1 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * target.def (hard_regno_nregs): New hook.
6 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7 * targhooks.h (default_hard_regno_nregs): Declare.
8 * targhooks.c (default_hard_regno_nregs): New function.
9 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10 (TARGET_HARD_REGNO_NREGS): ...this hook.
11 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12 (CLASS_MAX_NREGS): Likewise.
13 * doc/tm.texi: Regenerate.
14 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15 instead of HARD_REGNO_NREGS.
16 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
17 HARD_REGNO_NREGS in the comment.
18 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
19 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
20 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
21 Return an unsigned int.
22 (TARGET_HARD_REGNO_NREGS): Redefine.
23 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
24 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
25 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
26 (arc_hard_regno_nregs): New function.
27 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
28 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
29 (arm_hard_regno_nregs): New function.
30 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
31 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
32 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
33 (TARGET_HARD_REGNO_NREGS): Redefine.
34 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
35 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
36 (HARD_REGNO_NREGS): Delete.
37 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
38 (cr16_hard_regno_nregs): New function.
39 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
40 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
41 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
42 (cris_hard_regno_nregs): New function.
43 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
44 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
45 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
46 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
47 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
48 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
49 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
50 (frv_hard_regno_nregs): Make static. Take and return an
52 (frv_class_max_nregs): Remove outdated copy of documentation.
53 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
54 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
55 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
56 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
57 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
58 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
59 (TARGET_HARD_REGNO_NREGS): Redefine.
60 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
61 (CLASS_MAX_NREGS): Update comment.
62 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
63 (ia64_hard_regno_nregs): New function.
64 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
65 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
66 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
67 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
68 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
70 (m32c_hard_regno_nregs): Likewise. Make static.
71 (TARGET_HARD_REGNO_NREGS): Redefine.
72 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
73 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
74 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
75 (m68k_hard_regno_nregs): New function.
76 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
77 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
78 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
79 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
80 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
81 Take and return an unsigned int.
82 (TARGET_HARD_REGNO_NREGS): Redefine.
83 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
84 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
85 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
86 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
87 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
88 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
89 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
90 (msp430_hard_regno_nregs): Make static. Take and return an
92 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
93 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
94 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
95 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
96 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
97 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
98 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
99 (TARGET_HARD_REGNO_NREGS): Redefine.
100 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
101 (PA_HARD_REGNO_NREGS): ...this.
102 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
103 (PA_HARD_REGNO_NREGS): ...this.
104 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
105 (pa_hard_regno_nregs): New function.
106 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
107 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
108 (pdp11_hard_regno_nregs): New function.
109 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
110 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
111 (rs6000_hard_regno_nregs_hook): New function.
112 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
113 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
114 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
115 Take and return an unsigned int. Move earlier in file.
116 (TARGET_HARD_REGNO_NREGS): Redefine.
117 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
118 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
119 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
120 (rl78_hard_regno_nregs): Make static. Take and return an
122 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
123 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
124 (rs6000_hard_regno_nregs_hook): New function.
125 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
126 * config/rx/rx.c (rx_hard_regno_nregs): New function.
127 (TARGET_HARD_REGNO_NREGS): Redefine.
128 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
129 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
130 instead of HARD_REGNO_NREGS.
131 (s390_hard_regno_nregs): New function.
132 (s390_hard_regno_mode_ok): Add comment from s390.h.
133 (TARGET_HARD_REGNO_NREGS): Redefine.
134 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
135 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
136 (sh_hard_regno_nregs): New function.
137 (sh_pass_in_reg_p): Use it.
138 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
139 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
140 (sparc_hard_regno_nregs): New function.
141 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
142 * config/spu/spu.c (spu_hard_regno_nregs): New function.
143 (spu_function_arg_advance): Use it, supplying a valid register number.
144 (TARGET_HARD_REGNO_NREGS): Redefine.
145 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
146 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
147 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
148 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
149 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
150 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
151 (CLASS_MAX_NREGS): Remove copy of old documentation.
152 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
153 (visium_hard_regno_nregs): New function.
154 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
155 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
156 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
157 xtensa_hard_regno_nregs): New function.
158 * system.h (HARD_REGNO_NREGS): Poison.
160 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
162 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
163 hard_regno_nregs instead of HARD_REGNO_NREGS.
164 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
165 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
166 (c6x_expand_epilogue): Likewise.
167 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
168 (frv_read_iacc_argument): Likewise.
169 * config/sh/sh.c: Include regs.h.
170 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
171 (regs_used): Likewise.
172 (output_stack_adjust): Likewise.
173 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
174 * expmed.c: Include regs.h.
175 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
176 * ree.c: Include regs.h.
177 (combine_reaching_defs): Use hard_regno_nregs instead of
179 (add_removable_extension): Likewise.
181 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
183 * regs.h (hard_regno_nregs): Turn into a function.
184 (end_hard_regno): Update accordingly.
185 * caller-save.c (setup_save_areas): Likewise.
186 (save_call_clobbered_regs): Likewise.
187 (replace_reg_with_saved_mem): Likewise.
188 (insert_restore): Likewise.
189 (insert_save): Likewise.
190 * combine.c (can_change_dest_mode): Likewise.
191 (move_deaths): Likewise.
192 (distribute_notes): Likewise.
193 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
194 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
195 (rs6000_split_multireg_move): Likewise.
196 (rs6000_register_move_cost): Likewise.
197 (rs6000_memory_move_cost): Likewise.
198 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
199 (rs6000_split_multireg_move): Likewise.
200 (rs6000_register_move_cost): Likewise.
201 (rs6000_memory_move_cost): Likewise.
202 * cselib.c (cselib_reset_table): Likewise.
203 (cselib_lookup_1): Likewise.
204 * emit-rtl.c (set_mode_and_regno): Likewise.
205 * function.c (aggregate_value_p): Likewise.
206 * ira-color.c (setup_profitable_hard_regs): Likewise.
207 (check_hard_reg_p): Likewise.
208 (calculate_saved_nregs): Likewise.
209 (assign_hard_reg): Likewise.
210 (improve_allocation): Likewise.
211 (calculate_spill_cost): Likewise.
212 * ira-emit.c (modify_move_list): Likewise.
213 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
214 (ira_hard_reg_in_set_p): Likewise.
215 * ira.c (setup_reg_mode_hard_regset): Likewise.
216 (clarify_prohibited_class_mode_regs): Likewise.
217 (check_allocation): Likewise.
218 * lra-assigns.c (find_hard_regno_for_1): Likewise.
219 (lra_setup_reg_renumber): Likewise.
220 (setup_try_hard_regno_pseudos): Likewise.
221 (spill_for): Likewise.
222 (assign_hard_regno): Likewise.
223 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
224 * lra-constraints.c (in_class_p): Likewise.
225 (lra_constraint_offset): Likewise.
226 (simplify_operand_subreg): Likewise.
227 (lra_constraints): Likewise.
228 (split_reg): Likewise.
229 (split_if_necessary): Likewise.
230 (invariant_p): Likewise.
231 (inherit_in_ebb): Likewise.
232 * lra-lives.c (process_bb_lives): Likewise.
233 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
234 (get_hard_regs): Likewise.
235 (do_remat): Likewise.
236 * lra-spills.c (assign_spill_hard_regs): Likewise.
237 * mode-switching.c (create_pre_exit): Likewise.
238 * postreload.c (reload_combine_recognize_pattern): Likewise.
239 * recog.c (peep2_find_free_register): Likewise.
240 * regcprop.c (kill_value_regno): Likewise.
241 (set_value_regno): Likewise.
242 (copy_value): Likewise.
243 (maybe_mode_change): Likewise.
244 (find_oldest_value_reg): Likewise.
245 (copyprop_hardreg_forward_1): Likewise.
246 * regrename.c (check_new_reg_p): Likewise.
247 (regrename_do_replace): Likewise.
248 * reload.c (push_reload): Likewise.
249 (combine_reloads): Likewise.
250 (find_dummy_reload): Likewise.
251 (operands_match_p): Likewise.
252 (find_reloads): Likewise.
253 (find_equiv_reg): Likewise.
254 (reload_adjust_reg_for_mode): Likewise.
255 * reload1.c (count_pseudo): Likewise.
256 (count_spilled_pseudo): Likewise.
257 (find_reg): Likewise.
258 (clear_reload_reg_in_use): Likewise.
259 (free_for_value_p): Likewise.
260 (allocate_reload_reg): Likewise.
261 (choose_reload_regs): Likewise.
262 (reload_adjust_reg_for_temp): Likewise.
263 (emit_reload_insns): Likewise.
264 (delete_output_reload): Likewise.
265 * rtlanal.c (subreg_get_info): Likewise.
266 * sched-deps.c (sched_analyze_reg): Likewise.
267 * sel-sched.c (init_regs_for_mode): Likewise.
268 (mark_unavailable_hard_regs): Likewise.
269 (choose_best_reg_1): Likewise.
270 (verify_target_availability): Likewise.
271 * valtrack.c (dead_debug_insert_temp): Likewise.
272 * var-tracking.c (track_loc_p): Likewise.
273 (emit_note_insn_var_location): Likewise.
274 * varasm.c (make_decl_rtl): Likewise.
275 * reginfo.c (choose_hard_reg_mode): Likewise.
276 (init_reg_modes_target): Refer directly to
277 this_target_regs->x_hard_regno_nregs.
279 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
281 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
282 instead of hard_regno_nregs.
284 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
286 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
287 end_hard_regno instead of hard_regno_nregs.
288 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
289 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
290 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
291 * ira-color.c (improve_allocation): Likewise.
292 * lra-assigns.c (find_hard_regno_for_1): Likewise.
293 * lra-lives.c (mark_regno_live): Likewise.
294 (mark_regno_dead): Likewise.
295 * lra-remat.c (operand_to_remat): Likewise.
296 * lra.c (collect_non_operand_hard_regs): Likewise.
297 * postreload.c (reload_combine_note_store): Likewise.
298 (move2add_valid_value_p): Likewise.
299 * reload.c (regno_clobbered_p): Likewise.
301 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
303 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
305 * config/v850/v850.c (v850_reorg): Likewise.
306 * reload.c (refers_to_regno_for_reload_p): Likewise.
307 (find_equiv_reg): Likewise.
308 * reload1.c (reload_reg_reaches_end_p): Likewise.
310 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
312 * caller-save.c (add_used_regs): Use REG_NREGS instead of
314 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
315 * config/arm/arm.c (output_move_neon): Likewise.
316 (arm_attr_length_move_neon): Likewise.
317 (neon_split_vcombine): Likewise.
318 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
319 (c6x_mark_reg_written): Likewise.
320 (c6x_dwarf_register_span): Likewise.
321 * config/i386/i386.c (ix86_save_reg): Likewise.
322 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
323 (rws_access_reg): Likewise.
324 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
325 * mode-switching.c (create_pre_exit): Likewise.
326 * ree.c (combine_reaching_defs): Likewise.
327 (add_removable_extension): Likewise.
328 * regcprop.c (find_oldest_value_reg): Likewise.
329 (copyprop_hardreg_forward_1): Likewise.
330 * reload.c (reload_inner_reg_of_subreg): Likewise.
331 (push_reload): Likewise.
332 (combine_reloads): Likewise.
333 (find_dummy_reload): Likewise.
334 (reload_adjust_reg_for_mode): Likewise.
335 * reload1.c (find_reload_regs): Likewise.
336 (forget_old_reloads_1): Likewise.
337 (reload_reg_free_for_value_p): Likewise.
338 (reload_adjust_reg_for_temp): Likewise.
339 (emit_reload_insns): Likewise.
340 (delete_output_reload): Likewise.
341 * sel-sched.c (choose_best_reg_1): Likewise.
342 (choose_best_pseudo_reg): Likewise.
344 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
345 Alan Hayward <alan.hayward@arm.com>
346 David Sherwood <david.sherwood@arm.com>
348 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
349 * target.def (slow_unaligned_access): New hook.
350 * targhooks.h (default_slow_unaligned_access): Declare.
351 * targhooks.c (default_slow_unaligned_access): New function.
352 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
353 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
354 * doc/tm.texi: Regenerate.
355 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
356 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
357 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
359 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
360 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
362 (rs6000_slow_unaligned_access): New function.
363 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
364 (expand_block_compare): Likewise.
365 (expand_strn_compare): Likewise.
366 (rs6000_rtx_costs): Likewise.
367 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
368 (riscv_slow_unaligned_access): Likewise.
369 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
370 (riscv_slow_unaligned_access_p): ...this and make static.
371 (riscv_option_override): Update accordingly.
372 (riscv_slow_unaligned_access): New function.
373 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
374 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
375 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
376 (rs6000_slow_unaligned_access): New function.
377 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
378 (rs6000_rtx_costs): Likewise.
379 * config/rs6000/rs6000-string.c (expand_block_compare)
380 (expand_strn_compare): Use targetm.slow_unaligned_access instead
381 of SLOW_UNALIGNED_ACCESS.
382 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
383 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
384 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
385 of SLOW_UNALIGNED_ACCESS.
386 * expmed.c (simple_mem_bitfield_p): Likewise.
387 * expr.c (alignment_for_piecewise_move): Likewise.
388 (emit_group_load_1): Likewise.
389 (emit_group_store): Likewise.
390 (copy_blkmode_from_reg): Likewise.
391 (emit_push_insn): Likewise.
392 (expand_assignment): Likewise.
393 (store_field): Likewise.
394 (expand_expr_real_1): Likewise.
395 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
396 * lra-constraints.c (simplify_operand_subreg): Likewise.
397 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
398 * gimple-ssa-store-merging.c: Likewise in block comment at start
400 * tree-ssa-strlen.c: Include target.h.
401 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
402 of SLOW_UNALIGNED_ACCESS.
403 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
405 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
407 PR rtl-optimization/82185
408 * expmed.c (emit_store_flag_int): Only test tem if it has been
411 2017-09-12 Richard Biener <rguenther@suse.de>
414 * match.pd ((FTYPE) N CMP CST): Fix typo.
416 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
418 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
420 (mips_near_type_p): Add 'short_call' attribute as a synonym
422 * doc/extend.texi (short_call): Document new function attribute.
424 2017-09-12 Jakub Jelinek <jakub@redhat.com>
427 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
428 assertion check that in the condition.
429 (get_atomic_generic_size): Likewise. Before testing if parameter
430 has pointer type, if it has array type, call for C++
431 default_conversion to perform array-to-pointer conversion.
433 2017-09-12 Richard Biener <rguenther@suse.de>
435 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
436 for operations we cannot scalarize.
438 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
440 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
441 vectors heap vectors. Clean up comments.
442 Make visited_bbs a reference.
443 (profitable_jump_thread_path): Make GC
444 vectors heap vectors. Clean up comments.
446 (convert_and_register_jump_thread_path): Make GC vectors heap
448 (check_subpath_and_update_thread_path): Same. Clean up comments.
449 Make visited_bbs a reference.
450 (handle_phi): Abstract common code to to
451 register_jump_thread_path_if_profitable.
452 Rename VAR_BB to DEF_BB.
454 Make GC vectors heap vectors.
455 Make visited_bbs a reference.
456 (handle_assignment): Same.
457 (register_jump_thread_path_if_profitable): New.
458 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
460 Make GC vectors heap vectors. Clean up comments.
461 Make visited_bbs a reference.
462 (find_jump_threads_backwards): Make visited_bbs live in the stack.
463 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
466 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
469 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
470 words of E_DImode object are reachable by xtensa_uimm8x4 access.
472 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
474 Revert r251800 and r251799.
476 2017-09-11 Martin Jambor <mjambor@suse.cz>
479 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
480 arguments in advance.
481 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
482 use it to find predecessor edges.
483 (naive_outof_ssa): Collect vector of predecessors.
485 2017-09-08 Jason Merrill <jason@redhat.com>
487 PR c++/70029 - ICE with ref-qualifier and -flto
488 * langhooks.h (struct lang_hooks_for_types): Add
489 copy_lang_qualifiers.
490 * attribs.c (build_type_attribute_qual_variant): Use it.
491 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
493 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
494 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
496 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
499 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
500 (*mulsi3_sp64): New instruction.
501 (mulsi3): New expander.
503 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
505 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
507 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
509 * sancov.c: Include memmodel.h.
511 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
514 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
517 2017-09-07 Carl Love <cel@us.ibm.com>
519 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
520 the sldi instruction.
522 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
524 * sancov.c: Include tm_p.h.
526 2017-09-07 Jakub Jelinek <jakub@redhat.com>
529 * output.h (switch_to_other_text_partition): New declaration.
530 * varasm.c (switch_to_other_text_partition): New function.
531 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
532 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
533 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
534 to the other text partition before emitting LCL label and switch back
535 after emitting the word after it.
537 2017-09-07 Richard Biener <rguenther@suse.de>
539 * passes.def (pass_split_crit_edges): Remove instance before PRE.
540 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
541 critical edges here, after loop init.
542 (pass_data_pre): Remove PROP_no_crit_edges flags.
543 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
544 for valueization of call args to avoid leaking VN_TOP.
545 (visit_use): Assert we do not visit default defs.
546 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
547 Use error_mark_node to more easily detect leaking VN_TOP.
548 All default-defs are varying, not VN_TOP. Mark them visited.
549 (run_scc_vn): Make code match comment.
551 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
553 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
554 OPTION_MASK_FLOAT128_KEYWORD.
555 (POWERPC_MASKS): Likewise.
556 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
557 support for the -mfloat128-type option, and make -mfloat128
558 default on PowerPC Linux systems. Define or undefine
559 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
560 Define __float128 to be __ieee128 if IEEE 128-bit support is
561 enabled, or undefine it.
562 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
563 Delete defining __FLOAT128_TYPE__.
564 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
565 -mfloat128-type option and make -mfloat128 default on PowerPC
567 (TARGET_FLOAT128_TYPE): Likewise.
568 (-mfloat128-type): Likewise.
569 * config/rs6000/rs6000.c (rs6000_option_override_internal):
570 Delete the -mfloat128-type option and make -mfloat128 default on
571 PowerPC Linux systems. Always use __ieee128 to be the keyword for
572 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
573 128-bit floating point is enabled. Change tests from using
574 -mfloat128-type to -mfloat128.
575 (rs6000_mangle_type): Use the correct mangling for the __float128
576 type even if normal long double is restricted to 64-bits.
577 (floatn_mode): Enable the _Float128 type by default on VSX Linux
579 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
580 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
581 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
583 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
584 documentation for -mfloat128.
586 2017-09-06 Olivier Hainque <hainque@adacore.com>
588 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
590 2017-09-06 Wish Wu <wishwu007@gmail.com>
591 Jakub Jelinek <jakub@redhat.com>
593 * asan.c (initialize_sanitizer_builtins): Add
594 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
595 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
596 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
597 BT_FN_VOID_UINT64_PTR variables.
598 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
599 (BT_FN_VOID_UINT16_UINT16): Likewise.
600 (BT_FN_VOID_UINT32_UINT32): Likewise.
601 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
602 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
603 (BT_FN_VOID_UINT64_PTR): Likewise.
604 * common.opt (flag_sanitize_coverage): New variable.
605 (fsanitize-coverage=trace-pc): Remove.
606 (fsanitize-coverage=): Add.
607 * flag-types.h (enum sanitize_coverage_code): New enum.
608 * fold-const.c (fold_range_test): Disable non-short-circuit
609 optimization if flag_sanitize_coverage.
610 (fold_truth_andor): Likewise.
611 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
612 * opts.c (COVERAGE_SANITIZER_OPT): Define.
613 (coverage_sanitizer_opts): New array.
614 (get_closest_sanitizer_option): Add OPTS argument, handle also
615 OPT_fsanitize_coverage_.
616 (parse_sanitizer_options): Adjusted to also handle
617 OPT_fsanitize_coverage_.
618 (common_handle_option): Add OPT_fsanitize_coverage_.
619 * sancov.c (instrument_comparison, instrument_switch): New function.
620 (sancov_pass): Add trace-cmp support.
621 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
622 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
623 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
624 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
625 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
626 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
627 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
628 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
629 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
630 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
632 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
634 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
635 error. Only quit immediately if parsing is complete.
636 (BEGIN): Initialize fatal_err and parse_done.
637 (begin fpu, end fpu): Check number of arguments.
638 (begin arch, end arch): Likewise.
639 (begin cpu, end cpu): Likewise.
640 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
641 (optalias): Likewise.
643 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
645 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
646 * config/arm/arm-isa.h: Delete. Move definitions to ...
647 * arm-cpus.in: ... here. Use new feature and fgroup values.
648 * config/arm/arm.c (arm_option_override): Use lower case for feature
650 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
651 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
652 * config/arm/parsecpu.awk (END): Add new command 'isa'.
654 (print_isa_bits_for): New function.
655 (gen_isa): New function.
656 (gen_comm_data): Use print_isa_bits_for.
657 (define feature): New keyword.
658 (define fgroup): New keyword.
659 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
660 (arm-isa.h): Add rule to generate file.
661 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
662 case for feature bit names.
664 2017-09-06 Richard Biener <rguenther@suse.de>
666 * tree-ssa-pre.c (NECESSARY): Remove.
667 (create_expression_by_pieces): Do not touch pass-local flags.
668 (insert_into_preds_of_block): Likewise.
669 (do_pre_regular_insertion): Likewise.
670 (eliminate_insert): Likewise.
671 (eliminate_dom_walker::before_dom_children): Likewise.
672 (fini_eliminate): Do not look at inserted_exprs.
673 (mark_operand_necessary): Remove.
674 (remove_dead_inserted_code): Replace with simple work-list
675 algorithm based on inserted_exprs and SSA uses.
676 (pass_pre::execute): Re-order fini_eliminate and
677 remove_dead_inserted_code.
679 2017-09-06 Olivier Hainque <hainque@adacore.com>
681 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
682 for VxWorks 7. Adjust surrounding comments.
684 2017-09-06 Richard Biener <rguenther@suse.de>
686 * gimple-ssa-strength-reduction.c
687 (find_candidates_dom_walker::before_dom_children): Also allow
690 2017-09-06 Richard Biener <rguenther@suse.de>
692 PR tree-optimization/82108
693 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
694 for gap in the non-permutation SLP case.
696 2017-09-06 Martin Jambor <mjambor@suse.cz>
698 PR tree-optimization/82078
699 * tree-sra.c (sort_and_splice_var_accesses): Move call to
700 add_access_to_work_queue...
701 (build_accesses_from_assign): ...here.
702 (propagate_all_subaccesses): Make sure racc is the group
703 representative, if there is one.
705 2017-09-06 Jakub Jelinek <jakub@redhat.com>
708 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
711 2017-09-06 Richard Biener <rguenther@suse.de>
713 * gimple-ssa-strength-reduction.c
714 (find_candidates_dom_walker::before_doom_children): Use a
715 type and not a mode check.
717 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
720 * config/arm/predicates.md (arm_general_adddi_operand): Create new
722 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
724 2017-09-05 Jeff Law <law@redhat.com>
726 PR tree-optimization/64910
727 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
728 cases where we have 3 or more operands.
730 2017-09-05 Jakub Jelinek <jakub@redhat.com>
733 * omp-low.c (lower_omp_for): Recompute tree invariant if
734 gimple_omp_for_initial/final is ADDR_EXPR.
737 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
738 into gimple val before gimplification fo the COND_EXPR.
740 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
742 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
743 REGION_COPY argument.
744 (thread_through_all_blocks): Remove unused argument to
745 duplicate_thread_path.
747 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
748 Alan Hayward <alan.hayward@arm.com>
749 David Sherwood <david.sherwood@arm.com>
751 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
752 Take a scalar_mode rather than a machine_mode.
753 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
754 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
755 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
756 (aarch64_gen_adjusted_ldpstp): Likewise.
757 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
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_is_extend_from_extract):
764 Take a scalar_int_mode instead of a machine_mode.
765 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
766 (aarch64_output_scalar_simd_mov_immediate): Likewise.
767 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
768 (aarch64_simd_attr_length_rglist): Delete.
769 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
770 a scalar_int_mode instead of a machine_mode.
771 (aarch64_add_offset): Likewise.
772 (aarch64_internal_mov_immediate): Likewise
773 (aarch64_add_constant_internal): Likewise.
774 (aarch64_add_constant): Likewise.
775 (aarch64_movw_imm): Likewise.
776 (aarch64_rtx_arith_op_extract_p): Likewise.
777 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
778 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
779 Remove assert that the mode isn't a vector.
780 (aarch64_output_scalar_simd_mov_immediate): Likewise.
781 (aarch64_expand_mov_immediate): Update calls after above changes.
782 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
783 (aarch64_and_bitmask_imm): Check for scalar integer modes.
784 (aarch64_move_imm): Likewise.
785 (aarch64_can_const_movi_rtx_p): Likewise.
786 (aarch64_strip_extend): Likewise.
787 (aarch64_extr_rtx_p): Likewise.
788 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
789 a CONST_INT when the mode parameter is VOIDmode.
790 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
792 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
794 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
795 * stor-layout.c (bitwise_mode_for_mode): Likewise.
796 (bitwise_type_for_mode): Update accordingly.
798 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
800 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
801 * stor-layout.c (mode_for_size_tree): Likewise.
802 (mode_for_array): Update accordingly.
803 (layout_decl): Likewise.
804 (compute_record_mode): Likewise. Only set the mode once.
806 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
808 * target.def (get_mask_mode): Change return type to opt_mode.
810 * doc/tm.texi: Regenerate.
811 * targhooks.h (default_get_mask_mode): Return an opt_mode.
812 * targhooks.c (default_get_mask_mode): Likewise.
813 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
814 * optabs-query.c (can_vec_mask_load_store_p): Update use of
815 targetm.get_mask_mode.
816 * tree.c (build_truth_vector_type): Likewise.
818 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
820 * machmode.h (mode_for_vector): Return an opt_mode.
821 * stor-layout.c (mode_for_vector): Likewise.
822 (mode_for_int_vector): Update accordingly.
823 (layout_type): Likewise.
824 * config/i386/i386.c (emit_memmov): Likewise.
825 (ix86_expand_set_or_movmem): Likewise.
826 (ix86_expand_vector_init): Likewise.
827 (ix86_get_mask_mode): Likewise.
828 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
830 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
831 * expmed.c (extract_bit_field_1): Likewise.
832 * expr.c (expand_expr_real_2): Likewise.
833 * optabs-query.c (can_vec_perm_p): Likewise.
834 (can_vec_mask_load_store_p): Likewise.
835 * optabs.c (expand_vec_perm): Likewise.
836 * targhooks.c (default_get_mask_mode): Likewise.
837 * tree-vect-stmts.c (vectorizable_store): Likewise.
838 (vectorizable_load): Likewise.
839 (get_vectype_for_scalar_type_and_size): Likewise.
841 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
843 * machmode.h (mode_for_int_vector): New function.
844 * stor-layout.c (mode_for_int_vector): Likewise.
845 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
846 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
847 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
848 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
849 (s390_expand_vcond): Likewise.
851 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
853 * machmode.h (opt_machine_mode): New type.
854 (opt_mode<T>): Allow construction from anything that can be
856 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
857 (mode_for_size): Return an opt_machine_mode.
858 * stor-layout.c (mode_for_size): Likewise.
859 (mode_for_size_tree): Update call accordingly.
860 (bitwise_mode_for_mode): Likewise.
861 (make_fract_type): Likewise.
862 (make_accum_type): Likewise.
863 * caller-save.c (replace_reg_with_saved_mem): Update call
865 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
866 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
867 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
868 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
869 * expmed.c (extract_bit_field_1): Likewise.
870 * reload.c (get_secondary_mem): Likewise.
871 * varasm.c (assemble_integer): Likewise.
872 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
875 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
877 * machmode.h (decimal_float_mode_for_size): New function.
878 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
879 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
880 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
881 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
882 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
883 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
885 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
887 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
888 (get_builtin_sync_mode): Likewise.
889 (expand_ifn_atomic_compare_exchange): Likewise.
890 (expand_builtin_atomic_clear): Likewise.
891 (expand_builtin_atomic_test_and_set): Likewise.
892 (fold_builtin_atomic_always_lock_free): Likewise.
893 * calls.c (compute_argument_addresses): Likewise.
894 (emit_library_call_value_1): Likewise.
895 (store_one_arg): Likewise.
896 * combine.c (combine_instructions): Likewise.
897 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
898 * config/arm/arm.c (arm_function_value): Likewise.
899 (aapcs_allocate_return_reg): Likewise.
900 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
901 * config/i386/i386.c (construct_container): Likewise.
902 (ix86_gimplify_va_arg): Likewise.
903 (ix86_expand_sse_cmp): Likewise.
904 (emit_memmov): Likewise.
905 (emit_memset): Likewise.
906 (expand_small_movmem_or_setmem): Likewise.
907 (ix86_expand_pextr): Likewise.
908 (ix86_expand_pinsr): Likewise.
909 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
910 * config/microblaze/microblaze.c (microblaze_block_move_straight):
912 * config/mips/mips.c (mips_function_value_1) Likewise.
913 (mips_block_move_straight): Likewise.
914 (mips_expand_ins_as_unaligned_store): Likewise.
915 * config/powerpcspe/powerpcspe.c
916 (rs6000_darwin64_record_arg_advance_flush): Likewise.
917 (rs6000_darwin64_record_arg_flush): Likewise.
918 * config/rs6000/rs6000.c
919 (rs6000_darwin64_record_arg_advance_flush): Likewise.
920 (rs6000_darwin64_record_arg_flush): Likewise.
921 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
922 (sparc_function_value_1): Likewise.
923 * config/spu/spu.c (adjust_operand): Likewise.
924 (spu_emit_branch_or_set): Likewise.
925 (arith_immediate_p): Likewise.
926 * emit-rtl.c (gen_lowpart_common): Likewise.
927 * expr.c (expand_expr_real_1): Likewise.
928 * function.c (assign_parm_setup_block): Likewise.
929 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
930 * reload1.c (alter_reg): Likewise.
931 * stor-layout.c (mode_for_vector): Likewise.
932 (layout_type): Likewise.
934 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
936 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
937 (spu_convert_move): Likewise.
938 * lower-subreg.c (resolve_simple_move): Likewise.
940 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
943 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
944 define_insn to a define_expand.
945 (altivec_vsum2sws_direct): New define_insn.
946 (altivec_vsumsws): Convert from a define_insn to a define_expand.
948 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
950 * config/arm/arm.c (arm_option_params_internal): Improve setting of
953 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
958 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
959 realign stack if stack alignment needed is less than incoming
962 2017-09-05 Marek Polacek <polacek@redhat.com>
965 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
968 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
970 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
972 2017-09-05 Richard Biener <rguenther@suse.de>
974 PR tree-optimization/82084
975 * fold-const.c (can_native_encode_string_p): Handle wide characters.
977 2017-09-05 Richard Biener <rguenther@suse.de>
979 PR tree-optimization/82102
980 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
982 2017-09-05 Martin Liska <mliska@suse.cz>
984 PR tree-optimization/82032
985 * tree-cfg.c (generate_range_test): New function.
986 * tree-cfg.h (generate_range_test): Declared here.
987 * tree-cfgcleanup.c (convert_single_case_switch): New function.
988 (cleanup_control_expr_graph): Use it.
989 * tree-switch-conversion.c (try_switch_expansion): Remove
991 (emit_case_nodes): Use generate_range_test.
993 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
996 * config/i386/i386.md (*<btsc><mode>_mask): Add
997 TARGET_USE_BT to insn constraint.
998 (*btr<mode>_mask): Ditto.
1000 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1002 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1003 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1005 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1008 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1009 TARGET_NEON and TARGET_IWMMXT.
1010 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1011 TARGET_NEON and TARGET_IWMMXT.
1012 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1014 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1016 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1017 (ix86_rewrite_tls_address): Ditto.
1018 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1019 (ix86_rewrite_tls_address_1): Ditto.
1020 (ix86_rewrite_tls_address): Ditto.
1021 * config/i386/predicates.md (tls_address_pattern): New predicate.
1022 * config/i386/i386.md (TLS address splitter): New splitter.
1024 2017-09-04 Richard Biener <rguenther@suse.de>
1026 PR tree-optimization/82084
1027 * fold-const.h (can_native_encode_string_p): Declare.
1028 * fold-const.c (can_native_encode_string_p): Factor out from ...
1029 (native_encode_string): ... here.
1030 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1031 vectorizing stores from constants we later cannot handle.
1033 2017-09-04 Marek Polacek <polacek@redhat.com>
1036 * doc/invoke.texi: Update -Wtautological-compare documentation.
1038 2017-09-04 Jeff Law <law@redhat.com>
1040 PR tree-optimization/64910
1041 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1042 swap the first and last operand if the last is a constant.
1044 2017-09-04 Marek Polacek <polacek@redhat.com>
1047 * convert.c (do_narrow): When sanitizing signed integer overflows,
1048 bail out for signed types.
1049 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1051 2017-09-04 Richard Biener <rguenther@suse.de>
1053 PR tree-optimization/82060
1054 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1055 Move devirtualization after stmt folding and before EH/AB/noreturn
1056 cleanup to get the stmt refs canonicalized. Use a bool instead
1057 of gimple_modified_p since that doesn't work for NOPs. Schedule
1058 NOPs generated by folding for removal.
1060 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1061 Alan Hayward <alan.hayward@arm.com>
1062 David Sherwood <david.sherwood@arm.com>
1064 * coretypes.h (pad_direction): New enum.
1065 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1066 (FUNCTION_ARG_PADDING): Likewise.
1067 * target.def (function_arg_padding): New hook.
1068 * targhooks.h (default_function_arg_padding): Declare.
1069 * targhooks.c (default_function_arg_padding): New function.
1070 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1071 (TARGET_FUNCTION_ARG_PADDING): ...this.
1072 * doc/tm.texi: Regenerate.
1073 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1074 instead of direction.
1075 (compute_argument_addresses): Likewise.
1076 (load_register_parameters): Likewise.
1077 (emit_library_call_value_1): Likewise.
1078 (store_one_arg): Use targetm.calls.function_arg_padding instead
1079 of FUNCTION_ARG_PADDING.
1080 (must_pass_in_stack_var_size_or_pad): Likewise.
1081 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1082 (emit_group_store): Likewise.
1083 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1084 instead of FUNCTION_ARG_PADDING.
1085 (emit_push_insn): Likewise, and propagate enum change throughout
1087 * function.h (direction): Delete.
1088 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1090 * function.c (assign_parm_find_stack_rtl): Likewise.
1091 (assign_parm_setup_block_p): Likewise.
1092 (assign_parm_setup_block): Likewise.
1093 (gimplify_parameters): Likewise.
1094 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1095 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1097 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1098 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1099 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1100 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1101 (aarch64_function_arg_padding): ...this new function.
1102 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1103 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1104 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1105 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1106 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1107 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1108 (arm_pad_arg_upward): Replace with...
1109 (arm_function_arg_padding): ...this new function.
1110 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1112 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1113 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1114 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1115 (ia64_hpux_function_arg_padding): Replace with...
1116 (ia64_function_arg_padding): ...this new function. Use pad_direction
1117 instead of direction. Check for TARGET_HPUX.
1118 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1119 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1120 (iq2000_function_arg_padding): New function.
1121 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1122 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1123 (mips_function_arg_padding): ...this new function.
1124 (mips_pad_reg_upward): Update accordingly.
1125 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1126 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1127 targetm.calls.function_arg_padding.
1128 (FUNCTION_ARG_PADDING): Delete.
1129 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1130 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1131 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1132 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1133 (nios2_block_reg_padding): Return pad_direction instead of direction.
1134 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1135 instead of direction.
1136 (nios2_function_arg_padding): Likewise. Make static.
1137 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1138 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1139 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1140 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1141 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1142 (pa_function_arg_padding): Make static. Return pad_direction instead
1144 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1145 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1146 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1147 instead of direction. Use targetm.calls.function_arg_padding.
1148 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1149 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1150 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1151 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1152 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1154 (function_arg_padding): Rename to...
1155 (rs6000_function_arg_padding): ...this. Make static. Return
1156 pad_direction instead of direction.
1157 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1158 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1159 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1160 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1161 instead of direction. Use targetm.calls.function_arg_padding.
1162 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1163 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1164 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1165 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1166 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1167 (function_arg_padding): Rename to...
1168 (rs6000_function_arg_padding): ...this. Make static. Return
1169 pad_direction instead of direction.
1170 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1171 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1172 * config/s390/s390.c (s390_function_arg_padding): New function.
1173 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1174 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1175 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1176 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1177 (function_arg_padding): Rename to...
1178 (sparc_function_arg_padding): ...this. Make static. Return
1179 pad_direction instead of direction.
1180 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1181 * config/spu/spu.c (spu_function_arg_padding): New function.
1182 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1183 * system.h (FUNCTION_ARG_PADDING): Poison.
1185 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1186 Alan Hayward <alan.hayward@arm.com>
1187 David Sherwood <david.sherwood@arm.com>
1189 * target.def (modes_tieable_p): New hook.
1190 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1191 (TARGET_MODES_TIEABLE_P): ...this.
1192 * doc/tm.texi.in: Regenerate.
1193 * hooks.h (hook_bool_mode_mode_true): Declare.
1194 * hooks.c (hook_bool_mode_mode_true): New function.
1195 * combine.c (subst): Use targetm.modes_tieable_p instead of
1197 * dse.c (find_shift_sequence): Likewise.
1198 * expmed.c (extract_low_bits): Likewise.
1199 * lower-subreg.c: Include target.h.
1200 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1202 * rtlanal.c (rtx_cost): Likewise.
1203 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1204 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1205 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1206 (TARGET_MODES_TIEABLE_P): Redefine.
1207 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1208 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1209 (TARGET_MODES_TIEABLE_P): Redefine.
1210 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1211 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1212 (arc_modes_tieable_p): New function.
1213 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1214 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1215 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1216 (arm_modes_tieable_p): Make static.
1217 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1218 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1219 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1220 (TARGET_MODES_TIEABLE_P): Redefine.
1221 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1222 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1223 (TARGET_MODES_TIEABLE_P): Redefine.
1224 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1225 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1226 (cr16_modes_tieable_p): New function.
1227 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1228 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1229 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1230 (TRULY_NOOP_TRUNCATION): Update comment.
1231 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1232 (TRULY_NOOP_TRUNCATION): Update comment.
1233 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1234 (frv_modes_tieable_p): New function.
1235 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1236 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1237 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1238 (TARGET_MODES_TIEABLE_P): Redefine.
1239 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1240 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1241 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1242 (TARGET_MODES_TIEABLE_P): Redefine.
1243 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1244 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1245 (ia64_modes_tieable_p): New function.
1246 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1247 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1248 (iq2000_modes_tieable_p): New function.
1249 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1250 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1251 (lm32_modes_tieable_p): New function.
1252 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1253 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1254 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1255 (TARGET_MODES_TIEABLE_P): Redefine.
1256 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1257 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1258 (m32r_modes_tieable_p): New function.
1259 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1260 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1261 (m68k_modes_tieable_p): New function.
1262 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1263 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1264 (mcore_modes_tieable_p): New function.
1265 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1266 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1268 (TARGET_MODES_TIEABLE_P): Redefine.
1269 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1270 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1271 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1272 (TARGET_MODES_TIEABLE_P): Redefine.
1273 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1274 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1275 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1276 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1277 (mn10300_modes_tieable_p): ...this and make static.
1278 (TARGET_MODES_TIEABLE_P): Redefine.
1279 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1280 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1281 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1282 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1283 (msp430_modes_tieable_p): Make static.
1284 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1285 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1286 (TARGET_MODES_TIEABLE_P): Redefine.
1287 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1288 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1289 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1290 (TARGET_MODES_TIEABLE_P): Redefine.
1291 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1292 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1293 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1294 (TARGET_MODES_TIEABLE_P): Redefine.
1295 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1296 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1297 (pdp11_modes_tieable_p): New function.
1298 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1299 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1300 (rs6000_modes_tieable_p): New function.
1301 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1302 * config/powerpcspe/powerpcspe.md: Update comment.
1303 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1304 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1305 (TARGET_MODES_TIEABLE_P): Redefine.
1306 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1307 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1308 (rl78_modes_tieable_p): New function.
1309 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1310 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1311 (rs6000_modes_tieable_p): New function.
1312 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1313 * config/rs6000/rs6000.md: Update comment.
1314 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1315 * config/rx/rx.c (rx_modes_tieable_p): New function.
