for function we inline to.
24986 2017-01-13 Richard Biener <rguenther@suse.de>
24988 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24989 for dumping GIMPLE INTEGER_CSTs.
24991 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24993 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24994 to 201112L since C++17.
24996 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24999 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25000 if -fsanitize=kernel-address is present.
25002 2017-01-13 Richard Biener <rguenther@suse.de>
25004 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25005 as _Literal ( type ) number in case usual suffixes do not
25006 preserve all information.
25008 2017-01-13 Richard Biener <rguenther@suse.de>
25010 PR tree-optimization/77283
25011 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25012 and ssa-iterators.h.
25013 (is_feasible_trace): Implement a cost model based on joiner
25016 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25019 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25020 char or short to __float128/_Float128 directly.
25022 2017-01-12 Martin Sebor <msebor@redhat.com>
25024 to -Wformat-overflow.
25025 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25026 (min_bytes_remaining): Same.
25027 (get_string_length): Same.
25028 (format_string): Same.
25029 (format_directive): Same.
25031 (pass_sprintf_length::handle_gimple_call): Same.
25033 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25035 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25036 info.nowrite calls with no lhs that can't throw. Return bool
25037 whether gsi_remove has been called or not.
25038 (pass_sprintf_length::handle_gimple_call): Return bool whether
25039 try_substitute_return_value called gsi_remove. Formatting fix.
25040 (pass_sprintf_length::execute): Don't use gsi_remove if
25041 handle_gimple_call returned true.
25044 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25045 be removed due to side-effects, don't remove following barrier nor
25046 turn the successor edge into fallthru edge.
25048 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25051 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25052 element-reversing loads and stores as not swappable.
25054 2017-01-12 Nathan Sidwell <nathan@acm.org>
25055 Nicolai Stange <nicstange@gmail.com>
25057 * combine.c (try_combine): Don't ignore result of overlap checking
25058 loop. Combine overlap & asm check into single loop.
25060 2017-01-12 Richard Biener <rguenther@suse.de>
25062 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25063 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25065 2017-01-12 Richard Biener <rguenther@suse.de>
25067 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25068 and TS_TARGET_OPTION directly derive from TS_BASE.
25069 * tree-core.h (tree_optimization_option): Derive from tree_base.
25070 (tree_target_option): Likewise.
25072 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25074 * config/i386/i386.c (memory_address_length): Increase len
25075 only when rip_relative_addr_p returns false.
25077 2017-01-11 Julia Koval <julia.koval@intel.com>
25079 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25080 (OPTION_MASK_ISA_SGX_SET): New.
25081 (ix86_handle_option): Handle OPT_msgx.
25082 * config.gcc: Added sgxintrin.h.
25083 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25084 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25085 * config/i386/i386.c (ix86_target_string): Add -msgx.
25087 (ix86_option_override_internal): Handle new options.
25088 (ix86_valid_target_attribute_inner_p): Add sgx.
25089 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25090 * config/i386/i386.opt: Add msgx.
25091 * config/i386/sgxintrin.h: New file.
25092 * config/i386/x86intrin.h: Add sgxintrin.h.
25094 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25097 * fold-const.c (maybe_nonzero_address): Return 1 for function
25099 (tree_single_nonzero_warnv_p): Don't handle function local objects
25103 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25106 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25109 * opts.c: Include "spellcheck.h"
25110 (struct string_fragment): New struct.
25111 (struct edit_distance_traits<const string_fragment &>): New
25113 (get_closest_sanitizer_option): New function.
25114 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25116 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25118 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25120 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25121 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25122 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25123 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25124 for initial die_offset if dwarf_split_debug_info.
25125 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25126 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25128 (output_skeleton_debug_sections): Formatting fix. Use
25129 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25130 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25132 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25134 * config/arm/cortex-a53.md: Add bypasses for
25135 cortex_a53_r2f_cvt.
25136 (cortex_a53_r2f): Only use for transfers.
25137 (cortex_a53_f2r): Likewise.
25138 (cortex_a53_r2f_cvt): Add reservation for conversions.
25139 (cortex_a53_f2r_cvt): Likewise.
25141 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25143 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25144 to all inlined functions, change static to extern.
25146 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25149 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25151 (arm_assemble_integer): Likewise.
25153 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25155 * config.gcc: Use new awk script to check CPU, FPU and architecture
25156 parameters for --with-... options.
25157 * config/arm/parsecpu.awk: New file
25158 * config/arm/arm-cpus.in: New file.
25159 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25161 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25163 * config/arm/t-arm: Update dependency rules.