1316 (TARGET_MODES_TIEABLE_P): Redefine.
1317 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1318 * config/s390/s390.c (s390_modes_tieable_p): New function.
1319 (TARGET_MODES_TIEABLE_P): Redefine.
1320 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1321 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1322 (sh_modes_tieable_p): New function.
1323 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1324 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1325 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1326 (sparc_modes_tieable_p): Make static.
1327 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1328 * config/spu/spu.c (spu_modes_tieable_p): New function.
1329 (TARGET_MODES_TIEABLE_P): Redefine.
1330 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1331 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1332 (TARGET_MODES_TIEABLE_P): Redefine.
1333 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1334 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1335 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1336 * config/v850/v850.c (v850_modes_tieable_p): New function.
1337 (TARGET_MODES_TIEABLE_P): Redefine.
1338 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1339 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1340 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1341 (visium_modes_tieable_p): New function.
1342 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1343 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1344 (xtensa_modes_tieable_p): New function.
1345 * system.h (MODES_TIEABLE_P): Poison.
1347 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1348 Alan Hayward <alan.hayward@arm.com>
1349 David Sherwood <david.sherwood@arm.com>
1351 * target.def (hard_regno_mode_ok): New hook.
1352 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1353 (TARGET_HARD_REGNO_MODE_OK): ...this.
1354 * doc/tm.texi.in: Regenerate.
1355 * hooks.h (hook_bool_uint_mode_true): Declare.
1356 * hooks.c (hook_bool_uint_mode_true): New function.
1357 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1359 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1360 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1361 instead of HARD_REGNO_MODE_OK.
1362 * caller-save.c: Include target.h.
1363 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1365 * combine.c (can_combine_p): Likewise.
1366 (combinable_i3pat): Likewise.
1367 (can_change_dest_mode): Likewise.
1368 * expr.c (init_expr_target): Likewise.
1369 (convert_move): Likewise.
1370 (convert_modes): Likewise.
1371 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1372 (setup_prohibited_mode_move_regs): Likewise.
1373 * ira.h (target_ira): Likewise.
1374 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1375 * lra-constraints.c (process_alt_operands): Likewise.
1376 (split_reg): Likewise.
1377 * recog.c (peep2_find_free_register): Likewise.
1378 * ree.c (combine_reaching_defs): Likewise.
1379 * regcprop.c (maybe_mode_change): Likewise.
1380 * reginfo.c (init_reg_sets_1): Likewise.
1381 (choose_hard_reg_mode): Likewise.
1382 (simplifiable_subregs): Likewise.
1383 * regrename.c (check_new_reg_p): Likewise.
1384 * reload.c (find_valid_class): Likewise.
1385 (find_valid_class_1): Likewise.
1386 (reload_inner_reg_of_subreg): Likewise.
1387 (push_reload): Likewise.
1388 (combine_reloads): Likewise.
1389 (find_dummy_reload): Likewise.
1390 (find_reloads): Likewise.
1391 * reload1.c (find_reg): Likewise.
1392 (set_reload_reg): Likewise.
1393 (allocate_reload_reg): Likewise.
1394 (choose_reload_regs): Likewise.
1395 (reload_adjust_reg_for_temp): Likewise.
1396 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1397 (simplify_subreg_regno): Likewise.
1398 * sel-sched.c (init_regs_for_mode): Likewise.
1399 * varasm.c (make_decl_rtl): Likewise.
1400 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1401 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1403 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1404 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1405 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1406 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1407 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1408 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1409 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1410 (arc_mode_class): Delete.
1411 (HARD_REGNO_MODE_OK): Delete.
1412 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1413 (arc_hard_regno_mode_ok): Rename old array to...
1414 (arc_hard_regno_mode_ok_modes): ...this.
1415 (arc_conditional_register_usage): Update accordingly.
1416 (arc_mode_class): Make static.
1417 (arc_hard_regno_mode_ok): New function.
1418 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1419 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1420 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1421 (arm_hard_regno_mode_ok): Make static.
1422 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1424 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1425 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1426 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1428 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1429 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1430 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1431 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1432 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1433 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1434 * config/bfin/predicates.md (valid_reg_operand): Use
1435 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1436 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1437 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1438 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1439 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1440 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1441 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1442 (cr16_hard_regno_mode_ok): Make static and return a bool.
1443 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1444 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1445 (cris_hard_regno_mode_ok): New function.
1446 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1447 (epiphany_mode_class): Delete.
1448 (HARD_REGNO_MODE_OK): Delete.
1449 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1450 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1451 (hard_regno_mode_ok): Rename to...
1452 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1453 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1454 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1456 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1457 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1458 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1459 (frv_hard_regno_mode_ok): Make static and return a bool.
1460 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1462 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1463 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1464 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1465 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1467 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1468 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1469 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1470 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1472 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1473 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1474 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1475 (ia64_hard_regno_mode_ok): New function.
1476 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1477 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1478 (iq2000_hard_regno_mode_ok): New function.
1479 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1480 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1481 (lm32_hard_regno_mode_ok): New function.
1482 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1483 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1484 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1485 instead of HARD_REGNO_MODE_OK.
1486 (m32c_hard_regno_ok): Rename to...
1487 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1488 (m32c_cannot_change_mode_class): Update accordingly.
1489 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1490 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1491 (m32r_mode_class): Delete.
1492 (HARD_REGNO_MODE_OK): Delete.
1493 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1494 (m32r_hard_regno_mode_ok): Rename to...
1495 (m32r_hard_regno_modes): ...this.
1496 (m32r_mode_class): Make static.
1497 (m32r_hard_regno_mode_ok): New function.
1498 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1499 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1500 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1501 (m68k_hard_regno_mode_ok): Make static.
1502 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1503 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1504 (mcore_hard_regno_mode_ok): New function.
1505 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1506 (HARD_REGNO_MODE_OK): Delete.
1507 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1509 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1510 (microblaze_hard_regno_mode_ok): New function.
1511 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1512 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1513 (mips_hard_regno_mode_ok): Delete.
1514 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1515 (mips_hard_regno_mode_ok_p): ...this and make static.
1516 (mips_hard_regno_mode_ok_p): Rename to...
1517 (mips_hard_regno_mode_ok_uncached): ...this.
1518 (mips_hard_regno_mode_ok): New function.
1519 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1520 of HARD_REGNO_MODE_OK.
1521 (mips_option_override): Update after above name changes.
1522 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1523 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1524 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1525 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1526 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1527 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1528 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1529 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1530 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1531 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1532 (msp430_hard_regno_mode_ok): Make static and return a bool.
1533 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1534 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1535 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1537 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1538 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1539 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1540 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1541 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1542 (PA_HARD_REGNO_MODE_OK): ...this
1543 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1544 (PA_HARD_REGNO_MODE_OK): ...this.
1545 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1546 (pa_hard_regno_mode_ok): New function.
1547 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1548 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1549 (pdp11_hard_regno_mode_ok): New function.
1550 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1551 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1553 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1555 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1556 (rs6000_hard_regno_mode_ok): Rename to...
1557 (rs6000_hard_regno_mode_ok_uncached): ...this.
1558 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1559 (rs6000_hard_regno_mode_ok): New function.
1560 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1561 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1562 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1563 (riscv_hard_regno_mode_ok): ...this and make static.
1564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1565 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1566 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1567 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1568 (rl78_hard_regno_mode_ok): Make static and return bool.
1569 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1570 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1572 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1573 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1574 (rs6000_hard_regno_mode_ok): Rename to...
1575 (rs6000_hard_regno_mode_ok_uncached): ...this.
1576 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1577 (rs6000_hard_regno_mode_ok): New function.
1578 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1579 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1580 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1581 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1582 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1583 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1584 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1585 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1586 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1587 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1588 (sh_hard_regno_mode_ok): Make static.
1589 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1590 instead of HARD_REGNO_MODE_OK.
1591 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1592 (sparc_mode_class): Delete.
1593 (HARD_REGNO_MODE_OK): Delete.
1594 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1595 (hard_regno_mode_classes): Make static.
1596 (sparc_mode_class): Likewise.
1597 (sparc_hard_regno_mode_ok): New function.
1598 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1599 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1600 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1602 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1603 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1604 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1605 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1606 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1607 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1608 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1609 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1610 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1611 (visium_hard_regno_mode_ok): New function.
1612 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1613 instead of HARD_REGNO_MODE_OK.
1614 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1615 (HARD_REGNO_MODE_OK): Delete.
1616 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1617 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1618 (xtensa_option_override): Update accordingly.
1619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1620 (xtensa_hard_regno_mode_ok): New function.
1621 * system.h (HARD_REGNO_MODE_OK): Poison.
1623 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1624 Alan Hayward <alan.hayward@arm.com>
1625 David Sherwood <david.sherwood@arm.com>
1627 * target.def (hard_regno_call_part_clobbered): New hook.
1628 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1629 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1630 * doc/tm.texi: Regenerate.
1631 * hooks.h (hook_bool_uint_mode_false): Declare.
1632 * hooks.c (hook_bool_uint_mode_false): New function.
1633 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1634 * cselib.c (cselib_process_insn): Use
1635 targetm.hard_regno_call_part_clobbered instead of
1636 HARD_REGNO_CALL_PART_CLOBBERED.
1637 * ira-conflicts.c (ira_build_conflicts): Likewise.
1638 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1639 * lra-constraints.c (need_for_call_save_p): Likewise.
1640 * lra-lives.c: Include target.h.
1641 (check_pseudos_live_through_calls): Use
1642 targetm.hard_regno_call_part_clobbered instead of
1643 HARD_REGNO_CALL_PART_CLOBBERED.
1644 * regcprop.c: Include target.h.
1645 (copyprop_hardreg_forward_1): Use
1646 targetm.hard_regno_call_part_clobbered instead of
1647 HARD_REGNO_CALL_PART_CLOBBERED.
1648 * reginfo.c (choose_hard_reg_mode): Likewise.
1649 * regrename.c (check_new_reg_p): Likewise.
1650 * reload.c (find_equiv_reg): Likewise.
1651 * reload1.c (emit_reload_insns): Likewise.
1652 * sched-deps.c (deps_analyze_insn): Likewise.
1653 * sel-sched.c (init_regs_for_mode): Likewise.
1654 (mark_unavailable_hard_regs): Likewise.
1655 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1656 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1657 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1659 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1660 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1661 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1663 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1665 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1666 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1667 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1669 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1670 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1671 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1673 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1674 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1676 * config/powerpcspe/powerpcspe.c
1677 (rs6000_hard_regno_call_part_clobbered): New function.
1678 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1679 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1680 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1682 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1683 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1684 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1686 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1687 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1688 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1690 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1691 Alan Hayward <alan.hayward@arm.com>
1692 David Sherwood <david.sherwood@arm.com>
1694 * rtl.h (subreg_memory_offset): Declare.
1695 * emit-rtl.c (subreg_memory_offset): New function.
1696 * expmed.c (store_bit_field_1): Use it.
1697 * expr.c (undefined_operand_subword_p): Likewise.
1698 * simplify-rtx.c (simplify_subreg): Likewise.
1700 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1702 PR rtl-optimization/57448
1705 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1706 using atomic_load pattern.
1707 (expand_atomic_store): Likewise.
1709 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1712 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1713 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1714 handling. Use replace_call_with_value with NULL instead of
1715 gsi_replace, unlink_stmt_vdef and release_defs.
1717 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1720 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1722 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1725 * rtl.h (emit_library_call_value_1): Declare.
1726 (emit_library_call): Replace declaration with a series of overloads.
1727 Remove the parameter count argument.
1728 (emit_library_call_value): Likewise.
1729 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1730 with an "rtx_mode_t *".
1731 (emit_library_call_value): Delete.
1732 (emit_library_call): Likewise.
1733 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1734 (asan_emit_allocas_unpoison): Likewise.
1735 * builtins.c (expand_builtin_powi): Likewise.
1736 (expand_asan_emit_allocas_unpoison): Likewise.
1737 * cfgexpand.c (expand_main_function): Likewise.
1738 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1739 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1740 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1741 * config/arm/arm.c (arm_trampoline_init): Likewise.
1742 (arm_call_tls_get_addr): Likewise.
1743 (arm_expand_divmod_libfunc): Likewise.
1744 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1745 (smulsi3_highpart): Likewise.
1746 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1747 (c6x_expand_compare): Likewise.
1748 (c6x_expand_movmem): Likewise.
1749 * config/frv/frv.c (frv_trampoline_init): Likewise.
1750 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1751 (ix86_expand_divmod_libfunc): Likewise.
1752 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1753 (ia64_expand_compare): Likewise.
1754 (ia64_profile_hook): Likewise.
1755 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1756 (nonlocal_goto): Likewise.
1757 (restore_stack_nonlocal): Likewise.
1758 * config/m32r/m32r.c (block_move_call): Likewise.
1759 (m32r_trampoline_init): Likewise.
1760 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1761 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1762 (m68k_call_m68k_read_tp): Likewise.
1763 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1764 (microblaze_expand_divide): Likewise.
1765 * config/mips/mips.h (mips_args): Likewise.
1766 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1767 (MIPS_ICACHE_SYNC): Likewise.
1768 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1769 (nios2_trampoline_init): Likewise.
1770 * config/pa/pa.c (hppa_tls_call): Likewise.
1771 (pa_trampoline_init): Likewise.
1772 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1773 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1774 (expand_strn_compare): Likewise.
1775 (rs6000_generate_compare): Likewise.
1776 (rs6000_expand_float128_convert): Likewise.
1777 (output_profile_hook): Likewise.
1778 (rs6000_trampoline_init): Likewise.
1779 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1780 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1781 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1782 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1783 (rs6000_generate_compare): Likewise.
1784 (rs6000_expand_float128_convert): Likewise.
1785 (output_profile_hook): Likewise.
1786 (rs6000_trampoline_init): Likewise.
1787 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1788 * config/sh/sh.c (sh_trampoline_init): Likewise.
1789 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1790 (sparc_emit_float_lib_cmp): Likewise.
1791 (sparc32_initialize_trampoline): Likewise.
1792 (sparc64_initialize_trampoline): Likewise.
1793 (sparc_profile_hook): Likewise.
1794 * config/spu/spu.c (ea_load_store): Likewise.
1795 * config/spu/spu.md (floatunssidf2): Likewise.
1796 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1797 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1798 * config/visium/visium.c (expand_block_move_4): Likewise.
1799 (expand_block_move_2): Likewise.
1800 (expand_block_move_1): Likewise.
1801 (expand_block_set_4): Likewise.
1802 (expand_block_set_2): Likewise.
1803 (expand_block_set_1): Likewise.
1804 (visium_trampoline_init): Likewise.
1805 (visium_profile_hook): Likewise.
1806 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1807 (xtensa_setup_frame_addresses): Likewise.
1808 (xtensa_trampoline_init): Likewise.
1809 * except.c (sjlj_emit_function_enter): Likewise.
1810 (sjlj_emit_function_exit): Likewise.
1811 * explow.c (allocate_dynamic_stack_space): Likewise.
1812 (probe_stack_range): Likewise.
1813 * expr.c (convert_mode_scalar): Likewise.
1814 * optabs.c (expand_binop): Likewise.
1815 (expand_twoval_binop_libfunc): Likewise.
1816 (expand_unop): Likewise.
1817 (prepare_cmp_insn): Likewise.
1818 (prepare_float_lib_cmp): Likewise.
1819 (expand_float): Likewise.
1820 (expand_fix): Likewise.
1821 (expand_fixed_convert): Likewise.
1822 (maybe_emit_sync_lock_test_and_set): Likewise.
1823 (expand_atomic_compare_and_swap): Likewise.
1824 (expand_mem_thread_fence): Likewise.
1825 (expand_atomic_fetch_op): Likewise.
1827 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1829 * doc/generic.texi (OpenACC): Adjust URL.
1830 * doc/invoke.texi (C Dialect Options): Ditto.
1832 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1834 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1835 predicate for operand 1. Add (m,<S>) constraint.
1836 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1837 Prevent memory operand 1 with register operand 2.
1839 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1841 PR rtl-optimization/82024
1842 * combine.c (try_combine): If the combination result is a PARALLEL,
1843 and we only need to retain the SET in there that would be placed
1844 at I2, check that we can place that at I3 instead, before doing so.
1846 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1849 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1851 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1852 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1855 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1856 Jeff Law <law@redhat.com>
1858 * varasm.c (bss_initializer_p): Do not put constants into .bss
1859 (categorize_decl_for_section): Handle bss_initializer_p returning
1860 false when DECL_INITIAL is NULL.
1862 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1865 * config/s390/s390.c (s390_can_inline_p): New function.
1867 2017-09-01 Jeff Law <law@redhat.com>
1869 PR tree-optimization/82052
1870 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1871 Always initialize the returned slot after a hash table miss
1872 when INSERT is true.
1874 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1876 * config/s390/s390.md (mem_signal_fence): Remove.
1877 * doc/md.texi (mem_signal_fence): Remove.
1878 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1880 * target-insns.def (mem_signal_fence): Remove.
1882 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1885 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1888 * asan.c (create_odr_indicator): Strip name encoding from assembler
1889 name before appending it after __odr_asan_.
1891 2017-09-01 Martin Liska <mliska@suse.cz>
1893 PR tree-optimization/82059
1894 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1895 frequency only when an edge is redirected.
1897 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1899 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1900 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1901 (arc_conditional_register_usage): Remove ARC600 lp_count
1903 (arc_file_start): Emit Tag_ARC_CPU_variation.
1904 (arc_can_use_doloop_p): New conditions to use ZOLs.
1905 (hwloop_fail): New function.
1906 (hwloop_optimize): Likewise.
1907 (hwloop_pattern_reg): Likewise.
1908 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1909 (arc_reorg_loops): New function, calls reorg_loops.
1910 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1911 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1913 (arc_loop_hazard): Remove function, functionality moved into
1915 (arc_hazard): Remove arc_loop_hazard call.
1916 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1917 into hwloop_optimize.
1918 (arc_label_align): Remove ZOL handling.
1919 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1920 * config/arc/arc.md (doloop_begin): Remove pattern.
1921 (doloop_begin_i): Likewise.
1922 (doloop_end_i): Likewise.
1923 (doloop_fallback): Likewise.
1924 (doloop_fallback_m): Likewise.
1925 (doloop_end): Reimplement expand.
1926 (arc_lp): New pattern for LP instruction.
1927 (loop_end): New pattern.
1928 (loop_fail): Likewise.
1929 (decrement_and_branch_until_zero): Likewise.
1930 * config/arc/arc.opt (mlpc-width): New option.
1931 * doc/invoke.texi (mlpc-width): Document option.
1933 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1935 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1936 (arc_ccfsm_advance): Fix checking for delay slots.
1937 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1939 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1941 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1943 (movhi_insn): Update store instruction constraint which are saving
1944 6-bit short immediates.
1945 (movsi_insn): Consider also short scaled load operations.
1946 (zero_extendhisi2_i): Use Usd constraint instead of T.
1947 (extendhisi2_i): Add q constraint.
1948 (arc_clzsi2): Add type and length attributes.
1949 (arc_ctzsi2): Likewise.
1950 * config/arc/constraints.md (Usc): Update constraint, the
1951 assembler can parse two relocations for a single instruction.
1953 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1955 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1956 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1958 2017-08-31 Olivier Hainque <hainque@adacore.com>
1960 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1961 match as powerpc-wrs-vxworks*.
1963 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1965 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1966 register constraint for by-element operand.
1967 (aarch64_mls_elt_merge<mode>): Likewise.
1969 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1971 * config/arc/arc.c (arc_can_follow_jump): Check for short
1974 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1976 * config.gcc: Use g.opt for arc.
1977 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1978 functionality moved to ...
1979 (legitimate_scaled_address_p): New function, ...here.
1980 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1981 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1982 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1984 (arc_override_options): Handle G option.
1985 (arc_output_pic_addr_const): Correct function definition.
1986 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1987 (arc_decl_anon_ns_mem_p): Delete.
1988 (arc_in_small_data_p): Overhaul this function to take into
1989 consideration the value given via G option.
1990 (arc_rewrite_small_data_1): Renamed and corrected old
1991 arc_rewrite_small_data function.
1992 (arc_rewrite_small_data): New function.
1993 (small_data_pattern): Don't use pic_offset_table_rtx.
1994 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1995 * config/arc/simdext.md (movmisalignv2hi): Use
1996 prepare_move_operands function.
1998 (movmisalign*): Likewise.
1999 * doc/invoke.texi (ARC options): Document -G option.
2001 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2003 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2005 * config/arc/arc.c (arc_print_operand): Output scalled address for
2006 sdata whenever is possible.
2007 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2008 load/stores are available.
2009 (compact_sda_memory_operand): Check for the alignment required by
2010 code density instructions.
2011 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2013 * config/arc/constraints.md (Usd): Update constraint.
2014 (Us0): New constraint.
2015 (Usc): Update constraint.
2017 2017-08-31 Richard Biener <rguenther@suse.de>
2020 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2023 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2025 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2026 Resize type_signature.
2028 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2029 Alan Hayward <alan.hayward@arm.com>
2030 David Sherwood <david.sherwood@arm.com>
2032 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2033 subregs whose inner modes can be stored in GPRs.
2034 (aarch64_classify_index): Likewise.
2036 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2037 Alan Hayward <alan.hayward@arm.com>
2038 David Sherwood <david.sherwood@arm.com>
2040 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2041 (V_INT_EQUIV): ...this.
2042 (v_cmp_result): Rename to...
2043 (v_int_equiv): ...this.
2044 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2045 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2046 (copysign<mode>3): Likewise.
2047 (aarch64_simd_bsl<mode>_internal): Likewise.
2048 (aarch64_simd_bsl<mode>): Likewise.
2049 (vec_cmp<mode><mode>): Likewise.
2050 (vcond<mode><mode>): Likewise.
2051 (vcond<v_cmp_mixed><mode>): Likewise.
2052 (vcondu<mode><v_cmp_mixed>): Likewise.
2053 (aarch64_cm<optab><mode>): Likewise.
2054 (aarch64_cmtst<mode>): Likewise.
2055 (aarch64_fac<optab><mode>): Likewise.
2056 (vec_perm_const<mode>): Likewise.
2057 (vcond_mask_<mode><v_cmp_result>): Rename to...
2058 (vcond_mask_<mode><v_int_equiv>): ...this.
2059 (vec_cmp<mode><v_cmp_result>): Rename to...
2060 (vec_cmp<mode><v_int_equiv>): ...this.
2062 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2063 Alan Hayward <alan.hayward@arm.com>
2064 David Sherwood <david.sherwood@arm.com>
2066 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2068 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2069 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2070 (UNSPEC_LD4_DREG): New unspecs.
2071 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2072 (aarch64_ld2<mode>_dreg_be): Replace with...
2073 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2075 (aarch64_ld3<mode>_dreg_le)
2076 (aarch64_ld3<mode>_dreg_be): Replace with...
2077 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2079 (aarch64_ld4<mode>_dreg_le)
2080 (aarch64_ld4<mode>_dreg_be): Replace with...
2081 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2084 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2086 PR tree-optimization/81987
2087 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2088 insert an initializer in a location not dominated by the stride
2091 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2093 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2094 on the entire header of the finally block in the fallthru case.
2096 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2098 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2100 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2102 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2103 rs6000_emit_move_from_cr and call renamed function.
2104 (rs6000_emit_prologue): Call renamed functions.
2105 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2106 movesi_from_cr, remove volatile CRs.
2108 2017-08-30 Jon Beniston <jon@beniston.com>
2109 Richard Biener <rguenther@suse.de>
2111 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2112 of VECTOR_MODE_P check.
2113 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2114 element vector types.
2116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2118 * df.h (df_read_modify_subreg_p): Remove in favor of...
2119 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2120 const_rtx instead of an rtx.
2121 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2122 * df-problems.c (df_word_lr_mark_ref): Likewise.
2123 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2124 (mark_pseudo_reg_dead): Likewise.
2125 (mark_ref_dead): Likewise.
2126 * reginfo.c (init_subregs_of_mode): Likewise.
2127 * sched-deps.c (sched_analyze_1): Likewise.
2128 * df-scan.c (df_def_record_1): Likewise.
2129 (df_uses_record): Likewise.
2130 (df_read_modify_subreg_p): Remove in favor of...
2131 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2132 const_rtx instead of an rtx.
2134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2135 Alan Hayward <alan.hayward@arm.com>
2136 David Sherwood <david.sherwood@arm.com>
2138 * rtl.h (partial_subreg_p): New function.
2139 * caller-save.c (save_call_clobbered_regs): Use it.
2140 * calls.c (expand_call): Likewise.
2141 * combine.c (combinable_i3pat): Likewise.
2142 (simplify_set): Likewise.
2143 (make_extraction): Likewise.
2144 (make_compound_operation_int): Likewise.
2145 (gen_lowpart_or_truncate): Likewise.
2146 (force_to_mode): Likewise.
2147 (make_field_assignment): Likewise.
2148 (reg_truncated_to_mode): Likewise.
2149 (record_truncated_value): Likewise.
2150 (move_deaths): Likewise.
2151 * cse.c (record_jump_cond): Likewise.
2152 (cse_insn): Likewise.
2153 * cselib.c (cselib_lookup_1): Likewise.
2154 * expmed.c (extract_bit_field_using_extv): Likewise.
2155 * function.c (assign_parm_setup_reg): Likewise.
2156 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2157 * ira-build.c (create_insn_allocnos): Likewise.
2158 * lra-coalesce.c (merge_pseudos): Likewise.
2159 * lra-constraints.c (match_reload): Likewise.
2160 (simplify_operand_subreg): Likewise.
2161 (curr_insn_transform): Likewise.
2162 * lra-lives.c (process_bb_lives): Likewise.
2163 * lra.c (new_insn_reg): Likewise.
2164 (lra_substitute_pseudo): Likewise.
2165 * regcprop.c (mode_change_ok): Likewise.
2166 (maybe_mode_change): Likewise.
2167 (copyprop_hardreg_forward_1): Likewise.
2168 * reload.c (push_reload): Likewise.
2169 (find_reloads): Likewise.
2170 (find_reloads_subreg_address): Likewise.
2171 * reload1.c (alter_reg): Likewise.
2172 (eliminate_regs_1): Likewise.
2173 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2175 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2177 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2178 back to if statements, including unpack.
2180 2017-08-30 Martin Liska <mliska@suse.cz>
2183 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2185 (func_checker::compare_asm_inputs_outputs): ... this function.
2186 (func_checker::compare_gimple_asm): Use the function to compare
2187 also ASM constrains.
2188 * ipa-icf-gimple.h: Rename the function.
2190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2191 Alan Hayward <alan.hayward@arm.com>
2192 David Sherwood <david.sherwood@arm.com>
2194 * coretypes.h (complex_mode): New type.
2195 * gdbhooks.py (build_pretty_printer): Handle it.
2196 * machmode.h (complex_mode): New class.
2197 (complex_mode::includes_p): New function.
2198 (is_complex_int_mode): Likewise.
2199 (is_complex_float_mode): Likewise.
2200 * genmodes.c (get_mode_class): Handle complex mode classes.
2201 * function.c (expand_function_end): Use is_complex_int_mode.
2203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2204 Alan Hayward <alan.hayward@arm.com>
2205 David Sherwood <david.sherwood@arm.com>
2207 * coretypes.h (scalar_mode_pod): New typedef.
2208 * gdbhooks.py (build_pretty_printer): Handle it.
2209 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2210 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2211 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2212 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
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 * machmode.h (mode_for_vector): Take a scalar_mode instead
2221 * stor-layout.c (mode_for_vector): Likewise.
2222 * explow.c (promote_mode): Use as_a <scalar_mode>.
2223 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2226 Alan Hayward <alan.hayward@arm.com>
2227 David Sherwood <david.sherwood@arm.com>
2229 * target.def (preferred_simd_mode): Take a scalar_mode
2230 instead of a machine_mode.
2231 * targhooks.h (default_preferred_simd_mode): Likewise.
2232 * targhooks.c (default_preferred_simd_mode): Likewise.
2233 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2234 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2235 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2236 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2237 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2238 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2239 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2240 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2242 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2243 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2244 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2245 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2246 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2247 * doc/tm.texi: Regenerate.
2248 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2252 Alan Hayward <alan.hayward@arm.com>
2253 David Sherwood <david.sherwood@arm.com>
2255 * target.def (scalar_mode_supported_p): Take a scalar_mode
2256 instead of a machine_mode.
2257 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2258 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2259 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2260 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2261 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2262 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2263 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2264 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2265 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2266 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2267 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2268 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2269 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2270 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2272 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2273 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2274 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2275 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2276 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2278 * doc/tm.texi: Regenerate.
2280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2281 Alan Hayward <alan.hayward@arm.com>
2282 David Sherwood <david.sherwood@arm.com>
2284 * coretypes.h (opt_scalar_mode): New typedef.
2285 * gdbhooks.py (build_pretty_printers): Handle it.
2286 * machmode.h (mode_iterator::get_2xwider): Add overload for
2288 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2290 * expr.c (convert_mode_scalar): Likewise.
2291 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2292 * optabs.c (expand_float): Likewise.
2293 (expand_fix): Likewise.
2294 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2297 Alan Hayward <alan.hayward@arm.com>
2298 David Sherwood <david.sherwood@arm.com>
2300 * optabs.c (expand_float): Explicitly check for scalars before
2301 using a branching expansion.
2302 (expand_fix): Likewise.
2304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2305 Alan Hayward <alan.hayward@arm.com>
2306 David Sherwood <david.sherwood@arm.com>
2308 * expr.c (convert_mode): Split scalar handling out into...
2309 (convert_mode_scalar): ...this new function. Treat the modes
2312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2313 Alan Hayward <alan.hayward@arm.com>
2314 David Sherwood <david.sherwood@arm.com>
2316 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2318 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2321 Alan Hayward <alan.hayward@arm.com>
2322 David Sherwood <david.sherwood@arm.com>
2324 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2325 rather than a machine_mode.
2326 (fixed_from_string): Likewise.
2327 (fixed_convert): Likewise.
2328 (fixed_convert_from_int): Likewise.
2329 (fixed_convert_from_real): Likewise.
2330 (real_convert_from_fixed): Likewise.
2331 * fixed-value.c (fixed_from_double_int): Likewise.
2332 (fixed_from_string): Likewise.
2333 (fixed_convert): Likewise.
2334 (fixed_convert_from_int): Likewise.
2335 (fixed_convert_from_real): Likewise.
2336 (real_convert_from_fixed): Likewise.
2337 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2340 Alan Hayward <alan.hayward@arm.com>
2341 David Sherwood <david.sherwood@arm.com>
2343 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2344 of separate mode class checks. Do not allow vector modes here.
2345 (immed_wide_int_const): Use as_a <scalar_mode>.
2346 * explow.c (trunc_int_for_mode): Likewise.
2347 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2348 (wi::shwi): Likewise.
2349 (wi::min_value): Likewise.
2350 (wi::max_value): Likewise.
2351 * dwarf2out.c (loc_descriptor): Likewise.
2352 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2356 Alan Hayward <alan.hayward@arm.com>
2357 David Sherwood <david.sherwood@arm.com>
2359 * tree.h (SCALAR_TYPE_MODE): New macro.
2360 * expr.c (expand_expr_addr_expr_1): Use it.
2361 (expand_expr_real_2): Likewise.
2362 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2363 (fold_convert_const_fixed_from_int): Likewise.
2364 (fold_convert_const_fixed_from_real): Likewise.
2365 (native_encode_fixed): Likewise
2366 (native_encode_complex): Likewise
2367 (native_encode_vector): Likewise.
2368 (native_interpret_fixed): Likewise.
2369 (native_interpret_real): Likewise.
2370 (native_interpret_complex): Likewise.
2371 (native_interpret_vector): Likewise.
2372 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2373 (simd_clone_adjust_argument_types): Likewise.
2374 (simd_clone_init_simd_arrays): Likewise.
2375 (simd_clone_adjust): Likewise.
2376 * stor-layout.c (layout_type): Likewise.
2377 * tree.c (build_minus_one_cst): Likewise.
2378 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2379 * tree-inline.c (estimate_move_cost): Likewise.
2380 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2381 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2382 (vectorizable_reduction): Likewise.
2383 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2384 (vect_recog_mixed_size_cond_pattern): Likewise.
2385 (check_bool_pattern): Likewise.
2386 (adjust_bool_pattern): Likewise.
2387 (search_type_for_mask_1): Likewise.
2388 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2389 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2390 (vectorizable_load): Likewise.
2391 (vectorizable_store): Likewise.
2392 * ubsan.c (ubsan_encode_value): Likewise.
2393 * varasm.c (output_constant): Likewise.
2395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2396 Alan Hayward <alan.hayward@arm.com>
2397 David Sherwood <david.sherwood@arm.com>
2399 * coretypes.h (scalar_mode): New class.
2400 * machmode.h (scalar_mode): Likewise.
2401 (scalar_mode::includes_p): New function.
2402 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2403 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2404 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2405 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2406 * expmed.c (store_bit_field_1): Likewise.
2407 (extract_bit_field_1): Likewise.
2408 * expr.c (write_complex_part): Likewise.
2409 (read_complex_part): Likewise.
2410 (emit_move_complex_push): Likewise.
2411 (expand_expr_real_2): Likewise.
2412 * function.c (assign_parm_setup_reg): Likewise.
2413 (assign_parms_unsplit_complex): Likewise.
2414 * optabs.c (expand_binop): Likewise.
2415 * rtlanal.c (subreg_get_info): Likewise.
2416 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2417 * varasm.c (output_constant_pool_2): Likewise.
2419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2420 Alan Hayward <alan.hayward@arm.com>
2421 David Sherwood <david.sherwood@arm.com>
2423 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2425 (expmed_mult_highpart_optab): Likewise.
2426 (expmed_mult_highpart): Likewise.
2428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2429 Alan Hayward <alan.hayward@arm.com>
2430 David Sherwood <david.sherwood@arm.com>
2432 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2433 instead of a machine_mode.
2434 (builtin_memset_read_str): Likewise.
2435 * builtins.c (c_readstr): Likewise.
2436 (builtin_memcpy_read_str): Likewise.
2437 (builtin_strncpy_read_str): Likewise.
2438 (builtin_memset_read_str): Likewise.
2439 (builtin_memset_gen_str): Likewise.
2440 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2441 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2442 instead of a machine_mode.
2443 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2445 (make_extraction): Likewise.
2446 (try_widen_shift_mode): Take and return scalar_int_modes instead
2448 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2449 a scalar_int_mode instead of a machine_mode.
2450 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2451 (avr_addr_space_pointer_mode): Likewise.
2452 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2453 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2454 (msp430_unwind_word_mode): Likewise.
2455 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2456 (spu_addr_space_pointer_mode): Likewise.
2457 (spu_addr_space_address_mode): Likewise.
2458 (spu_libgcc_cmp_return_mode): Likewise.
2459 (spu_libgcc_shift_count_mode): Likewise.
2460 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2461 (rl78_addr_space_pointer_mode): Likewise.
2462 (fl78_unwind_word_mode): Likewise.
2463 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2465 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2466 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2467 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2468 (mips_valid_pointer_mode): Likewise.
2469 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2470 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2471 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2473 (ft32_addr_space_address_mode): Likewise.
2474 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2475 scalar_int_mode instead of a machine_mode.
2476 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2478 (m32c_addr_space_address_mode): Likewise.
2479 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2480 (rs6000_eh_return_filter_mode): Likewise.
2481 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2482 (rs6000_eh_return_filter_mode): Likewise.
2483 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2484 (s390_libgcc_shift_count_mode): Likewise.
2485 (s390_unwind_word_mode): Likewise.
2486 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2488 * target.def (mode_rep_extended): Likewise.
2489 (valid_pointer_mode): Likewise.
2490 (addr_space.valid_pointer_mode): Likewise.
2491 (eh_return_filter_mode): Return a scalar_int_mode rather than
2493 (libgcc_cmp_return_mode): Likewise.
2494 (libgcc_shift_count_mode): Likewise.
2495 (unwind_word_mode): Likewise.
2496 (addr_space.pointer_mode): Likewise.
2497 (addr_space.address_mode): Likewise.
2498 * doc/tm.texi: Regenerate.
2499 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2501 (do_jump): Use scalar_int_mode for local variables.
2502 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2503 rather than a machine_mode.
2504 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2505 (scompare_loc_descriptor_wide): Likewise.
2506 (scompare_loc_descriptor_narrow): Likewise.
2507 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2509 * except.c (sjlj_emit_dispatch_table): Likewise.
2510 (expand_builtin_eh_copy_values): Likewise.
2511 * explow.c (convert_memory_address_addr_space_1): Likewise.
2512 Take a scalar_int_mode rather than a machine_mode.
2513 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2514 than a machine_mode.
2515 (memory_address_addr_space): Use scalar_int_mode for local variables.
2516 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2517 rather than a machine_mode.
2518 * expmed.c (mask_rtx): Likewise.
2519 (init_expmed_one_conv): Likewise.
2520 (expand_mult_highpart_adjust): Likewise.
2521 (extract_high_half): Likewise.
2522 (expmed_mult_highpart_optab): Likewise.
2523 (expmed_mult_highpart): Likewise.
2524 (expand_smod_pow2): Likewise.
2525 (expand_sdiv_pow2): Likewise.
2526 (emit_store_flag_int): Likewise.
2527 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2529 (extract_low_bits): Likewise.
2530 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2532 * expr.c (pieces_addr::adjust): Likewise.
2533 (can_store_by_pieces): Likewise.
2534 (store_by_pieces): Likewise.
2535 (clear_by_pieces_1): Likewise.
2536 (expand_expr_addr_expr_1): Likewise.
2537 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2538 (expand_expr_real_1): Likewise.
2539 (try_casesi): Likewise.
2540 * final.c (shorten_branches): Likewise.
2541 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2542 type of "mode" to machine_mode.
2543 * internal-fn.c (expand_arith_overflow_result_store): Take a
2544 scalar_int_mode rather than a machine_mode.
2545 (expand_mul_overflow): Use scalar_int_mode for local variables.
2546 * loop-doloop.c (doloop_modify): Likewise.
2547 (doloop_optimize): Likewise.
2548 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2549 than a machine_mode.
2550 (expand_doubleword_shift_condmove): Likewise.
2551 (expand_doubleword_shift): Likewise.
2552 (expand_doubleword_clz): Likewise.
2553 (expand_doubleword_popcount): Likewise.
2554 (expand_doubleword_parity): Likewise.
2555 (expand_absneg_bit): Use scalar_int_mode for local variables.
2556 (prepare_float_lib_cmp): Likewise.
2557 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2558 rather than a machine_mode.
2559 (convert_memory_address_addr_space): Likewise.
2560 (get_mode_bounds): Likewise.
2561 (get_address_mode): Return a scalar_int_mode rather than a
2563 * rtlanal.c (get_address_mode): Likewise.
2564 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2565 than a machine_mode.
2566 * targhooks.c (default_mode_rep_extended): Likewise.
2567 (default_valid_pointer_mode): Likewise.
2568 (default_addr_space_valid_pointer_mode): Likewise.
2569 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2570 than a machine_mode.
2571 (default_libgcc_cmp_return_mode): Likewise.
2572 (default_libgcc_shift_count_mode): Likewise.
2573 (default_unwind_word_mode): Likewise.
2574 (default_addr_space_pointer_mode): Likewise.
2575 (default_addr_space_address_mode): Likewise.
2576 * targhooks.h (default_eh_return_filter_mode): Likewise.
2577 (default_libgcc_cmp_return_mode): Likewise.
2578 (default_libgcc_shift_count_mode): Likewise.
2579 (default_unwind_word_mode): Likewise.
2580 (default_addr_space_pointer_mode): Likewise.
2581 (default_addr_space_address_mode): Likewise.
2582 (default_mode_rep_extended): Take a scalar_int_mode rather than
2584 (default_valid_pointer_mode): Likewise.
2585 (default_addr_space_valid_pointer_mode): Likewise.
2586 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2588 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2589 rather than a machine_mode.
2590 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2591 for local variables.
2592 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2593 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2594 than a machine_mode.
2596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2597 Alan Hayward <alan.hayward@arm.com>
2598 David Sherwood <david.sherwood@arm.com>
2600 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2601 the mode argument to scalar_int_mode.
2602 (do_jump_by_parts_zero_rtx): Likewise.
2603 (do_jump_by_parts_equality_rtx): Likewise.
2604 (do_jump_by_parts_greater): Take a mode argument.
2605 (do_jump_by_parts_equality): Likewise.
2606 (do_jump_1): Update calls accordingly.
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 * is-a.h (safe_dyn_cast): New function.
2613 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2614 (jump_table_for_label): Likewise.
2615 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2616 instead of an rtx_insn *.
2617 (shorten_branches): Use dyn_cast instead of LABEL_P and
2618 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2619 rtx_jump_table_data::get_data_mode.
2620 (final_scan_insn): Likewise.
2622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2623 Alan Hayward <alan.hayward@arm.com>
2624 David Sherwood <david.sherwood@arm.com>
2626 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2627 trying to combine a full-register integer set with a subreg
2630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2631 Alan Hayward <alan.hayward@arm.com>
2632 David Sherwood <david.sherwood@arm.com>
2634 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2635 that is always either address_mode or pointer_mode.
2637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2638 Alan Hayward <alan.hayward@arm.com>
2639 David Sherwood <david.sherwood@arm.com>
2641 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2642 when the two are known to be equal.
2644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2645 Alan Hayward <alan.hayward@arm.com>
2646 David Sherwood <david.sherwood@arm.com>
2648 * simplify-rtx.c (simplify_const_unary_operation): Use
2649 is_a <scalar_int_mode> instead of checking for a nonzero
2650 precision. Forcibly convert op_mode to a scalar_int_mode
2651 in that case. More clearly differentiate the operand and
2652 result modes and use the former when deciding what the value
2653 of a count-bits operation should be. Use is_int_mode instead
2654 of checking for a MODE_INT. Remove redundant check for whether
2655 this mode has a zero precision.
2657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2658 Alan Hayward <alan.hayward@arm.com>
2659 David Sherwood <david.sherwood@arm.com>
2661 * optabs.c (widen_leading): Change the type of the mode argument
2662 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2663 (widen_bswap): Likewise.
2664 (expand_parity): Likewise.
2665 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2666 (expand_ffs): Likewise.
2667 (epand_unop): Check for scalar integer modes before calling the
2670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2671 Alan Hayward <alan.hayward@arm.com>
2672 David Sherwood <david.sherwood@arm.com>
2674 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2676 (expand_expr_real_1): Update call accordingly.
2678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2679 Alan Hayward <alan.hayward@arm.com>
2680 David Sherwood <david.sherwood@arm.com>
2682 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2683 value_mode arguments. Use scalar_int_mode internally.
2684 (store_bit_field_1): Rename the new integer mode from imode
2685 to op0_mode and use it instead of GET_MODE (op0). Update calls
2686 to store_split_bit_field, store_bit_field_using_insv and
2687 store_fixed_bit_field.
2688 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2689 Use scalar_int_mode internally. Use a bit count rather than a mode
2690 when calculating the largest bit size for get_best_mode.
2691 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2692 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2693 Remove assertion that OP0 has a scalar integer mode.
2694 (store_split_bit_field): Add op0_mode and value_mode arguments.
2695 Update calls to extract_fixed_bit_field.
2696 (extract_bit_field_using_extv): Add an op0_mode argument.
2697 Use scalar_int_mode internally.
2698 (extract_bit_field_1): Rename the new integer mode from imode to
2699 op0_mode and use it instead of GET_MODE (op0). Update calls to
2700 extract_split_bit_field, extract_bit_field_using_extv and
2701 extract_fixed_bit_field.
2702 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2703 to extract_split_bit_field and extract_fixed_bit_field_1.
2704 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2705 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2707 (extract_split_bit_field): Add an op0_mode argument. Update call
2708 to extract_fixed_bit_field.
2710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2711 Alan Hayward <alan.hayward@arm.com>
2712 David Sherwood <david.sherwood@arm.com>
2714 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2715 * explow.c (hard_function_value): Likewise.
2716 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2717 convert_to_mode call outside the loop.
2718 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2719 for the mode iterator. Require the mode specified by max_pieces
2721 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2723 (copy_blkmode_to_reg): Likewise.
2724 (set_storage_via_setmem): Likewise.
2725 * optabs.c (prepare_cmp_insn): Likewise.
2726 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2727 * stor-layout.c (finish_bitfield_representative): Likewise.
2729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2730 Alan Hayward <alan.hayward@arm.com>
2731 David Sherwood <david.sherwood@arm.com>
2733 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2734 * expr.c (convert_move): Use them.
2735 (convert_modes): Likewise.
2736 (store_expr_with_bounds): Likewise.
2738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2739 Alan Hayward <alan.hayward@arm.com>
2740 David Sherwood <david.sherwood@arm.com>
2742 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2743 parameter for the mode of "x". Remove the "known_x", "known_mode"
2744 and "known_ret" arguments. Change the type of the mode argument
2746 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2747 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2748 (reg_num_sign_bit_copies_for_combine): Likewise.
2749 * rtlanal.c (nonzero_bits1): Likewise.
2750 (num_sign_bit_copies1): Likewise.
2751 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2752 (reg_num_sign_bit_copies_general): Likewise.
2753 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2754 (reg_nonzero_bits_general): Likewise.
2756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2757 Alan Hayward <alan.hayward@arm.com>
2758 David Sherwood <david.sherwood@arm.com>
2760 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2761 than in subroutines. Return 1 for non-integer modes.
2762 (cached_num_sign_bit_copies): Change the type of the mode parameter
2764 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2765 classes. Handle CONST_INT_P first and then check whether X also
2766 has a scalar integer mode. Check the same thing for inner registers
2767 of a SUBREG and for values that are being extended or truncated.
2769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2770 Alan Hayward <alan.hayward@arm.com>
2771 David Sherwood <david.sherwood@arm.com>
2773 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2774 in subroutines. Return the mode mask for non-integer modes.
2775 (cached_nonzero_bits): Change the type of the mode parameter
2777 (nonzero_bits1): Likewise. Remove early exit for other mode
2778 classes. Handle CONST_INT_P first and then check whether X
2779 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2786 Return a scalar_int_mode and assert that the size is greater than
2788 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2789 (op_by_pieces_d::op_by_pieces_d): Likewise.
2790 (op_by_pieces_d::run): Likewise.
2791 (can_store_by_pieces): Likewise.
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 * combine.c (extract_left_shift): Add a mode argument and update
2799 (make_compound_operation_int): Change the type of the mode parameter
2800 to scalar_int_mode and update the call to extract_left_shift.
2802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2803 Alan Hayward <alan.hayward@arm.com>
2804 David Sherwood <david.sherwood@arm.com>
2806 * combine.c (simplify_and_const_int): Change the type of the mode
2807 parameter to scalar_int_mode.
2808 (simplify_and_const_int_1): Likewise. Update recursive call.
2810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2811 Alan Hayward <alan.hayward@arm.com>
2812 David Sherwood <david.sherwood@arm.com>
2814 * combine.c (simplify_compare_const): Check that the mode is a
2815 scalar_int_mode (rather than VOIDmode) before testing its
2817 (simplify_comparison): Move COMPARISON_P handling out of the
2818 loop and restrict the latter part of the loop to scalar_int_modes.
2819 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2820 and when considering SUBREG_REGs. Use is_int_mode instead of
2821 checking GET_MODE_CLASS against MODE_INT.
2823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2824 Alan Hayward <alan.hayward@arm.com>
2825 David Sherwood <david.sherwood@arm.com>
2827 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2828 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2829 shift_unit_mode and for modes involved in scalar shifts.
2831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2832 Alan Hayward <alan.hayward@arm.com>
2833 David Sherwood <david.sherwood@arm.com>
2835 * combine.c (force_int_to_mode): New function, split out from...
2836 (force_to_mode): ...here. Keep xmode up-to-date and use it
2837 instead of GET_MODE (x).
2839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2840 Alan Hayward <alan.hayward@arm.com>
2841 David Sherwood <david.sherwood@arm.com>
2843 * optabs-query.h (extraction_insn::struct_mode): Change type to
2844 opt_scalar_int_mode and update comment.
2845 (extraction_insn::field_mode): Change type to scalar_int_mode.
2846 (extraction_insn::pos_mode): Likewise.
2847 * combine.c (make_extraction): Update accordingly.
2848 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2849 (get_optab_extraction_insn): Likewise.
2850 * recog.c (simplify_while_replacing): Likewise.
2851 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2852 parameter to opt_scalar_int_mode.
2854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2855 Alan Hayward <alan.hayward@arm.com>
2856 David Sherwood <david.sherwood@arm.com>
2858 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2859 to a scalar_int_mode instead of a machine_mode.
2860 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2861 (get_best_mode): Return a boolean and use a pointer argument to store
2862 the selected mode. Replace the limit mode parameter with a bit limit.
2863 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2864 for the values returned by bit_field_mode_iterator::next_mode.
2865 (store_bit_field): Update call to get_best_mode.
2866 (store_fixed_bit_field): Likewise.
2867 (extract_fixed_bit_field): Likewise.
2868 * expr.c (optimize_bitfield_assignment_op): Likewise.
2869 * fold-const.c (optimize_bit_field_compare): Likewise.
2870 (fold_truth_andor_1): Likewise.
2871 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2872 Update for new type of m_mode.
2873 (get_best_mode): As above.
2875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2876 Alan Hayward <alan.hayward@arm.com>
2877 David Sherwood <david.sherwood@arm.com>
2879 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2880 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2881 (store_bit_field): Check is_a <scalar_int_mode> before calling
2882 strict_volatile_bitfield_p.
2883 (extract_bit_field): Likewise.
2885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2886 Alan Hayward <alan.hayward@arm.com>
2887 David Sherwood <david.sherwood@arm.com>
2889 * target.def (cstore_mode): Return a scalar_int_mode.
2890 * doc/tm.texi: Regenerate.
2891 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2892 * targhooks.h (default_cstore_mode): Likewise.
2893 * targhooks.c (default_cstore_mode): Likewise, using a forced
2895 * expmed.c (emit_cstore): Expect the target of the cstore to be
2898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2899 Alan Hayward <alan.hayward@arm.com>
2900 David Sherwood <david.sherwood@arm.com>
2902 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2904 (niter_desc): Likewise mode.
2905 (iv_analyze): Add a mode parameter.
2907 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2908 and change its type to scalar_int_mode.
2909 * loop-iv.c: Update commentary at head of file.
2910 (iv_constant): Pass the mode paraeter before the rtx it describes
2911 and change its type to scalar_int_mode. Remove VOIDmode handling.
2912 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2913 (iv_extend): Likewise.
2914 (shorten_into_mode): Likewise.
2915 (iv_add): Use scalar_int_mode.
2916 (iv_mult): Likewise.
2917 (iv_shift): Likewise.
2918 (canonicalize_iv_subregs): Likewise.
2919 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2920 it describes and change its mode to scalar_int_mode. Also change
2921 the type of the returned inner_mode to scalar_int_mode.
2922 (get_biv_step): Likewise, turning outer_mode from a pointer
2923 into a direct parameter. Update call to get_biv_step_1.
2924 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2925 iv_constant and get_biv_step.
2926 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2927 and change its type to scalar_int_mode. Don't initialise iv->mode
2928 to VOIDmode and remove later checks for its still being VOIDmode.
2929 Update calls to iv_analyze_op and iv_analyze_expr. Check
2930 is_a <scalar_int_mode> when changing the mode under consideration.
2931 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2932 Update call to iv_analyze_expr.
2933 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2934 inner register is not also a scalar_int_mode. Update call to
2936 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2937 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2938 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2939 separate mode class checks. Update calls to iv_analyze. Remove
2940 fix-up of VOIDmodes after iv_analyze_biv.
2941 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2942 don't have a scalar_int_mode. Update call to biv_p.
2944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2945 Alan Hayward <alan.hayward@arm.com>
2946 David Sherwood <david.sherwood@arm.com>
2948 * cfgexpand.c (convert_debug_memory_address): Use
2949 as_a <scalar_int_mode>.
2950 * combine.c (expand_compound_operation): Likewise.
2951 (make_extraction): Likewise.
2952 (change_zero_ext): Likewise.
2953 (simplify_comparison): Likewise.
2954 * cse.c (cse_insn): Likewise.
2955 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2956 (mem_loc_descriptor): Likewise.
2957 (loc_descriptor): Likewise.
2958 * expmed.c (init_expmed_one_mode): Likewise.
2959 (synth_mult): Likewise.
2960 (emit_store_flag_1): Likewise.
2961 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2962 of a comparison with size.
2963 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2964 (reduce_to_bit_field_precision): Likewise.
2965 * function.c (expand_function_end): Likewise.
2966 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2967 * loop-doloop.c (doloop_modify): Likewise.
2968 * optabs.c (expand_binop): Likewise.
2969 (expand_unop): Likewise.
2970 (expand_copysign_absneg): Likewise.
2971 (prepare_cmp_insn): Likewise.
2972 (maybe_legitimize_operand): Likewise.
2973 * recog.c (const_scalar_int_operand): Likewise.
2974 * rtlanal.c (get_address_mode): Likewise.
2975 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2976 (simplify_cond_clz_ctz): Likewise.
2977 * tree-nested.c (get_nl_goto_field): Likewise.
2978 * tree.c (build_vector_type_for_mode): Likewise.
2979 * var-tracking.c (use_narrower_mode): Likewise.
2981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2982 Alan Hayward <alan.hayward@arm.com>
2983 David Sherwood <david.sherwood@arm.com>
2985 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2986 * builtins.c (expand_builtin_signbit): Use it.
2987 * cfgexpand.c (expand_debug_expr): Likewise.
2988 * dojump.c (do_jump): Likewise.
2989 (do_compare_and_jump): Likewise.
2990 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2991 * expmed.c (make_tree): Likewise.
2992 * expr.c (expand_expr_real_2): Likewise.
2993 (expand_expr_real_1): Likewise.
2994 (try_casesi): Likewise.
2995 * fold-const-call.c (fold_const_call_ss): Likewise.
2996 * fold-const.c (unextend): Likewise.
2997 (extract_muldiv_1): Likewise.
2998 (fold_single_bit_test): Likewise.
2999 (native_encode_int): Likewise.
3000 (native_encode_string): Likewise.
3001 (native_interpret_int): Likewise.
3002 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3003 * internal-fn.c (expand_addsub_overflow): Likewise.
3004 (expand_neg_overflow): Likewise.
3005 (expand_mul_overflow): Likewise.
3006 (expand_arith_overflow): Likewise.
3007 * match.pd: Likewise.
3008 * stor-layout.c (layout_type): Likewise.
3009 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3010 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3011 * tree-ssanames.c (get_range_info): Likewise.
3012 * tree-switch-conversion.c (array_value_type) Likewise.
3013 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3014 (vect_recog_divmod_pattern): Likewise.
3015 (vect_recog_mixed_size_cond_pattern): Likewise.
3016 * tree-vrp.c (extract_range_basic): Likewise.
3017 (simplify_float_conversion_using_ranges): Likewise.
3018 * tree.c (int_fits_type_p): Likewise.
3019 * ubsan.c (instrument_bool_enum_load): Likewise.
3020 * varasm.c (mergeable_string_section): Likewise.
3021 (narrowing_initializer_constant_valid_p): Likewise.
3022 (output_constant): Likewise.
3024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3025 Alan Hayward <alan.hayward@arm.com>
3026 David Sherwood <david.sherwood@arm.com>
3028 * machmode.h (NARROWEST_INT_MODE): New macro.
3029 * expr.c (alignment_for_piecewise_move): Use it instead of
3030 GET_CLASS_NARROWEST_MODE (MODE_INT).
3031 (push_block): Likewise.
3032 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3034 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * postreload.c (move2add_valid_value_p): Change the type of the
3041 mode parameter to scalar_int_mode.
3042 (move2add_use_add2_insn): Add a mode parameter and use it instead
3044 (move2add_use_add3_insn): Likewise.
3045 (reload_cse_move2add): Update accordingly.
3047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3048 Alan Hayward <alan.hayward@arm.com>
3049 David Sherwood <david.sherwood@arm.com>
3051 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3053 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3054 * optabs.c (expand_unop): Likewise.
3056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3057 Alan Hayward <alan.hayward@arm.com>
3058 David Sherwood <david.sherwood@arm.com>
3060 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3061 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3062 (popcount_loc_descriptor): Likewise.
3063 (bswap_loc_descriptor): Likewise.
3064 (rotate_loc_descriptor): Likewise.
3065 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3066 calling the functions above.
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 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3074 (try_combine): Likewise.
3075 (simplify_if_then_else): Likewise.
3076 * cse.c (cse_insn): Likewise.
3077 * dwarf2out.c (mem_loc_descriptor): Likewise.
3078 * emit-rtl.c (gen_lowpart_common): Likewise.
3079 * simplify-rtx.c (simplify_truncation): Likewise.
3080 (simplify_binary_operation_1): Likewise.
3081 (simplify_const_relational_operation): Likewise.
3082 (simplify_ternary_operation): Likewise.
3083 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3086 Alan Hayward <alan.hayward@arm.com>
3087 David Sherwood <david.sherwood@arm.com>
3089 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3090 * reload.c (push_reload): Likewise.
3091 (find_reloads): Likewise.
3093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3094 Alan Hayward <alan.hayward@arm.com>
3095 David Sherwood <david.sherwood@arm.com>
3097 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3098 (make_compound_operation_int): Likewise.
3099 (change_zero_ext): Likewise.
3100 * expr.c (convert_move): Likewise.
3101 (convert_modes): Likewise.
3102 * fwprop.c (forward_propagate_subreg): Likewise.
3103 * loop-iv.c (get_biv_step_1): Likewise.
3104 * optabs.c (widen_operand): Likewise.
3105 * postreload.c (move2add_valid_value_p): Likewise.
3106 * recog.c (simplify_while_replacing): Likewise.
3107 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3108 (simplify_binary_operation_1): Likewise. Remove redundant
3109 mode equality check.
3111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3112 Alan Hayward <alan.hayward@arm.com>
3113 David Sherwood <david.sherwood@arm.com>
3115 * combine.c (combine_simplify_rtx): Add checks for
3116 is_a <scalar_int_mode>.
3117 (simplify_if_then_else): Likewise.
3118 (make_field_assignment): Likewise.
3119 (simplify_comparison): Likewise.
3120 * ifcvt.c (noce_try_bitop): Likewise.
3121 * loop-invariant.c (canonicalize_address_mult): Likewise.
3122 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3125 Alan Hayward <alan.hayward@arm.com>
3126 David Sherwood <david.sherwood@arm.com>
3128 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3129 is_a <scalar_int_mode> instead of != BLKmode.
3131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3132 Alan Hayward <alan.hayward@arm.com>
3133 David Sherwood <david.sherwood@arm.com>
3135 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3136 instead of != VOIDmode.
3137 * combine.c (if_then_else_cond): Likewise.
3138 (change_zero_ext): Likewise.
3139 * dwarf2out.c (mem_loc_descriptor): Likewise.
3140 (loc_descriptor): Likewise.
3141 * rtlanal.c (canonicalize_condition): Likewise.
3142 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3145 Alan Hayward <alan.hayward@arm.com>
3146 David Sherwood <david.sherwood@arm.com>
3148 * simplify-rtx.c (simplify_binary_operation_1): Use
3149 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3152 Alan Hayward <alan.hayward@arm.com>
3153 David Sherwood <david.sherwood@arm.com>
3155 * wide-int.h (int_traits<unsigned char>) New class.
3156 (int_traits<unsigned short>) Likewise.
3157 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3158 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3160 * combine.c (set_nonzero_bits_and_sign_copies): Use
3161 is_a <scalar_int_mode>.
3162 (find_split_point): Likewise.
3163 (combine_simplify_rtx): Likewise.
3164 (simplify_logical): Likewise.
3165 (expand_compound_operation): Likewise.
3166 (expand_field_assignment): Likewise.
3167 (make_compound_operation): Likewise.
3168 (extended_count): Likewise.
3169 (change_zero_ext): Likewise.
3170 (simplify_comparison): Likewise.
3171 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3172 (ucompare_loc_descriptor): Likewise.
3173 (minmax_loc_descriptor): Likewise.
3174 (mem_loc_descriptor): Likewise.
3175 (loc_descriptor): Likewise.
3176 * expmed.c (init_expmed_one_mode): Likewise.
3177 * lra-constraints.c (lra_constraint_offset): Likewise.
3178 * optabs.c (prepare_libcall_arg): Likewise.
3179 * postreload.c (move2add_note_store): Likewise.
3180 * reload.c (operands_match_p): Likewise.
3181 * rtl.h (load_extend_op): Likewise.
3182 * rtlhooks.c (gen_lowpart_general): Likewise.
3183 * simplify-rtx.c (simplify_truncation): Likewise.
3184 (simplify_unary_operation_1): Likewise.
3185 (simplify_binary_operation_1): Likewise.
3186 (simplify_const_binary_operation): Likewise.
3187 (simplify_const_relational_operation): Likewise.
3188 (simplify_subreg): Likewise.
3189 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3190 * var-tracking.c (adjust_mems): Likewise.
3191 (prepare_call_arguments): Likewise.
3193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3194 Alan Hayward <alan.hayward@arm.com>
3195 David Sherwood <david.sherwood@arm.com>
3197 * machmode.h (is_int_mode): New fuction.
3198 * combine.c (find_split_point): Use it.
3199 (combine_simplify_rtx): Likewise.
3200 (simplify_if_then_else): Likewise.
3201 (simplify_set): Likewise.
3202 (simplify_shift_const_1): Likewise.
3203 (simplify_comparison): Likewise.
3204 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3205 * cse.c (notreg_cost): Likewise.
3206 (cse_insn): Likewise.
3207 * cselib.c (cselib_lookup_1): Likewise.
3208 * dojump.c (do_jump_1): Likewise.
3209 (do_compare_rtx_and_jump): Likewise.
3210 * dse.c (get_call_args): Likewise.
3211 * dwarf2out.c (rtl_for_decl_init): Likewise.
3212 (native_encode_initializer): Likewise.
3213 * expmed.c (emit_store_flag_1): Likewise.
3214 (emit_store_flag): Likewise.
3215 * expr.c (convert_modes): Likewise.
3216 (store_field): Likewise.
3217 (expand_expr_real_1): Likewise.
3218 * fold-const.c (fold_read_from_constant_string): Likewise.
3219 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3220 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3221 * optabs.c (expand_binop): Likewise.
3222 (expand_unop): Likewise.
3223 (expand_abs_nojump): Likewise.
3224 (expand_one_cmpl_abs_nojump): Likewise.
3225 * simplify-rtx.c (mode_signbit_p): Likewise.
3226 (val_signbit_p): Likewise.
3227 (val_signbit_known_set_p): Likewise.
3228 (val_signbit_known_clear_p): Likewise.
3229 (simplify_relational_operation_1): Likewise.
3230 * tree.c (vector_type_mode): Likewise.
3232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3233 Alan Hayward <alan.hayward@arm.com>
3234 David Sherwood <david.sherwood@arm.com>
3236 * machmode.h (smallest_mode_for_size): Fix formatting.
3237 (smallest_int_mode_for_size): New function.
3238 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3239 instead of smallest_mode_for_size.
3240 * combine.c (make_extraction): Likewise.
3241 * config/arc/arc.c (arc_expand_movmem): Likewise.
3242 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3243 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3244 * config/s390/s390.c (s390_expand_insv): Likewise.
3245 * config/sparc/sparc.c (assign_int_registers): Likewise.
3246 * config/spu/spu.c (spu_function_value): Likewise.
3247 (spu_function_arg): Likewise.
3248 * coverage.c (get_gcov_type): Likewise.
3249 (get_gcov_unsigned_t): Likewise.
3250 * dse.c (find_shift_sequence): Likewise.
3251 * expmed.c (store_bit_field_1): Likewise.
3252 * expr.c (convert_move): Likewise.
3253 (store_field): Likewise.
3254 * internal-fn.c (expand_arith_overflow): Likewise.
3255 * optabs-query.c (get_best_extraction_insn): Likewise.
3256 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3257 * stor-layout.c (layout_type): Likewise.
3258 (initialize_sizetypes): Likewise.
3259 * targhooks.c (default_get_mask_mode): Likewise.
3260 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3263 Alan Hayward <alan.hayward@arm.com>
3264 David Sherwood <david.sherwood@arm.com>
3266 * machmode.h (opt_mode::else_blk): New function.
3267 (int_mode_for_mode): Declare.
3268 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3269 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3271 * cfgexpand.c (expand_debug_expr): Likewise.
3272 * combine.c (gen_lowpart_or_truncate): Likewise.
3273 (gen_lowpart_for_combine): Likewise.
3274 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3275 * config/avr/avr.c (avr_to_int_mode): Likewise.
3276 (avr_out_plus_1): Likewise.
3277 (avr_out_plus): Likewise.
3278 (avr_out_round): Likewise.
3279 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3280 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3281 (s390_expand_vcond): Likewise.
3282 * config/spu/spu.c (spu_split_immediate): Likewise.
3283 (spu_expand_mov): Likewise.
3284 * dse.c (get_stored_val): Likewise.
3285 * expmed.c (store_bit_field_1): Likewise.
3286 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3288 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3289 (extract_low_bits): Likewise.
3290 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3291 handling rather than repeating the check.
3292 (emit_group_store): Likewise.
3293 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3294 * optabs.c (expand_absneg_bit): Likewise.
3295 (expand_copysign_absneg): Likewise.
3296 (expand_copysign_bit): Likewise.
3297 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3298 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3299 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3300 * var-tracking.c (prepare_call_arguments): Likewise.
3301 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3302 int_mode_for_mode instead of mode_for_size.
3303 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3306 Alan Hayward <alan.hayward@arm.com>
3307 David Sherwood <david.sherwood@arm.com>
3309 * machmode.h (int_mode_for_size): New function.
3310 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3311 instead of mode_for_size.
3312 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3314 * combine.c (expand_field_assignment): Use int_mode_for_size
3315 instead of mode_for_size.
3316 (make_extraction): Likewise.
3317 (simplify_shift_const_1): Likewise.
3318 (simplify_comparison): Likewise.
3319 * dojump.c (do_jump): Likewise.
3320 * dwarf2out.c (mem_loc_descriptor): Likewise.
3321 * emit-rtl.c (init_derived_machine_modes): Likewise.
3322 * expmed.c (flip_storage_order): Likewise.
3323 (convert_extracted_bit_field): Likewise.
3324 * expr.c (copy_blkmode_from_reg): Likewise.
3325 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3326 * internal-fn.c (expand_mul_overflow): Likewise.
3327 * lower-subreg.c (simple_move): Likewise.
3328 * optabs-libfuncs.c (init_optabs): Likewise.
3329 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3330 * tree.c (vector_type_mode): Likewise.
3331 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3332 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3333 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3334 * tree-vect-stmts.c (vectorizable_load): Likewise.
3335 (vectorizable_store): Likewise.
3337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3338 Alan Hayward <alan.hayward@arm.com>
3339 David Sherwood <david.sherwood@arm.com>
3341 * coretypes.h (pod_mode): New type.
3342 (scalar_int_mode_pod): New typedef.
3343 * machmode.h (pod_mode): New class.
3344 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3345 * genmodes.c (emit_mode_int_n): Update accordingly.
3346 * lower-subreg.h (target_lower_subreg): Change type to
3347 scalar_int_mode_pod.
3348 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3349 scalar_int_mode_pod.
3351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3352 Alan Hayward <alan.hayward@arm.com>
3353 David Sherwood <david.sherwood@arm.com>
3355 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3356 machine_mode to scalar_int_mode.
3357 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3358 (rs6000_option_override_internal): Remove cast to int.
3359 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3360 machine_mode to scalar_int_mode.
3361 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3362 (rs6000_option_override_internal): Remove cast to int.
3363 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3364 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3366 * config/s390/s390.c (s390_expand_builtin): Likewise.
3367 * coretypes.h (scalar_int_mode): New type.
3368 (opt_scalar_int_mode): New typedef.
3369 * machmode.h (scalar_int_mode): New class.
3370 (scalar_int_mode::includes_p): New function.
3371 (byte_mode): Change type to scalar_int_mode.
3372 (word_mode): Likewise.
3373 (ptr_mode): Likewise.
3374 * emit-rtl.c (byte_mode): Likewise.
3375 (word_mode): Likewise.
3376 (ptr_mode): Likewise.
3377 (init_derived_machine_modes): Update accordingly.
3378 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3379 and MODE_PARTIAL_INT.
3380 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3381 opt_scalar_int_mode.
3383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3384 Alan Hayward <alan.hayward@arm.com>
3385 David Sherwood <david.sherwood@arm.com>
3387 * target.def (libgcc_floating_mode_supported_p): Take a
3389 * doc/tm.texi: Regenerate.
3390 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3392 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3393 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3397 Alan Hayward <alan.hayward@arm.com>
3398 David Sherwood <david.sherwood@arm.com>
3400 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3401 * doc/tm.texi: Regenerate.
3402 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3403 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3404 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3405 * targhooks.h (default_floatn_mode): Likewise.
3406 * targhooks.c (default_floatn_mode): Likewise.
3407 * tree.c (build_common_tree_nodes): Update accordingly.
3409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3410 Alan Hayward <alan.hayward@arm.com>
3411 David Sherwood <david.sherwood@arm.com>
3413 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3414 (mode_iterator::iterate_p): Likewise.
3415 (mode_iterator::get_wider): Likewise.
3416 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3419 Alan Hayward <alan.hayward@arm.com>
3420 David Sherwood <david.sherwood@arm.com>
3422 * coretypes.h (opt_scalar_float_mode): New typedef.
3423 * machmode.h (float_mode_for_size): New function.
3424 * emit-rtl.c (double_mode): Delete.
3425 (init_emit_once): Use float_mode_for_size.
3426 * stor-layout.c (layout_type): Likewise.
3427 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3430 Alan Hayward <alan.hayward@arm.com>
3431 David Sherwood <david.sherwood@arm.com>
3433 * output.h (assemble_real): Take a scalar_float_mode.
3434 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3435 * config/arm/arm.md (consttable_4): Likewise.
3436 (consttable_8): Likewise.
3437 (consttable_16): Likewise.
3438 * config/mips/mips.md (consttable_float): Likewise.
3439 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3440 * varasm.c (assemble_real): Take a scalar_float_mode.
3441 (output_constant_pool_2): Update accordingly.
3442 (output_constant): Likewise.
3444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3445 Alan Hayward <alan.hayward@arm.com>
3446 David Sherwood <david.sherwood@arm.com>
3448 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3449 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3450 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3451 (native_encode_real): Likewise.
3452 (native_interpret_real): Likewise.
3453 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3454 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3461 <scalar_float_mode>. Simplify.
3462 (gen_extend_conv_libfunc): Likewise.
3464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3465 Alan Hayward <alan.hayward@arm.com>
3466 David Sherwood <david.sherwood@arm.com>
3468 * coretypes.h (scalar_float_mode): New type.
3469 * machmode.h (mode_traits::from_int): Use machine_mode if
3470 USE_ENUM_MODES is defined.
3471 (is_a): New function.
3473 (dyn_cast): Likewise.
3474 (scalar_float_mode): New class.
3475 (scalar_float_mode::includes_p): New function.
3476 (is_float_mode): Likewise.
3477 * gdbhooks.py (MachineModePrinter): New class.
3478 (build_pretty_printer): Use it for scalar_float_mode.
3479 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3480 (format_helper::format_helper): Turn into a template.
3481 * genmodes.c (get_mode_class): New function.
3482 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3483 or machine_mode if none.
3484 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3485 as_a <scalar_float_mode>.
3486 * dwarf2out.c (mem_loc_descriptor): Likewise.
3487 (insert_float): Likewise.
3488 (add_const_value_attribute): Likewise.
3489 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3490 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3491 (expand_unop): Update accordingly.
3492 (expand_abs_nojump): Likewise.
3493 (expand_copysign_absneg): Take a scalar_float_mode.
3494 (expand_copysign_bit): Likewise.
3495 (expand_copysign): Update accordingly.
3497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3498 Alan Hayward <alan.hayward@arm.com>
3499 David Sherwood <david.sherwood@arm.com>
3501 * coretypes.h (opt_mode): New class.
3502 * machmode.h (opt_mode): Likewise.
3503 (opt_mode::else_void): New function.
3504 (opt_mode::require): Likewise.
3505 (opt_mode::exists): Likewise.
3506 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3507 (GET_MODE_2XWIDER_MODE): Likewise.
3508 (mode_iterator::get_wider): Update accordingly.
3509 (mode_iterator::get_2xwider): Likewise.
3510 (mode_iterator::get_known_wider): Likewise, turning into a template.
3511 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3512 forcing a wider mode to exist.
3513 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3514 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3515 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3516 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3517 * lower-subreg.c (init_lower_subreg): Likewise.
3518 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3519 on the final iteration.
3520 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3521 a wider mode exists before asking for a move pattern.
3522 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3523 forcing a wider mode to exist.
3524 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3525 returning false if no such mode exists.
3526 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3527 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3528 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3529 Avoid checking for a MODE_INT if we already know the mode is not a
3531 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3532 forcing a wider mode to exist.
3533 (expmed_mult_highpart_optab): Likewise.
3534 (expmed_mult_highpart): Likewise.
3535 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3537 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3538 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3539 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3540 * internal-fn.c (expand_mul_overflow): Update use of
3541 GET_MODE_2XWIDER_MODE.
3542 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3543 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3544 GET_MODE_WIDER_MODE.
3545 (convert_plusminus_to_widen): Likewise.
3546 * tree-switch-conversion.c (array_value_type): Likewise.
3547 * var-tracking.c (emit_note_insn_var_location): Likewise.
3548 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3549 Return false inside rather than outside the loop if no wider mode
3551 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3552 and GET_MODE_2XWIDER_MODE
3553 (can_compare_p): Use else_void.
3554 * gdbhooks.py (OptMachineModePrinter): New class.
3555 (build_pretty_printer): Use it for opt_mode.
3557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3558 Alan Hayward <alan.hayward@arm.com>
3559 David Sherwood <david.sherwood@arm.com>
3561 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3562 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3565 Alan Hayward <alan.hayward@arm.com>
3566 David Sherwood <david.sherwood@arm.com>
3568 * machmode.h (mode_traits): New structure.
3569 (get_narrowest_mode): New function.
3570 (mode_iterator::start): Likewise.
3571 (mode_iterator::iterate_p): Likewise.
3572 (mode_iterator::get_wider): Likewise.
3573 (mode_iterator::get_known_wider): Likewise.
3574 (mode_iterator::get_2xwider): Likewise.
3575 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3576 (FOR_EACH_MODE): Likewise.
3577 (FOR_EACH_MODE_FROM): Likewise.
3578 (FOR_EACH_MODE_UNTIL): Likewise.
3579 (FOR_EACH_WIDER_MODE): Likewise.
3580 (FOR_EACH_2XWIDER_MODE): Likewise.
3581 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3582 * combine.c (simplify_comparison): Likewise
3583 * config/i386/i386.c (type_natural_mode): Likewise.
3584 * cse.c (cse_insn): Likewise.
3585 * dse.c (find_shift_sequence): Likewise.
3586 * emit-rtl.c (init_derived_machine_modes): Likewise.
3587 (init_emit_once): Likewise.
3588 * explow.c (hard_function_value): Likewise.
3589 * expmed.c (extract_fixed_bit_field_1): Likewise.
3590 (extract_bit_field_1): Likewise.
3591 (expand_divmod): Likewise.
3592 (emit_store_flag_1): Likewise.
3593 * expr.c (init_expr_target): Likewise.
3594 (convert_move): Likewise.
3595 (alignment_for_piecewise_move): Likewise.
3596 (widest_int_mode_for_size): Likewise.
3597 (emit_block_move_via_movmem): Likewise.
3598 (copy_blkmode_to_reg): Likewise.
3599 (set_storage_via_setmem): Likewise.
3600 (compress_float_constant): Likewise.
3601 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3602 * optabs-query.c (get_best_extraction_insn): Likewise.
3603 * optabs.c (expand_binop): Likewise.
3604 (expand_twoval_unop): Likewise.
3605 (expand_twoval_binop): Likewise.
3606 (widen_leading): Likewise.
3607 (widen_bswap): Likewise.
3608 (expand_parity): Likewise.
3609 (expand_unop): Likewise.
3610 (prepare_cmp_insn): Likewise.
3611 (prepare_float_lib_cmp): Likewise.
3612 (expand_float): Likewise.
3613 (expand_fix): Likewise.
3614 (expand_sfix_optab): Likewise.
3615 * postreload.c (move2add_use_add2_insn): Likewise.
3616 * reg-stack.c (reg_to_stack): Likewise.
3617 * reginfo.c (choose_hard_reg_mode): Likewise.
3618 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3619 * stor-layout.c (mode_for_size): Likewise.
3620 (smallest_mode_for_size): Likewise.
3621 (mode_for_vector): Likewise.
3622 (finish_bitfield_representative): Likewise.
3623 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3624 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3625 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3626 * var-tracking.c (prepare_call_arguments): Likewise.
3628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3629 Alan Hayward <alan.hayward@arm.com>
3630 David Sherwood <david.sherwood@arm.com>
3632 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3633 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3634 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3635 * machmode.h (mode_size): Move earlier in file.
3636 (mode_precision): Likewise.
3637 (mode_inner): Likewise.
3638 (mode_nunits): Likewise.
3639 (mode_unit_size): Likewise.
3640 (unit_unit_precision): Likewise.
3641 (mode_wider): Likewise.
3642 (mode_2xwider): Likewise.
3643 (machine_mode): New class.
3644 (mode_to_bytes): New function.
3645 (mode_to_bits): Likewise.
3646 (mode_to_precision): Likewise.
3647 (mode_to_inner): Likewise.
3648 (mode_to_unit_size): Likewise.
3649 (mode_to_unit_precision): Likewise.
3650 (mode_to_nunits): Likewise.
3651 (GET_MODE_SIZE): Use mode_to_bytes.
3652 (GET_MODE_BITSIZE): Use mode_to_bits.
3653 (GET_MODE_PRECISION): Use mode_to_precision.
3654 (GET_MODE_INNER): Use mode_to_inner.
3655 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3656 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3657 (GET_MODE_NUNITS): Use mode_to_nunits.
3658 * system.h (ALWAYS_INLINE): New macro.
3659 * config/powerpcspe/powerpcspe-c.c
3660 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3661 int for arg1_mode and arg2_mode.
3663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3664 Alan Hayward <alan.hayward@arm.com>
3665 David Sherwood <david.sherwood@arm.com>
3667 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3668 Prefix mode names with E_ in case statements.
3669 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3670 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3671 (aarch64_split_simd_move): Likewise.
3672 (aarch64_gen_storewb_pair): Likewise.
3673 (aarch64_gen_loadwb_pair): Likewise.
3674 (aarch64_gen_store_pair): Likewise.
3675 (aarch64_gen_load_pair): Likewise.
3676 (aarch64_get_condition_code_1): Likewise.
3677 (aarch64_constant_pool_reload_icode): Likewise.
3678 (get_rsqrte_type): Likewise.
3679 (get_rsqrts_type): Likewise.
3680 (get_recpe_type): Likewise.
3681 (get_recps_type): Likewise.
3682 (aarch64_gimplify_va_arg_expr): Likewise.
3683 (aarch64_simd_container_mode): Likewise.
3684 (aarch64_emit_load_exclusive): Likewise.
3685 (aarch64_emit_store_exclusive): Likewise.
3686 (aarch64_expand_compare_and_swap): Likewise.
3687 (aarch64_gen_atomic_cas): Likewise.
3688 (aarch64_emit_bic): Likewise.
3689 (aarch64_emit_atomic_swap): Likewise.
3690 (aarch64_emit_atomic_load_op): Likewise.
3691 (aarch64_evpc_trn): Likewise.
3692 (aarch64_evpc_uzp): Likewise.
3693 (aarch64_evpc_zip): Likewise.
3694 (aarch64_evpc_ext): Likewise.
3695 (aarch64_evpc_rev): Likewise.
3696 (aarch64_evpc_dup): Likewise.
3697 (aarch64_gen_ccmp_first): Likewise.
3698 (aarch64_gen_ccmp_next): Likewise.
3699 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3700 (alpha_emit_xfloating_libcall): Likewise.
3701 (emit_insxl): Likewise.
3702 (alpha_arg_type): Likewise.
3703 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3704 (arc_preferred_simd_mode): Likewise.