25164 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25165 of processing .def files.
25166 * config/arm/genopt.sh: Deleted.
25167 * config/arm/gentune.sh: Deleted.
25168 * config/arm/arm-cores.def: Deleted.
25169 * config/arm/arm-arches.def: Deleted.
25170 * config/arm/arm-fpus.def: Deleted.
25171 * config/arm/arm-tune.md: Regenerated.
25172 * config/arm/arm-tables.opt: Regenerated.
25173 * config/arm/arm-cpu.h: New generated file.
25174 * config/arm/arm-cpu-data.h: New generated file.
25175 * config/arm/arm-cpu-cdata.h: New generated file.
25177 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25180 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25182 (input_varpool_node): Unpack dynamically_initialized bit.
25184 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25186 PR rtl-optimization/79032
25187 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25188 the alignment of the adjusted memory reference against that of MODE,
25189 instead of the alignment of the original memory reference.
25191 2017-01-11 Martin Jambor <mjambor@suse.cz>
25193 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25195 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25196 decorated functions.
25198 2017-01-11 Richard Biener <rguenther@suse.de>
25200 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25201 set range/nonnull info for PHI results. Do not set it on
25202 stmts marked for removal.
25204 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25206 * expr.c (store_field): In the bitfield case, fetch the return value
25207 from the registers before applying a single big-endian adjustment.
25208 Always do a final load for a BLKmode value not larger than a word.
25210 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25213 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25214 that we correctly handle column numbers greater than
25215 LINE_MAP_MAX_COLUMN_NUMBER.
25217 2017-01-10 Martin Sebor <msebor@redhat.com>
25219 PR middle-end/78245
25220 * gimple-ssa-sprintf.c (get_destination_size): Call
25221 {init,fini}object_sizes.
25222 * tree-object-size.c (addr_object_size): Adjust.
25223 (pass_through_call): Adjust.
25224 (pass_object_sizes::execute): Adjust.
25225 * tree-object-size.h (fini_object_sizes): Declare.
25227 2017-01-10 Martin Sebor <msebor@redhat.com>
25229 PR tree-optimization/78775
25230 * builtins.c (get_size_range): Move...
25231 * calls.c: ...to here.
25232 (alloc_max_size): Accept zero argument.
25233 (operand_signed_p): Remove.
25234 (maybe_warn_alloc_args_overflow): Call get_size_range.
25235 * calls.h (get_size_range): Declare.
25237 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25239 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25240 from TI's devices.csv file as of September 2016.
25241 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25243 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25245 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25246 * doc/invoke.texi: Likewise.
25247 * doc/md.texi: Likewise.
25248 * doc/objc.texi: Likewise.
25250 2017-01-10 Joshua Conner <joshconner@google.com>
25252 * config/arm/fuchsia-elf.h: New file.
25253 * config/fuchsia.h: New file.
25254 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25255 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25257 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25259 2016-01-10 Richard Biener <rguenther@suse.de>
25261 PR tree-optimization/79034
25262 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25263 Propagate out degenerate PHIs in the joiner.
25265 2017-01-10 Martin Liska <mliska@suse.cz>
25267 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25268 (sort_congruence_classes_by_decl_uid): Likewise.
25269 (sort_congruence_class_groups_by_decl_uid): Likewise.
25270 (sem_item_optimizer::merge_classes): Sort class, groups in these
25271 classes and members in the groups by DECL_UID of declarations.
25272 This would make merge operations stable.
25274 2017-01-10 Martin Liska <mliska@suse.cz>
25276 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25277 usage of m_classes_vec.
25278 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25279 (sem_item_optimizer::get_group_by_hash): Likewise.
25280 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25281 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25282 (sem_item_optimizer::verify_classes): Likewise.
25283 (sem_item_optimizer::process_cong_reduction): Likewise.
25284 (sem_item_optimizer::dump_cong_classes): Likewise.
25285 (sem_item_optimizer::merge_classes): Likewise.
25286 * ipa-icf.h (congruence_class_hash): Rename from
25287 congruence_class_group_hash. Remove declaration of m_classes_vec.
25289 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25291 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25292 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25293 * config.gcc: Add avx512vpopcntdqintrin.h.
25294 * config/i386/avx512vpopcntdqintrin.h: New.
25295 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25296 * config/i386/i386-builtin-types.def: Add new types.
25297 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25298 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25299 __builtin_ia32_vpopcountq_v8di_mask): New.
25300 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25301 __AVX512VPOPCNTDQ__.