3705 (arc_secondary_reload): Likewise.
3706 (get_arc_condition_code): Likewise.
3707 (arc_print_operand): Likewise.
3708 (arc_legitimate_constant_p): Likewise.
3709 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3710 * config/arc/arc.md (casesi_load): Likewise.
3711 (casesi_compact_jump): Likewise.
3712 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3713 (cc_use_register): Likewise.
3714 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3715 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3716 (arm_init_iwmmxt_builtins): Likewise.
3717 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3718 (neon_expand_vector_init): Likewise.
3719 (arm_attr_length_move_neon): Likewise.
3720 (maybe_get_arm_condition_code): Likewise.
3721 (arm_emit_vector_const): Likewise.
3722 (arm_preferred_simd_mode): Likewise.
3723 (arm_output_iwmmxt_tinsr): Likewise.
3724 (thumb1_output_casesi): Likewise.
3725 (thumb2_output_casesi): Likewise.
3726 (arm_emit_load_exclusive): Likewise.
3727 (arm_emit_store_exclusive): Likewise.
3728 (arm_expand_compare_and_swap): Likewise.
3729 (arm_evpc_neon_vuzp): Likewise.
3730 (arm_evpc_neon_vzip): Likewise.
3731 (arm_evpc_neon_vrev): Likewise.
3732 (arm_evpc_neon_vtrn): Likewise.
3733 (arm_evpc_neon_vext): Likewise.
3734 (arm_validize_comparison): Likewise.
3735 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3736 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3737 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3738 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3739 (c6x_preferred_simd_mode): Likewise.
3740 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3741 (epiphany_rtx_costs): Likewise.
3742 * config/epiphany/predicates.md (proper_comparison_operator):
3744 * config/frv/frv.c (condexec_memory_operand): Likewise.
3745 (frv_emit_move): Likewise.
3746 (output_move_single): Likewise.
3747 (output_condmove_single): Likewise.
3748 (frv_hard_regno_mode_ok): Likewise.
3749 (frv_matching_accg_mode): Likewise.
3750 * config/h8300/h8300.c (split_adds_subs): Likewise.
3751 (h8300_rtx_costs): Likewise.
3752 (h8300_print_operand): Likewise.
3753 (compute_mov_length): Likewise.
3754 (output_logical_op): Likewise.
3755 (compute_logical_op_length): Likewise.
3756 (compute_logical_op_cc): Likewise.
3757 (h8300_shift_needs_scratch_p): Likewise.
3758 (output_a_shift): Likewise.
3759 (compute_a_shift_length): Likewise.
3760 (compute_a_shift_cc): Likewise.
3761 (expand_a_rotate): Likewise.
3762 (output_a_rotate): Likewise.
3763 * config/i386/i386.c (classify_argument): Likewise.
3764 (function_arg_advance_32): Likewise.
3765 (function_arg_32): Likewise.
3766 (function_arg_64): Likewise.
3767 (function_value_64): Likewise.
3768 (ix86_gimplify_va_arg): Likewise.
3769 (ix86_legitimate_constant_p): Likewise.
3770 (put_condition_code): Likewise.
3771 (split_double_mode): Likewise.
3772 (ix86_avx256_split_vector_move_misalign): Likewise.
3773 (ix86_expand_vector_logical_operator): Likewise.
3774 (ix86_split_idivmod): Likewise.
3775 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3776 (ix86_build_const_vector): Likewise.
3777 (ix86_build_signbit_mask): Likewise.
3778 (ix86_match_ccmode): Likewise.
3779 (ix86_cc_modes_compatible): Likewise.
3780 (ix86_expand_branch): Likewise.
3781 (ix86_expand_sse_cmp): Likewise.
3782 (ix86_expand_sse_movcc): Likewise.
3783 (ix86_expand_int_sse_cmp): Likewise.
3784 (ix86_expand_vec_perm_vpermi2): Likewise.
3785 (ix86_expand_vec_perm): Likewise.
3786 (ix86_expand_sse_unpack): Likewise.
3787 (ix86_expand_int_addcc): Likewise.
3788 (ix86_split_to_parts): Likewise.
3789 (ix86_vectorize_builtin_gather): Likewise.
3790 (ix86_vectorize_builtin_scatter): Likewise.
3791 (avx_vpermilp_parallel): Likewise.
3792 (inline_memory_move_cost): Likewise.
3793 (ix86_tieable_integer_mode_p): Likewise.
3794 (x86_maybe_negate_const_int): Likewise.
3795 (ix86_expand_vector_init_duplicate): Likewise.
3796 (ix86_expand_vector_init_one_nonzero): Likewise.
3797 (ix86_expand_vector_init_one_var): Likewise.
3798 (ix86_expand_vector_init_concat): Likewise.
3799 (ix86_expand_vector_init_interleave): Likewise.
3800 (ix86_expand_vector_init_general): Likewise.
3801 (ix86_expand_vector_set): Likewise.
3802 (ix86_expand_vector_extract): Likewise.
3803 (emit_reduc_half): Likewise.
3804 (ix86_emit_i387_round): Likewise.
3805 (ix86_mangle_type): Likewise.
3806 (ix86_expand_round_sse4): Likewise.
3807 (expand_vec_perm_blend): Likewise.
3808 (canonicalize_vector_int_perm): Likewise.
3809 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3810 (expand_vec_perm_1): Likewise.
3811 (expand_vec_perm_interleave3): Likewise.
3812 (expand_vec_perm_even_odd_pack): Likewise.
3813 (expand_vec_perm_even_odd_1): Likewise.
3814 (expand_vec_perm_broadcast_1): Likewise.
3815 (ix86_vectorize_vec_perm_const_ok): Likewise.
3816 (ix86_expand_vecop_qihi): Likewise.
3817 (ix86_expand_mul_widen_hilo): Likewise.
3818 (ix86_expand_sse2_abs): Likewise.
3819 (ix86_expand_pextr): Likewise.
3820 (ix86_expand_pinsr): Likewise.
3821 (ix86_preferred_simd_mode): Likewise.
3822 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3823 * config/i386/sse.md (*andnot<mode>3): Likewise.
3824 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3825 (*<code><mode>3): Likewise.
3826 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3827 (ia64_expand_atomic_op): Likewise.
3828 (ia64_arg_type): Likewise.
3829 (ia64_mode_to_int): Likewise.
3830 (ia64_scalar_mode_supported_p): Likewise.
3831 (ia64_vector_mode_supported_p): Likewise.
3832 (expand_vec_perm_broadcast): Likewise.
3833 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3834 (iq2000_function_arg_advance): Likewise.
3835 (iq2000_function_arg): Likewise.
3836 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3837 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3838 (m68k_libcall_value): Likewise.
3839 (m68k_function_value): Likewise.
3840 (sched_attr_op_type): Likewise.
3841 * config/mcore/mcore.c (mcore_output_move): Likewise.
3842 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3844 (microblaze_function_arg): Likewise.
3845 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3846 (mips_print_operand): Likewise.
3847 (mips_mode_ok_for_mov_fmt_p): Likewise.
3848 (mips_vector_mode_supported_p): Likewise.
3849 (mips_preferred_simd_mode): Likewise.
3850 (mips_expand_vpc_loongson_even_odd): Likewise.
3851 (mips_expand_vec_unpack): Likewise.
3852 (mips_expand_vi_broadcast): Likewise.
3853 (mips_expand_vector_init): Likewise.
3854 (mips_expand_vec_reduc): Likewise.
3855 (mips_expand_msa_cmp): Likewise.
3856 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3857 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3858 (cc_flags_for_mode): Likewise.
3859 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3860 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3861 (nds32_output_casesi_pc_relative): Likewise.
3862 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3863 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3864 (nvptx_gen_unpack): Likewise.
3865 (nvptx_gen_pack): Likewise.
3866 (nvptx_gen_shuffle): Likewise.
3867 (nvptx_gen_wcast): Likewise.
3868 (nvptx_preferred_simd_mode): Likewise.
3869 * config/pa/pa.c (pa_secondary_reload): Likewise.
3870 * config/pa/predicates.md (base14_operand): Likewise.
3871 * config/powerpcspe/powerpcspe-c.c
3872 (altivec_resolve_overloaded_builtin): Likewise.
3873 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3875 (rs6000_preferred_simd_mode): Likewise.
3876 (output_vec_const_move): Likewise.
3877 (rs6000_expand_vector_extract): Likewise.
3878 (rs6000_split_vec_extract_var): Likewise.
3879 (reg_offset_addressing_ok_p): Likewise.
3880 (rs6000_legitimate_offset_address_p): Likewise.
3881 (rs6000_legitimize_address): Likewise.
3882 (rs6000_emit_set_const): Likewise.
3883 (rs6000_const_vec): Likewise.
3884 (rs6000_emit_move): Likewise.
3885 (spe_build_register_parallel): Likewise.
3886 (rs6000_darwin64_record_arg_recurse): Likewise.
3887 (swap_selector_for_mode): Likewise.
3888 (spe_init_builtins): Likewise.
3889 (paired_init_builtins): Likewise.
3890 (altivec_init_builtins): Likewise.
3891 (do_load_for_compare): Likewise.
3892 (rs6000_generate_compare): Likewise.
3893 (rs6000_expand_float128_convert): Likewise.
3894 (emit_load_locked): Likewise.
3895 (emit_store_conditional): Likewise.
3896 (rs6000_output_function_epilogue): Likewise.
3897 (rs6000_handle_altivec_attribute): Likewise.
3898 (rs6000_function_value): Likewise.
3899 (emit_fusion_gpr_load): Likewise.
3900 (emit_fusion_p9_load): Likewise.
3901 (emit_fusion_p9_store): Likewise.
3902 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3903 (fusion_gpr_mem_load): Likewise.
3904 (fusion_addis_mem_combo_load): Likewise.
3905 (fusion_addis_mem_combo_store): Likewise.
3906 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3907 (fusion_gpr_mem_load): Likewise.
3908 (fusion_addis_mem_combo_load): Likewise.
3909 (fusion_addis_mem_combo_store): Likewise.
3910 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3912 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3913 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3914 (rs6000_preferred_simd_mode): Likewise.
3915 (output_vec_const_move): Likewise.
3916 (rs6000_expand_vector_extract): Likewise.
3917 (rs6000_split_vec_extract_var): Likewise.
3918 (reg_offset_addressing_ok_p): Likewise.
3919 (rs6000_legitimate_offset_address_p): Likewise.
3920 (rs6000_legitimize_address): Likewise.
3921 (rs6000_emit_set_const): Likewise.
3922 (rs6000_const_vec): Likewise.
3923 (rs6000_emit_move): Likewise.
3924 (rs6000_darwin64_record_arg_recurse): Likewise.
3925 (swap_selector_for_mode): Likewise.
3926 (paired_init_builtins): Likewise.
3927 (altivec_init_builtins): Likewise.
3928 (rs6000_expand_float128_convert): Likewise.
3929 (emit_load_locked): Likewise.
3930 (emit_store_conditional): Likewise.
3931 (rs6000_output_function_epilogue): Likewise.
3932 (rs6000_handle_altivec_attribute): Likewise.
3933 (rs6000_function_value): Likewise.
3934 (emit_fusion_gpr_load): Likewise.
3935 (emit_fusion_p9_load): Likewise.
3936 (emit_fusion_p9_store): Likewise.
3937 * config/rx/rx.c (rx_gen_move_template): Likewise.
3938 (flags_from_mode): Likewise.
3939 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3940 (s390_slb_comparison): Likewise.
3941 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3942 (s390_vector_mode_supported_p): Likewise.
3943 (s390_cc_modes_compatible): Likewise.
3944 (s390_match_ccmode_set): Likewise.
3945 (s390_canonicalize_comparison): Likewise.
3946 (s390_emit_compare_and_swap): Likewise.
3947 (s390_branch_condition_mask): Likewise.
3948 (s390_rtx_costs): Likewise.
3949 (s390_secondary_reload): Likewise.
3950 (__SECONDARY_RELOAD_CASE): Likewise.
3951 (s390_expand_cs): Likewise.
3952 (s390_preferred_simd_mode): Likewise.
3953 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3954 * config/sh/sh.c (sh_print_operand): Likewise.
3955 (dump_table): Likewise.
3956 (sh_secondary_reload): Likewise.
3957 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3958 * config/sh/sh.md (casesi_worker_1): Likewise.
3959 (casesi_worker_2): Likewise.
3960 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3961 (fcc_comparison_operator): Likewise.
3962 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3963 (emit_soft_tfmode_cvt): Likewise.
3964 (sparc_preferred_simd_mode): Likewise.
3965 (output_cbranch): Likewise.
3966 (sparc_print_operand): Likewise.
3967 (sparc_expand_vec_perm_bmask): Likewise.
3968 (vector_init_bshuffle): Likewise.
3969 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3970 (spu_vector_mode_supported_p): Likewise.
3971 (spu_expand_insv): Likewise.
3972 (spu_emit_branch_or_set): Likewise.
3973 (spu_handle_vector_attribute): Likewise.
3974 (spu_builtin_splats): Likewise.
3975 (spu_builtin_extract): Likewise.
3976 (spu_builtin_promote): Likewise.
3977 (spu_expand_sign_extend): Likewise.
3978 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3979 (tilegx_simd_int): Likewise.
3980 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3981 (tilepro_simd_int): Likewise.
3982 * config/v850/v850.c (const_double_split): Likewise.
3983 (v850_print_operand): Likewise.
3984 (ep_memory_offset): Likewise.
3985 * config/vax/vax.c (vax_rtx_costs): Likewise.
3986 (vax_output_int_move): Likewise.
3987 (vax_output_int_add): Likewise.
3988 (vax_output_int_subtract): Likewise.
3989 * config/visium/predicates.md (visium_branch_operator): Likewise.
3990 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3991 (visium_print_operand_address): Likewise.
3992 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3993 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3994 (xtensa_expand_conditional_branch): Likewise.
3995 (xtensa_copy_incoming_a7): Likewise.
3996 (xtensa_output_literal): Likewise.
3997 * dfp.c (decimal_real_maxval): Likewise.
3998 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4001 Alan Hayward <alan.hayward@arm.com>
4002 David Sherwood <david.sherwood@arm.com>
4004 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4005 (mode_nunits_inline): Likewise.
4006 (mode_inner_inline): Likewise.
4007 (mode_unit_size_inline): Likewise.
4008 (mode_unit_precision_inline): Likewise.
4009 (emit_insn_modes_h): Likewise. Also emit a #define of the
4011 (emit_mode_wider): Add an E_ prefix to mode names.
4012 (emit_mode_complex): Likewise.
4013 (emit_mode_inner): Likewise.
4014 (emit_mode_adjustments): Likewise.
4015 (emit_mode_int_n): Likewise.
4016 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4017 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4018 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4019 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4020 (CRC32_BUILTIN, ENTRY): Likewise.
4021 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4022 (aarch64_pop_regs): Likewise.
4023 (aarch64_process_components): Likewise.
4024 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4025 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4026 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4027 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4028 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4029 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4031 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4032 (pcmpistr): Likewise.
4033 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4034 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4035 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4037 * config/rl78/rl78.c (mduc_regs): Likewise.
4038 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4039 (htm_expand_builtin): Likewise.
4040 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4041 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4042 * config/xtensa/xtensa.c (print_operand): Likewise.
4043 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4044 (NUM_MODE_VECTOR_INT): Likewise.
4045 * genoutput.c (null_operand): Likewise.
4046 (output_operand_data): Likewise.
4047 * genrecog.c (print_parameter_value): Likewise.
4048 * lra.c (debug_operand_data): Likewise.
4050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4052 * dwarf2out.c (scompare_loc_descriptor_wide)
4053 (scompare_loc_descriptor_narrow): New functions, split out from...
4054 (scompare_loc_descriptor): ...here.
4055 * expmed.c (emit_store_flag_int): New function, split out from...
4056 (emit_store_flag): ...here.
4058 2017-08-30 Richard Biener <rguenther@suse.de>
4060 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4061 (dwarf2out_early_finish): Move setting of AT_pubnames from
4062 early debug output to early finish.
4064 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4066 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4067 and -mdata-region to the assembler.
4069 2017-08-30 Richard Biener <rguenther@suse.de>
4071 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4073 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4074 (dwarf2out_early_global_decl): Only generate a DIE for the
4075 abstract origin if it doesn't already exist or is a declaration DIE.
4076 (resolve_addr): Do not add the linkage name twice when
4077 generating a stub DIE for the DW_TAG_GNU_call_site target.
4079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4081 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4082 Use machine_mode rather than int for arg1_mode.
4084 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4087 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4088 that the second argument of the built-in functions to unpack
4089 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4090 switch statement instead a lot of if statements.
4091 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4092 Allow 64-bit values to be in Altivec registers as well as
4093 traditional floating point registers.
4094 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4096 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4098 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4099 MAX_REGS_PER_ADDRESS == 1.
4101 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4103 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4104 * config/i386/i386.c (ix86_option_override_internal): Simplify
4105 setting of opts->x_flag_entry.
4107 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4108 Jakub Jelinek <jakub@redhat.com>
4109 Richard Biener <rguenther@suse.de>
4111 PR tree-optimization/81503
4112 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4113 folded constant fits in the target type; reorder tests for clarity.
4115 2017-08-29 Martin Liska <mliska@suse.cz>
4117 * passes.def: Include pass_lower_switch.
4118 * stmt.c (dump_case_nodes): Remove and move to
4119 tree-switch-conversion.
4120 (case_values_threshold): Likewise.
4121 (expand_switch_as_decision_tree_p): Likewise.
4122 (emit_case_decision_tree): Likewise.
4123 (expand_case): Likewise.
4124 (balance_case_nodes): Likewise.
4125 (node_has_low_bound): Likewise.
4126 (node_has_high_bound): Likewise.
4127 (node_is_bounded): Likewise.
4128 (emit_case_nodes): Likewise.
4129 (struct simple_case_node): New struct.
4130 (add_case_node): Remove.
4131 (emit_case_dispatch_table): Use vector instead of case_list.
4132 (reset_out_edges_aux): Remove.
4133 (compute_cases_per_edge): Likewise.
4134 (expand_case): Build list of simple_case_node.
4135 (expand_sjlj_dispatch_table): Use it.
4136 * tree-switch-conversion.c (struct case_node): Moved from
4137 stmt.c and adjusted.
4138 (emit_case_nodes): Likewise.
4139 (node_has_low_bound): Likewise.
4140 (node_has_high_bound): Likewise.
4141 (node_is_bounded): Likewise.
4142 (case_values_threshold): Likewise.
4143 (reset_out_edges_aux): Likewise.
4144 (compute_cases_per_edge): Likewise.
4145 (add_case_node): Likewise.
4146 (dump_case_nodes): Likewise.
4147 (balance_case_nodes): Likewise.
4148 (expand_switch_as_decision_tree_p): Likewise.
4149 (emit_jump): Likewise.
4150 (emit_case_decision_tree): Likewise.
4151 (try_switch_expansion): Likewise.
4152 (do_jump_if_equal): Likewise.
4153 (emit_cmp_and_jump_insns): Likewise.
4154 (fix_phi_operands_for_edge): New function.
4155 (record_phi_operand_mapping): Likewise.
4156 (class pass_lower_switch): New pass.
4157 (pass_lower_switch::execute): New function.
4158 (make_pass_lower_switch): Likewise.
4159 (conditional_probability):
4160 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4161 * tree-pass.h: Add make_pass_lower_switch.
4163 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4166 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4169 2017-08-29 Richard Biener <rguenther@suse.de>
4171 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4172 we do not add a DW_AT_inline attribute twice.
4173 (gen_subprogram_die): Remove code setting DW_AT_inline on
4174 DECL_ABSTRACT_P nodes.
4176 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4178 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4179 calls to internal functions.
4180 (gimplify_modify_expr): Likewise.
4181 * tree-call-cdce.c (use_internal_fn): Likewise.
4182 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4183 (convert_to_divmod): Set the nothrow flag.
4184 * tree-if-conv.c (predicate_mem_writes): Likewise.
4185 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4186 (vectorizable_call): Likewise.
4187 (vectorizable_store): Likewise.
4188 (vectorizable_load): Likewise.
4189 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4190 (vect_recog_mask_conversion_pattern): Likewise.
4192 2017-08-29 Martin Liska <mliska@suse.cz>
4195 * gcc.c (driver_handle_option): Add new argument.
4196 * opts-common.c (handle_option): Pass
4197 target_option_override_hook.
4198 * opts-global.c (lang_handle_option): Add new option.
4199 (set_default_handlers): Add new argument.
4200 (decode_options): Likewise.
4201 * opts.c (target_handle_option): Likewise.
4202 (common_handle_option): Call target_option_override_hook.
4203 * opts.h (struct cl_option_handler_func): Add hook for
4204 target option override.
4205 (struct cl_option_handlers): Likewise.
4206 (set_default_handlers): Add new argument.
4207 (decode_options): Likewise.
4208 (common_handle_option): Likewise.
4209 (target_handle_option): Likewise.
4210 * toplev.c (toplev::main): Pass targetm.target_option.override
4213 2017-08-29 Richard Biener <rguenther@suse.de>
4214 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4216 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4217 life to the active subtree.
4219 2017-08-28 Jeff Law <law@redhat.com>
4221 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4222 derive_equivalences.
4223 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4225 (edge_info::derive_equivalences): New private member function
4227 * tree-ssa-dom.c (class edge_info): Changed from a struct
4228 to a class. Add ctor/dtor, methods and data members.
4229 (edge_info::edge_info): Renamed from allocate_edge_info.
4230 Initialize additional members.
4231 (edge_info::~edge_info): New.
4232 (free_dom_edge_info): Delete the edge info.
4233 (record_edge_info): Use new class & associated member functions.
4234 Tighten forms for testing for edge equivalences.
4235 (record_temporary_equivalences): Iterate over the simple
4236 equivalences rather than assuming there's only one per edge.
4237 (cprop_into_successor_phis): Iterate over the simple
4238 equivalences rather than assuming there's only one per edge.
4239 (optimize_stmt): Use operand_equal_p rather than pointer
4240 equality for mini-DSE code.
4242 2017-08-28 Nathan Sidwell <nathan@acm.org>
4244 * gcc.c (execute): Fold SIGPIPE handling into switch
4245 statement. Adjust internal error message.
4247 2017-08-28 Richard Biener <rguenther@suse.de>
4250 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4251 Do nothing for removed DIEs.
4253 2017-08-28 Richard Biener <rguenther@suse.de>
4255 PR tree-optimization/81977
4256 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4259 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4262 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4263 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4264 using targetm.gen_mem_thread_fence.
4266 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4269 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4270 predicate to register_operand. Reorder operands.
4271 (*btr<mode>): Ditto.
4272 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4273 (*btr<mode>_mask): Ditto.
4275 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4277 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4278 * config/rs6000/xmmintrin.h: New file.
4279 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4281 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4284 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4285 parameter and_insn and return it.
4286 (recombine_lvx_pattern): Insert a copy to ensure availability of
4287 the base register of the copied masking operation at the point of
4288 the instruction replacement.
4289 (recombine_stvx_pattern): Likewise.
4291 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4293 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4294 undocumented switches.
4295 (-mpower9-dform-vector): Likewise.
4296 (-mpower9-dform): Likewise.
4297 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4298 comments to delete references to -mpower9-dform* switches.
4299 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4300 Delete reference to -mpower9-dform* switches, test for
4301 -mpower9-vector instead.
4302 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4303 (OTHER_P9_VECTOR_MASKS): Likewise.
4304 (POWERPC_MASKS): Likewise.
4305 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4306 tests against -mpower9-dform* to -mpower9-vector. Delete code
4307 that checked for -mpower9-dform* consistancy with other options.
4308 Add test for -mpower9-misc to enable other power9 switches.
4309 (rs6000_init_hard_regno_mode_ok): Likewise.
4310 (rs6000_option_override_internal): Likewise.
4311 (rs6000_emit_prologue): Likewise.
4312 (rs6000_emit_epilogue): Likewise.
4313 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4314 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4315 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4317 (emit_fusion_p9_store): Likewise.
4318 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4319 resetting these macros if the assembler does not support ISA 3.0
4321 (TARGET_P9_DFORM_VECTOR): Likewise.
4322 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4323 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4325 2017-08-25 Alan Modra <amodra@gmail.com>
4327 PR rtl-optimization/81747
4328 * cse.c (cse_extended_basic_block): Don't attempt to record
4329 equivalences for degenerate conditional branches that branch
4330 to their fall-through.
4332 2017-08-24 Martin Sebor <msebor@redhat.com>
4335 * gimple-fold.c (size_must_be_zero_p): New function.
4336 (gimple_fold_builtin_memory_op): Call it.
4338 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4340 * config/rs6000/mm_malloc.h: New file.
4342 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4344 PR tree-optimization/81913
4345 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4346 analysis when either IVs in condition can wrap.
4348 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4350 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4351 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4353 2017-08-24 Richard Biener <rguenther@suse.de>
4356 * targhooks.c (default_target_can_inline_p): Properly
4357 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4358 is present and always compare.
4359 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4360 infer -mfpmath=sse from TARGET_SSE_P.
4361 (ix86_can_inline_p): Properly use target_option_default_node when
4362 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4364 2017-08-24 Richard Biener <rguenther@suse.de>
4367 * dwarf2out.c (output_die): Handle flag_generate_offload like
4369 (output_comp_unit): Likewise.
4370 (gen_array_type_die): Likewise.
4371 (dwarf2out_early_finish): Likewise.
4372 (note_variable_value_in_expr): Likewise.
4373 (dwarf2out_finish): Likewise. Adjust assert.
4374 * cgraphunit.c (symbol_table::compile): Move setting of
4375 flag_generate_offload earlier ...
4376 (symbol_table::finalize_compilation_unit): ... here, before
4377 early debug finalization.
4379 2017-08-24 Richard Biener <rguenther@suse.de>
4381 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4382 and ipa-fnsummary.h.
4383 (ix86_can_inline_p): When ix86_fpmath flags do not match
4384 check whether the callee uses FP math at all.
4386 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4389 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4390 instead of TYPE_PRECISION.
4392 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4393 Alan Hayward <alan.hayward@arm.com>
4394 David Sherwood <david.sherwood@arm.com>
4396 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4398 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4400 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4402 (ix86_frame::stack_realign_allocate): New field.
4403 (struct machine_frame_state): Modify comments.
4404 (machine_frame_state::sp_realigned_fp_end): New field.
4405 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4407 (sp_valid_at): Add assertion to assure no attempt to access invalid
4408 offset of a realigned stack.
4409 (fp_valid_at): Likewise.
4410 (choose_baseaddr): Modify comments.
4411 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4412 ix86_expand_prologue.
4413 (ix86_expand_prologue): Modify stack realignment and allocation.
4414 (ix86_expand_epilogue): Modify comments.
4415 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4416 avx2_runtime, avx512f, and avx512f_runtime.
4418 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4420 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4421 (mstackrealign): Do not init to -1.
4422 * config/i386/i386.c (ix86_option_override_internal):
4423 Check opts_set, not opts when setting default value of
4424 opts->x_ix86_force_align_arg_pointer.
4426 2017-08-23 Richard Biener <rguenther@suse.de>
4428 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4429 lang_hooks.decl_printable_name.
4430 * print-rtl-function.c (print_rtx_function): Likewise.
4431 * tree-pretty-print.c (dump_function_header): Likewise.
4433 2017-08-23 Richard Biener <rguenther@suse.de>
4436 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4437 -g0 at compile-time.
4439 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4442 * doc/sourcebuild.texi (Other hardware attributes):
4445 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4448 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4449 Fix single-use check.
4451 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4453 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4455 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4457 * doc/install.texi: Modify to add more details on running selected
4460 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4462 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4463 is combined with -mabi=ms.
4464 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4467 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4469 PR tree-optimization/81488
4470 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4471 and cached_basis fields.
4472 (MAX_SPREAD): New constant.
4473 (alloc_cand_and_find_basis): Initialize new fields.
4474 (clear_visited): New function.
4475 (create_phi_basis_1): Rename from create_phi_basis, set visited
4476 and cached_basis fields.
4477 (create_phi_basis): New wrapper function.
4478 (phi_add_costs_1): Rename from phi_add_costs, add spread
4479 parameter, set visited field, short-circuit when limits reached.
4480 (phi_add_costs): New wrapper function.
4481 (record_phi_increments_1): Rename from record_phi_increments, set
4483 (record_phi_increments): New wrapper function.
4484 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4485 (phi_incr_cost): New wrapper function.
4486 (all_phi_incrs_profitable_1): Rename from
4487 all_phi_incrs_profitable, set visited field.
4488 (all_phi_incrs_profitable): New wrapper function.
4490 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4491 Alan Hayward <alan.hayward@arm.com>
4492 David Sherwood <david.sherwood@arm.com>
4494 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4495 that takes the outer and inner modes.
4496 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4497 comparison as the canonical test for a paradoxical subreg.
4498 * combine.c (simplify_set): Use paradoxical_subreg_p.
4499 (make_extraction): Likewise.
4500 (force_to_mode): Likewise.
4501 (rtx_equal_for_field_assignment_p): Likewise.
4502 (gen_lowpart_for_combine): Likewise.
4503 (simplify_comparison): Likewise.
4504 * cse.c (equiv_constant): Likewise.
4505 * expmed.c (store_bit_field_1): Likewise.
4506 * final.c (alter_subreg): Likewise.
4507 * fwprop.c (propagate_rtx): Likewise.
4508 (forward_propagate_subreg): Likewise.
4509 * ira-conflicts.c (ira_build_conflicts): Likewise.
4510 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4511 * lra-constraints.c (curr_insn_transform): Likewise.
4512 (split_reg): Likewise.
4513 * lra-eliminations.c (move_plus_up): Likewise.
4514 (lra_eliminate_regs_1): Likewise.
4515 * recog.c (general_operand): Likewise.
4516 * ree.c (combine_reaching_defs): Likewise.
4517 * reload.c (push_reload): Likewise.
4518 (find_reloads): Likewise.
4519 * reload1.c (elimination_effects): Likewise.
4520 (compute_reload_subreg_offset): Likewise.
4521 (choose_reload_regs): Likewise.
4522 * rtlanal.c (subreg_lsb_1): Likewise.
4523 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4524 (simplify_subreg): Likewise.
4525 * var-tracking.c (track_loc_p): Likewise.
4526 * emit-rtl.c (byte_lowpart_offset): Likewise.
4527 (paradoxical_subreg_p): Delete out-of-line definition.
4529 2017-08-22 Jeff Law <law@redhat.com>
4531 PR tree-optimization/81741
4532 PR tree-optimization/71947
4533 * tree-ssa-dom.c: Include tree-inline.h.
4534 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4535 equivalences if one is more expensive to compute than the other.
4536 * tree-ssa-scopedtables.h (class const_or_copies): Make
4537 record_const_or_copy_raw method private.
4538 (class avail_exprs_stack): New method simplify_binary_operation.
4539 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4540 avail_exprs_stack::simplify_binary_operation as needed.
4541 (avail_exprs_stack::simplify_binary_operation): New function.
4543 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4545 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4546 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4547 (DOT_SYMBOLS): Likewise.
4548 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4549 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4550 (RS6000_ABI_NAME): Likewise.
4551 (TARGET_CMODEL): Likewise.
4552 (TOC_SECTION_ASM_OP): Likewise.
4553 (SET_CMODEL): New macro.
4554 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4556 2017-08-22 Richard Biener <rguenther@suse.de>
4558 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4559 to allow for free-lang-data replacements similar to verify_type_variant.
4561 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4562 Alan Hayward <alan.hayward@arm.com>
4563 David Sherwood <david.sherwood@arm.com>
4565 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4566 VOIDmode for the LABEL_REF.
4568 2017-08-22 Richard Biener <rguenther@suse.de>
4570 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4572 2017-08-22 Richard Biener <rguenther@suse.de>
4574 * common.opt (feliminate-dwarf2-dups): Ignore.
4575 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4576 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4577 same_die_p_wrap, compute_section_prefix,
4578 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4579 (comdat_symbol_id, comdat_symbol_number): Likewise.
4580 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4582 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4583 (output_die): Mark unreachable path unreachable.
4584 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4585 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4586 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4587 (dwarf2out_early_finish): Likewise.
4589 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4591 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4593 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4596 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4597 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4598 (avr_attribute_table) <io, io_low, address>: Initialize
4599 .decl_required with true.
4601 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4603 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4604 undocumented debugging options.
4605 (-mvsx-scalar-double): Likewise.
4606 (-mallow-df-permute): Likewise.
4607 (-mvectorize-builtins): Likewise.
4608 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4609 (rs6000_builtin_vectorized_function): Likewise.
4610 (rs6000_builtin_md_vectorized_function): Likewise.
4611 (rs6000_opt_vars): Likewise.
4613 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4616 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4617 (*btrq_imm): Rename from *btrq.
4618 (*btcq_imm): Rename from *btcq.
4619 (btsc): New code attribute.
4620 (*<btsc><mode>): New insn pattern.
4621 (*btr<mode>): Ditto.
4622 (*<btsc><mode>_mask): New insn_and_split pattern.
4623 (*btr<mode>_mask): Ditto.
4625 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4626 Alan Hayward <alan.hayward@arm.com>
4627 David Sherwood <david.sherwood@arm.com>
4629 * function.c (pad_below): Simplify padding calculation.
4631 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4632 Alan Hayward <alan.hayward@arm.com>
4633 David Sherwood <david.sherwood@arm.com>
4635 * target.def (function_prologue): Remove frame size argument.
4636 (function_epilogue): Likewise.
4637 * doc/tm.texi: Regenerate.
4638 * final.c (final_start_function): Update call to function_prologue.
4639 (final_end_function): Update call to function_epilogue.
4640 (default_function_pro_epilogue): Remove frame size argument.
4641 * output.h (default_function_pro_epilogue): Likewise.
4642 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4643 (arm_output_function_prologue): Likewise.
4644 * config/frv/frv.c (frv_function_prologue): Likewise.
4645 (frv_function_epilogue): Likewise.
4646 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4647 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4648 (ia64_output_function_epilogue): Likewise.
4649 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4650 (m32r_output_function_epilogue): Likewise.
4651 * config/microblaze/microblaze.c (microblaze_function_prologue)
4652 (microblaze_function_epilogue): Likewise.
4653 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4654 (mips_output_function_epilogue): Likewise.
4655 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4656 (mmix_target_asm_function_epilogue): Likewise.
4657 * config/msp430/msp430.c (msp430_start_function): Likewise.
4658 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4659 (nds32_asm_function_epilogue): Likewise.
4660 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4661 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4662 (pa_output_function_epilogue): Likewise.
4663 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4664 (rs6000_output_function_epilogue): Likewise.
4665 * config/rl78/rl78.c (rl78_start_function): Likewise.
4666 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4667 (rs6000_output_function_epilogue): Likewise.
4668 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4669 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4670 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4671 (sparc_asm_function_epilogue): Likewise.
4673 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4675 * tree.h (type_has_mode_precision_p): New function.
4676 * convert.c (convert_to_integer_1): Use it.
4677 * expr.c (expand_expr_real_2): Likewise.
4678 (expand_expr_real_1): Likewise.
4679 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4680 * match.pd: Likewise.
4681 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4682 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4683 * tree-tailcall.c (process_assignment): Likewise.
4684 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4685 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4686 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4687 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4688 (vectorizable_assignment): Likewise.
4689 (vectorizable_shift): Likewise.
4690 (vectorizable_operation): Likewise.
4691 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4693 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4695 * match.pd: Add pow (C, x) simplification.
4697 2017-08-21 Richard Biener <rguenther@suse.de>
4699 PR tree-optimization/81900
4700 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4701 for blocks with abnormal predecessors.
4702 (compute_antic): Do not set visited flag prematurely.
4704 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4707 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4709 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4711 * stor-layout.h (vector_type_mode): Move to...
4712 * tree.h (vector_type_mode): ...here.
4713 * stor-layout.c (vector_type_mode): Move to...
4714 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4716 2017-08-21 Richard Biener <rguenther@suse.de>
4718 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4719 register_external_die hooks.
4720 (debug_false_tree_charstarstar_uhwistar): Declare.
4721 (debug_nothing_tree_charstar_uhwi): Likewise.
4722 * debug.c (do_nothing_debug_hooks): Adjust.
4723 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4724 (debug_nothing_tree_charstar_uhwi): Likewise.
4725 * dbxout.c (dbx_debug_hooks): Adjust.
4726 (xcoff_debug_hooks): Likewise.
4727 * sdbout.c (sdb_debug_hooks): Likewise.
4728 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4729 * dwarf2out.c (macinfo_label_base): New global.
4730 (dwarf2out_register_external_die): New function for the
4731 register_external_die hook.
4732 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4733 (dwarf2_debug_hooks): Use them.
4734 (dwarf2_lineno_debug_hooks): Adjust.
4735 (struct die_struct): Add with_offset flag.
4736 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4737 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4738 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4739 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4740 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4741 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4742 defining section names for the early LTO debug variants.
4743 (reset_indirect_string): New helper.
4744 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4745 (print_dw_val): Add support for offsetted symbol references.
4746 (get_ultimate_context): Split out from is_cxx.
4747 (is_cxx): Use get_ultimate_context.
4748 (is_fortran): Add decl overload.
4749 (compute_comp_unit_symbol): Split out worker from
4750 compute_section_prefix.
4751 (compute_section_prefix): Call compute_comp_unit_symbol and
4752 set comdat_type_p here.
4753 (output_die): Skip DIE symbol output for the LTO added one.
4754 Handle DIE symbol references with offset.
4755 (output_comp_unit): Guard section name mangling properly.
4756 For LTO debug sections emit a symbol at the section beginning
4757 which we use to refer to its DIEs.
4758 (add_abstract_origin_attribute): For DIEs registered via
4759 dwarf2out_register_external_die directly refer to the early
4760 DIE rather than indirectly through the shadow one we created.
4761 Remove obsolete call to dwarf2out_abstract_function for
4762 non-function/block origins.
4763 (gen_array_type_die): When generating early LTO debug do
4764 not emit DW_AT_string_length.
4765 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4766 late when in LTO. As suggested place a gcc_unreachable for
4767 the DECL_ABSTRACT_P case.
4768 (gen_subprogram_die): Avoid another specification DIE
4769 for early built declarations/definitions for the late LTO case.
4770 (gen_variable_die): Add type references for late duplicated VLA dies
4772 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4773 we have the abstract instance already.
4774 (process_scope_var): Adjust decl DIE contexts in LTO which
4775 first puts them in limbo.
4776 (gen_decl_die): Do not generate type DIEs late apart from
4777 types for VLAs or for decls we do not yet have a DIE. Do not
4778 call dwarf2out_abstract_function late.
4779 (dwarf2out_early_global_decl): Make sure to create DIEs
4780 for abstract instances of a decl first.
4781 (dwarf2out_late_global_decl): Adjust comment.
4782 (output_macinfo_op): With multiple macro sections use
4783 macinfo_label_base to distinguish labels.
4784 (output_macinfo): Likewise. Update macinfo_label_base.
4785 Pass in the line info label.
4786 (note_variable_value_in_expr): When generating LTO resolve
4787 all variable values here by generating DIEs as needed.
4788 (init_sections_and_labels): Add early LTO debug flag parameter
4789 and generate different sections and names if set. Add generation
4790 counter for the labels so we can have multiple of them.
4791 (reset_dies): Helper to allow DIEs to be output multiple times.
4792 (dwarf2out_finish): When outputting DIEs to the fat part of an
4793 LTO object first reset DIEs.
4794 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4795 (modified_type_die): Check for decl_ultimate_origin being self
4797 (gen_type_die_with_usage): Likewise.
4798 (gen_typedef_die): Allow decl_ultimate_origin being self.
4799 (set_decl_abstract_flags): Remove.
4800 (set_block_abstract_flags): Likewise.
4801 (dwarf2out_abstract_function): Treat the early generated DIEs
4802 as the abstract copy and only add DW_AT_inline and
4803 DW_AT_artificial here and call set_decl_origin_self.
4804 If the DIE has an abstract origin don't do anything.
4805 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4806 if we have none yet (Go fails to build one, PR78628).
4807 (variably_modified_type_p): Prevent endless recursion for Ada
4808 cyclic pointer types.
4809 * lto-streamer-in.c: Include debug.h.
4810 (dref_queue): New global.
4811 (lto_read_tree_1): Stream in DIE references.
4812 (lto_input_tree): Register DIE references.
4813 (input_function): Stream DECL_DEBUG_ARGS.
4814 * lto-streamer-out.c: Include debug.h.
4815 (lto_write_tree_1): Output DIE references.
4816 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4817 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4818 (output_function): Stream DECL_DEBUG_ARGS.
4819 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4820 Stream DECL_ABSTRACT_ORIGIN.
4821 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4822 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4823 DECL_CONTEXT for file-scope decls.
4824 * lto-streamer.h (struct dref_entry): Declare.
4825 (dref_queue): Likewise.
4826 * cfgexpand.c (pass_expand::execute): Do not call the
4827 outlining_inline_function hook here.
4828 * lto-wrapper.c (debug_obj): New global.
4829 (tool_cleanup): Unlink it if required.
4830 (debug_objcopy): New function.
4831 (run_gcc): Handle early debug sections in the IL files by
4832 extracting them to separate files, partially linkin them and
4833 feeding the result back as result to the linker.
4834 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4835 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4836 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4837 sections into a separate segment.
4838 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4840 (darwin_asm_dwarf_section): Likewise.
4841 (darwin_asm_output_dwarf_offset): Likewise.
4842 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4844 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4845 Alan Hayward <alan.hayward@arm.com>
4846 David Sherwood <david.sherwood@arm.com>
4848 * read-md.h (md_reader::record_potential_iterator_use): Replace
4849 pointer argument with an rtx and an index.
4850 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4851 (apply_mode_iterator): Likewise.
4852 (apply_code_iterator): Likewise.
4853 (apply_int_iterator): Likewise.
4854 (apply_subst_iterator): Likewise.
4855 (record_iterator_use): Likewise.
4856 (record_attribute_use): Likewise.
4857 (md_reader::record_potential_iterator_use): Likewise. Update calls
4858 to record_iterator_use and apply_iterator.
4859 (iterator_use): Replace ptr with x and index.
4860 (attribute_use): Likewise.
4861 (apply_attribute_uses): Update calls to apply_iterator.
4862 (apply_iterators): Likewise. Update initialization of iterator_use.
4863 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4864 and record_potential_iterator_use.
4865 (rtx_reader::read_rtx_operand): Likewise.
4867 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4868 Alan Hayward <alan.hayward@arm.com>
4869 David Sherwood <david.sherwood@arm.com>
4871 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4874 2017-08-21 Richard Biener <rguenther@suse.de>
4877 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4878 at struct end conservatively when comparing common bases.
4880 2017-08-21 Richard Biener <rguenther@suse.de>
4882 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4883 (mem_ref_in_stmt): Remove.
4884 (determine_max_movement): Use ref index to get at the reference.
4885 (invariantness_dom_walker::before_dom_children): Deal with
4886 lim data already initialized.
4887 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4889 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4891 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4892 (TARGET_ISA_ROUND): Ditto.
4893 (TARGET_ROUND): Ditto.
4894 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4895 * config/i386/i386.md: Ditto.
4896 * config/i386/sse.md: Ditto.
4897 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4898 with OPTION_MASK_ISA_SSE4_1.
4900 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4903 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4904 __builtin_ia32_lzcnt_u16.
4906 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4909 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4910 (rs6000_set_current_function): Rewrite function to use it.
4912 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4915 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4916 and TYPE_WARN_IF_NOT_ALIGN.
4917 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4918 (handle_warn_if_not_align): New.
4919 (place_union_field): Call handle_warn_if_not_align.
4920 (place_field): Call handle_warn_if_not_align.
4921 Copy TYPE_WARN_IF_NOT_ALIGN.
4922 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4923 (layout_type): Likewise.
4924 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4926 (tree_decl_common): Add warn_if_not_align.
4927 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4928 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4929 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4930 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4931 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4932 * doc/extend.texi: Document warn_if_not_aligned attribute.
4933 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4935 2017-08-17 Martin Liska <mliska@suse.cz>
4938 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4939 (get_data_dependence): Use it as pointer type.
4940 (distribute_loop): Likewise.
4942 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4944 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4945 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4946 (p8_vmrgow_<mode>_direct): New define_insn.
4947 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4948 handle endianness for vmrgew and vmrgow permute patterns.
4950 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4952 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4953 * config/rs6000/rs6000-cpus.def: Remove comment.
4954 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4955 (POWERPC_MASKS): Likewise.
4956 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4957 use of TARGET_VSX_TIMODE.
4958 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4959 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4960 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4961 (rs6000_option_override_internal): Remove dead code.
4962 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4963 (rs6000_legitimize_reload_address): Likewise.
4964 (rs6000_legitimate_address_p): Likewise.
4965 (rs6000_opt_masks): Delete "vsx-timode".
4966 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4967 from function comment.
4968 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4969 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4970 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4972 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4973 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4974 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4977 (define_peephole2): Likewise.
4979 2017-08-17 Martin Sebor <msebor@redhat.com>
4982 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4983 past the end of an array.
4984 (test_pp_format): Add test cases.
4986 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4988 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4989 missing ECF_NOTHROW flags.
4991 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4994 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4995 operands residing in integer registers.
4996 (*vsx_le_perm_load_<mode>): Likewise.
4997 (*vsx_le_perm_store_<mode>): Likewise.
4998 (define_peephole2): Add peepholes to optimize the above.
5000 2017-08-17 Marek Polacek <polacek@redhat.com>
5003 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5004 to mimic what shorten_compare did. Change the return type to bool.
5005 (fold_cond_expr_with_comparison): Update call to
5006 operand_equal_for_comparison_p.
5007 (fold_ternary_loc): Likewise.
5009 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5011 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5013 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5015 2017-08-17 Richard Biener <rguenther@suse.de>
5017 * tree-ssa-structalias.c (solve_graph): When propagating
5018 to successors update the graphs succ edges and avoid duplicate work.
5020 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5023 * config/i386/i386.c (ix86_option_override_internal): Save target
5024 specific options after ix86_stack_protector_guard_reg was changed.
5026 2017-08-17 Richard Biener <rguenther@suse.de>
5028 PR tree-optimization/81827
5029 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5030 (new_var_info): Initialize it conservatively.
5031 (get_call_vi): Mark register vars.
5032 (new_scalar_tmp_constraint_exp): Likewise.
5033 (handle_rhs_call): Likewise.
5034 (handle_const_call): Likewise.
5035 (create_function_info_for): Likewise.
5036 (solve_constraints): Sort varinfos to separate register from
5037 non-register vars to pack points-to solution bitmaps during
5040 2017-08-17 Marek Polacek <polacek@redhat.com>
5042 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5044 2017-08-17 Richard Biener <rguenther@suse.de>
5046 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5047 to true when overflow is undefined and we saturated the result.
5049 2017-08-17 Alan Modra <amodra@gmail.com>
5052 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5053 Don't use store multiple if only one reg needs saving.
5054 (interesting_frame_related_regno): New function.
5055 (rs6000_frame_related): Don't emit frame info for regs that
5057 (rs6000_emit_epilogue): Likewise.
5059 2017-08-16 Nathan Sidwell <nathan@acm.org>
5061 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5062 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5063 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5064 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5065 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5066 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5067 (verify_type): Adjust for TYPE_BINFO move.
5068 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5069 process TYPE_BINFO directly.
5070 (hash_tree): Likewise.
5071 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5073 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5076 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5078 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5080 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5083 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5084 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5085 Add ix86_binary_operator_ok to insn constraint.
5086 (*iordi_1_bts): Ditto.
5087 (*xordi_1_btc): Ditto.
5088 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5089 Update corresponding peephole2 pattern.
5093 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5095 PR tree-optimization/81832
5096 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5097 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5099 2017-08-16 Marek Polacek <polacek@redhat.com>
5102 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5103 regarding size_zero_node.
5105 2017-08-16 Martin Liska <mliska@suse.cz>
5108 * config.gcc: Respect previously set extra_objs in case
5111 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5113 PR tree-optimization/81835
5114 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5115 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5116 not depend on the phi.
5118 2017-08-16 Alan Modra <amodra@gmail.com>
5120 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5123 2017-08-16 Alan Modra <amodra@gmail.com>
5125 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5126 (save_reg_p): ..into this. Update all callers.
5127 (first_reg_to_save): Simplify.
5129 2017-08-16 Alan Modra <amodra@gmail.com>
5131 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5134 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5138 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5139 constant count if that count is less than 32.
5141 2017-08-15 Nathan Sidwell <nathan@acm.org>
5143 * gcc.c (execute): Emit friendlier message if inferior is killed
5144 by an external cause.
5146 2017-08-15 Richard Biener <rguenther@suse.de>
5148 PR tree-optimization/81790
5149 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5150 CONSTRUCTORs from simplifying and VN.
5152 2017-08-14 Martin Sebor <msebor@redhat.com>
5154 * builtin-attrs.def: Add comments.
5156 2017-08-14 Martin Sebor <msebor@redhat.com>
5159 * doc/extend.texi (attribute nonstring): Document new attribute.
5161 2017-08-14 Martin Sebor <msebor@redhat.com>
5164 * tree-diagnostic.c (default_tree_printer): Handle %G.
5165 * gimple-pretty-print.h (percent_G_format): Declare new function.
5166 * gimple-pretty-print.c (percent_G_format): Define.
5167 * tree-pretty-print.c (percent_K_format): Add argument.
5169 2017-08-14 Martin Sebor <msebor@redhat.com>
5171 PR translation/79998
5172 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5173 Remove a stray space.
5175 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5178 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5179 (*iordi_1_bts): Ditto.
5180 (*xordi_1_btc): Ditto.
5182 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5185 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5186 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5188 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5189 quoted strings, and make more translator-friendly.
5190 (darwin_rs6000_override_options): Likewise.
5191 (rs6000_option_override_internal): Likewise.
5192 (rs6000_return_in_memory): Fix overlong line.
5193 (init_cmulative_args): Use quoted strings, and make more
5194 translator-friendly.
5195 (rs6000_pass_by_reference): Fix overlong line.
5196 (def_builtin): Use quoted strings.
5197 (altivec_expand_predicate_builtin): Use quoted strings, and make
5198 more translator-friendly.
5199 (htm_expand_builtin): Use quoted strings.
5200 (cpu_expand_builtin): Use quoted strings, and make more
5201 translator-friendly.
5202 (altivec_expand_builtin): Likewise.
5203 (paired_expand_predicate_builtin): Likewise.
5204 (rs6000_invalid_builtin): Likewise.
5205 (builtin_function_type): Use quoted strings.
5206 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5207 more translator-friendly.
5208 (rs6000_trampoline_init): Likewise.
5209 (rs6000_handle_altivec_attribute): Likewise.
5210 (rs6000_inner_target_options): Use quoted strings.
5211 (rs6000_disable_incompatible_switches): Likewise.
5212 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5213 strings, and make more translator-friendly.
5214 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5216 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5218 PR tree-optimization/81799
5219 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5220 cond_expr to simple gimple operand.
5222 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5225 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5227 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5231 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5232 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5233 TARGET_GASISR_PROLOGUES.
5234 * config/avr/avr.c (avr_option_override): Same.
5235 (avr_pass_pre_proep::execute): Same.
5237 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5240 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5241 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5243 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5245 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5246 define_insn_and_split. Split to a memory load from 0 in
5247 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5248 using PTR mode iterator.
5249 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5250 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5251 (*add_tp_<mode>): Redefine as define_insn_and_split.
5252 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5253 address space. Merge with *add_tp_x32 using PTR mode iterator.
5254 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5255 Split to an add with a memory load from 0 in
5256 DEFAULT_TLS_SEG_REG address space.
5258 2017-08-12 Andrew Pinski <apinski@cavium.com>
5260 * config/aarch64/aarch64-option-extensions.def (rdma):
5261 Fix feature string to what Linux prints out in /proc/cpuinfo.
5263 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5266 * dwarf2out.c (modified_type_die): For C typedef types that have
5267 an ultimate origin, process the ultimate origin instead of the
5269 (gen_typedef_die): Assert that input DECLs have no ultimate
5271 (gen_type_die_with_usage): For typedef variants that have an
5272 ultimate origin, just call gen_decl_die on the original DECL.
5273 (process_scope_var): Avoid creating DIEs for local typedefs and
5274 concrete static variables.
5276 2017-08-12 Alan Modra <amodra@gmail.com>
5280 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5281 match gnu-user.h startfile.
5282 (ENDFILE_LINUX_SPEC): Similarly.
5284 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5287 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5289 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5291 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5292 * config/aarch64/aarch64.md (mov<mode>): Change.
5293 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5294 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5295 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5297 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5299 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5300 for storage order barriers.
5302 2017-08-11 Martin Liska <mliska@suse.cz>
5305 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5306 variables of void type.
5308 2017-08-11 Martin Liska <mliska@suse.cz>
5310 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5311 TARGET_SUPPORTS_ALIASES.
5312 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5313 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5314 (optimize_weakref): Likewise.
5315 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5316 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5317 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5319 2017-08-11 Martin Liska <mliska@suse.cz>
5322 * config/i386/i386.c (make_resolver_func): Do complete
5323 refactoring of the function.
5325 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5328 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5329 * config/i386/i386.c (ix86_stack_protect_guard): Use
5330 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5331 * doc/invoke.texi (x86 Options): Document
5332 -mstack-protector-guard-symbol= option.
5334 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5336 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5337 * config/i386/i386.c (ix86_split_stack_guard): New function.
5338 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5339 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5340 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5341 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5342 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5343 (split_stack_space_check): Call ix86_split_stack_guard.
5345 2017-08-10 Martin Sebor <msebor@redhat.com>
5347 * print-tree.c (print_node): Print location using the established
5349 Replace spaces with colons.
5350 (debug_raw, debug): Ditto.
5352 2017-08-10 Martin Sebor <msebor@redhat.com>
5355 * pretty-print.c (pp_format): Correct the handling of %s precision.
5357 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5360 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5362 (ix86_finalize_stack_frame_flags): This. Also clear
5363 frame_pointer_needed if -fno-omit-frame-pointer is used without
5365 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5366 with ix86_finalize_stack_frame_flags.
5367 (ix86_expand_epilogue): Likewise.
5368 (ix86_expand_split_stack_prologue): Likewise.
5369 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5371 2017-08-10 Martin Liska <mliska@suse.cz>
5374 * c-attribs.c (handle_target_attribute):
5375 Report warning for an empty string argument of target attribute.
5377 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5380 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5382 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5383 or DECL_NONLOCAL labels.
5384 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5385 or DECL_NONLOCAL labels here.
5387 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5389 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5390 to indicate when early gimple folding has been disabled.
5391 (rs6000_gimple_fold_builtin): Add debug content.
5392 (rs6000_invalid_builtin): Fix whitespace.
5393 (rs6000_expand_builtin): Fix whitespace.
5394 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5396 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5399 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5400 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5402 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5404 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5405 (qdf24xx): Likewise.
5406 * config/aarch64/aarch64.md: Include falkor.md.
5407 * config/aarch64/falkor.md: New.
5409 2017-08-09 Marek Polacek <polacek@redhat.com>
5412 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5413 * diagnostic.c (emit_diagnostic): Add a comment.
5414 (emit_diagnostic_valist): New function.
5416 2017-08-09 Marek Polacek <polacek@redhat.com>
5419 * input.c (make_location): New overload.
5420 * input.h (make_location): Declare.
5422 2017-08-08 Alan Modra <amodra@gmail.com>
5423 H.J. Lu <hongjiu.lu@intel.com>
5426 * gcc.c (NO_PIE_SPEC): Delete.
5427 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5429 (LINK_PIE_SPEC): ..to here.
5430 (LINK_COMMAND_SPEC): Support -no-pie.
5431 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5432 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5434 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5435 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5436 (ENDFILE_CRTEND_SPEC): Similarly.
5438 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5441 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5442 (mstack-protector-guard-offset=): Ditto.
5443 * config/i386/i386.c (ix86_option_override): Handle
5444 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5446 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5447 ix86_stack_protect_guard_offset variables.
5448 (TARGET_STACK_PROTECT_GUARD): Always define.
5449 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5450 and -mstack-protector-guard-offset= options.
5452 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5454 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5455 boundary case for the last candidate.
5457 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5459 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5460 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5462 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5465 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5466 * config/aarch64/aarch64-builtins.c
5467 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5468 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5469 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5471 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5472 Andrew Pinski <pinskia@gmail.com>
5475 * internal-fn.def (XORSIGN): New.
5476 * optabs.def (xorsign_optab): New.
5477 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5478 (convert_expand_mult_copysign): New.
5479 (pass_optimize_widening_mul::execute): Call
5480 convert_expand_mult_copysign.
5482 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5484 PR tree-optimization/81354
5485 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5486 Insert on edges rather than explicitly creating landing pads.
5487 (analyze_candidates_and_replace): Commit edge inserts.
5489 2017-08-08 Richard Biener <rguenther@suse.de>
5492 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5493 (expand_simple_operations): Also look through ADDR_EXPRs with
5494 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5496 2017-08-08 Richard Biener <rguenther@suse.de>
5498 PR tree-optimization/81723
5499 * tree-vect-slp.c (struct bst_traits): New hash traits.
5500 (bst_fail): New global.
5501 (vect_build_slp_tree_2): New worker, split out from ...
5502 (vect_build_slp_tree): ... this now wrapping it with using
5503 bst_fail set to cache SLP tree build fails. Properly handle
5505 (vect_analyze_slp_instance): Allocate and free bst_fail.
5507 2017-08-08 Martin Liska <mliska@suse.cz>
5510 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5511 LABEL_DECLs that can be from a different function.
5513 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5515 PR tree-optimization/81744
5516 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5517 loop's number of iterations.
5519 2017-08-08 Martin Liska <mliska@suse.cz>
5521 * asan.c: Include header files.
5522 * attribs.c (build_decl_attribute_variant): New function moved
5524 (build_type_attribute_qual_variant): Likewise.
5525 (cmp_attrib_identifiers): Likewise.
5526 (simple_cst_list_equal): Likewise.
5527 (omp_declare_simd_clauses_equal): Likewise.
5528 (attribute_value_equal): Likewise.
5529 (comp_type_attributes): Likewise.
5530 (build_type_attribute_variant): Likewise.
5531 (lookup_ident_attribute): Likewise.
5532 (remove_attribute): Likewise.
5533 (merge_attributes): Likewise.
5534 (merge_type_attributes): Likewise.
5535 (merge_decl_attributes): Likewise.
5536 (merge_dllimport_decl_attributes): Likewise.
5537 (handle_dll_attribute): Likewise.
5538 (attribute_list_equal): Likewise.
5539 (attribute_list_contained): Likewise.
5540 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5541 (lookup_attribute_by_prefix): Likewise.
5542 * bb-reorder.c: Include header files.
5543 * builtins.c: Likewise.
5544 * calls.c: Likewise.
5545 * cfgexpand.c: Likewise.
5546 * cgraph.c: Likewise.
5547 * cgraphunit.c: Likewise.
5548 * convert.c: Likewise.
5549 * dwarf2out.c: Likewise.
5550 * final.c: Likewise.
5551 * fold-const.c: Likewise.
5552 * function.c: Likewise.
5553 * gimple-expr.c: Likewise.
5554 * gimple-fold.c: Likewise.
5555 * gimple-pretty-print.c: Likewise.
5556 * gimple.c: Likewise.
5557 * gimplify.c: Likewise.
5558 * hsa-common.c: Likewise.
5559 * hsa-gen.c: Likewise.
5560 * internal-fn.c: Likewise.
5561 * ipa-chkp.c: Likewise.
5562 * ipa-cp.c: Likewise.
5563 * ipa-devirt.c: Likewise.
5564 * ipa-fnsummary.c: Likewise.
5565 * ipa-inline.c: Likewise.
5566 * ipa-visibility.c: Likewise.
5568 * lto-cgraph.c: Likewise.
5569 * omp-expand.c: Likewise.
5570 * omp-general.c: Likewise.
5571 * omp-low.c: Likewise.
5572 * omp-offload.c: Likewise.
5573 * omp-simd-clone.c: Likewise.
5574 * opts-global.c: Likewise.
5575 * passes.c: Likewise.
5576 * predict.c: Likewise.
5577 * sancov.c: Likewise.
5578 * sanopt.c: Likewise.
5579 * symtab.c: Likewise.
5580 * toplev.c: Likewise.
5581 * trans-mem.c: Likewise.
5582 * tree-chkp.c: Likewise.
5583 * tree-eh.c: Likewise.
5584 * tree-into-ssa.c: Likewise.
5585 * tree-object-size.c: Likewise.
5586 * tree-parloops.c: Likewise.
5587 * tree-profile.c: Likewise.
5588 * tree-ssa-ccp.c: Likewise.
5589 * tree-ssa-live.c: Likewise.
5590 * tree-ssa-loop.c: Likewise.
5591 * tree-ssa-sccvn.c: Likewise.
5592 * tree-ssa-structalias.c: Likewise.
5593 * tree-ssa.c: Likewise.
5594 * tree-streamer-in.c: Likewise.
5595 * tree-vectorizer.c: Likewise.
5596 * tree-vrp.c: Likewise.
5598 * ubsan.c: Likewise.
5599 * varasm.c: Likewise.
5600 * varpool.c: Likewise.
5601 * tree.c: Remove functions moved to attribs.[ch].
5603 * config/aarch64/aarch64.c: Add attrs.h header file.
5604 * config/alpha/alpha.c: Likewise.
5605 * config/arc/arc.c: Likewise.
5606 * config/arm/arm.c: Likewise.
5607 * config/avr/avr.c: Likewise.
5608 * config/bfin/bfin.c: Likewise.
5609 * config/c6x/c6x.c: Likewise.
5610 * config/cr16/cr16.c: Likewise.
5611 * config/cris/cris.c: Likewise.
5612 * config/darwin.c: Likewise.
5613 * config/epiphany/epiphany.c: Likewise.
5614 * config/fr30/fr30.c: Likewise.
5615 * config/frv/frv.c: Likewise.
5616 * config/ft32/ft32.c: Likewise.
5617 * config/h8300/h8300.c: Likewise.
5618 * config/i386/winnt.c: Likewise.
5619 * config/ia64/ia64.c: Likewise.
5620 * config/iq2000/iq2000.c: Likewise.
5621 * config/lm32/lm32.c: Likewise.
5622 * config/m32c/m32c.c: Likewise.
5623 * config/m32r/m32r.c: Likewise.
5624 * config/m68k/m68k.c: Likewise.
5625 * config/mcore/mcore.c: Likewise.
5626 * config/microblaze/microblaze.c: Likewise.
5627 * config/mips/mips.c: Likewise.
5628 * config/mmix/mmix.c: Likewise.
5629 * config/mn10300/mn10300.c: Likewise.
5630 * config/moxie/moxie.c: Likewise.
5631 * config/msp430/msp430.c: Likewise.
5632 * config/nds32/nds32-isr.c: Likewise.
5633 * config/nds32/nds32.c: Likewise.
5634 * config/nios2/nios2.c: Likewise.
5635 * config/nvptx/nvptx.c: Likewise.
5636 * config/pa/pa.c: Likewise.
5637 * config/pdp11/pdp11.c: Likewise.
5638 * config/powerpcspe/powerpcspe.c: Likewise.
5639 * config/riscv/riscv.c: Likewise.
5640 * config/rl78/rl78.c: Likewise.
5641 * config/rx/rx.c: Likewise.
5642 * config/s390/s390.c: Likewise.
5643 * config/sh/sh.c: Likewise.
5644 * config/sol2.c: Likewise.
5645 * config/sparc/sparc.c: Likewise.
5646 * config/spu/spu.c: Likewise.
5647 * config/stormy16/stormy16.c: Likewise.
5648 * config/tilegx/tilegx.c: Likewise.
5649 * config/tilepro/tilepro.c: Likewise.
5650 * config/v850/v850.c: Likewise.
5651 * config/vax/vax.c: Likewise.
5652 * config/visium/visium.c: Likewise.
5653 * config/xtensa/xtensa.c: Likewise.
5655 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5658 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5659 constraints since the -mupper-regs-* switches have been
5661 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5662 into a vector from a double word element that was extracted from
5663 another vector, and eliminate extra XXPERMDI instructions.
5664 (vsx_concat_<mode>_2): Likewise.
5665 (vsx_concat_<mode>_3): Likewise.
5666 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5667 concat to allow optimizing inserts from previous extracts.
5669 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5671 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5672 memory reference to a SSP offset in TLS address space.
5673 (ix86_print_operand) <case '@'>: Remove.
5674 (ix86_print_operand_punct_valid_p): Remove '@' code.
5675 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5677 (stack_tls_protect_set_<mode>): Remove.
5678 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5679 (stack_tls_protect_test_<mode>): Remove.
5680 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5682 2017-08-07 Olivier Hainque <hainque@adacore.com>
5685 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5687 2017-08-07 Douglas Rupp <rupp@adacore.com>
5689 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5690 variable was referenced as multidir in command.
5692 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5695 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5698 2017-08-07 Martin Liska <mliska@suse.cz>
5700 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5701 * config/rl78/rl78.c: Add include of attribs.h.
5702 * config/sh/sh.c: Likewise.
5703 * config/v850/v850.c: Likewise.
5705 2017-08-07 Tom de Vries <tom@codesourcery.com>
5708 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5710 2017-08-07 Martin Liska <mliska@suse.cz>
5712 * config/mips/mips.c: Include attribs.h.
5714 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5717 * doc/invoke.texi: Document change in behvaior for -Ofast for
5720 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5722 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5724 (aarch64_pop_regs): Likewise.
5725 (aarch64_gen_load_pair): Likewise.
5726 (aarch64_save_callee_saves): Likewise.
5727 (aarch64_restore_callee_saves): Likewise.
5729 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5731 * config/i386/i386.c: Revert the last change.
5733 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5736 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5738 (ix86_finalize_stack_frame_flags): This. Also clear
5739 frame_pointer_needed if -fno-omit-frame-pointer is used without
5741 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5742 with ix86_finalize_stack_frame_flags.
5743 (ix86_expand_epilogue): Likewise.
5744 (ix86_expand_split_stack_prologue): Likewise.
5746 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5749 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5750 to P_AES for Westmere.
5752 2017-08-07 Jonathan Yong <10walls@gmail.com>
5754 * config/i386/mingw.opt (fset-stack-executable): Removed.
5755 * config/i386/cygming.opt (fset-stack-executable): Moved
5757 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5759 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5761 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5763 2017-08-07 Marek Polacek <polacek@redhat.com>
5766 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5768 2017-08-07 Martin Liska <mliska@suse.cz>
5770 * attribs.h (canonicalize_attr_name): New function.
5771 (cmp_attribs): Move from c-format.c and adjusted.
5772 (is_attribute_p): Moved from tree.h.
5773 * tree-inline.c: Add new includes.
5774 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5775 (private_is_attribute_p): Remove.
5776 (private_lookup_attribute): Likewise.
5777 (private_lookup_attribute_by_prefix): Simplify.
5778 (remove_attribute): Use is_attribute_p.
5779 * tree.h: Remove removed declarations.
5781 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5784 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5785 instead of computing it in the function. Formatting fix.
5786 (expand_case): Don't rely on default_edge being the first edge,
5787 clear it if removing it, pass default_edge to
5788 emit_case_dispatch_table.
5789 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5792 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5794 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5795 insn in the function, emit NOP after the insn.
5797 2017-08-06 Tom de Vries <tom@codesourcery.com>
5799 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5802 2017-08-06 Tom de Vries <tom@codesourcery.com>
5804 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5807 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5809 PR tree-optimization/57371
5810 * match.pd: New pattern.
5812 2017-08-04 Marek Polacek <polacek@redhat.com>
5815 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5816 perform the computation in offset_int.
5818 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5820 PR tree-optimization/81136
5821 * tree-vectorizer.h: Include tree-hash-traits.h.
5822 (vec_base_alignments): New typedef.
5823 (vec_info): Add a base_alignments field.
5824 (vect_record_base_alignments): Declare.
5825 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5827 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5828 (create_data_ref): Add an is_conditional_in_stmt argument.
5829 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5830 the is_conditional_in_stmt field.
5831 (data_ref_loc): Add an is_conditional_in_stmt field.
5832 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5833 (find_data_references_in_stmt): Update call to create_data_ref.
5834 (graphite_find_data_references_in_stmt): Likewise.
5835 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5836 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5837 (vect_record_base_alignment): New function.
5838 (vect_record_base_alignments): Likewise.
5839 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5840 for nested statements even if we fail to compute a misalignment.
5841 Use pooled base alignments for unconditional references.
5842 (vect_find_same_alignment_drs): Compare base addresses instead
5844 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5845 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5847 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5849 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5850 Add an explicit name for the enum. Use auto_vec for slp_instances
5852 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5854 (_bb_vec_info): Add a constructor and destructor.
5855 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5856 (destroy_loop_vec_info): Delete.
5857 (vect_destroy_datarefs): Likewise.
5858 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5859 (vec_info::vec_info): New function.
5860 (vec_info::~vec_info): Likewise.
5861 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5862 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5863 destroy_loop_vec_info.
5864 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5865 (_loop_vec_info::_loop_vec_info): ...this.
5866 (destroy_loop_vec_info): Replace with...
5867 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5868 the stmt_vec_infos. Leave handling of vec_info information to its
5869 destructor. Remove explicit vector releases.
5870 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5871 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5872 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5873 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5874 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5875 (destroy_bb_vec_info): Replace with...
5876 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5877 information to its destructor.
5878 (vect_slp_analyze_bb_1): Use new and delete instead of
5879 new_bb_vec_info and destroy_bb_vec_info.
5880 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5883 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5885 * tree-data-ref.h (subscript): Add access_fn field.
5886 (data_dependence_relation): Add could_be_independent_p.
5887 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5888 (same_access_functions): Move to tree-data-ref.c.
5889 * tree-data-ref.c (ref_contains_union_access_p): New function.
5890 (access_fn_component_p): Likewise.
5891 (access_fn_components_comparable_p): Likewise.
5892 (dr_analyze_indices): Add a reference to access_fn_component_p.
5893 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5895 (constant_access_functions): Likewise.
5896 (add_other_self_distances): Likewise.
5897 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5898 (initialize_data_dependence_relation): Use XCNEW and remove
5899 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5900 of access functions that have the same type. Allow the
5901 subsequence to end with different bases in some circumstances.
5902 Record the chosen access functions in SUB_ACCESS_FN.
5903 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5904 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5905 (subscript_dependence_tester_1): Likewise dra and drb.
5906 (build_classic_dist_vector): Update calls accordingly.
5907 (subscript_dependence_tester): Likewise.
5908 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5909 DDR_COULD_BE_INDEPENDENT_P.
5910 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5911 comp_alias_ddrs instead of may_alias_ddrs.
5912 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5914 (vect_analyze_data_ref_dependence): Use it if
5915 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5916 distance vectors if that fails.
5917 (dependence_distance_ge_vf): New function.
5918 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5919 LOOP_VINFO_MAY_ALIAS_DDRS.
5921 2017-08-04 Richard Biener <rguenther@suse.de>
5924 * fold-const.c (fold_binary_loc): Properly restrict
5925 minus_var0 && minus_var1 case when associating undefined overflow
5928 2017-08-04 Tom de Vries <tom@codesourcery.com>
5930 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5932 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5934 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5935 Don't start diagnostic messages with a capital letter.
5936 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5938 (rs6000_invalid_builtin): Likewise.
5939 (rs6000_trampoline_init): Likewise.
5941 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5944 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5945 after setting changeable df flags.
5947 2017-08-03 Richard Biener <rguenther@suse.de>
5949 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5950 up if the use is in USE - X.
5952 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5954 * toplev.c (dumpfile.h): New include.
5955 (internal_error_reentered): New static function. Use it...
5956 (internal_error_function): ...here to handle reentered internal_error.
5958 2017-08-03 Richard Biener <rguenther@suse.de>
5961 * fold-const.c (split_tree): Add minus_var and minus_con
5962 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5963 here but always use minus_*.
5964 (associate_trees): Assert we never associate with MINUS_EXPR
5965 and NULL first operand. Do not recurse for PLUS_EXPR operands
5966 when associating as MINUS_EXPR either.
5967 (fold_binary_loc): Track minus_var and minus_con.
5969 2017-08-03 Tom de Vries <tom@codesourcery.com>
5972 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5973 ACCEL_COMPILER, apply finish_options on
5974 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5976 2017-08-03 Tom de Vries <tom@codesourcery.com>
5979 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5980 function_entry_patch_area_size > 0.
5982 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5985 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5986 instead of 10??LU, perform unit multiplication in wide_int,
5987 don't change alloc_object_size_limit if the limit is larger
5990 PR tree-optimization/81655
5991 PR tree-optimization/81588
5992 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5993 the case when ranges[i].low and high are 1 for unsigned type with
5997 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5998 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6000 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6002 * tree-vrp.h: Add include guard.
6004 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6007 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6008 (ud2): New insn pattern.
6009 * config/i386/i386.c (ix86_expand_epilogue):
6010 For naked functions, generate ud2 instead of trap insn.
6012 2017-08-02 Marek Polacek <polacek@redhat.com>
6015 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6017 2017-08-02 Tom de Vries <tom@codesourcery.com>
6018 Cesar Philippidis <cesar@codesourcery.com>
6020 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6021 Add missing edge probabilities.
6023 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6025 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6028 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6031 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6032 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6033 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6034 prologue_seq sequences - if any.
6036 2017-08-02 Richard Biener <rguenther@suse.de>
6038 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6039 via vectors if supported, integer extracts via punning if supported
6040 or otherwise vector extracts.
6042 2017-08-02 Richard Biener <rguenther@suse.de>
6044 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6046 (bitmap_insert_into_set): ... this.
6048 2017-08-02 Richard Biener <rguenther@suse.de>
6050 PR tree-optimization/81633
6052 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6054 PR tree-optimization/71752
6055 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6057 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6059 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6060 (machine_function::call_ms2sysv_pad_out): Remove field.
6061 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6062 (ix86_compute_frame_layout): Likewise.
6064 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6067 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6068 attribute with interrupt attribute.
6070 2017-08-01 Andrew Pinski <apinski@cavium.com>
6072 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6073 BIT_INSERT_EXPR's operand 1
6074 to see if the types precision matches.
6076 2017-08-01 Martin Liska <mliska@suse.cz>
6079 * builtins.c (expand_builtin_memcpy_args): Remove.
6080 (expand_builtin_memcpy): Call newly added function
6081 expand_builtin_memory_copy_args.
6082 (expand_builtin_memcpy_with_bounds): Likewise.
6083 (expand_builtin_mempcpy): Remove last argument.
6084 (expand_builtin_mempcpy_with_bounds): Likewise.
6085 (expand_builtin_memory_copy_args): New function created from
6086 expand_builtin_mempcpy_args with small modifications.
6087 (expand_builtin_mempcpy_args): Remove.
6088 (expand_builtin_stpcpy): Remove unused argument.
6089 (expand_builtin): Likewise.
6090 (expand_builtin_with_bounds): Likewise.
6092 2017-08-01 Martin Liska <mliska@suse.cz>
6095 Make mempcpy more optimal (PR middle-end/70140).
6097 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6100 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6101 __builtin_vec_cmpne verify both arguments are compatible vectors
6102 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6103 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6104 move computation of aligned to after checking the argument types.
6108 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6109 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6112 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6113 Doug Rupp <rupp@adacore.com>
6114 Olivier Hainque <hainque@adacore.com>
6116 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6117 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6119 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6120 for TARGET_OS_CPP_BUILTIN.
6121 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6122 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6124 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6125 passing required abi options to the assembler for EABI configurations.
6126 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6127 of .text.hot and .text.unlikely sections for kernel modules when
6128 using ARM style exceptions.
6129 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6130 options. Add EXTRA_CC1_SPEC.
6131 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6133 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6135 (ARM_TARGET2_DWARF_FORMAT): Define.
6136 * config/arm/t-vxworks: Adjust multilib control to removal of the
6137 Diab command line options.
6139 2017-08-01 Martin Liska <mliska@suse.cz>
6141 PR gcov-profile/81561
6142 * gcov.c (unblock): Make unblocking safe as we need to preserve
6143 index correspondence of blocks and block_lists.
6145 2017-08-01 Richard Biener <rguenther@suse.de>
6147 PR tree-optimization/81181
6148 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6149 (compute_antic): ... end of iteration here.
6151 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6153 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6154 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6155 (ftree-slp-vectorize): Likewise.
6156 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6157 can no longer be set independent of flag_tree_loop_vectorize.
6158 * omp-general.c (emp_max_vf): Likewise.
6159 * opts.c (enable_fdo_optimizations): Remove references to
6160 flag_tree_vectorize, these are now implicit.
6161 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6162 and leave it for the options machinery.
6164 2017-08-01 Martin Liska <mliska@suse.cz>
6167 * builtins.c (expand_builtin_memcpy_args): Remove.
6168 (expand_builtin_memcpy): Call newly added function
6169 expand_builtin_memory_copy_args.
6170 (expand_builtin_memcpy_with_bounds): Likewise.
6171 (expand_builtin_mempcpy): Remove last argument.
6172 (expand_builtin_mempcpy_with_bounds): Likewise.
6173 (expand_builtin_memory_copy_args): New function created from
6174 expand_builtin_mempcpy_args with small modifications.
6175 (expand_builtin_mempcpy_args): Remove.
6176 (expand_builtin_stpcpy): Remove unused argument.
6177 (expand_builtin): Likewise.
6178 (expand_builtin_with_bounds): Likewise.
6180 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6183 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6184 print "ds:" only for immediates in generic address space.
6186 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6189 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6190 (ix86_function_ok_for_sibcall): Return false for naked functions.
6192 2017-08-01 Richard Biener <rguenther@suse.de>
6194 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6195 (compute_antic): Seed worklist with exit block predecessors.
6196 * cfganal.c (dfs_find_deadend): For a cycle return the source
6197 of the edge closing it.
6199 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6201 * config/aarch64/aarch64.c
6202 (aarch64_can_const_movi_rtx_p): Move 0 check.
6204 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6206 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6207 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6209 * match.pd: Ditto in address comparison pattern.
6211 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6213 PR tree-optimization/81627
6214 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6215 closed ssa form for store-store chain.
6217 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6219 PR tree-optimization/81620
6220 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6221 for store-store chain.
6223 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6225 PR tree-optimization/81588
6226 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6227 ranges[i].in_p, invert comparison code ccode. For >/>=,
6228 swap rhs1 and rhs2 and comparison code unconditionally,
6229 for </<= don't do that. Don't swap rhs1/rhs2 again if
6230 ranges[i].in_p, instead invert comparison code ccode if
6231 opcode or oe->rank is BIT_IOR_EXPR.
6234 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6235 a direct optab to conversion optab.
6236 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6237 with GET_MODE_INNER as last argument instead of optab_handler.
6238 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6239 vector extraction if possible and optab is available.
6240 * expr.c (store_constructor): Use convert_optab_handler instead
6241 of optab_handler. Use vector initialization from smaller
6242 vectors if possible and optab is available.
6243 * tree-vect-stmts.c (vectorizable_load): Likewise.
6244 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6246 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6247 of vec_init from half-sized vectors with the same element mode.
6248 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6249 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6250 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6251 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6252 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6253 after mode in gen_vec_extract* calls.
6254 (vec_extract<mode>): Renamed to ...
6255 (vec_extract<mode><ssescalarmodelower>): ... this.
6256 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6257 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6258 element mode after mode in gen_vec_init* calls.
6259 (VEC_INIT_HALF_MODE): New mode iterator.
6260 (vec_init<mode>): Renamed to ...
6261 (vec_init<mode><ssescalarmodelower>): ... this.
6262 (vec_init<mode><ssehalfvecmodelower>): New expander.
6263 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6264 (vec_extractv2sfsf): ... this.
6265 (vec_initv2sf): Renamed to ...
6266 (vec_initv2sfsf): ... this.
6267 (vec_extractv2si): Renamed to ...
6268 (vec_extractv2sisi): ... this.
6269 (vec_initv2si): Renamed to ...
6270 (vec_initv2sisi): ... this.
6271 (vec_extractv4hi): Renamed to ...
6272 (vec_extractv4hihi): ... this.
6273 (vec_initv4hi): Renamed to ...
6274 (vec_initv4hihi): ... this.
6275 (vec_extractv8qi): Renamed to ...