25302 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25303 (PTA_AVX512VPOPCNTDQ): Define.
25304 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25305 TARGET_AVX512VPOPCNTDQ_P): Define.
25306 * config/i386/i386.opt: Add mavx512vpopcntdq.
25307 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25308 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25310 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25312 PR middle-end/77484
25313 * predict.def (PRED_CALL): Set to 67.
25315 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25317 * expr.c (store_field): In the bitfield case, if the value comes from
25318 a function call and is of an aggregate type returned in registers, do
25319 not modify the field mode; extract the value in all cases if the mode
25320 is BLKmode and the size is not larger than a word.
25322 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25325 * config/i386/cpuid.h: Fix undefined behavior.
25327 2017-01-04 Jeff Law <law@redhat.com>
25329 PR tree-optimization/79007
25330 PR tree-optimization/67955
25331 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25332 conservative for pt.null when flag_non_call_exceptions is on.
25334 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25336 PR translation/79019
25337 PR translation/79020
25338 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25339 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25340 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25342 * config/avr/avr.opt (maccumulate-args): Likewise.
25343 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25344 * common.opt (freport-bug): Likewise.
25345 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25346 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25347 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25348 translatable string.
25349 * config/i386/i386.c (function_value_32): Likewise.
25350 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25351 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25353 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25354 * common/config/msp430/msp430-common.c (msp430_handle_option):
25356 * symtab.c (symtab_node::verify_base): Likewise.
25357 * opts.c (set_debug_level): Likewise.
25358 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25359 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25360 missing whitespace to translatable strings.
25361 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25362 * config/sh/superh.h: Likewise.
25363 * config/i386/xopintrin.h: Likewise.
25364 * config/i386/znver1.md: Likewise.
25365 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25366 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25367 * double-int.h (struct double_int): Likewise.
25368 * double-int.c (div_and_round_double): Likewise.
25369 * wide-int.cc: Likewise.
25370 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25371 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25372 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25373 (crossjumps_occurred): ... this.
25374 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25377 PR tree-optimization/78899
25378 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25379 returning bool return struct loop *, NULL for failure and the new
25381 (versionable_outer_loop_p): Don't version outer loop if it has
25382 dont_vectorized bit set.
25383 (tree_if_conversion): When versioning outer loop, ensure
25384 tree_if_conversion is performed also on the inner loop of the
25385 non-vectorizable outer loop copy.
25386 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25387 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25388 prevent vectorization of it.
25389 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25390 the outer loop vectorization of the non-scalar version is attempted
25391 before vectorization of the inner loop in scalar version. If
25392 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25393 vectorization of its inner loop.
25394 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25395 has 2 inner loops, rename also on edges from bb whose single pred
25396 is outer_loop->header. Fix typo in function comment.
25398 2017-01-09 Martin Sebor <msebor@redhat.com>
25401 * asan.c (asan_emit_stack_protection): Increase local buffer size
25402 to avoid snprintf truncation warning.
25404 2017-01-09 Andrew Pinski <apinski@cavium.com>
25406 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25407 to reference thunderx2t99 for the tuning structure
25408 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25410 (thunderx2t99_extra_costs): This.
25411 * config/aarch64/aarch64-tune.md: Regenerate.
25412 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25413 (vulcan_addrcost_table): This.
25414 (vulcan_regmove_cost): Rename to ...
25415 (thunderx2t99_regmove_cost): This.
25416 (vulcan_vector_cost): Rename to ...
25417 (thunderx2t99_vector_cost): this.
25418 (vulcan_branch_cost): Rename to ...
25419 (thunderx2t99_branch_cost): This.
25420 (vulcan_tunings): Rename to ...
25421 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25422 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25424 2017-01-09 Martin Jambor <mjambor@suse.cz>
25428 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25429 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25430 (propagate_vr_accross_jump_function): Use the above function for all
25431 value range computations for pass-through jump functions and type
25432 converasion from explicit value range values.
25433 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25434 parameters from TYPE_ARG_TYPES.
25435 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25436 (ipa_write_node_info): Stream type of the actual argument.
25437 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25439 2017-01-09 Martin Liska <mliska@suse.cz>
25442 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25443 (lookup_compiler): Do not show error message with have_E.
25445 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25447 PR tree-optimization/78938
25448 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25449 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25450 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25451 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25454 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25456 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25459 2017-01-09 Richard Biener <rguenther@suse.de>
25461 PR tree-optimization/78997
25462 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25463 name condition properly.