6276 (vec_extractv8qiqi): ... this.
6277 (vec_initv8qi): Renamed to ...
6278 (vec_initv8qiqi): ... this.
6279 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6280 (vec_init<mode>): Renamed to ...
6281 (vec_init<mode><VEC_base_l>): ... this.
6282 (vec_extract<mode>): Renamed to ...
6283 (vec_extract<mode><VEC_base_l>): ... this.
6284 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6285 (vec_initv2sfsf): ... this.
6286 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6287 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6288 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6289 element mode after mode in gen_vec_init* calls.
6290 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6291 (vec_init<mode><Vel>): ... this.
6292 (vec_extract<mode>): Renamed to ...
6293 (vec_extract<mode><Vel>): ... this.
6294 * config/aarch64/iterators.md (Vel): New mode attribute.
6295 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6296 Add element mode after mode in gen_vec_extract* calls.
6297 * config/s390/vector.md (non_vec_l): New mode attribute.
6298 (vec_extract<mode>): Renamed to ...
6299 (vec_extract<mode><non_vec_l>): ... this.
6300 (vec_init<mode>): Renamed to ...
6301 (vec_init<mode><non_vec_l>): ... this.
6302 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6303 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6305 * config/arm/iterators.md (V_elem_l): New mode attribute.
6306 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6307 (vec_extract<mode><V_elem_l>): ... this.
6308 (vec_extractv2di): Renamed to ...
6309 (vec_extractv2didi): ... this.
6310 (vec_init<mode>): Renamed to ...
6311 (vec_init<mode><V_elem_l>): ... this.
6312 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6313 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6314 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6315 Add element mode after gen_vec_extract* calls.
6316 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6317 (vec_init<mode><unitmode>): ... this.
6318 (vec_extract<mode>): Renamed to ...
6319 (vec_extract<mode><unitmode>): ... this.
6320 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6321 (vec_init<mode><unitmode>): ... this.
6322 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6323 (vec_initv2sfsf): ... this.
6324 (vec_extractv2sf): Renamed to ...
6325 (vec_extractv2sfsf): ... this.
6326 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6327 Add element mode after gen_vec_extract* calls.
6328 * config/mips/mips.md (unitmode): New mode iterator.
6329 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6330 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6331 * config/spu/spu.md (inner_l): New mode attribute.
6332 (vec_init<mode>): Renamed to ...
6333 (vec_init<mode><inner_l>): ... this.
6334 (vec_extract<mode>): Renamed to ...
6335 (vec_extract<mode><inner_l>): ... this.
6336 * config/sparc/sparc.md (veltmode): New mode iterator.
6337 (vec_init<VMALL:mode>): Renamed to ...
6338 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6339 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6340 (vec_initv2sisi): ... this.
6341 (vec_initv2sf): Renamed to ...
6342 (vec_initv2sfsf): ... this.
6343 (vec_extractv2sf): Renamed to ...
6344 (vec_extractv2sfsf): ... this.
6345 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6346 (vec_init<mode>): Renamed to ...
6347 (vec_init<mode><VEC_base_l>): ... this.
6348 (vec_extract<mode>): Renamed to ...
6349 (vec_extract<mode><VEC_base_l>): ... this.
6350 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6351 (vec_initv2sfsf): ... this.
6352 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6353 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6354 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6355 gen_vec_init* calls.
6357 2017-08-01 Richard Biener <rguenther@suse.de>
6359 PR tree-optimization/81297
6360 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6361 TREE_OVERFLOW from INTEGER_CSTs.
6363 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6365 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6367 2017-07-31 Carl Love <cel@us.ibm.com>
6369 * config/rs6000/rs6000-c: Add support for built-in functions
6370 vector signed char vec_xl_be (signed long long, signed char *);
6371 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6372 vector signed int vec_xl_be (signed long long, signed int *);
6373 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6374 vector signed long long vec_xl_be (signed long long, signed long long *);
6375 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6376 vector signed short vec_xl_be (signed long long, signed short *);
6377 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6378 vector double vec_xl_be (signed long long, double *);
6379 vector float vec_xl_be (signed long long, float *);
6380 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6381 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6382 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6384 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6385 (altivec_expand_builtin): Add switch statement to call
6386 altivec_expand_xl_be for each builtin.
6387 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6388 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6389 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6390 __builtin_vsx_le_be_v16qi.
6391 * doc/extend.texi: Update the built-in documentation file for the
6392 new built-in functions.
6394 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6397 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6399 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6401 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6403 * config.gcc: Add z14.
6404 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6405 CPU model numbers for z13s and z14.
6406 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6408 * config/s390/s390-opts.h (enum processor_type): Rename
6409 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6410 * config/s390/s390.c (processor_table): Add field for CPU name to
6411 be passed to Binutils.
6412 (s390_asm_output_machine_for_arch): Use the new field in
6413 processor_table for Binutils.
6414 (s390_expand_builtin): Replace arch12 with z14.
6415 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6416 (s390_get_sched_attrmask): Likewise.
6417 (s390_get_unit_mask): Likewise.
6418 * config/s390/s390.opt: Add z14 to processor_type enum.
6420 2017-07-31 Martin Jambor <mjambor@suse.cz>
6423 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6424 regardless of optimization level.
6426 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6427 Martin Liska <mliska@suse.cz>
6429 * predict.def: Remove old comment and adjust probability.
6430 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6433 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6436 * config/i386/i386.c (ix86_function_naked): New function.
6437 (ix86_can_use_return_insn_p): Return false for naked functions.
6438 (ix86_expand_prologue): Skip prologue for naked functions.
6439 (ix86_expand_epilogue): Skip epilogue for naked functions
6440 and emit trap instruction.
6441 (ix86_warn_func_return): New function.
6442 (ix86_attribute_table): Add "naked" attribute specification.
6443 (TARGET_WARN_FUNC_RETURN): Define.
6444 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6446 2017-07-31 Martin Liska <mliska@suse.cz>
6448 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6449 (dump_gimple_bb_header): Always dump BB info.
6450 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6452 2017-07-31 Martin Liska <mliska@suse.cz>
6455 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6456 also with current_function_decl non-null equality.
6458 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6461 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6462 change type to the element type, instead add eltype variable and
6463 use it where we are interested in the element type.
6465 PR tree-optimization/81603
6466 * ipa-polymorphic-call.c
6467 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6468 offset arithmetic in offset_int, bail out if the resulting bit offset
6469 doesn't fit into shwi.
6471 2017-07-31 Martin Liska <mliska@suse.cz>
6473 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6474 (gimplify_save_expr): Fix comment.
6476 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6479 * config/i386/i386.c (ix86_function_arg): Update arguments for
6481 (ix86_compute_frame_layout): Set the initial stack offset to
6482 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6483 INCOMING_FRAME_SP_OFFSET.
6484 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6485 stack before exception handler returns.
6486 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6487 the 'ERROR_CODE' for exception handler.
6489 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6491 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6492 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6493 (ASM_OUTPUT_REG_POP): Ditto.
6494 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6495 instead of asm_fprintf to output pure string.
6497 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6499 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6500 to imported_module_or_decl hook.
6501 (debug_nothing_tree_tree_tree_bool): Remove.
6502 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6503 * debug.c (do_nothing_debug_hooks): Use
6504 debug_nothing_tree_tree_tree_bool_bool instead of
6505 debug_nothing_tree_tree_tree_bool.
6506 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6507 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6508 * sdbout.c (sdb_debug_hooks): Likewise.
6509 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6510 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6512 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6513 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6514 attribute, don't add anything.
6516 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6518 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6519 (fold_build2_stat_loc): Likewise.
6520 (fold_build3_stat_loc): Likewise.
6521 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6522 (fold_build1_loc): Remove macro.
6523 (fold_build2_loc): Likewise.
6524 (fold_build3_loc): Likewise.
6526 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6528 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6529 (gimple_build_debug_bind_source_stat): Likewise.
6530 * gimple.h (gimple_build_debug_bind): Remove macro.
6531 (gimple_build_debug_bind_source): Likewise.
6533 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6535 * bitmap.c (bitmap_alloc): Adjust.
6536 (bitmap_gc_alloc): Likewise.
6537 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6541 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6542 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6543 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6544 (bitmap_gc_alloc_stat): Likewise.
6545 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6547 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6549 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6550 * rtl.h (shallow_copy_rtx): Remove macro.
6552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6554 * emit-rtl.c (gen_raw_REG): Adjust.
6555 * gengenrtl.c (gendef): Likewise.
6556 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6557 * rtl.h (rtx_alloc): Remove macro.
6559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6561 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6562 (build_tree_list_stat): Likewise.
6563 * tree.h (build_tree_list): Remove macro.
6564 (build_tree_list_vec): Likewise.
6566 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6568 * tree.c (make_vector_stat): Remove _stat from name.
6569 (build_vector_stat): Likewise.
6570 * tree.h (make_vector_stat): Remove macro.
6571 (build_vector_stat): Likewise.
6573 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6575 * tree.h (build_var_debug_value): Remove prototype.
6577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6579 * tree.c (tree_cons_stat): Remove _stat from name.
6580 * tree.h (tree_cons): Remove macro.
6582 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6584 * tree.c (build_vl_exp_stat): Remove _stat from name.
6585 * tree.h (build_vl_exp): Remove macro.
6587 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6589 * tree.c (build_decl_stat): Remove _stat from name.
6590 * tree.h (build_decl): Remove macro.
6592 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6594 * gimple.c (gimple_build_with_ops_stat): Adjust.
6595 (gimple_alloc_stat): Remove _stat from name.
6596 * gimple.h (gimple_alloc): Remove macro.
6598 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6600 * tree.c (make_tree_vec_stat): Remove _stat from name.
6601 (grow_tree_vec_stat): Likewise.
6602 * tree.h (make_tree_vec_stat): Adjust prototype.
6603 (grow_tree_vec_stat): Likewise.
6604 (make_tree_vec): Remove macro.
6605 (grow_tree_vec): Likewise.
6607 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6609 * fold-const.c (fold_build1_stat_loc): Adjust.
6610 (fold_build2_stat_loc): Likewise.
6611 (fold_build3_stat_loc): Likewise.
6612 * tree.c (build0_stat): Remove _stat from name.
6613 (build1_stat): Likewise.
6614 (build2_stat): Likewise.
6615 (build3_stat): Likewise.
6616 (build4_stat): Likewise.
6617 (build5_stat): Likewise.
6618 * tree.h (build1_loc): Remove macro, and rename _stat function
6620 (build2_loc): Likewise.
6621 (build3_loc): Likewise.
6622 (build4_loc): Likewise.
6623 (build5_loc): Likewise.
6625 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6627 * tree.c (make_int_cst_stat): Remove _stat from name.
6628 * tree.h (make_int_cst_stat): Adjust prototype.
6629 (make_int_cst): Remove macro.
6631 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6633 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6634 * tree.h (make_tree_binfo_stat): Adjust prototype.
6635 (make_tree_binfo): Remove.
6637 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6639 * tree.c (copy_node_stat): Rename to copy_node.
6640 (build_distinct_type_copy): Adjust.
6641 * tree.h (copy_node_stat): Adjust prototype.
6642 (copy_node): Remove macro.
6644 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6646 * tree.c (make_node_stat): rename to make_node.
6647 (build_tree_list_stat): Adjust.
6648 (build0_stat): Likewise.
6649 (build2_stat): Likewise.
6650 (build3_stat): Likewise.
6651 (build4_stat): Likewise.
6652 (build5_stat): Likewise.
6653 (build_decl_stat): Likewise.
6654 * tree.h (make_node_stat): Adjust prototype.
6655 (make_node): remove macro.
6657 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6659 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6660 (PPC_FEATURE2_SCV): Likewise.
6661 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6663 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6665 * config/aarch64/aarch64.c
6666 (aarch64_internal_mov_immediate): Add new special pattern.
6667 * config/aarch64/aarch64.md (*movdi_aarch64):
6668 Add reg/32bit const mov case.
6670 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6671 Richard Sandiford <richard.sandiford@linaro.org>
6673 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6674 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6675 Add integer and movi cases.
6676 (movi-split-hf-df-sf split, fp16): New.
6677 (enabled): Added TARGET_FP_F16INST.
6678 * config/aarch64/iterators.md (GPF_HF): New.
6679 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6681 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6683 * config/aarch64/aarch64.c
6684 (aarch64_simd_container_mode): Add prototype.
6685 (aarch64_expand_mov_immediate): Add HI support.
6686 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6687 (aarch64_can_const_movi_rtx_p): New.
6688 (aarch64_preferred_reload_class):
6689 Remove restrictions of using FP registers for certain SIMD operations.
6690 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6691 (aarch64_valid_floating_const): Add integer move validation.
6692 (aarch64_simd_imm_scalar_p): Remove.
6693 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6694 (aarch64_legitimate_constant_p): Expand list of supported cases.
6695 * config/aarch64/aarch64-protos.h
6696 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6697 (aarch64_reinterpret_float_as_int): New.
6698 (aarch64_simd_imm_scalar_p): Remove.
6699 * config/aarch64/constraints.md (Uvi): New.
6700 (Dd): Split into Ds and new Dd.
6701 * config/aarch64/aarch64.md (*movsi_aarch64):
6703 (*movdi_aarch64): Add SIMD mov case.
6705 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6707 * tree-predcom.c: (struct chain): Handle store-store chain in which
6708 stores for elimination only store loop invariant values.
6709 (execute_pred_commoning_chain): Ditto.
6710 (prepare_initializers_chain_store_elim): Ditto.
6711 (prepare_finalizers): Ditto.
6712 (is_inv_store_elimination_chain): New function.
6713 (initialize_root_vars_store_elim_1): New function.
6715 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6717 * tree-predcom.c: Revise general description of the pass.
6718 (enum chain_type): New enum type for store elimination.
6719 (struct chain): New field supporting store elimination.
6720 (struct component): Ditto.
6721 (dump_chain): Dump store-stores chain.
6722 (release_chain): Release resources.
6723 (split_data_refs_to_components): Compute and create component
6724 contains only stores for elimination.
6725 (get_chain_last_ref_at): New function.
6726 (make_invariant_chain): Initialization.
6727 (make_rooted_chain): Specify chain type in parameter and record it.
6728 (add_looparound_copies): Skip for store-stores chain.
6729 (determine_roots_comp): Compute type of chain and pass it to
6731 (initialize_root_vars_store_elim_2): New function.
6732 (finalize_eliminated_stores): New function.
6733 (remove_stmt): Handle store for elimination.
6734 (execute_pred_commoning_chain): Execute predictive commoning on
6736 (determine_unroll_factor): Skip unroll for store-stores chain.
6737 (prepare_initializers_chain_store_elim): New function.
6738 (prepare_initializers_chain): Hanlde store-store chain.
6739 (prepare_finalizers_chain, prepare_finalizers): New function.
6740 (tree_predictive_commoning_loop): Return integer value indicating
6741 if loop is unrolled or lcssa form is corrupted.
6742 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6744 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6746 * tree-predcom.c (initialize_root): Delete.
6747 (execute_pred_commoning_chain): Initialize root vars and replace
6748 reference of non-combined chain directly, rather than call above
6751 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6753 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6754 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6756 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6758 * tree-predcom.c (struct chain): New field init_seq.
6759 (release_chain): Release init_seq.
6760 (prepare_initializers_chain): Record intialization stmts in above
6762 (insert_init_seqs): New function.
6763 (tree_predictive_commoning_loop): Call insert_init_seqs.
6765 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6767 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6769 2017-07-28 Richard Biener <rguenther@suse.de>
6771 * match.pd: Remove superfluous :c.
6772 * genmatch.c (simplify::id): Add member.
6773 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6775 (current_id): New global.
6776 (dt_node::parent): Move from ...
6777 (dt_operand::parent): ... here. Add for_id member.
6778 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6779 (decision_tree::find_node): Relax order requirement when
6780 merging DT_TRUE nodes to ones inbetween the current simplify
6781 and the one we try to merge with. Add diagnostic whenever
6782 we need to enforce pattern order by not merging.
6783 (decision_tree::insert): Set current_id.
6784 (decision_tree::print_node): Dump parent node and for_id.
6785 (parser::last_id): Add member.
6786 (parser::push_simplify): Assign unique id.
6787 (parser::parser): Initialize last_id.
6789 2017-07-28 Martin Liska <mliska@suse.cz>
6792 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6793 gimple_build_debug_bind.
6795 2017-07-28 Richard Biener <rguenther@suse.de>
6797 PR tree-optimization/81502
6798 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6800 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6802 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6803 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6804 for BIT_FIELD_REF args.
6805 * fold-const.c (make_bit_field_ref): Likewise.
6806 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6808 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6811 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6812 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6813 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6814 Or it into SANITIZER_UNDEFINED.
6815 * ubsan.c: Include gimple-fold.h and varasm.h.
6816 (ubsan_expand_ptr_ifn): New function.
6817 (instrument_pointer_overflow): New function.
6818 (maybe_instrument_pointer_overflow): New function.
6819 (instrument_object_size): Formatting fix.
6820 (pass_ubsan::execute): Call instrument_pointer_overflow
6821 and maybe_instrument_pointer_overflow.
6822 * internal-fn.c (expand_UBSAN_PTR): New function.
6823 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6824 * sanitizer.def (__ubsan_handle_pointer_overflow,
6825 __ubsan_handle_pointer_overflow_abort): New builtins.
6826 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6827 * internal-fn.def (UBSAN_PTR): New internal function.
6828 * opts.c (sanitizer_opts): Add pointer-overflow.
6829 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6830 * fold-const.c (build_range_check): Compute pointer range check in
6831 integral type if pointer arithmetics would be needed. Formatting
6834 2017-07-28 Martin Liska <mliska@suse.cz>
6837 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6838 parameters that are of a variable-length.
6840 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6842 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6844 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6845 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6846 (CRT_CALL_STATIC_FUNCTION): Likewise.
6847 (ASM_DEFAULT_SPEC): New define.
6848 (ASM_SPEC32): Likewise.
6849 (ASM_SPEC64): Likewise.
6850 (ASM_SPEC_COMMON): Likewise.
6851 (ASM_SPEC): Likewise.
6852 (INVALID_64BIT): Likewise.
6853 (LINK_OS_DEFAULT_SPEC): Likewise.
6854 (LINK_OS_SPEC32): Likewise.
6855 (LINK_OS_SPEC64): Likewise.
6856 (POWERPC_LINUX): Likewise.
6857 (PTRDIFF_TYPE): Likewise.
6858 (RESTORE_FP_PREFIX): Likewise.
6859 (RESTORE_FP_SUFFIX): Likewise.
6860 (SAVE_FP_PREFIX): Likewise.
6861 (SAVE_FP_SUFFIX): Likewise.
6862 (SIZE_TYPE): Likewise.
6863 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6864 (TARGET_64BIT): Likewise.
6865 (TARGET_64BIT): Likewise.
6866 (TARGET_AIX): Likewise.
6867 (WCHAR_TYPE_SIZE): Likewise.
6868 (WCHAR_TYPE): Undefine.
6869 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6870 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6871 (CPP_OS_RTEMS_SPEC): Delete.
6872 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6873 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6875 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6877 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6879 PR tree-optimization/81578
6880 * tree-parloops.c (build_new_reduction): Bail out if
6881 reduction_code isn't one of the standard OpenMP reductions.
6882 Move the details printing after that decision.
6884 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6886 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6887 related to reload_in_progress.
6888 (splat_input_operand): Likewise.
6889 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6891 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6893 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6894 (TARGET_INSTANTIATE_DECLS): Likewise.
6895 (legitimate_indexed_address_p): Delete reload_in_progress code.
6896 (rs6000_debug_legitimate_address_p): Likewise.
6897 (rs6000_eliminate_indexed_memrefs): Likewise.
6898 (rs6000_emit_le_vsx_store): Likewise.
6899 (rs6000_emit_move_si_sf_subreg): Likewise.
6900 (rs6000_emit_move): Likewise.
6901 (register_to_reg_type): Likewise.
6902 (rs6000_pre_atomic_barrier): Likewise.
6903 (rs6000_machopic_legitimize_pic_address): Likewise.
6904 (rs6000_allocate_stack_temp): Likewise.
6905 (rs6000_address_for_fpconvert): Likewise.
6906 (rs6000_address_for_altivec): Likewise.
6907 (rs6000_secondary_memory_needed_rtx): Delete function.
6908 (rs6000_check_sdmode): Likewise.
6909 (rs6000_alloc_sdmode_stack_slot): Likewise.
6910 (rs6000_instantiate_decls): Likewise.
6911 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6912 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6913 Delete reload_in_progress.
6914 (*vec_reload_and_plus_<mptrsize>): Likewise.
6915 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6916 (vsx_div_v2di): Likewise.
6917 (vsx_udiv_v2di): Likewise.
6919 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6921 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6922 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6923 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6924 (rs6000_debug_reg_global): Delete print of LRA status.
6925 (rs6000_option_override_internal): Delete dead LRA related code.
6926 (rs6000_lra_p): Delete function.
6927 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6929 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6931 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6932 * config/riscv/rtems.h: New file.
6934 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6935 Sudakshina Das <sudi.das@arm.com>
6937 * config/aarch64/aarch64.md
6938 (define_split for and<mode>3nr_compare): Move
6939 non aarch64_logical_operand to a register.
6940 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6941 register immediate operand to a register.
6942 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6944 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6947 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6949 2017-07-27 Richard Biener <rguenther@suse.de>
6951 PR tree-optimization/81573
6952 PR tree-optimization/81494
6953 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6954 multi defuse cycle case.
6956 2017-07-27 Richard Biener <rguenther@suse.de>
6958 PR tree-optimization/81571
6959 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6962 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6964 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6965 earlier and only if MASK_FPU is set. Adjust formatting.
6967 2017-07-27 Martin Liska <mliska@suse.cz>
6969 * opt-functions.awk: Add validation of value of Init.
6970 * optc-gen.awk: Pass new argument.
6972 2017-07-27 Martin Liska <mliska@suse.cz>
6974 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6975 Fix wrong condition.
6977 2017-07-27 Martin Liska <mliska@suse.cz>
6979 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6980 BBs and edges seen by autoFDO.
6982 2017-07-27 Richard Biener <rguenther@suse.de>
6984 PR tree-optimization/81502
6985 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6986 with incompatible but same sized type.
6987 (execute_update_addresses_taken): Likewise.
6989 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6991 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6992 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6994 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6997 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6998 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6999 Change s_operand to memory_operand.
7001 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7003 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7004 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7005 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7006 Emit instructions rather than returning an expression. Handle TFmode
7007 and KFmode by casting to TImode.
7008 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7009 (rs6000_emit_le_vsx_store): Likewise.
7010 * config/rs6000/vsx.md (VSX_TI): New iterator.
7011 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7012 (*vsx_le_undo_permute_<mode>): Likewise.
7013 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7014 emit the split sequence.
7015 (*vsx_le_perm_store_<mode>): Likewise.
7017 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7019 PR tree-optimization/81555
7020 PR tree-optimization/81556
7021 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7022 if true, force CHANGED for the recursive invocation.
7023 (reassociate_bb): Remember original length of ops array, pass
7024 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7026 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7027 attributes for noipa attribute. For naked attribute use
7028 lookup_attribute first before lookup_attribute_spec.
7029 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7031 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7032 for functions with noipa attribute.
7033 (cgraph_externally_visible_p): Return true for functions with noipa
7035 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7036 for functions with noipa attribute.
7037 * doc/extend.texi: Document noipa function attribute.
7038 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7039 also for functions with noipa attribute.
7040 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7042 2017-07-26 Andrew Pinski <apinski@cavium.com>
7044 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7045 vec_unalign_load_cost and vec_unalign_store_cost.
7047 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7049 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7050 -mvsx-small-integer option.
7051 (ISA_3_0_MASKS_IEEE): Likewise.
7052 (OTHER_VSX_VECTOR_MASKS): Likewise.
7053 (POWERPC_MASKS): Likewise.
7054 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7055 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7056 code, only testing for DImode being allowed in non-VSX floating
7058 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7059 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7061 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7062 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7063 TARGET_P8_VECTOR test.
7064 (rs6000_secondary_reload_simple_move): Likewise.
7065 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7066 since TARGET_P9_VECTOR was already tested.
7067 (rs6000_opt_masks): Remove -mvsx-small-integer.
7068 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7069 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7071 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7072 test for TARGET_VEXTRACTUB was used, and that uses
7074 (p9 extract splitter): Likewise.
7075 (vsx_extract_<mode>_di_p9): Likewise.
7076 (vsx_extract_<mode>_store_p9): Likewise.
7077 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7078 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7079 the elimination of TARGET_VSX_SMALL_INTEGER.
7080 (vsx_extract_<mode>_p8): Likewise.
7081 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7082 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7083 (vsx_set_<mode>_p9): Likewise.
7084 (vsx_set_v4sf_p9): Likewise.
7085 (vsx_set_v4sf_p9_zero): Likewise.
7086 (vsx_insert_extract_v4sf_p9): Likewise.
7087 (vsx_insert_extract_v4sf_p9_2): Likewise.
7088 * config/rs6000/rs6000.md (sign extend splitter): Change
7089 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7090 (floatsi<mode>2_lfiwax_mem): Likewise.
7091 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7092 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7093 since a test for TARGET_P9_VECTOR was used.
7094 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7095 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7096 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7097 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7098 TARGET_P8_VECTOR test.
7099 (fix_trunc<mode>si2_stfiwx): Likewise.
7100 (fix_trunc<mode>si2_internal): Likewise.
7101 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7102 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7104 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7105 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7106 TARGET_P8_VECTOR test.
7107 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7108 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7109 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7111 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7112 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7113 since a test for TARGET_P9_VECTOR was used.
7114 (splitter for loading small constants): Likewise.
7116 2017-07-26 Andrew Pinski <apinski@cavium.com>
7118 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7121 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7124 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7125 (fp_valid_at): Likewise.
7127 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7129 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7130 (qdf24xx_addrcost_table): Likewise.
7131 (cortexa57_tunings): Update to use generic_branch_cost.
7132 (cortexa72_tunings): Likewise.
7133 (cortexa73_tunings): Likewise.
7134 (qdf24xx_tunings): Likewise.
7136 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7138 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7139 (thunderx2t99_branch_cost): Likewise.
7140 (cortexa35_tunings): Update to use generic_branch_cost.
7141 (cortexa53_tunings): Likewise.
7142 (cortexa57_tunings): Likewise.
7143 (cortexa72_tunings): Likewise.
7144 (cortexa73_tunings): Likewise.
7145 (thunderx2t99_tunings): Likewise.
7147 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7149 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7150 (sparc_option_override): Honour MASK_FSMULD.
7151 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7152 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7153 * config/sparc/sparc.opt (mfsmuld): New option.
7154 * doc/invoke.texi (mfsmuld): Document option.
7156 2017-07-26 Marek Polacek <polacek@redhat.com>
7159 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7161 2017-07-26 Richard Biener <rguenther@suse.de>
7163 * gimple-match-head.c (do_valueize): Return OP if valueize
7165 (get_def): New helper to get at the def stmt of a SSA name
7167 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7168 do_valueize to get at the def stmt.
7169 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7171 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7174 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7176 2017-07-26 Martin Liska <mliska@suse.cz>
7179 * function.c (expand_function_start): Make expansion of
7180 nonlocal_goto_save_area after parm_birth_insn.
7182 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7184 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7185 from all CPU target flags enable members.
7187 2017-07-26 Richard Biener <rguenther@suse.de>
7189 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7190 (decision_tree::gen): Make 'type' const.
7191 (write_predicate): Likewise.
7193 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7195 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7196 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7197 (rs6000_option_override_internal): Likewise.
7198 (rs6000_expand_vector_set): Likewise.
7199 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7200 (TARGET_UPPER_REGS_SF): Likewise.
7201 (TARGET_UPPER_REGS_DI): Likewise.
7202 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7203 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7204 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7205 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7206 (Splitters for DI constants in Altivec registers): Likewise.
7207 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7208 (vsx_set_v4sf_p9): Likewise.
7209 (vsx_set_v4sf_p9_zero): Likewise.
7210 (vsx_insert_extract_v4sf_p9): Likewise.
7211 (vsx_insert_extract_v4sf_p9_2): Likewise.
7213 2017-07-25 Carl Love <cel@us.ibm.com>
7215 * doc/extend.texi: Update the built-in documentation file for the
7216 existing built-in functions
7217 vector signed char vec_cnttz (vector signed char);
7218 vector unsigned char vec_cnttz (vector unsigned char);
7219 vector signed short vec_cnttz (vector signed short);
7220 vector unsigned short vec_cnttz (vector unsigned short);
7221 vector signed int vec_cnttz (vector signed int);
7222 vector unsigned int vec_cnttz (vector unsigned int);
7223 vector signed long long vec_cnttz (vector signed long long);
7224 vector unsigned long long vec_cnttz (vector unsigned long long);
7226 2017-07-25 Andrew Pinski <apinski@cavium.com>
7228 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7229 accesses where the use is for the first operand of a BIT_INSERT.
7231 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7234 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7235 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7237 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7239 * config/i386/gstabs.h: Delete.
7240 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7242 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7244 * config/i386/i386.c (ix86_decompose_address): Do not check for
7245 register RTX when looking at index_reg or base_reg.
7246 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7248 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7250 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7251 to update EH info here.
7253 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7255 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7257 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7259 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7261 2017-07-25 Torsten Duwe <duwe@suse.de>
7263 * common.opt: Introduce -fpatchable-function-entry
7264 command line option, and its variables function_entry_patch_area_size
7265 and function_entry_patch_area_start.
7266 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7267 including a two-value parser.
7268 * target.def (print_patchable_function_entry): New target hook.
7269 * targhooks.h (default_print_patchable_function_entry): New function.
7270 * targhooks.c (default_print_patchable_function_entry): Likewise.
7271 * toplev.c (process_options): Switch off IPA-RA if
7272 patchable function entries are being generated.
7273 * varasm.c (assemble_start_function): Look at the
7274 patchable-function-entry command line switch and current
7275 function attributes and maybe generate NOP instructions by
7276 calling the print_patchable_function_entry hook.
7277 * doc/extend.texi: Document patchable_function_entry attribute.
7278 * doc/invoke.texi: Document -fpatchable_function_entry
7279 command line option.
7280 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7282 * doc/tm.texi: Re-generate.
7284 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7287 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7288 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7290 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7292 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7294 2017-07-25 Richard Biener <rguenther@suse.de>
7296 PR tree-optimization/81455
7297 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7298 not walk in cycles when looking for guards.
7300 2017-07-25 Richard Biener <rguenther@suse.de>
7302 PR tree-optimization/81529
7303 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7304 when optimizing backedge uses.
7306 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7308 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7310 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7311 to dl_section_ref. On AIX, append an expression to subtract
7312 the size of the section length to dl_section_ref.
7314 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7316 * configure.ac: If any of the config.* scripts fail, exit 1.
7317 * configure: Regenerate.
7319 2017-07-25 Richard Biener <rguenther@suse.de>
7322 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7323 of immediate uses, be more verbose on errors.
7325 2017-07-25 Richard Biener <rguenther@suse.de>
7327 PR tree-optimization/81510
7328 * tree-vect-loop.c (vect_is_simple_reduction): When the
7329 reduction stmt is not inside the loop bail out.
7331 2017-07-25 Richard Biener <rguenther@suse.de>
7333 PR tree-optimization/81303
7334 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7335 profitability check against LOOP_VINFO_NITERSM1.
7337 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7339 * domwalk.c (cmp_bb_postorder): Simplify.
7340 (sort_bbs_postorder): New function. Use it...
7341 (dom_walker::walk): ...here to optimize common cases.
7343 2017-07-25 Martin Liska <mliska@suse.cz>
7346 * ipa-visibility.c (function_and_variable_visibility): Make the
7347 redirection just on target that supports aliasing.
7348 Fix GNU coding style.
7350 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7353 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7355 * config/arm/rtems.h: Likewise.
7356 * config/bfin/rtems.h: Likewise.
7357 * config/i386/rtemself.h: Likewise.
7358 * config/lm32/rtems.h: Likewise.
7359 * config/m32c/rtems.h: Likewise.
7360 * config/m68k/rtemself.h: Likewise.
7361 * config/microblaze/rtems.h: Likewise.
7362 * config/mips/rtems.h: Likewise.
7363 * config/moxie/rtems.h: Likewise.
7364 * config/nios2/rtems.h: Likewise.
7365 * config/powerpcspe/rtems.h: Likewise.
7366 * config/rs6000/rtems.h: Likewise.
7367 * config/rtems.h: Likewise.
7368 * config/sh/rtems.h: Likewise.
7369 * config/sh/rtemself.h: Likewise.
7370 * config/sparc/rtemself.h: Likewise.
7372 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7375 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7376 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7377 * tree-ssa-structalias.c (alias_get_name): Same.
7379 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7382 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7383 instructions if no du chain is found.
7385 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7387 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7389 2017-07-25 Richard Biener <rguenther@suse.de>
7392 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7395 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7397 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7399 (ISA_2_7_MASKS_SERVER): Likewise.
7400 (ISA_3_0_MASKS_IEEE): Likewise.
7401 (OTHER_P8_VECTOR_MASKS): Likewise.
7402 (OTHER_VSX_VECTOR_MASKS): Likewise.
7403 (POWERPC_MASKS): Likewise.
7404 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7405 duplicate list of options.
7406 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7407 explicit -mupper-regs options.
7408 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7409 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7410 alias for -mupper-regs-df.
7411 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7412 (rs6000_init_hard_regno_mode_ok): Likewise.
7413 (rs6000_option_override_internal): Likewise.
7414 (rs6000_opt_masks): Likewise.
7415 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7416 options in terms of whether -mvsx or -mpower8-vector was used.
7417 (TARGET_UPPER_REGS_DI): Likewise.
7418 (TARGET_UPPER_REGS_SF): Likewise.
7419 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7420 -mupper-regs-* options.
7422 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7424 * passes.c (emergency_dump_function): Print some empty lines and a
7425 header before the RTL dump.
7427 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7429 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7431 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7434 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7435 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7437 2017-07-24 Carl Love <cel@us.ibm.com>
7439 * config/rs6000/rs6000-c.c: Add support for built-in functions
7440 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7441 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7442 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7443 vec_extract_fp_from_shortl): Add defines for the two builtins.
7444 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7445 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7447 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7448 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7449 * doc/extend.texi: Update the built-in documentation file for the
7450 new built-in function.
7452 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7455 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7457 * doc/generic.texi: Likewise.
7458 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7459 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7461 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7463 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7464 (aarch64_mls_elt_merge<mode>): Likewise.
7466 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7468 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7469 having __cxa_atexit.
7471 2017-07-23 Michael Collison <michael.collison@arm.com>
7473 * config/arm/arm.c (arm_option_override): Deprecate
7474 use of -mstructure-size-boundary.
7475 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7476 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7478 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7481 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7482 Reduce cost estimate for direct moves.
7484 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7487 * config/i386/i386.c (ix86_option_override_internal): Disable
7488 BMI, BMI2 and TBM instructions for -m16.
7490 2017-07-21 Carl Love <cel@us.ibm.com>
7492 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7493 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7494 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7495 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7496 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7497 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7498 VMULOSW): New enum "unspec" values.
7499 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7500 altivec_vmulosw): New patterns.
7501 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7502 VMULOSW): Add definitions.
7504 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7506 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7507 (qdf24xx): Likewise.
7508 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7509 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7510 (AARCH64_FL_V8_1): Renumber.
7511 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7512 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7513 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7514 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7515 rdma to feature modifiers list.
7517 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7520 * ipa-visibility (function_and_variable_visibility): Convert
7521 recursive PLT call to direct call if appropriate.
7523 2017-07-21 Andrew Pinski <apinski@cavium.com>
7525 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7526 operand 1 to see if the types precision matches.
7527 * fold-const.c (operand_equal_p): Likewise.
7529 2017-07-21 Richard Biener <rguenther@suse.de>
7531 PR tree-optimization/81303
7532 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7533 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7534 (vect_peeling_hash_get_lowest_cost): Adjust.
7535 (vect_enhance_data_refs_alignment): Likewise. Use
7536 vect_get_peeling_costs_all_drs to compute the penalty for no
7537 peeling to match up costs.
7539 2017-07-21 Richard Biener <rguenther@suse.de>
7541 PR tree-optimization/81500
7542 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7543 we didn't identify a reduction path.
7545 2017-07-21 Tom de Vries <tom@codesourcery.com>
7546 Cesar Philippidis <cesar@codesourcery.com>
7548 PR gcov-profile/81442
7549 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7552 2017-07-21 Tom de Vries <tom@codesourcery.com>
7555 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7557 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7558 nvptx_override_options_after_change.
7560 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7562 * dwarf2out.c (output_file_names): Avoid double testing for
7565 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7567 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7569 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7571 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7573 (try_crossjump_to_edge): Do not punt on partitioned functions.
7575 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7577 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7578 Put all BBs reachable only via paths crossing cold region to cold
7580 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7582 2016-07-21 Richard Biener <rguenther@suse.de>
7584 PR tree-optimization/81303
7585 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7586 into account prologue and epilogue iterations when raising
7587 min_profitable_iters to sth at least covering one vector iteration.
7589 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7591 * config/arm/arm.c (arm_test_cpu_arch_dat):
7594 2017-07-20 Nathan Sidwell <nathan@acm.org>
7596 Remove TYPE_METHODS.
7597 * tree.h (TYPE_METHODS): Delete.
7598 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7599 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7600 (dbxout_type_methods): Scan TYPE_FIELDS.
7601 (dbxout_type): Don't check TYPE_METHODS here.
7602 * function.c (use_register_for_decl): Always ignore register for
7603 class types when not optimizing.
7604 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7605 * tree.c (free_lang_data_in_type): Stitch out member functions and
7606 templates from TYPE_FIELDS.
7607 (build_distinct_type_copy, verify_type_variant,
7608 verify_type): Member fns are on TYPE_FIELDS.
7609 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7610 * tree-pretty-print.c (dump_generic_node): Likewise.
7612 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7615 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7616 V2TImode and V4TImode.
7617 (ix86_expand_vector_extract): Likewise.
7618 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7619 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7620 (ssescalarmode): Handle V4TImode and V2TImode.
7621 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7622 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7623 (VEXTRACTI128_MODE): New mode iterator.
7624 (splitter for *vec_extractv?ti first element): New.
7625 (VEC_INIT_MODE): New mode iterator.
7626 (vec_init<mode>): Consolidate 3 expanders into one using
7627 VEC_INIT_MODE mode iterator.
7629 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7631 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7632 non_spilled_static_chain_regno_p.
7634 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7636 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7638 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7640 * bb-reorder.c (connect_traces): Allow copying of blocks within
7643 2017-07-20 Richard Biener <rguenther@suse.de>
7645 * gimple.h (gimple_phi_result): Add gphi * overload.
7646 (gimple_phi_result_ptr): Likewise.
7647 (gimple_phi_arg): Likewise. Adjust index assert to only
7648 allow actual argument accesses rather than all slots available
7650 (gimple_phi_arg_def): Add gphi * overload.
7651 * tree-phinodes.c (make_phi_node): Initialize only actual
7653 (resize_phi_node): Clear memory not covered by old node,
7654 do not initialize excess argument slots.
7655 (reserve_phi_args_for_new_edge): Initialize new argument slot
7658 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7660 PR tree-optimization/81388
7662 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7664 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7665 by removing computation of may_be_zero.
7667 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7668 Tom de Vries <tom@codesourcery.com>
7671 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7672 when gimple level profile disagrees with what RTL expander did.
7674 2017-07-20 Richard Biener <rguenther@suse.de>
7676 PR tree-optimization/61171
7677 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7678 (vect_analyze_stmt): Add slp instance parameter.
7679 (vectorizable_reduction): Likewise.
7680 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7681 (vect_is_simple_reduction): Deal with chains not detected
7682 as SLP reduction chain, specifically not properly associated
7683 chains containing a mix of plus/minus.
7684 (get_reduction_op): Remove.
7685 (get_initial_defs_for_reduction): Simplify, pass in whether
7686 this is a reduction chain, pass in the SLP node for the PHIs.
7687 (vect_create_epilog_for_reduction): Get the SLP instance as
7689 (vectorizable_reduction): Get the SLP instance as arg.
7690 During analysis remember the SLP node with the PHIs in the
7691 instance. Simplify getting at the vectorized reduction PHIs.