25465 2017-01-09 Richard Biener <rguenther@suse.de>
25468 * dwarf2out.c (is_cxx): New overload with context.
25469 (is_naming_typedef_decl): Use it.
25471 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25473 * invoke.texi (Option Summary): Correct spacing in option lists
25474 and add line breaks to fix over-long lines.
25476 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25478 PR middle-end/17660
25480 * extend.texi (Common Variable Attributes): Add xref to GCC
25481 Internals manual to explain mode attribute keywords.
25483 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25486 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25487 and Preprocessor Options.
25488 (Options for Linking): Document -pthread here....
25489 (RS/6000 and PowerPC Options): ...not here.
25490 (Solaris 2 Options): ...or here.
25491 * doc/cppopts.texi: Document -pthread.
25493 2017-01-08 Martin Sebor <msebor@redhat.com>
25495 PR middle-end/77708
25496 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25497 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25498 New member functions.
25499 (format_directive): Used them.
25501 (pass_sprintf_length::handle_gimple_call): Same.
25502 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25503 to avoid truncation for any argument.
25504 (extract_affine_mul): Same.
25505 * tree.c (get_file_function_name): Same.
25507 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25509 PR middle-end/77484
25510 * predict.def (PRED_INDIR_CALL): Set to 86.
25512 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25514 PR preprocessor/54124
25515 * doc/cppopts.texi: Reformat -d subtable to list the full name
25516 of the options. Add cross-reference to the docs for the general
25517 compiler -d options.
25518 * doc/invoke.texi (Developer Options): Add cross-reference to the
25519 preprocessor-specific -d option documentation.
25521 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25523 PR preprocessor/13498
25524 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25525 redudant material, and reflect new command-line options.
25526 (System Headers): Likewise.
25528 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25530 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25531 -isystem, and -idirafter. Copy-edit.
25532 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25533 default for -ftrack-macro-expansion. Delete obsolete and
25534 badly-formatted implementation details about -fdebug-cpp output.
25535 * doc/cppwarnopts.texi: Copy-edit.
25537 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25540 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25541 that the transition from a max line width >= 1<<10 to narrower
25542 lines works correctly.
25544 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25546 * doc/options.texi (PerFunction): New.
25547 * opt-functions.awk (switch_flags): Map both Optimization and
25548 PerFunction to CL_OPTIMIZATION.
25549 * opth-gen.awk: Test for PerFunction flag along with
25551 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25552 it only when the latter is present. Skip those that don't in
25553 the hash function generator.
25554 * common.opt (fvar-tracking): Mark as PerFunction instead of
25556 (fvar-tracking-assignments): Likewise.
25557 (fvar-tracking-assignments-toggle): Likewise.
25558 (fvar-tracking-uninit): Likewise.
25560 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25562 PR translation/79018
25563 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25566 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25569 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25570 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25571 TARGET_LONG_JUMP_TABLE_OFFSETS.
25572 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25573 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25574 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25575 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25576 * config/m68k/m68k.md (tablejump expander): Likewise.
25577 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25578 TARGET_LONG_JUMP_TABLE_OFFSETS.
25579 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25580 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25582 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25583 David Holsgrove <david.holsgrove@xilinx.com>
25585 * common/config/microblaze/microblaze-common.c
25586 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25587 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25589 * config/microblaze/microblaze.c (microblaze_must_save_register)
25590 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25592 (microblaze_eh_return): New function.
25593 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25594 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25595 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25596 * config/microblaze/microblaze.md (eh_return): New pattern.
25598 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25600 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25601 GCC_DIAGNOSTIC_STRINGIFY): Define.
25603 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25605 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25607 * config/arm/arm.md (<mcrr>): New.
25609 * config/arm/arm.c (arm_arch5te): New.
25610 (arm_option_override): Set arm_arch5te.
25611 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25613 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25614 (arm_mcrr_qualifiers): ... this. New.
25615 (MRRC_QUALIFIERS): Define to...
25616 (arm_mrrc_qualifiers): ... this. New.
25617 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25619 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25620 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25621 (MRRCI, mrrc, MRRC): New.
25622 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25623 VUNSPEC_MRRC2): New.
25625 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25627 * config/arm/arm.md (<mcr>): New.
25629 * config/arm/arm.c (arm_coproc_builtin_available): Add
25630 support for mcr, mrc, mcr2 and mrc2.
25631 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25632 (arm_mcr_qualifiers): ... this. New.
25633 (MRC_QUALIFIERS): Define to ...
25634 (arm_mrc_qualifiers): ... this. New.
25635 (MCR_QUALIFIERS): Define to ...
25636 (arm_mcr_qualifiers): ... this. New.
25637 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25639 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25640 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25641 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25642 VUNSPEC_MRC2): New.
25644 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25646 * config/arm/arm.md (*ldc): New.
25650 * config/arm/arm.c (arm_coproc_builtin_available): Add
25651 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25652 (arm_coproc_ldc_stc_legitimate_address): New.
25653 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25654 'qualifier_const_pointer'.
25655 (LDC_QUALIFIERS): Define to...
25656 (arm_ldc_qualifiers): ... this. New.
25657 (STC_QUALIFIERS): Define to...
25658 (arm_stc_qualifiers): ... this. New.
25659 * config/arm/arm-protos.h
25660 (arm_coproc_ldc_stc_legitimate_address): New.
25661 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25662 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25663 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25664 stc2, stcl, stc2l): New.
25665 * config/arm/constraints.md (Uz): New.
25666 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25667 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25668 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25669 VUNSPEC_STC2L): New.
25671 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25673 * config/arm/arm.md (<cdp>): New.
25674 * config/arm/arm.c (neon_const_bounds): Rename this ...
25675 (arm_const_bounds): ... this.
25676 (arm_coproc_builtin_available): New.
25677 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25678 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25679 (CDP_QUALIFIERS): Define to...
25680 (arm_cdp_qualifiers): ... this. New.
25682 (arm_expand_builtin_args): Add case for 6 arguments.
25683 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25684 (arm_const_bounds): ... this.
25685 (arm_coproc_builtin_available): New.
25686 * config/arm/arm_acle.h (__arm_cdp): New.
25688 * config/arm/arm_acle_builtins.def (cdp): New.
25690 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25691 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25692 'arm_const_bounds'.
25693 * config/arm/types.md (coproc): New.
25694 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25695 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25696 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25697 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25699 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25701 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25702 (UBINOP_QUALIFIERS): New.
25704 (acle_builtin_data): New. Change comment.
25705 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25706 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25707 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25708 arm_acle_builtins.def.
25709 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25710 (arm_init_acle_builtins): New.
25711 (CRC32_BUILTIN): Remove.
25712 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25713 crc32cb, crc32ch and crc32cw.
25714 (arm_init_crc32_builtins): Remove.
25715 (arm_init_builtins): Use arm_init_acle_builtins rather
25716 than arm_init_crc32_builtins.
25717 (arm_expand_acle_builtin): New.
25718 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25719 * config/arm/arm_acle_builtins.def: New.
25721 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25723 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25724 (arm_builtin_datum): ... this.
25725 (arm_init_neon_builtin): Rename to ...
25726 (arm_init_builtin): ... this. Add a new parameters PREFIX
25727 and USE_SIG_IN_NAME.
25728 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25729 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25730 'arm_builtin_datum'.
25731 (arm_init_vfp_builtins): Likewise.
25732 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25733 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25734 (arm_expand_neon_args): Rename to ...
25735 (arm_expand_builtin_args): ... this. Rename builtin_arg
25736 enum values and differentiate between ARG_BUILTIN_MEMORY
25737 and ARG_BUILTIN_NEON_MEMORY.
25738 (arm_expand_neon_builtin_1): Rename to ...
25739 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25740 values, arm_expand_builtin_args and add bool parameter NEON.
25741 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25742 (arm_expand_vfp_builtin): Likewise.
25743 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25745 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25747 PR middle-end/77484
25748 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25749 * predict.c (tree_estimate_probability_bb): Reverse direction of
25750 polymorphic call predictor.
25752 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25754 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25755 (determine_pass_name_match): ...this new function and...
25756 (should_skip_pass_p): ...this new function.
25758 2017-01-06 Nathan Sidwell <nathan@acm.org>
25760 * ipa-visibility.c (function_and_variable_visibility): Reformat
25761 comments and long lines. Remove extrneous if.
25762 * symtab.c (symtab_node::make_decl_local): Fix code format.
25763 (symtab_node::set_section_for_node): Fix comment typo.
25765 2017-01-06 Martin Liska <mliska@suse.cz>
25768 * lra-constraints.c: Rename invariant to lra_invariant.
25769 * predict.c (set_even_probabilities): Initialize e to NULL.
25771 2017-01-05 Martin Sebor <msebor@redhat.com>
25773 PR tree-optimization/78910
25774 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25775 (format_integer): Correct off-by-one error in the handling
25776 of precision with negative numbers in signed conversions..