7692 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7693 through SLP instance.
7694 (vect_slp_analyze_operations): Likewise.
7695 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7696 (vect_transform_stmt): Likewise.
7698 2017-07-20 Tom de Vries <tom@codesourcery.com>
7700 PR tree-optimization/81489
7701 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7702 read of phi arg location to before loop that modifies phi.
7704 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7706 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7709 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7712 * except.c (execute): Fix ordering issue.
7714 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7716 PR rtl-optimization/81423
7717 * combine.c (make_compound_operation_int): Don't try to optimize
7718 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7720 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7722 PR rtl-optimization/81423
7723 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7724 with a constant that is -1 in the truncated to mode.
7726 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7728 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7729 (determine_unlikely_bbs): ... here.
7730 * predict.h (propagate_unlikely_bbs_forward): Declare.
7731 * cfgexpand.c (pass_expand::execute): Use it.
7732 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7734 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7735 propagate_unlikely_bbs_forward.
7737 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7740 * except.c (maybe_add_nop_after_section_switch): New function.
7743 2017-07-19 Tom de Vries <tom@codesourcery.com>
7745 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7747 2017-07-19 Tom de Vries <tom@codesourcery.com>
7749 * gimple.h (gimple_phi_arg): Make assert more strict.
7751 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7753 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7754 * config/rs6000/mmintrin.h: New file.
7755 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7757 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7759 PR tree-optimization/81346
7760 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7762 2017-07-19 Tom de Vries <tom@codesourcery.com>
7764 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7766 2017-07-19 Tom de Vries <tom@codesourcery.com>
7768 * config/nvptx/nvptx-modes.def: Add V2DImode.
7769 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7770 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7771 (nvptx_output_mov_insn): Handle lack of mov.b128.
7772 (nvptx_print_operand): Handle 'H' and 'L' codes.
7773 (nvptx_vector_mode_supported): Allow V2DImode.
7774 (nvptx_preferred_simd_mode): New function.
7775 (nvptx_data_alignment): New function.
7776 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7777 nvptx_preferred_simd_mode.
7778 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7780 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7782 2017-07-19 Tom de Vries <tom@codesourcery.com>
7784 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7785 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7786 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7787 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7788 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7789 (mov<VECIM>_insn): New define_insn.
7790 (define_expand "mov<VECIM>): New define_expand.
7792 2017-07-19 Tom de Vries <tom@codesourcery.com>
7794 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7796 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7798 PR tree-optimization/81346
7799 * fold-const.h (fold_div_compare, range_check_type): Declare.
7800 * fold-const.c (range_check_type): New function.
7801 (build_range_check): Use range_check_type.
7802 (fold_div_compare): No longer static, rewritten into
7803 a match.pd helper function.
7804 (fold_comparison): Don't call fold_div_compare here.
7805 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7808 2017-07-19 Nathan Sidwell <nathan@acm.org>
7810 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7811 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7812 * tree.c (find_decls_types_r, verify_type): Use
7813 TYPE_{MIN,MAX}_VALUE_RAW.
7814 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7815 (hash_tree): Likewise.
7816 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7818 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7821 2017-07-18 Tom de Vries <tom@codesourcery.com>
7824 * omp-expand.c (expand_omp_for_static_chunk): Handle
7825 equal-argument loop exit phi.
7827 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7830 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7831 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7832 operand 2 predicate.
7833 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7834 operand 2 predicate.
7835 (ror,rol -> rorx splitters): Use const_int_operand as
7836 operand 2 predicate.
7838 2017-06-18 Richard Biener <rguenther@suse.de>
7840 PR tree-optimization/81410
7841 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7842 the gap in the ! slp_perm SLP case after each group.
7844 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7847 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7850 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7853 * predict.c (set_even_probabilities): Cleanup; do not affect
7854 probabilities that are already known.
7855 (combine_predictions_for_bb): Call even when count is set.
7857 2017-07-18 Nathan Sidwell <nathan@acm.org>
7859 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7862 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7865 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7866 optimization for loop niter analysis.
7868 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7871 * config/avr/avr.c (avr_optimize_casesi): Don't use
7872 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7874 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7876 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7877 body_cost_vec from _vect_peel_extended_info.
7878 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7879 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7882 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7884 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7886 2017-07-18 Richard Biener <rguenther@suse.de>
7888 PR tree-optimization/80620
7889 PR tree-optimization/81403
7890 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7891 info when re-using a VN table entry.
7893 2017-07-18 Richard Biener <rguenther@suse.de>
7895 PR tree-optimization/81418
7896 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7897 vectype_in. Verify that with lane-reducing reduction operations
7898 we have a single def-use cycle.
7900 2017-07-17 Carl Love <cel@us.ibm.com>
7902 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7904 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7905 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7906 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7908 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7909 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7910 VMULOSW): New enum "unspec" values.
7911 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7912 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7913 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7914 altivec_vmulosw): New patterns.
7915 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7916 VMULOSW): Add definitions.
7917 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7919 * config/alpha/alpha.c: Include predict.h.
7921 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7923 * tree-vrp.c (compare_assert_loc): Fix comparison function
7924 to return predictable results.
7926 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7928 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7931 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7932 * doc/invoke.texi (mexpand-adddi): Update text.
7934 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7936 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7937 that also clobbers the CC register. The old expand code is moved
7939 (*arc_clzsi2): ... here.
7940 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7941 the CC register. The old expand code is moved to ...
7942 (arc_ctzsi2): ... here.
7944 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7946 * config/arc/arc.opt (mindexed-loads): Use initial value
7947 TARGET_INDEXED_LOADS_DEFAULT.
7948 (mauto-modify-reg): Use initial value
7949 TARGET_AUTO_MODIFY_REG_DEFAULT.
7950 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7951 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7952 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7953 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7955 2017-07-17 Martin Liska <mliska@suse.cz>
7958 * opts.c (finish_options): Do not allow -fgnu-tm
7959 w/ -fsanitize={kernel-,}address. Say sorry.
7961 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7964 * tree-loop-distribution.c (classify_partition): Only assert on
7965 numer of iterations.
7966 (merge_dep_scc_partitions): Delete prameter. Update function call.
7967 (distribute_loop): Remove code handling loop with unknown niters.
7968 (pass_loop_distribution::execute): Skip loop with unknown niters.
7970 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7973 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7974 function sort_partitions_by_post_order.
7976 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7978 PR tree-optimization/81374
7979 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7980 the max index of basic blocks, rather than number of basic blocks.
7982 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7984 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7986 (arc_legitimate_pic_operand_p): Likewise.
7987 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7989 (arc_needs_pcl_p): Likewise.
7990 (arc_legitimate_pc_offset_p): Likewise.
7991 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7992 function is also used in constrains.md.
7993 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7994 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7995 PLUS. Only return true/false in known cases, otherwise assert.
7996 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7997 is already called in arc_legitimate_constant_p.
7998 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8000 (LEGITIMATE_PIC_OPERAND_P): Use
8001 arc_raw_symbolic_reference_mentioned_p function.
8002 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8007 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8008 Andrew Burgess <andrew.burgess@embecosm.com>
8010 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8011 (arc_return_address_register): New function.
8012 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8013 (arc_handle_fndecl_attribute): Add naked attribute.
8014 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8015 (TARGET_WARN_FUNC_RETURN): Likewise.
8016 (arc_allocate_stack_slots_for_args): New function.
8017 (arc_warn_func_return): Likewise.
8018 (machine_function): Change type fn_type.
8019 (arc_compute_function_type): Consider new naked function type,
8020 change function return type.
8021 (arc_must_save_register): Adapt to handle new
8022 arc_compute_function_type's return type.
8023 (arc_expand_prologue): Likewise.
8024 (arc_expand_epilogue): Likewise.
8025 (arc_return_address_regs): Delete.
8026 (arc_return_address_register): New function.
8027 (arc_epilogue_uses): Use above function.
8028 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8029 (arc_function_type): Change encoding, add naked type.
8030 (ARC_INTERRUPT_P): Change to handle the new encoding.
8031 (ARC_FAST_INTERRUPT_P): Likewise.
8032 (ARC_NORMAL_P): Define.
8033 (ARC_NAKED_P): Likewise.
8034 (arc_compute_function_type): Delete prototype.
8035 * config/arc/arc.md (in_ret_delay_slot): Use
8036 arc_return_address_register function.
8037 (simple_return): Likewise.
8038 (p_return_i): Likewise.
8040 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8042 PR tree-optimization/81428
8043 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8044 can't be built for those types.
8046 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8048 Remove stuff dead since r239246.
8050 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8051 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8052 (avr_inform_devices): Remove dead stuff.
8054 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8056 * config/arm/arm_neon.h: Fix softp typo.
8058 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8060 PR tree-optimization/81365
8061 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8062 aggregate moves onto bb predecessor edges, make sure there are no
8063 loads that could alias the lhs in between the start of bb and the
8066 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8069 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8070 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8071 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8073 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8075 PR tree-optimization/81396
8076 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8077 (init_symbolic_number): Initialize it to 1.
8078 (perform_symbolic_merge): Add n_ops from both operands into the new
8080 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8081 without base_addr as useless if they need more than one operation.
8082 (bswap_replace): Handle !bswap case for NULL base_addr.
8084 2017-07-17 Tom de Vries <tom@codesourcery.com>
8087 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8090 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8092 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8093 conditional builtin define __FIX_LEON3FT_B2BST.
8095 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8097 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8098 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8101 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8103 PR rtl-optimization/81424
8104 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8105 to remove potential trapping from operands if -fnon-call-exceptions.
8107 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8109 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8110 profile_proability for scalling.
8111 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8113 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8115 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8117 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8119 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8120 fixpoint arithmetics.
8122 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8124 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8125 fixpoint arithmetics.
8127 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8129 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8130 fixpoint arithmetics.
8132 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8134 * profile-count.h (profile_probability::from_reg_br_prob_note,
8135 profile_probability::to_reg_br_prob_note): New functions.
8136 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8137 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8138 * predict.c (probability_reliable_p): Update.
8139 (edge_probability_reliable_p): Update.
8140 (br_prob_note_reliable_p): Update.
8141 (invert_br_probabilities): Update.
8142 (add_reg_br_prob_note): New function.
8143 (combine_predictions_for_insn): Update.
8144 * asan.c (asan_clear_shadow): Update.
8145 * cfgbuild.c (compute_outgoing_frequencies): Update.
8146 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8147 (update_br_prob_note): Update.
8148 (rtl_verify_edges): Update.
8149 (purge_dead_edges): Update.
8150 (fixup_reorder_chain): Update.
8151 * emit-rtl.c (try_split): Update.
8152 * ifcvt.c (cond_exec_process_insns): Update.
8153 (cond_exec_process_if_block): Update.
8154 (dead_or_predicable): Update.
8155 * internal-fn.c (expand_addsub_overflow): Update.
8156 (expand_neg_overflow): Update.
8157 (expand_mul_overflow): Update.
8158 * loop-doloop.c (doloop_modify): Update.
8159 * loop-unroll.c (compare_and_jump_seq): Update.
8160 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8161 * predict.h: Update.
8162 * reorg.c (mostly_true_jump): Update.
8164 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8165 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8166 * config/arc/arc.c: (emit_unlikely_jump): Update.
8167 * config/arm/arm.c: (emit_unlikely_jump): Update.
8168 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8169 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8170 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8171 (ix86_print_operand): Update.
8172 (ix86_split_fp_branch): Update.
8173 (predict_jump): Update.
8174 * config/ia64/ia64.c (ia64_print_operand): Update.
8175 * config/mmix/mmix.c (mmix_print_operand): Update.
8176 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8177 (rs6000_expand_split_stack_prologue): Update.
8178 * config/rs6000/rs6000.c: Update.
8179 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8180 (s390_expand_vec_movstr): Update.
8181 (s390_expand_cs_tdsi): Update.
8182 (s390_expand_split_stack_prologue): Update.
8183 * config/sh/sh.c (sh_print_operand): Update.
8184 (expand_cbranchsi4): Update.
8185 (expand_cbranchdi4): Update.
8186 * config/sparc/sparc.c (output_v9branch): Update.
8187 * config/spu/spu.c (get_branch_target): Update.
8188 (ea_load_store_inline): Update.
8189 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8190 * config/tilepro/tilepro.c: Update.
8192 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8194 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8195 (gimplify_save_expr): Likewise.
8197 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8199 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8201 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8203 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8205 * timevar.def (TV_IPA_FNSUMMARY): Define.
8207 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8209 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8210 to back store errata sensitive sequence from being generated.
8211 (sqrtdf2_fix): Likewise.
8213 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8215 * tree-ssa-threadupdate.c (compute_path_counts,
8216 update_joiner_offpath_counts): Use profile_probability.
8218 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8221 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8223 * config/arm/arm-c.c (arm_cpu_builtins): Define
8224 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8226 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8228 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8229 array entries to represent __ieee128 versions of the
8230 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8231 scalar_extract_sig, and scalar_insert_exp built-in functions.
8232 (altivec_resolve_overloaded_builtin): Add special case handling
8233 for the __builtin_scalar_insert_exp function, as represented by
8234 the P9V_BUILTIN_VEC_VSIEDP constant.
8235 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8236 exponent support for __ieee128 argument.
8237 (VSESQP): Add scalar extract signature support for __ieee128
8239 (VSTDCNQP): Add scalar test negative support for __ieee128
8241 (VSIEQP): Add scalar insert exponent support for __int128 argument
8242 with __ieee128 result.
8243 (VSIEQPF): Add scalar insert exponent support for __ieee128
8244 argument with __ieee128 result.
8245 (VSTDCQP): Add scalar test data class support for __ieee128
8247 (VSTDCNQP): Add overload support for scalar test negative with
8249 (VSTDCQP): Add overload support for scalar test data class
8251 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8253 (UNSPEC_VSX_SIEXPQP): New constant.
8254 (xsxexpqp): New insn for VSX scalar extract exponent quad
8256 (xsxsigqp): New insn for VSX scalar extract significand quad
8258 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8259 precision with floating point argument.
8260 (xststdcqp): New expand for VSX scalar test data class quad
8262 (xststdcnegqp): New expand for VSX scalar test negative quad
8264 (xststdcqp): New insn to match expansions for VSX scalar test data
8265 class quad precision and VSX scalar test negative quad precision.
8266 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8267 special case operand checking to enforce that second operand of
8268 VSX scalar test data class with quad precision argument is a 7-bit
8270 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8271 prototypes and descriptions of __ieee128 versions of
8272 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8273 scalar_test_data_class, and scalar_test_neg built-in functions.
8275 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8277 PR tree-optimization/81162
8278 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8279 replace a negate with an add.
8281 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8283 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8285 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8287 * config/arm/arm-c.c (arm_cpu_builtins): Define
8288 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8290 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8292 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8293 (armv8-r): Set ARM Cortex-R52 as default CPU.
8294 * config/arm/arm-tables.opt: Regenerate.
8295 * config/arm/arm-tune.md: Regenerate.
8296 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8298 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8299 extension for -mcpu=cortex-r52.
8301 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8303 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8304 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8305 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8306 (fp-armv8): Define it as FP_ARMv8 only.
8307 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8308 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8310 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8311 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8312 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8313 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8314 than TARGET_FPU_ARMV8.
8315 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8316 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8317 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8319 * config/arm/neon.md (neon_vrint): Likewise.
8320 (neon_vcvt): Likewise.
8321 (neon_<fmaxmin_op><mode>): Likewise.
8322 (<fmaxmin><mode>3): Likewise.
8323 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8324 * config/arm/predicates.md (arm_cond_move_operator): Check against
8325 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8327 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8329 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8332 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8334 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8335 loop in comment with memset.
8337 2017-07-14 Martin Liska <mliska@suse.cz>
8339 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8340 * dwarf2out.c (is_java): Remove the function.
8341 (output_pubname): Remove usage of the function.
8342 (lower_bound_default): Remove usage of DW_LANG_Java.
8343 (gen_compile_unit_die): Likewise.
8344 * gcc.c: Remove compiler defaults for .java and .zip files.
8345 * gimple-expr.c (remove_suffix): Change as there's no longer
8346 extension than 4-letter one.
8347 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8348 (gimplify_save_expr): Likewise.
8349 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8350 as it's possible even for other languages than Java.
8351 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8352 * lto-opts.c (lto_write_options): Remove reference to Java.
8353 * opts.c (strip_off_ending): Update file extension handling.
8354 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8355 * tree-eh.c (lower_resx): Likewise.
8356 * tree.c (free_lang_data_in_type): Remove dead code.
8357 (find_decls_types_r): Likewise.
8358 (build_common_builtin_nodes): Remove Java from a comment.
8359 (verify_type): Remove dead code.
8360 * varasm.c (assemble_external): Remove Java from a comment.
8362 2017-07-14 Martin Liska <mliska@suse.cz>
8364 * opts.c (finish_options): Add quotes.
8365 (common_handle_option): Likewise.
8367 2017-07-14 Martin Liska <mliska@suse.cz>
8369 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8370 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8372 * dwarf2out.c (lower_bound_default): Do not handle
8374 (gen_compile_unit_die): Likewise.
8375 * gcc.c: Remove default extension binding for GNU Pascal.
8376 * stmt.c: Remove Pascal language from a comment.
8377 * xcoffout.c: Likewise.
8379 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8382 * diagnostic-show-locus.c (fixit_cmp): New function.
8383 (layout::layout): Sort m_fixit_hints.
8384 (column_range::column_range): Assert that the values are valid.
8385 (struct char_span): New struct.
8386 (correction::overwrite): New method.
8387 (struct source_line): New struct.
8388 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8389 calls in terms of classes source_line and char_span, and
8390 correction::overwrite.
8391 (selftest::test_overlapped_fixit_printing_2): New function.
8392 (selftest::diagnostic_show_locus_c_tests): Call it.
8394 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8396 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8397 early if there is no lhs.
8399 2017-07-13 Martin Liska <mliska@suse.cz>
8401 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8402 (gen_reference_type_die): Likewise.
8403 * stor-layout.c: Remove Pascal-related comment.
8405 2017-07-13 Martin Liska <mliska@suse.cz>
8407 * opts.c (finish_options): Add quotes to error messages.
8408 (parse_sanitizer_options): Likewise.
8410 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8412 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8414 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8416 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8418 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8420 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8422 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8424 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8427 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8428 provides the hardware capability bits, define the macro
8429 __BUILTIN_CPU_SUPPORTS__.
8430 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8431 if GLIBC does not provide the hardware capability bits. Add a
8432 gcc_unreachable call if the built-in cpu function is neither
8433 __builtin_cpu_is nor __builtin_cpu_supports.
8434 (rs6000_get_function_versions_dispatcher): Change the warning
8435 that an old GLIBC is used which does not export the capability
8436 bits to be an error.
8437 * doc/extend.texi (target_clones attribute): Document the
8438 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8439 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8440 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8441 the macros defined by GCC if the newer GLIBC is available.
8443 2017-07-12 Jeff Law <law@redhat.com>
8445 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8446 remaining includes slightly.
8447 * config/riscv/riscv-builtins.c: Include profile-count.h.
8449 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8452 * config/avr/avr.c (avr_set_current_function): In diagnostic
8453 messages: Quote keywords and (parts of) identifiers.
8454 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8457 2017-07-12 Carl Love <cel@us.ibm.com>
8459 * config/rs6000/rs6000-c.c: Add support for built-in functions
8460 vector bool char vec_revb (vector bool char);
8461 vector bool short vec_revb (vector short char);
8462 vector bool int vec_revb (vector bool int);
8463 vector bool long long vec_revb (vector bool long long);
8464 * doc/extend.texi: Update the built-in documentation file for the
8465 new built-in functions.
8467 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8469 * config/s390/s390.md: Remove movcc splitter.
8471 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8473 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8474 load/store on condition.
8476 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8479 * config/avr/avr.c (avr_encode_section_info)
8480 [progmem && !TREE_READONLY]: Error if progmem object needs
8483 2017-07-11 Michael Collison <michael.collison@arm.com>
8485 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8488 2017-07-11 Carl Love <cel@us.ibm.com>
8490 * config/rs6000/rs6000-c.c: Add support for builtins
8491 vector unsigned int vec_parity_lsbb (vector signed int);
8492 vector unsigned int vec_parity_lsbb (vector unsigned int);
8493 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8494 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8495 vector unsigned long long vec_parity_lsbb (vector signed long long);
8496 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8497 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8498 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8499 * doc/extend.texi: Update the built-in documentation file for the
8500 new built-in functions.
8502 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8504 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8505 (layout::m_primary_loc): New field.
8506 (layout::layout): Initialize new field. Move location filtering
8507 logic from here to...
8508 (layout::maybe_add_location_range): ...this new method. Add
8509 support for filtering to just the lines already specified by other
8511 (layout::will_show_line_p): New method.
8512 (gcc_rich_location::add_location_if_nearby): New method.
8513 (selftest::test_add_location_if_nearby): New test function.
8514 (selftest::diagnostic_show_locus_c_tests): Call it.
8515 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8518 2017-07-11 Tom de Vries <tom@codesourcery.com>
8520 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8521 (bb_first_real_insn): New function.
8522 (nvptx_single): Add extra initialization of broadcasted condition
8525 2017-07-11 Nathan Sidwell <nathan@acm.org>
8527 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8529 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8531 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8532 Binutils doc as TEXI will mess them up.
8533 * doc/invoke.texi (AVR Options): Same here.
8535 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8537 * config/sparc/sparc.opt (mfix-ut700): New option.
8538 (mfix-gr712rc): Likewise.
8539 (sparc_fix_b2bst): New variable.
8540 * doc/invoke.texi (SPARC options): Document them.
8541 (ARM options): Fix warnings.
8542 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8543 instructions to prevent sequences that can trigger the store-store
8544 errata for certain LEON3FT processors.
8545 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8546 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8547 * config/sparc/sparc.md (fix_b2bst): New attribute.
8548 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8550 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8553 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8555 (*rsqrtsf2_sse): Ditto.
8557 (div<mode>3): Macroize insn from divdf3 and divsf3
8558 using MODEF mode iterator.
8560 2017-07-10 Martin Sebor <msebor@redhat.com>
8562 PR tree-optimization/80397
8563 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8564 instead of testing for equality to INTEGER_TYPE.
8566 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8568 * config.gcc: Remove uclibc from arc target spec.
8570 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8572 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8574 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8577 * lto-wrapper.c (remove_option): New function.
8578 (merge_and_complain): Merge PIC/PIE options more realistically.
8580 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8582 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8586 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8587 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8588 * config.in: Regenerate.
8589 * configure: Regenerate.
8590 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8591 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8592 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8593 (TARGET_GASISR_PROLOGUES): ...target mask.
8594 * common/config/avr/avr-common.c
8595 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8596 Set -mgas-isr-prologues.
8597 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8598 INSERT_PASS_BEFORE for it.
8599 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8600 * config/avr/avr.c (avr_option_override)
8601 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8602 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8603 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8604 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8605 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8607 (make_avr_pass_pre_proep): New function.
8608 (emit_push_sfr) <treg>: Add argument to function and use it
8610 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8611 and set machine->gasisr.yes.
8612 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8613 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8614 __gcc_isr.n_pushed to .L__stack_usage.
8615 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8616 (avr_asm_final_postscan_insn): ...this new static function.
8617 * config/avr/avr.h (machine_function)
8618 <is_no_gccisr, use_L__stack_usage>: New fields.
8619 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8620 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8621 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8622 (gasisr, *gasisr): New expander and insn.
8623 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8624 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8625 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8627 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8629 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8632 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8634 Move jump-tables out of .text again.
8637 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8638 (ASM_OUTPUT_ADDR_VEC): New function.
8639 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8640 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8641 INSN_ADDRESSes as asm comment.
8642 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8643 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8644 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8645 * config/avr/avr.md (*tablejump): Adjust comment.
8646 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8647 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8649 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8650 (avr_output_addr_vec): New proto.
8651 (avr_log_t) <insn_addresses>: New field.
8653 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8656 * config/i386/i386.c (ix86_function_arg_advance): Set
8657 outgoing_args_on_stack to true if there are outgoing arguments
8659 (ix86_function_arg): Likewise.
8660 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8661 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8662 * config/i386/i386.h (machine_function): Add
8663 outgoing_args_on_stack.
8665 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8667 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8669 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8671 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8673 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8674 (REAL_H): Remove $(MACHMODE_H).
8675 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8677 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8678 $(MACHMODE_H) and double-int.h.
8679 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8681 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8684 2017-07-07 Andrew Pinski <apinski@cavium.com>
8686 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8687 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8689 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8691 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8692 Add warning if GCC was not configured to link against a GLIBC that
8693 exports the hardware capability bits.
8694 (make_resolver_func): Make resolver function private and not a
8695 COMDAT function. Create the name with clone_function_name instead
8696 of make_unique_name.
8699 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8700 correct operand in doing the split.
8702 2017-07-07 Carl Love <cel@us.ibm.com>
8704 * config/rs6000/rs6000-c: Add support for built-in function
8705 vector unsigned short vec_pack_to_short_fp32 (vector float,
8707 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8708 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8709 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8710 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8711 (convert_4f32_8i16): Add define_expand.
8712 * doc/extend.texi: Update the built-in documentation file for the
8713 new built-in function.
8715 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8717 * config/sparc/m8.md: New file.
8718 * config/sparc/sparc.md: Include m8.md.
8720 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8722 * config/sparc/sparc.opt: New option -mvis4b.
8723 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8724 (sparc_option_override): Handle VIS4B.
8725 (enum sparc_builtins): Define
8726 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8727 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8728 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8729 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8730 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8731 (check_constant_argument): New function.
8732 (sparc_vis_init_builtins): Define builtins
8733 __builtin_vis_dictunpack{8,16,32},
8734 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8735 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8736 __builtin_vis_fpcmpde{8,16,32}shl and
8737 __builtin_vis_fpcmpur{8,16,32}shl.
8738 (sparc_expand_builtin): Check that the constant operands to
8739 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8740 constant and in range.
8741 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8743 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8744 (SPARC_IMM5_P): Likewise.
8745 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8746 (enabled): Handle vis4b.
8747 (UNSPEC_DICTUNPACK): New unspec.
8748 (UNSPEC_FPCMPSHL): Likewise.
8749 (UNSPEC_FPUCMPSHL): Likewise.
8750 (UNSPEC_FPCMPDESHL): Likewise.
8751 (UNSPEC_FPCMPURSHL): Likewise.
8752 (cpu_feature): New CPU feature `vis4b'.
8753 (dictunpack{8,16,32}): New insns.
8754 (FPCSMODE): New mode iterator.
8755 (fpcscond): New code iterator.
8756 (fpcsucond): Likewise.
8757 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8758 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8759 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8760 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8761 * config/sparc/constraints.md: Define constraints `q' for unsigned
8762 2-bit integer constants and `t' for unsigned 5-bit integer
8764 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8766 (imm5_operand_dictunpack16): Likewise.
8767 (imm5_operand_dictunpack32): Likewise.
8768 (imm2_operand): Likewise.
8769 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8770 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8771 ditunpack* and fpcmp*shl builtins.
8773 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8775 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8776 * config.in: Add HAVE_AS_SPARC6 define.
8777 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8779 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8781 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8782 (CPP_CPU_SPEC): Handle m8.
8783 (ASM_CPU_SPEC): Likewise.
8784 * config/sparc/sparc-opts.h (enum processor_type): Add
8786 * config/sparc/sparc.c (m8_costs): New struct.
8787 (sparc_option_override): Handle TARGET_CPU_m8.
8788 (sparc32_initialize_trampoline): Likewise.
8789 (sparc64_initialize_trampoline): Likewise.
8790 (sparc_issue_rate): Likewise.
8791 (sparc_register_move_cost): Likewise.
8792 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8793 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8794 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8795 (CPP_CPU_SPEC): Handle M8.
8796 (ASM_CPU_SPEC): Likewise.
8797 (AS_M8_FLAG): Define.
8798 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8799 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8800 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8802 * configure: Regenerate.
8803 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8806 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8808 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8810 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8811 ("*movdi_insn_sp32"): Do not set v3pipe.
8812 ("*movsi_insn"): Likewise.
8813 ("*movdi_insn_sp64"): Likewise.
8814 ("*movsf_insn"): Likewise.
8815 ("*movdf_insn_sp32"): Likewise.
8816 ("*movdf_insn_sp64"): Likewise.
8817 ("*zero_extendsidi2_insn_sp64"): Likewise.
8818 ("*sign_extendsidi2_insn"): Likewise.
8819 ("*mov<VM32:mode>_insn"): Likewise.
8820 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8821 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8822 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8823 ("<vlop:code><VL:mode>3"): Likewise.
8824 ("*not_<vlop:code><VL:mode>3"): Likewise.
8825 ("*nand<VL:mode>_vis"): Likewise.
8826 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8827 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8828 ("one_cmpl<VL:mode>2"): Likewise.
8829 ("faligndata<VM64:mode>_vis"): Likewise.
8830 ("alignaddrsi_vis"): Likewise.
8831 ("alignaddrdi_vis"): Likweise.
8832 ("alignaddrlsi_vis"): Likewise.
8833 ("alignaddrldi_vis"): Likewise.
8834 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8835 ("bmaskdi_vis"): Likewise.
8836 ("bmasksi_vis"): Likewise.
8837 ("bshuffle<VM64:mode>_vis"): Likewise.
8838 ("cmask8<P:mode>_vis"): Likewise.
8839 ("cmask16<P:mode>_vis"): Likewise.
8840 ("cmask32<P:mode>_vis"): Likewise.
8841 ("pdistn<P:mode>_vis"): Likewise.
8842 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8844 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8846 * config/sparc/sparc.md ("subtype"): New insn attribute.
8847 ("*wrgsr_sp64"): Set insn subtype.
8848 ("*rdgsr_sp64"): Likewise.
8849 ("alignaddrsi_vis"): Likewise.
8850 ("alignaddrdi_vis"): Likewise.
8851 ("alignaddrlsi_vis"): Likewise.
8852 ("alignaddrldi_vis"): Likewise.
8853 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8854 ("fexpand_vis"): Likewise.
8855 ("fpmerge_vis"): Likewise.
8856 ("faligndata<VM64:mode>_vis"): Likewise.
8857 ("bshuffle<VM64:mode>_vis"): Likewise.
8858 ("cmask8<P:mode>_vis"): Likewise.
8859 ("cmask16<P:mode>_vis"): Likewise.
8860 ("cmask32<P:mode>_vis"): Likewise.
8861 ("fchksm16_vis"): Likewise.
8862 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8863 ("fmean16_vis"): Likewise.
8864 ("fp<plusminus_insn>64_vis"): Likewise.
8865 ("<plusminus_insn>v8qi3"): Likewise.
8866 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8867 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8868 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8869 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8870 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8871 ("*movqi_insn"): Likewise.
8872 ("*movhi_insn"): Likewise.
8873 ("*movsi_insn"): Likewise.
8874 ("movsi_pic_gotdata_op"): Likewise.
8875 ("*movdi_insn_sp32"): Likewise.
8876 ("*movdi_insn_sp64"): Likewise.
8877 ("movdi_pic_gotdata_op"): Likewise.
8878 ("*movsf_insn"): Likewise.
8879 ("*movdf_insn_sp32"): Likewise.
8880 ("*movdf_insn_sp64"): Likewise.
8881 ("*zero_extendhisi2_insn"): Likewise.
8882 ("*zero_extendqihi2_insn"): Likewise.
8883 ("*zero_extendqisi2_insn"): Likewise.
8884 ("*zero_extendqidi2_insn"): Likewise.
8885 ("*zero_extendhidi2_insn"): Likewise.
8886 ("*zero_extendsidi2_insn_sp64"): Likewise.
8887 ("ldfsr"): Likewise.
8888 ("prefetch_64"): Likewise.
8889 ("prefetch_32"): Likewise.
8890 ("tie_ld32"): Likewise.
8891 ("tie_ld64"): Likewise.
8892 ("*tldo_ldub_sp32"): Likewise.
8893 ("*tldo_ldub1_sp32"): Likewise.
8894 ("*tldo_ldub2_sp32"): Likewise.
8895 ("*tldo_ldub_sp64"): Likewise.
8896 ("*tldo_ldub1_sp64"): Likewise.
8897 ("*tldo_ldub2_sp64"): Likewise.
8898 ("*tldo_ldub3_sp64"): Likewise.
8899 ("*tldo_lduh_sp32"): Likewise.
8900 ("*tldo_lduh1_sp32"): Likewise.
8901 ("*tldo_lduh_sp64"): Likewise.
8902 ("*tldo_lduh1_sp64"): Likewise.
8903 ("*tldo_lduh2_sp64"): Likewise.
8904 ("*tldo_lduw_sp32"): Likewise.
8905 ("*tldo_lduw_sp64"): Likewise.
8906 ("*tldo_lduw1_sp64"): Likewise.
8907 ("*tldo_ldx_sp64"): Likewise.
8908 ("*mov<VM32:mode>_insn"): Likewise.
8909 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8910 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8912 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8914 * config/sparc/sparc.md ("type"): New insn type viscmp.
8915 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8917 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8918 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8919 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8920 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8922 ("n7_vis_logical_11cycle"): Likewise.
8923 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8924 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8925 * config/sparc/niagara.md ("niag_vis"): Likewise.
8926 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8927 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8929 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8931 * config/sparc/sparc.md: New instruction type `bmask'.
8932 (bmaskdi_vis): Use the `bmask' type.
8933 (bmasksi_vis): Likewise.
8934 * config/sparc/ultra3.md (us3_array): Likewise.
8935 * config/sparc/niagara7.md (n7_array): Likewise.
8936 * config/sparc/niagara4.md (n4_array): Likewise.
8937 * config/sparc/niagara2.md (niag2_vis): Likewise.
8938 (niag3_vis): Likewise.
8939 * config/sparc/niagara.md (niag_vis): Likewise.
8941 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8943 * ipa-comdats.c: Remove optimize check from gate.
8944 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8945 for functions not optimized.
8946 (ipa_fn_summary_read): Skip optimize check.
8947 (ipa_fn_summary_write): Likewise.
8948 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8950 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8952 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8953 (check_callers): Give up on caller which is not optimized.
8954 (inline_small_functions): Likewise.
8955 (ipa_inline): Do not give up when not optimizing.
8956 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8957 away unoptimizes cdtors.
8958 (whole_program_function_and_variable_visibility): Do
8959 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8960 * ipa.c (process_references): Do not check optimize.
8961 (symbol_table::remove_unreachable_nodes): Update optimize check.
8962 (set_writeonly_bit): Update optimize check.
8963 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8964 (pass_ipa_single_use::gate): Remove.
8966 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8968 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8969 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8970 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8971 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8972 permute_load, permute_store, adjust_extract, adjust_splat,
8973 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8974 replace_swap_with_copy, dump_swap_insn_table,
8975 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8976 recombine_lvx_pattern, recombine_stvx_pattern,
8977 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8978 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8979 to file rs6000-p8swap.c.
8980 * config/rs6000/rs6000-p8swap.c: New file.
8981 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8982 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8983 and rs6000*-*-* targets.
8985 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8987 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8989 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8991 * lto-wrapper.c (merge_and_complain): Do not merge
8992 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8993 fsigned_zeros, ftrapping_math, fwrapv.
8994 (append_compiler_options): Do not track these options.
8995 (append_linker_options): Likewie
8997 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8999 * cgraphunit.c (cgraph_node::finalize_function): When
9000 !flag_toplevel_reorde set no_reorder flag.
9001 (varpool_node::finalize_decl): Likewise.
9002 (symbol_table::compile): Drop no toplevel reorder path.
9004 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9006 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9007 edges; zero probability is not better than uninitialized.
9009 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9011 * asan.h (asan_sanitize_allocas_p): Declare.
9012 * asan.c (asan_sanitize_allocas_p): New function.
9013 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9014 (handle_builtin_alloca): Likewise.
9015 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9016 if !asan_sanitize_allocas_p.
9017 * params.def (asan-instrument-allocas): Add new option.
9018 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9019 * opts.c (common_handle_option): Disable allocas sanitization for
9022 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9024 * asan.c: Include gimple-fold.h.
9025 (get_last_alloca_addr): New function.
9026 (handle_builtin_stackrestore): Likewise.
9027 (handle_builtin_alloca): Likewise.
9028 (asan_emit_allocas_unpoison): Likewise.
9029 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9030 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9031 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9032 (instrument_builtin_call): Pass gimple iterator to
9033 get_mem_refs_of_builtin_call.
9034 (last_alloca_addr): New global.
9035 * asan.h (asan_emit_allocas_unpoison): Declare.
9036 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9037 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9038 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9039 if function calls alloca.
9040 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9041 * gimple-fold.h (replace_call_with_value): Declare.
9042 * internal-fn.c: Include asan.h.
9043 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9044 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9046 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9048 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9049 (C_SELFTEST_FLAGS): New.
9050 (CPP_SELFTEST_FLAGS): New.
9051 (SELFTEST_DEPS): New, from deps of s-selftest.
9052 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9053 (CPP_SELFTEST_DEPS): New.
9054 (selftest): Add dependency on s-selftest-c++.
9055 (s-selftest): Rename to...
9056 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9057 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9058 than SELFTEST_FLAGS.
9059 (selftest-gdb): Rename to...
9060 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9062 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9063 (selftest-valgrind): Rename to...
9064 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9066 (selftest-valgrind): Reintroduce as an alias for
9067 selftest-c-valgrind.
9068 (s-selftest-c++): New.
9069 (selftest-c++-gdb): New.
9070 (selftest-c++-valgrind): New.
9072 2017-07-06 Olivier Hainque <hainque@adacore.com>
9074 * gcc.c (process_command): When deciding if undefined variables
9075 should be ignored when processing specs, accept "gcc -v" as well.
9077 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9079 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9080 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9082 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9084 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9085 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9086 * config/arm/arm-tables.opt: Regenerate.
9087 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9089 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9091 2017-07-06 Carl Love <cel@us.ibm.com>
9093 * ChangeLog: Clean up from mid air collision
9095 2017-07-06 Carl Love <cel@us.ibm.com>
9097 * config/rs6000/rs6000-c.c: Add support for built-in functions
9098 vector signed int vec_subc (vector signed int, vector signed int);
9099 vector signed __int128 vec_subc (vector signed __int128,
9100 vector signed __int128);
9101 vector unsigned __int128 vec_subc (vector unsigned __int128,
9102 vector unsigned __int128);
9103 vector signed int vec_sube (vector signed int, vector signed int,
9105 vector unsigned int vec_sube (vector unsigned int,
9106 vector unsigned int,
9107 vector unsigned int);
9108 vector signed __int128 vec_sube (vector signed __int128,
9109 vector signed __int128,
9110 vector signed__int128);
9111 vector unsigned __int128 vec_sube (vector unsigned __int128,
9112 vector unsigned __int128,
9113 vector unsigned __int128);
9114 vector signed int vec_subec (vector signed int, vector signed int,
9116 vector unsigned int vec_subec (vector unsigned int,
9117 vector unsigned int,
9118 vector unsigned int);
9119 vector signed __int128 vec_subec (vector signed __int128,
9120 vector signed __int128,
9121 vector signed__int128);
9122 vector unsigned __int128 vec_subec (vector unsigned __int128,
9123 vector unsigned __int128,
9124 vector unsigned __int128);
9125 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9126 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9127 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9128 BU_ALTIVEC_OVERLOAD_X definitions.
9129 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9130 * doc/extend.texi: Update the built-in documentation file for the new
9133 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9136 * diagnostic-show-locus.c (layout::layout): Use start and finish
9137 spelling location for the start and finish of each range.
9138 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9139 Add unused aspect param.
9140 * input.c (expand_location_1): Add "aspect" param, and use it
9141 to access the correct part of the location.
9142 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9144 (expand_location_to_spelling_point): Likewise.
9145 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9146 param, and pass it to expand_location_1.
9148 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9150 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9151 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9152 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9153 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9154 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9155 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9156 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9157 _mm_maskz_getmant_ss): New intrinsics.
9158 (__builtin_ia32_getexpss128_mask): Changed to ...
9159 __builtin_ia32_getexpss128_round ... this.
9160 (__builtin_ia32_getexpsd128_mask): Changed to ...
9161 __builtin_ia32_getexpsd128_round ... this.
9162 * config/i386/i386-builtin-types.def
9163 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9164 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9165 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9166 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9167 __builtin_ia32_getmantss_mask_round): New builtins.