25778 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25780 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25782 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25784 PR tree-optimization/71016
25785 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25786 factor_out_conditional_conversion. Formatting fix.
25787 (factor_out_conditional_conversion): Add cond_stmt argument.
25788 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25789 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25792 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25794 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25795 read-rtl-function.o, and selftest-rtl.o.
25796 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25797 (selftest::aarch64_test_loading_full_dump): New function.
25798 (selftest::aarch64_run_selftests): New function.
25799 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25800 selftest::aarch64_run_selftests.
25801 * config/i386/i386.c
25802 (selftest::ix86_test_loading_dump_fragment_1): New function.
25803 (selftest::ix86_test_loading_call_insn): New function.
25804 (selftest::ix86_test_loading_full_dump): New function.
25805 (selftest::ix86_test_loading_unspec): New function.
25806 (selftest::ix86_run_selftests): Call the new functions.
25807 * emit-rtl.c (maybe_set_max_label_num): New function.
25808 * emit-rtl.h (maybe_set_max_label_num): New decl.
25809 * function.c (instantiate_decls): Guard call to
25810 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25811 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25813 * gensupport.c (gen_reader::gen_reader): Pass "false"
25814 for new "compact" param of rtx_reader.
25815 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25816 rather than an empty string for NULL strings.
25817 * read-md.c: Potentially include config.h rather than bconfig.h.
25818 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25819 (have_error): New global, copied from errors.c.
25820 (md_reader::read_name): Rename to...
25821 (md_reader::read_name_1): ...this, adding "out_loc" param,
25822 and converting "missing name or number" to returning false, rather
25824 (md_reader::read_name): Reimplement in terms of read_name_1.
25825 (md_reader::read_name_or_nil): New function.
25826 (md_reader::read_string): Handle "(nil)" by returning NULL.
25827 (md_reader::md_reader): Add new param "compact".
25828 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25829 (md_reader::read_file): New method.
25830 * read-md.h (md_reader::md_reader): Add new param "compact".
25831 (md_reader::read_file): New method.
25832 (md_reader::is_compact): New accessor.
25833 (md_reader::read_name): Convert return type from void to file_location.
25834 (md_reader::read_name_or_nil): New decl.
25835 (md_reader::read_name_1): New decl.
25836 (md_reader::m_compact): New field.
25837 (noop_reader::noop_reader): Pass "false" for new "compact" param
25839 (rtx_reader::rtx_reader): Add new "compact" param.
25840 (rtx_reader::read_rtx_operand): Make virtual and convert return
25841 type from void to rtx.
25842 (rtx_reader::read_until): New decl.
25843 (rtx_reader::handle_any_trailing_information): New virtual function.
25844 (rtx_reader::postprocess): New virtual function.
25845 (rtx_reader::finalize_string): New virtual function.
25846 (rtx_reader::m_in_call_function_usage): New field.
25847 (rtx_reader::m_reuse_rtx_by_id): New field.
25848 * read-rtl-function.c: New file.
25849 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25850 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25851 (selftest::verify_three_block_rtl_cfg): New decl.
25852 * read-rtl-function.h: New file.
25853 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25854 For host, include function.h, memmodel.h, and emit-rtl.h.
25855 (one_time_initialization): New function.
25856 (struct compact_insn_name): New struct.
25857 (compact_insn_names): New array.
25858 (find_code): Handle insn codes in compact dumps.
25859 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25860 (bind_subst_iter_and_attr): Likewise.
25861 (add_condition_to_string): Likewise.
25862 (add_condition_to_rtx): Likewise.
25863 (apply_attribute_uses): Likewise.
25864 (add_current_iterators): Likewise.
25865 (apply_iterators): Likewise.
25866 (initialize_iterators): Guard usage of apply_subst_iterator with
25867 #ifdef GENERATOR_FILE.
25868 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25869 (md_reader::read_mapping): Likewise.
25870 (add_define_attr_for_define_subst): Likewise.
25871 (add_define_subst_attr): Likewise.
25872 (read_subst_mapping): Likewise.
25873 (check_code_iterator): Likewise.
25874 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25876 (one_time_initialization): New function.
25877 (rtx_reader::read_until): New method.
25878 (read_flags): New function.
25879 (parse_reg_note_name): New function.
25880 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25881 Handle reuse_rtx ids.
25882 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25883 Add parsing support for RTL dumps, mirroring the special-cases in
25884 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25885 values, and calling handle_any_trailing_information.
25886 (rtx_reader::read_rtx_operand): Convert return type from void
25887 to rtx, returning return_rtx. Handle case 'e'. Call
25888 finalize_string on XSTR and XTMPL fields.
25889 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25890 "(nil)" values were omitted. Call the postprocess vfunc on the
25892 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25893 class ctor. Initialize m_in_call_function_usage. Call
25894 one_time_initialization.
25895 * rtl-tests.c (selftest::test_uncond_jump): Call
25896 set_new_first_and_last_insn.
25897 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25898 * selftest-rtl.c: New file.
25899 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25900 (selftest::get_insn_by_uid): New decl.
25901 * selftest-run-tests.c (selftest::run_tests): Call
25902 read_rtl_function_c_tests.
25903 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25904 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25907 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25909 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25910 operands in a special way. Assert that pos+len <= mode precision.
25912 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25914 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25915 3 argument Alias with unlimited for the negative form.
25916 (fno-vect-cost-model): Removed.
25918 2017-01-05 Martin Liska <mliska@suse.cz>
25920 * hsa-gen.c (gen_hsa_divmod): New function.
25921 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25923 2017-01-05 Martin Liska <mliska@suse.cz>
25926 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25929 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25931 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25932 small constant length operands.
25934 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25936 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25937 between loop iterations.
25939 2017-01-05 Martin Liska <mliska@suse.cz>
25942 * gimplify.c (gimplify_decl_expr): Compare to
25943 asan_poisoned_variables instread of checking flags.
25944 (gimplify_target_expr): Likewise.
25945 (gimplify_expr): Likewise.
25946 (gimplify_function_tree): Conditionally initialize
25947 asan_poisoned_variables.
25949 2017-01-04 Jeff Law <law@redhat.com>
25951 PR tree-optimizatin/78812
25952 * rtl.h (contains_mem_rtx_p): Prototype.
25953 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25954 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25955 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25956 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25957 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25959 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25961 * input.c (assert_char_at_range): Default-initialize actual_range.
25963 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25965 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25966 to match the caller.
25968 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25970 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25971 insns after final jump in test to emit dummy move.
25973 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25975 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25976 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25978 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25980 * multiple_target.c (create_dispatcher_calls): Init e_next.
25981 * tree-ssa-loop-split.c (split_loop): Init border.
25982 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25985 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25990 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25991 (altivec_register_operand): Do not return true if the operand
25992 contains a SUBREG mixing SImode and SFmode.
25993 (vsx_register_operand): Likewise.
25994 (vsx_reg_sfsubreg_ok): New predicate.
25995 (vfloat_operand): Do not return true if the operand contains a
25996 SUBREG mixing SImode and SFmode.
25997 (vint_operand): Likewise.
25998 (vlogical_operand): Likewise.
25999 (gpc_reg_operand): Likewise.
26000 (int_reg_operand): Likewise.
26001 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26002 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26003 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26005 (rs6000_emit_move_si_sf_subreg): New helper function.
26006 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26007 fixup SUBREGs involving SImode and SFmode.
26008 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26009 numbers for the new peephole2 optimization.
26010 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26011 the GLIBC math library that do AND/IOR/XOR operations on single
26012 precision floating point.
26013 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26014 target macros to say whether we need to avoid SUBREGs mixing
26016 (TARGET_ALLOW_SF_SUBREG): Likewise.
26017 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26018 (UNSPEC_SI_FROM_SF): Likewise.
26019 (iorxor): Change spacing.
26020 (and_ior_xor): New iterator for AND, IOR, and XOR.
26021 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26022 (movdi_from_sf_zero_ext): Likewise.
26023 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26024 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26025 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26026 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26027 (fms<mode>4): Likewise.
26028 (fnma<mode>4): Likewise.
26029 (fnms<mode>4): Likewise.
26030 (nfma<mode>4): Likewise.
26031 (nfms<mode>4): Likewise.
26033 2017-01-04 Marek Polacek <polacek@redhat.com>
26036 * doc/invoke.texi: Document -Wpointer-compare.
26038 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26040 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26043 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26044 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26047 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26050 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26051 documentation of the powerpc_popcntb_ok attribute.
26052 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26053 code to issue warning messages if a requested CPU configuration is
26054 not supported by the binary (assembler and loader) toolchain.
26055 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26056 made to define a built-in function that has been disabled.
26057 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26058 made to define a built-in function that has been disabled.
26059 (altivec_init_builtins): Add comment explaining why definition
26060 of the DST built-in functions is not preceded by an assertion
26061 check. Add assertions to prevent ICE if attempts are made to
26062 define an altivec predicate or an abs* built-in function that has
26064 (htm_init_builtins): Add comment explaining why definition of the
26065 htm built-in functions is not preceded by an assertion check.
26067 2017-01-04 Jeff Law <law@redhat.com>
26069 PR tree-optimizatin/67955
26070 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26071 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26072 the points-to solution does not include pt_null. Use DECL_PT_UID
26075 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26077 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26078 Use gen_int_mode instead of gen_lopwart for const_int operands.
26080 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26082 PR tree-optimization/71563
26083 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26084 out of range value - has low bits known to be zero.
26086 2017-01-04 Alan Modra <amodra@gmail.com>
26088 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26089 * configure: Regenerate.
26090 * config.in: Regenerate.
26092 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26095 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26096 a substring of the message, but strcmp with the whole message. Ifdef
26097 ENABLE_NLS, translate the message first using dgettext.
26099 2017-01-03 Jeff Law <law@redhat.com>
26101 PR tree-optimizatin/78856
26102 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26103 (mark_threaded_blocks): Remove code to truncate thread paths that
26104 cross multiple loop headers. Instead invalidate the cached loop
26105 iteration information and handle case of a thread path walking
26106 into an irreducible region.
26108 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26111 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26112 assertions. Add support for doing the signbit if the IEEE 128-bit
26113 floating point value is in a GPR.
26114 * config/rs6000/rs6000.md (Fsignbit): Delete.
26115 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26116 Update the length attribute if the value is in a GPR.
26117 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26118 the sign or zero extension instruction, since the value is always 0/1.
26119 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26122 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26123 extracting SImode to a GPR register so that we can generate a
26124 store, limit the vector to be in a traditional Altivec register
26125 for the vextuwrx instruction.
26127 2017-01-03 Ian Lance Taylor <iant@google.com>
26129 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26131 2017-01-03 Martin Sebor <msebor@redhat.com>
26133 PR tree-optimization/78696
26134 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26135 precision. Use MPFR for %f for greater fidelity. Correct handling
26137 (pass_sprintf_length::compute_format_length): Set width and precision
26138 specified by asrerisk to void_node for vararg functions.
26139 (try_substitute_return_value): Adjust dump output.
26141 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26143 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26145 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26147 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26148 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26149 -mlra/-mno-lra was passed to the compiler.
26151 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26153 PR rtl-optimization/65618
26154 * emit-rtl.c (try_split): Move initialization of "before" and
26155 "after" to just before the call to emit_insn_after_setloc.
26157 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26159 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26161 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26163 * dwarf2out.c (gen_enumeration_type_die): When
26164 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26166 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26168 PR tree-optimization/78965
26169 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26170 Change first argument from const call_info & to call_info &. For %n
26171 set info.nowrite to false.
26173 PR middle-end/78901
26174 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26175 possibly throwing calls.
26177 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26178 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26179 and fns handling, rather than in a separate case SSA_NAME.
26181 2017-01-02 Jeff Law <law@redhat.com>
26183 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26184 fixes for first_period and second_period variables.
26186 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26189 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26190 (*insvqi_1): New insn pattern.
26191 (*insvqi_1_mem_rex64): Ditto.
26192 (*insvqi_2): Ditto.
26193 (*insvqi_3): Rename from *insvqi.
26195 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26197 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26199 * doc/cfg.texi (Edges): Remove reference to Java.
26200 (Maintaining the CFG): Ditto.
26202 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26204 PR middle-end/77674
26205 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26206 transparent aliases.
26208 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26210 PR middle-end/77484
26211 * predict.def (PRED_CALL): Update hitrate.
26212 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26213 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26214 into direct/indirect/polymorphic variants.
26216 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26218 Update copyright years.
26220 * gcc.c (process_command): Update copyright notice dates.
26221 * gcov-dump.c (print_version): Ditto.
26222 * gcov.c (print_version): Ditto.
26223 * gcov-tool.c (print_version): Ditto.
26224 * gengtype.c (create_file): Ditto.
26225 * doc/cpp.texi: Bump @copying's copyright year.
26226 * doc/cppinternals.texi: Ditto.
26227 * doc/gcc.texi: Ditto.
26228 * doc/gccint.texi: Ditto.
26229 * doc/gcov.texi: Ditto.
26230 * doc/install.texi: Ditto.
26231 * doc/invoke.texi: Ditto.
26233 Copyright (C) 2017 Free Software Foundation, Inc.
26235 Copying and distribution of this file, with or without modification,
26236 are permitted in any medium without royalty provided the copyright
26237 notice and this notice are preserved.