9168 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9169 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9170 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9171 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9172 * config/i386/sse.md
9173 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9174 avx512f_sgetexp<mode><mask_scalar_name>
9175 <round_saeonly_scalar_name> ... this.
9176 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9177 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9178 vgetexp<ssescalarmodesuffix>
9179 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9180 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9181 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9182 avx512f_vgetmant<mode><mask_scalar_name>
9183 <round_saeonly_scalar_name> ... this.
9184 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9185 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9186 vgetmant<ssescalarmodesuffix>
9187 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9188 %0<mask_scalar_operand4>, %1, %2
9189 <round_saeonly_scalar_mask_op4>, %3} ... this.
9190 * config/i386/subst.md (mask_scalar_operand4,
9191 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9192 round_saeonly_scalar_nimm_predicate): New subst attributes.
9194 2017-07-06 Julia Koval <julia.koval@intel.com>
9196 * config/i386/i386.c (ix86_erase_embedded_rounding):
9197 Remove code for old rounding pattern.
9199 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9201 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9203 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9205 * doc/sourcebuild.texi (Test Directives, Variants of
9206 dg-require-support): Add documentation for dg-require-stack-check.
9208 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9210 * config/i386/subst.md (mask_scalar, round_scalar,
9211 round_saeonly_scalar): New meta-templates.
9212 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9213 round_scalar_mask_operand3, round_scalar_mask_op3,
9214 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9215 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9216 round_saeonly_scalar_constraint,
9217 round_saeonly_scalar_prefix): New subst attribute.
9218 * config/i386/sse.md
9219 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9220 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9221 <round_scalar_name> ... this.
9222 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9223 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9224 <round_scalar_name> ... this.
9225 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9226 <sse>_vm<code><mode>3<mask_scalar_name>
9227 <round_saeonly_scalar_name> ... this.
9228 (v<plusminus_mnemonic><ssescalarmodesuffix>
9229 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9230 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9231 v<plusminus_mnemonic><ssescalarmodesuffix>
9232 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9233 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9234 (v<multdiv_mnemonic><ssescalarmodesuffix>
9235 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9236 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9237 v<multdiv_mnemonic><ssescalarmodesuffix>
9238 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9239 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9240 (v<maxmin_float><ssescalarmodesuffix>
9241 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9242 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9243 v<maxmin_float><ssescalarmodesuffix>
9244 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9245 %0<mask_scalar_operand3>, %1, %<iptr>2
9246 <round_saeonly_scalar_mask_op3>} ... this.
9248 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9250 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9251 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9253 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9254 Alan Hayward <alan.hayward@arm.com>
9255 David Sherwood <david.sherwood@arm.com>
9257 * combine.c (simplify_if_then_else): Remove "enum" before
9259 * compare-elim.c (can_eliminate_compare): Likewise.
9260 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9262 (aarch64_lookup_simd_builtin_type): Likewise.
9263 (aarch64_simd_builtin_type): Likewise.
9264 (aarch64_init_simd_builtin_types): Likewise.
9265 (aarch64_simd_expand_args): Likewise.
9266 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9268 (aarch64_reverse_mask): Likewise.
9269 (aarch64_simd_emit_reg_reg_move): Likewise.
9270 (aarch64_gen_adjusted_ldpstp): Likewise.
9271 (aarch64_ccmp_mode_to_code): Likewise.
9272 (aarch64_operands_ok_for_ldpstp): Likewise.
9273 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9274 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9276 (aarch64_min_divisions_for_recip_mul): Likewise.
9277 (aarch64_reassociation_width): Likewise.
9278 (aarch64_get_condition_code_1): Likewise.
9279 (aarch64_simd_emit_reg_reg_move): Likewise.
9280 (aarch64_simd_attr_length_rglist): Likewise.
9281 (aarch64_reverse_mask): Likewise.
9282 (aarch64_operands_ok_for_ldpstp): Likewise.
9283 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9284 (aarch64_gen_adjusted_ldpstp): Likewise.
9285 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9287 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9288 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9289 (arm_lookup_simd_builtin_type): Likewise.
9290 (arm_simd_builtin_type): Likewise.
9291 (arm_init_simd_builtin_types): Likewise.
9292 (arm_expand_builtin_args): Likewise.
9293 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9294 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9295 (ft32_setup_incoming_varargs): Likewise.
9296 (ft32_function_arg): Likewise.
9297 (ft32_function_arg_advance): Likewise.
9298 (ft32_pass_by_reference): Likewise.
9299 (ft32_arg_partial_bytes): Likewise.
9300 (ft32_valid_pointer_mode): Likewise.
9301 (ft32_addr_space_pointer_mode): Likewise.
9302 (ft32_addr_space_legitimate_address_p): Likewise.
9303 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9305 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9306 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9307 (iamcu_alignment): Likewise.
9308 (canonicalize_vector_int_perm): Likewise.
9309 (ix86_noce_conversion_profitable_p): Likewise.
9310 (ix86_mpx_bound_mode): Likewise.
9311 (ix86_operands_ok_for_move_multiple): Likewise.
9312 * config/microblaze/microblaze-protos.h
9313 (microblaze_expand_conditional_branch_reg): Likewise.
9314 * config/microblaze/microblaze.c
9315 (microblaze_expand_conditional_branch_reg): Likewise.
9316 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9318 (rs6000_reassociation_width): Likewise.
9319 (rs6000_invalid_binary_op): Likewise.
9320 (fusion_p9_p): Likewise.
9321 (emit_fusion_p9_load): Likewise.
9322 (emit_fusion_p9_store): Likewise.
9323 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9325 (riscv_hard_regno_mode_ok_p): Likewise.
9326 (riscv_address_insns): Likewise.
9327 (riscv_split_symbol): Likewise.
9328 (riscv_legitimize_move): Likewise.
9329 (riscv_function_value): Likewise.
9330 (riscv_hard_regno_nregs): Likewise.
9331 (riscv_expand_builtin): Likewise.
9332 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9333 (riscv_build_integer): Likewise.
9334 (riscv_split_integer): Likewise.
9335 (riscv_legitimate_constant_p): Likewise.
9336 (riscv_cannot_force_const_mem): Likewise.
9337 (riscv_regno_mode_ok_for_base_p): Likewise.
9338 (riscv_valid_base_register_p): Likewise.
9339 (riscv_valid_offset_p): Likewise.
9340 (riscv_valid_lo_sum_p): Likewise.
9341 (riscv_classify_address): Likewise.
9342 (riscv_legitimate_address_p): Likewise.
9343 (riscv_address_insns): Likewise.
9344 (riscv_load_store_insns): Likewise.
9345 (riscv_force_binary): Likewise.
9346 (riscv_split_symbol): Likewise.
9347 (riscv_force_address): Likewise.
9348 (riscv_legitimize_address): Likewise.
9349 (riscv_move_integer): Likewise.
9350 (riscv_legitimize_const_move): Likewise.
9351 (riscv_legitimize_move): Likewise.
9352 (riscv_address_cost): Likewise.
9353 (riscv_subword): Likewise.
9354 (riscv_output_move): Likewise.
9355 (riscv_canonicalize_int_order_test): Likewise.
9356 (riscv_emit_int_order_test): Likewise.
9357 (riscv_function_arg_boundary): Likewise.
9358 (riscv_pass_mode_in_fpr_p): Likewise.
9359 (riscv_pass_fpr_single): Likewise.
9360 (riscv_pass_fpr_pair): Likewise.
9361 (riscv_get_arg_info): Likewise.
9362 (riscv_function_arg): Likewise.
9363 (riscv_function_arg_advance): Likewise.
9364 (riscv_arg_partial_bytes): Likewise.
9365 (riscv_function_value): Likewise.
9366 (riscv_pass_by_reference): Likewise.
9367 (riscv_setup_incoming_varargs): Likewise.
9368 (riscv_print_operand): Likewise.
9369 (riscv_elf_select_rtx_section): Likewise.
9370 (riscv_save_restore_reg): Likewise.
9371 (riscv_for_each_saved_reg): Likewise.
9372 (riscv_register_move_cost): Likewise.
9373 (riscv_hard_regno_mode_ok_p): Likewise.
9374 (riscv_hard_regno_nregs): Likewise.
9375 (riscv_class_max_nregs): Likewise.
9376 (riscv_memory_move_cost): Likewise.
9377 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9378 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9379 (rl78_addr_space_address_mode): Likewise.
9380 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9382 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9383 (rs6000_reassociation_width): Likewise.
9384 (rs6000_invalid_binary_op): Likewise.
9385 (fusion_p9_p): Likewise.
9386 (emit_fusion_p9_load): Likewise.
9387 (emit_fusion_p9_store): Likewise.
9388 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9389 (ok_for_simple_move_operands): Likewise.
9390 (ok_for_simple_move_strict_operands): Likewise.
9391 (ok_for_simple_arith_logic_operands): Likewise.
9392 (visium_legitimize_reload_address): Likewise.
9393 (visium_select_cc_mode): Likewise.
9394 (output_cbranch): Likewise.
9395 (visium_split_double_move): Likewise.
9396 (visium_expand_copysign): Likewise.
9397 (visium_expand_int_cstore): Likewise.
9398 (visium_expand_fp_cstore): Likewise.
9399 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9400 (visium_function_arg): Likewise.
9401 (visium_function_arg_advance): Likewise.
9402 (visium_libcall_value): Likewise.
9403 (visium_setup_incoming_varargs): Likewise.
9404 (visium_legitimate_constant_p): Likewise.
9405 (visium_legitimate_address_p): Likewise.
9406 (visium_legitimize_address): Likewise.
9407 (visium_secondary_reload): Likewise.
9408 (visium_register_move_cost): Likewise.
9409 (visium_memory_move_cost): Likewise.
9410 (prepare_move_operands): Likewise.
9411 (ok_for_simple_move_operands): Likewise.
9412 (ok_for_simple_move_strict_operands): Likewise.
9413 (ok_for_simple_arith_logic_operands): Likewise.
9414 (visium_function_value_1): Likewise.
9415 (rtx_ok_for_offset_p): Likewise.
9416 (visium_legitimize_reload_address): Likewise.
9417 (visium_split_double_move): Likewise.
9418 (visium_expand_copysign): Likewise.
9419 (visium_expand_int_cstore): Likewise.
9420 (visium_expand_fp_cstore): Likewise.
9421 (visium_split_cstore): Likewise.
9422 (visium_select_cc_mode): Likewise.
9423 (visium_split_cbranch): Likewise.
9424 (output_cbranch): Likewise.
9425 (visium_print_operand_address): Likewise.
9426 * expmed.c (flip_storage_order): Likewise.
9427 * expmed.h (emit_cstore): Likewise.
9428 (flip_storage_order): Likewise.
9429 * genrecog.c (validate_pattern): Likewise.
9430 * hsa-gen.c (gen_hsa_addr): Likewise.
9431 * internal-fn.c (expand_arith_overflow): Likewise.
9432 * ira-color.c (allocno_copy_cost_saving): Likewise.
9433 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9434 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9435 (process_invariant_for_inheritance): Likewise.
9436 * lra-eliminations.c (move_plus_up): Likewise.
9437 * omp-low.c (lower_oacc_reductions): Likewise.
9438 * simplify-rtx.c (simplify_subreg): Likewise.
9439 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9440 (TARGET_CHKP_BOUND_MODE): Likewise..
9441 * targhooks.c (default_chkp_bound_mode): Likewise.
9442 (default_setup_incoming_vararg_bounds): Likewise.
9443 * targhooks.h (default_chkp_bound_mode): Likewise.
9444 (default_setup_incoming_vararg_bounds): Likewise.
9445 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9446 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9447 (have_whole_vector_shift): Likewise.
9448 * tree-vect-stmts.c (vectorizable_load): Likewise.
9449 * doc/tm.texi: Regenerate.
9451 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9453 Graceful degrade if Binutils PR21472 is not available.
9456 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9457 .rodata in flash test fails.
9458 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9459 * confgure: Regenerate.
9460 * config.in: Regenerate.
9461 * config/avr/avr.c (avr_asm_named_section)
9462 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9463 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9464 (avr_asm_init_sections): Same.
9466 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9468 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9469 (fma<VH:mode>4_intrinsic): Likewise.
9470 (*fmsub<VCVTF:mode>4): Likewise.
9471 (*fmsub<VH:mode>4_intrinsic): Likewise.
9473 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9476 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9477 Don't depend on "optimize > 0".
9478 (out_movhi_r_mr, out_movqi_mr_r): Same.
9479 (out_movhi_mr_r, out_movqi_r_mr): Same.
9480 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9481 io_address_operand on "optimize > 0".
9483 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9485 * tree-loop-distribution.c: Add general explanantion on the pass.
9486 (generate_loops_for_partition): Mark distributed loop.
9487 (pg_add_dependence_edges): New parameter. Handle alias data
9488 dependence specially and record it in the parameter if asked.
9489 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9490 (init_partition_graph_vertices, add_partition_graph_edge): New.
9491 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9492 (free_partition_graph_vdata, build_partition_graph): New.
9493 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9494 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9495 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9496 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9497 (version_for_distribution_p, finalize_partitions): New.
9498 (distribute_loop): Handle alias data dependence specially. Factor
9499 out loop fusion code as functions and call these functions.
9501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9503 * tree-loop-distribution.c (classify_partition): New parameter and
9504 better handle reduction statement.
9505 (rdg_build_partitions): Revise comment.
9506 (distribute_loop): Compute statements in all partitions and pass it
9507 to classify_partition.
9509 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9511 * tree-loop-distribution.c (enum partition_type): New.
9512 (struct partition): New field type.
9513 (partition_merge_into): Add parameter. Update partition type.
9514 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9515 (build_rdg_partition_for_vertex): Compute partition type.
9516 (rdg_build_partitions): Dump partition type.
9517 (distribute_loop): Update calls to partition_merge_into.
9519 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9521 * tree-loop-distribution.c (struct ddr_hasher): New.
9522 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9524 (classify_partition): Call get_data_dependence.
9525 (pg_add_dependence_edges): Ditto.
9526 (distribute_loop): Release data dependence hash table.
9528 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9530 * tree-loop-distribution.c (ref_base_address): Delete.
9531 (similar_memory_accesses): Rename ...
9532 (share_memory_accesses): ... to this. Check if partitions access
9533 the same memory reference.
9534 (distribute_loop): Call share_memory_accesses.
9536 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9538 * tree-loop-distribution.c (struct partition): New field recording
9540 (partition_alloc, partition_free): Init and release data refs.
9541 (partition_merge_into): Merge data refs.
9542 (build_rdg_partition_for_vertex): Collect data refs for partition.
9543 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9545 (distribute_loop): Remve data refs from vertice data of partition
9548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9550 * tree-loop-distribution.c (params.h): Include header file.
9551 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9552 (datarefs_vec): New global var.
9553 (create_rdg_vertices): Use datarefs_vec directly.
9554 (free_rdg): Don't free data references.
9555 (build_rdg): Update use. Don't free data references.
9556 (distribute_loop): Compute global variable for data references.
9557 Bail out if there are too many data references.
9559 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9561 * tree-loop-distribution.c (loop_nest): New global var.
9562 (build_rdg): Use loop directly, rather than loop nest.
9563 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9565 (distribute_loop): Compute global variable loop nest. Update use.
9567 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9569 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9570 (partition_merge_into): New parameter. Dump reason for fusion.
9571 (distribute_loop): Update use of partition_merge_into.
9573 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9575 * tree-loop-distribution.c (bb_top_order_index): New.
9576 (bb_top_order_index_size, bb_top_order_cmp): New.
9577 (stmts_from_loop): Use topological order.
9578 (pass_loop_distribution::execute): Compute and release topological
9579 order for basic blocks.
9581 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9583 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9586 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9588 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9589 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9590 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9591 * internal-fn.def (LOOP_DIST_ALIAS): New.
9592 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9593 (fold_loop_internal_call): ... this.
9594 (vect_loop_dist_alias_call): New function.
9595 (set_uid_loop_bbs): Call fold_loop_internal_call.
9596 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9599 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9602 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9603 Require dead FLAGS_REG at the beginning of a peephole.
9605 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9608 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9609 arguments in the call to __builtin_ia32_sbb_u32.
9610 (_subborrow_u64): Swap _X and _Y arguments in the call to
9611 __builtin_ia32_sbb_u64.
9613 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9616 * tree-vrp.c (compare_assert_loc): Turn into a function template
9617 with stable template parameter. Only test if a->e is NULL,
9618 !a->e == !b->e has been verified already. Use e == NULL or
9619 e != NULL instead of e or ! e tests. If stable is true, don't use
9620 iterative_hash_expr, on the other side allow a or b or both NULL
9621 and sort the NULLs last.
9622 (process_assert_insertions): Sort using compare_assert_loc<false>
9623 instead of compare_assert_loc, later sort using
9624 compare_assert_loc<true> before calling process_assert_insertions_for
9625 in a loop. Use break instead of continue once seen NULL pointer.
9627 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9629 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9630 Cortex-R7 and Cortex-R8 processors.
9632 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9634 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9635 uninitialized while src is not.
9637 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9639 * common/config/arm/arm-common.c: Adjust include path for
9641 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9642 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9643 (arm-cpu-data.h): Likewise.
9644 (arm-cpu-cdata.h): Likewise.
9645 * config/arm/arm-cpu.h: Delete.
9646 * config/arm/arm-cpu-cdata.h: Delete.
9647 * config/arm/arm-cpu-data.h: Delete.
9649 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9651 * config/arm/arm-cpus.in (cortex-a55): New.
9652 (cortex-a75): Likewise.
9653 (cortex-a75.cortex-a55): Likewise.
9654 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9656 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9657 * config/arm/arm-cpu-cdata.h: Regenerate.
9658 * config/arm/arm-cpu-data.h: Regenerate.
9659 * config/arm/arm-cpu.h: Regenerate.
9660 * config/arm/arm-tables.opt: Regenerate.
9661 * config/arm/arm-tune.md: Regenerate.
9663 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9665 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9667 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9669 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9672 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9674 PR tree-optimization/81292
9675 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9676 full_string_p, also call adjust_related_strinfos if the adjustment
9677 is simple, otherwise invalidate related strinfos.
9679 2017-07-04 Martin Liska <mliska@suse.cz>
9682 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9683 newly created variable as DECL_IGNORED_P.
9685 2017-07-04 Martin Liska <mliska@suse.cz>
9688 * ipa-inline.c (inline_small_functions):
9689 Use xstrdup_for_dump.
9691 2017-07-04 Tom de Vries <tom@codesourcery.com>
9693 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9695 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9698 * config/darwin.c (darwin_function_switched_text_sections):
9701 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9703 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9705 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9707 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9709 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9711 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9712 min_profitable_iters, and th as inclusive lower bounds.
9713 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9714 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9715 for min_profitable_iters and min_profitable_estimate.
9716 (vect_transform_loop): Treat th as an inclusive lower bound.
9717 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9719 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9722 * config/darwin.c (darwin_function_switched_text_sections):
9723 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9724 in two pieces, and suppress the use of buf.
9726 2017-07-03 Nathan Sidwell <nathan@acm.org>
9728 * hash-table.h (hash_table_mod1): Fix indentation.
9730 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9733 * predict.c (force_edge_cold): Be more careful about propagation
9735 * profile-count.h (profile_probability::guessed,
9736 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9738 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9740 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9742 * doc/invoke.texi (rcpc architecture extension): Document it.
9744 2017-07-03 Richard Biener <rguenther@suse.de>
9746 PR tree-optimization/60510
9747 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9748 the scalar reduction PHI and use it.
9749 (vectorizable_reduction): Properly guard the single_defuse_cycle
9750 path for non-SLP reduction chains where we cannot use it.
9751 Rework reduc_def/index and vector type deduction. Rework
9752 vector operand gathering during reduction op code-gen.
9753 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9754 chains dissolve the chain and leave it to non-SLP reduction
9757 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9759 * tree-data-ref.h (dr_alignment): Declare.
9760 * tree-data-ref.c (dr_alignment): New function.
9761 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9762 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9764 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9766 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9768 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9769 and base_misalignment fields.
9770 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9771 * tree-data-ref.c: Include builtins.h.
9772 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9773 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9774 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9775 * tree-vect-data-refs.c: Include tree-cfg.h.
9776 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9777 fields instead of calculating an alignment here.
9778 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9779 innermost_loop_behavior fields.
9781 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9783 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9785 (DR_STEP_ALIGNMENT): New macro.
9786 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9787 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9788 (create_data_ref): Print it.
9789 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9790 to tell whether the step preserves vector (mis)alignment.
9791 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9792 Move the check for an integer step and generalise to all INTEGER_CST.
9793 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9794 Print the outer step alignment.
9796 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9798 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9799 with offset_alignment.
9800 (DR_ALIGNED_TO): Delete.
9801 (DR_OFFSET_ALIGNMENT): New macro.
9802 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9803 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9804 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9805 (create_data_ref): Likewise.
9806 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9807 (vect_analyze_data_refs): Likewise.
9808 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9809 creating dummy innermost behavior.
9811 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9813 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9814 with a "innermost_loop_behavior *" and refeence tree.
9815 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9816 (create_data_ref): Update call accordingly.
9817 * tree-predcom.c (find_looparound_phi): Likewise.
9819 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9821 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9822 fields with dr_wrt_vec_loop.
9823 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9824 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9825 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9826 (vect_dr_behavior): New function.
9827 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9828 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9829 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9830 track whether the step preserves the misalignment.
9831 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9832 Use vect_dr_behavior.
9833 (vect_setup_realignment): Update call accordingly.
9834 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9835 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9836 call to vect_create_addr_base_for_vector_ref.
9837 (vect_create_cond_for_align_checks): Likewise.
9838 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9839 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9840 (vect_recog_mask_conversion_pattern): Likewise.
9841 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9842 (new_stmt_vec_info): Remove redundant zeroing.
9844 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9846 * common/config/arm/arm-common.c (arm_be8_option): New function.
9847 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9848 (ISA_ARMv6): Add isa_bit_be8.
9849 * config/arm/arm.h (arm_be8_option): Add prototype.
9850 (BE8_SPEC_FUNCTION): New define.
9851 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9852 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9853 (mlittle-endian): Similarly.
9854 (mbe8, mbe32): New options.
9855 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9856 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9858 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9860 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9862 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9864 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9865 (cleanup_tree_cfg_bb): Use it.
9866 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9868 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9870 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9873 * loop-doloop.c (add_test): Update profile.
9875 2017-07-03 Martin Liska <mliska@suse.cz>
9878 * sanopt.c (rewrite_usage_of_param): New function.
9879 (sanitize_rewrite_addressable_params): Likewise.
9880 (pass_sanopt::execute): Call rewrite_usage_of_param.
9882 2017-07-03 Richard Biener <rguenther@suse.de>
9884 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9885 back to using VIEW_CONVERT_EXPR.
9887 2017-07-03 Martin Liska <mliska@suse.cz>
9890 * doc/extend.texi: Document when a resolver function is
9891 generated for target_clones.
9893 2017-07-03 Martin Liska <mliska@suse.cz>
9895 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9896 and shadow memory of auto variables which are subject of
9897 use-after-scope sanitization.
9898 (asan_expand_mark_ifn): Add do set only when is_poison.
9900 2016-07-03 Richard Biener <rguenther@suse.de>
9902 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9904 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9905 (vectorizable_reduction): Perform reduction PHI creation when
9906 visiting a reduction PHI and adjust and simplify code generation
9907 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9908 (vect_transform_loop): Visit reduction PHIs.
9909 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9910 defs into the SLP tree.
9911 (vect_build_slp_tree): Reduction defs terminate the recursion.
9912 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9914 (vect_get_vec_defs_for_stmt_copy): Export.
9915 (vect_get_vec_defs): Likewise.
9916 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9918 (vect_get_vec_defs_for_stmt_copy): Declare.
9919 (vect_get_vec_defs): Likewise.
9921 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9923 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9924 parameter with a "loop" parameter and use it instead of the
9925 loop containing DR_STMT. Don't check simple_iv when doing
9926 BB analysis. Describe the two analysis modes in the comment.
9928 2017-07-03 Tom de Vries <tom@codesourcery.com>
9930 PR tree-optimization/69468
9931 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9932 (find_same_succ_bb): Handle ignore_edge_flags.
9934 2017-07-03 Tom de Vries <tom@codesourcery.com>
9936 PR tree-optimization/81192
9937 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9939 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9941 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9943 2017-07-03 Tom de Vries <tom@codesourcery.com>
9945 PR tree-optimization/81192
9946 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9947 BB_SAME_SUCC (bb) == NULL.
9949 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9951 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9954 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9956 * dumpfile.c: Include profile-count.h
9957 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9959 (insert_cond_bb): Update profile.
9960 * tree-cfg.h (insert_cond_bb): Update prototype.
9961 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9962 * tree-dump.c: Do not include tree-cfg.
9964 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9966 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9968 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9970 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9973 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9975 * tree-complex.c (expand_complex_div_wide): update profile.
9977 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9978 Alan Hayward <alan.hayward@arm.com>
9979 David Sherwood <david.sherwood@arm.com>
9981 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9982 (CORETYPES_H): New define.
9983 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9984 (insn-modes-inline.h, s-modes-inline-h): New rules.
9985 (generated_files): Add insn-modes-inline.h.
9986 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9987 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9988 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9989 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9990 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9991 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9992 (build/gencodes.o, build/genconditions.o): Likewise.
9993 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9994 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9995 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9996 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9997 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9998 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9999 * coretypes.h: Include everything up to real.h for generators.
10000 Include insn-modes.h first. Include wide-int-print.h after
10001 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10002 * machmode.h: Don't include insn-modes.h here.
10003 * function-tests.c: Remove includes of signop.h, machmode.h,
10004 double-int.h and wide-int.h.
10006 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10008 * optc-save-gen.awk: Likewise.
10009 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10010 * godump.c: Remove include of wide-int-print.h.
10011 * pretty-print.h: Likewise.
10012 * wide-int-print.cc: Likewise.
10013 * wide-int.cc: Likewise.
10014 * hash-map-tests.c: Remove include of signop.h.
10015 * hash-set-tests.c: Likewise.
10016 * rtl-tests.c: Likewise.
10017 * mkconfig.sh: Remove include of machmode.h.
10018 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10020 (emit_insn_modes_inline_h): ...this new function. Emit the code
10021 into an insn-modes-inline.h header file, adding appropriate
10022 include guards and end comments.
10023 (emit_insn_modes_c_header): Remove include of machmode.h.
10024 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10026 (main): Handle -i flag and call emit_insn_modes_inline_h when
10029 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10031 * tree-ssa-strlen.c (strinfo): Rename the length field to
10032 nonzero_chars. Add a full_string_p field.
10033 (compare_nonzero_chars, zero_length_string_p): New functions.
10034 (get_addr_stridx): Add an offset_out parameter.
10035 Use compare_nonzero_chars.
10036 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10037 (new_strinfo): Update after above changes to strinfo.
10038 (set_endptr_and_length): Set full_string_p.
10039 (get_string_length): Update after above changes to strinfo.
10040 (unshare_strinfo): Update call to new_strinfo.
10041 (maybe_invalidate): Likewise.
10042 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10043 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10044 as a uhwi instead of an shwi. Update after above changes to
10045 strinfo and new_strinfo.
10046 (zero_length_string): Assert that chainsi contains full strings.
10047 Use zero_length_string_p. Update call to new_strinfo.
10048 (adjust_related_strinfos): Update after above changes to strinfo.
10049 Copy full_string_p from origsi.
10050 (adjust_last_stmt): Use zero_length_string_p.
10051 (handle_builtin_strlen): Update after above changes to strinfo and
10052 new_strinfo. Install the lhs as the string length if the previous
10053 entry didn't describe a full string.
10054 (handle_builtin_strchr): Update after above changes to strinfo
10056 (handle_builtin_strcpy): Likewise.
10057 (handle_builtin_strcat): Likewise.
10058 (handle_builtin_malloc): Likewise.
10059 (handle_pointer_plus): Likewise.
10060 (handle_builtin_memcpy): Likewise. Track nonzero characters
10061 that aren't necessarily followed by a nul terminator.
10062 (handle_char_store): Likewise.
10064 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10066 PR tree-optimization/80769
10067 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10068 for malloc and calloc. Document the new invariant that all related
10069 strinfos have delayed lengths or none do.
10070 (verify_related_strinfos): Move earlier in file.
10071 (set_endptr_and_length): New function, split out from...
10072 (get_string_length): ...here. Also set the lengths of related
10074 (zero_length_string): Assert that chainsi has known (rather than
10076 (adjust_related_strinfos): Likewise.
10078 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10080 PR tree-optimization/81136
10081 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10082 assert that two references with the same misalignment have the same
10083 compile-time misalignment if those compile-time misalignments
10086 2017-07-01 Andi Kleen <ak@linux.intel.com>
10088 * print-tree.c (print_node): Print all attributes.
10090 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10092 * cfg.c (scale_bbs_frequencies): New function.
10093 * cfg.h (scale_bbs_frequencies): Declare it.
10094 * cfgloopanal.c (single_likely_exit): Cleanup.
10095 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10097 (scale_loop_profile): Likewise.
10098 (loop_version): Likewise.
10099 (create_empty_loop_on_edge): Update.
10100 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10101 scale_loop_frequencies, scale_loop_profile, loopify,
10102 loop_version): Update prototypes.
10103 * modulo-sched.c (sms_schedule): Update.
10104 * predict.c (unlikely_executed_edge_p): Also check probability.
10105 (probably_never_executed_edge_p): Fix typo.
10106 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10107 * tree-parloops.c (gen_parallel_loop): Update.
10108 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10109 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10110 * tree-ssa-loop-split.c (split_loop): Update.
10111 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10112 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10113 (vect_loop_versioning): Update.
10114 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10116 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10118 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10120 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10122 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10123 to keep profile consistent.
10125 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10127 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10128 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10129 * profile-count.h (max_safe_multiplier): Make unsigned.
10130 (profile_count::guessed_zero): New.
10132 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10134 * bb-reorder.c (fix_up_crossing_landing_pad,
10135 fix_crossing_conditional_branches): Use make_single_succ_edge
10136 to keep profile consistent.
10138 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10140 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10143 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10146 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10147 the right scopes, make sure cond_jump isn't preserved between multiple
10148 iterations. Search for fallthru edge whenever there are 3+ edges and
10149 use find_fallthru_edge for it.
10151 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10153 Patch by Alexander Monakov <amonakov@ispras.ru>
10154 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10155 probabilities consistently.
10157 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10159 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10160 emit_cmp_and_jump_insns.
10162 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10165 * tree-inline.c (expand_call_inline): Combine profile statuses.
10167 2017-06-30 Andrew Pinski <apinski@cavium.com>
10169 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10170 fold_stmt returned true.
10172 2017-06-30 Nathan Sidwell <nathan@acm.org>
10174 * ggc.h (empty_string): Delete.
10175 * cfgexpand.c (expand_asm_stmt): Use plain "".
10176 * optabs.c (expand_asm_memory_barrier): Likewise.
10177 * stringpool.c (empty_string): Delete.
10178 (digit_vector, digit_string): Delete.
10179 (ggc_alloc_string): Use plain "", don't optimize single digit
10180 strings. Use ggc_alloc_atomic.
10182 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10184 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10185 comparison set and one other set, use the cost of the non-comparison
10188 2017-06-30 Nathan Sidwell <nathan@acm.org>
10190 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10193 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10195 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10196 loops. Remove now unneeded calls to gimple_switch_set_label() that
10197 just set removed labels to NULL_TREE.
10199 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10201 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10202 (set_range_info): ...here. Only call set_range_info_raw if domain
10204 (set_nonzero_bits): Call set_range_info_raw.
10205 * tree-ssanames.h (set_range_info_raw): New.
10207 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10210 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10211 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10212 of nonimmediate_operand and <store_mask_constraint> instead of m
10213 for the input operand. For V8FI iterator, always split if input
10214 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10215 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10216 <store_mask_predicate> instead of register_operand and
10217 <store_mask_constraint> instead of v for the input operand. Make
10218 sure both operands aren't MEMs for if not <mask_applied>.
10220 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10222 * lto-wrapper.c (copy_file) Close both file descriptors before
10225 2017-06-30 Martin Liska <mliska@suse.cz>
10228 * multiple_target.c (create_dispatcher_calls): Make ifunc
10229 also for function that don't have calls or are not referenced.
10231 2017-06-30 Richard Biener <rguenther@suse.de>
10233 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10234 analyze the first scalar stmt. Move vector type computation
10235 for the BB case here from ...
10236 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10237 live operation processing in the SLP case properly.
10239 2017-06-30 Richard Biener <rguenther@suse.de>
10241 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10243 2017-06-30 Martin Liska <mliska@suse.cz>
10246 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10247 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10249 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10251 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10254 2017-06-30 Richard Biener <rguenther@suse.de>
10256 PR tree-optimization/81249
10257 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10258 condition reduction result to original scalar type.
10260 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10262 * profile-count.h (enum profile_quality): Fix typos and whitespace
10265 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10267 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10268 type for branch probabilities.
10270 2017-06-29 Julian Brown <julian@codesourcery.com>
10271 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10273 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10274 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10275 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10276 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10278 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10280 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10281 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10282 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10283 CC usage from generic code to here.
10284 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10285 CC usage into the target macros.
10287 2017-06-29 Maya Rashish <coypu@sdf.org>
10289 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10292 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10294 * arm/arm-builtins.c: Include profile-count.h
10295 * except.c (sjlj_emit_function_enter): Use
10296 profile_probability::unlikely.
10298 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10300 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10301 and tocrel_offset be pointer args rather than implicitly using
10303 (legitimate_constant_pool_address_p, rs6000_emit_move,
10304 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10305 tocrel_offset and use in toc_relative_expr_p call.
10306 (print_operand, print_operand_address): Use static tocrel_base_oac
10307 and tocrel_offset_oac.
10308 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10311 2017-06-29 Maya Rashish <coypu@sdf.org>
10313 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10315 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10317 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10318 objects, take into account only the alignment of 'op0' and 'mode1' if
10321 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10323 * ccmp.c (ccmp_tree_comparison_p): New function.
10324 (ccmp_candidate_p): Update to use above function.
10325 (get_compare_parts): New function.
10326 (expand_ccmp_next): Update to use new functions.
10327 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10329 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10330 take mode as argument.
10331 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10332 * expr.c (expand_expr_real_1): Pass mode as argument.
10334 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10336 * combine.c (combine_instructions): Print insns to dump_file, together
10339 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10341 * asan.c (asan_emit_stack_protection): Update.
10342 (create_cond_insert_point): Update.
10343 * auto-profile.c (afdo_propagate_circuit): Update.
10344 * basic-block.h (struct edge_def): Turn probability to
10345 profile_probability.
10346 (EDGE_FREQUENCY): Update.
10347 * bb-reorder.c (find_traces_1_round): Update.
10348 (better_edge_p): Update.
10349 (sanitize_hot_paths): Update.
10350 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10351 (make_single_succ_edge): Update.
10352 (check_bb_profile): Update.
10353 (dump_edge_info): Update.
10354 (update_bb_profile_for_threading): Update.
10355 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10356 probabilitycount to 0.
10357 * cfgbuild.c (compute_outgoing_frequencies): Update.
10358 * cfgcleanup.c (try_forward_edges): Update.
10359 (outgoing_edges_match): Update.
10360 (try_crossjump_to_edge): Update.
10361 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10362 (expand_gimple_tailcall): Update.
10363 (construct_init_block): Use make_single_succ_edge.
10364 (construct_exit_block): Use make_single_succ_edge.
10365 * cfghooks.c (verify_flow_info): Update.
10366 (redirect_edge_succ_nodup): Update.
10367 (split_edge): Update.
10368 (account_profile_record): Update.
10369 * cfgloopanal.c (single_likely_exit): Update.
10370 * cfgloopmanip.c (scale_loop_profile): Update.
10371 (set_zero_probability): Remove.
10372 (duplicate_loop_to_header_edge): Update.
10373 * cfgloopmanip.h (loop_version): Update prototype.
10374 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10375 (force_nonfallthru_and_redirect): Update.
10376 (update_br_prob_note): Update.
10377 (rtl_verify_edges): Update.
10378 (purge_dead_edges): Update.
10379 (rtl_lv_add_condition_to_bb): Update.
10380 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10381 * cgraphunit.c (init_lowered_empty_function): Update.
10382 (cgraph_node::expand_thunk): Update.
10383 * cilk-common.c: Include profile-count.h
10384 * dojump.c (inv): Remove.
10385 (jumpifnot): Update.
10386 (jumpifnot_1): Update.
10387 (do_jump_1): Update.
10389 (do_jump_by_parts_greater_rtx): Update.
10390 (do_compare_rtx_and_jump): Update.
10391 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10392 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10393 * dwarf2cfi.c: Include profile-count.h
10394 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10395 (sjlj_emit_dispatch_table): Likewise.
10396 * explow.c: Include profile-count.h
10397 * expmed.c (emit_store_flag_force): Update.
10398 (do_cmp_and_jump): Update.
10399 * expr.c (compare_by_pieces_d::generate): Update.
10400 (compare_by_pieces_d::finish_mode): Update.
10401 (emit_block_move_via_loop): Update.
10402 (store_expr_with_bounds): Update.
10403 (store_constructor): Update.
10404 (expand_expr_real_2): Update.
10405 (expand_expr_real_1): Update.
10406 * expr.h (try_casesi, try_tablejump): Update prototypes.
10407 * gimple-pretty-print.c (dump_probability): Update.
10408 (dump_profile): New.
10409 (dump_gimple_label): Update.
10410 (dump_gimple_bb_header): Update.
10411 * graph.c (draw_cfg_node_succ_edges): Update.
10412 * hsa-gen.c (convert_switch_statements): Update.
10413 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10414 (find_if_case_1): Update.
10415 (find_if_case_2): Update.
10416 * internal-fn.c (expand_arith_overflow_result_store): Update.
10417 (expand_addsub_overflow): Update.
10418 (expand_neg_overflow): Update.
10419 (expand_mul_overflow): Update.
10420 (expand_vector_ubsan_overflow): Update.
10421 * ipa-cp.c (good_cloning_opportunity_p): Update.
10422 * ipa-split.c (split_function): Use make_single_succ_edge.
10423 * ipa-utils.c (ipa_merge_profiles): Update.
10424 * loop-doloop.c (add_test): Update.
10425 (doloop_modify): Update.
10426 * loop-unroll.c (compare_and_jump_seq): Update.
10427 (unroll_loop_runtime_iterations): Update.
10428 * lra-constraints.c (lra_inheritance): Update.
10429 * lto-streamer-in.c (input_cfg): Update.
10430 * lto-streamer-out.c (output_cfg): Update.
10431 * mcf.c (adjust_cfg_counts): Update.
10432 * modulo-sched.c (sms_schedule): Update.
10433 * omp-expand.c (expand_omp_for_init_counts): Update.
10434 (extract_omp_for_update_vars): Update.
10435 (expand_omp_ordered_sink): Update.
10436 (expand_omp_for_ordered_loops): Update.
10437 (expand_omp_for_generic): Update.
10438 (expand_omp_for_static_nochunk): Update.
10439 (expand_omp_for_static_chunk): Update.
10440 (expand_cilk_for): Update.
10441 (expand_omp_simd): Update.
10442 (expand_omp_taskloop_for_outer): Update.
10443 (expand_omp_taskloop_for_inner): Update.
10444 * omp-simd-clone.c (simd_clone_adjust): Update.
10445 * optabs.c (expand_doubleword_shift): Update.
10446 (expand_abs): Update.
10447 (emit_cmp_and_jump_insn_1): Update.
10448 (expand_compare_and_swap_loop): Update.
10449 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10450 * predict.c (predictable_edge_p): Update.
10451 (edge_probability_reliable_p): Update.
10452 (set_even_probabilities): Update.
10453 (combine_predictions_for_insn): Update.
10454 (combine_predictions_for_bb): Update.
10455 (propagate_freq): Update.
10456 (estimate_bb_frequencies): Update.
10457 (force_edge_cold): Update.
10458 * profile-count.c (profile_count::dump): Add missing space into dump.
10459 (profile_count::debug): Add newline.
10460 (profile_count::differs_from_p): Explicitly convert to unsigned.
10461 (profile_count::stream_in): Update.
10462 (profile_probability::dump): New member function.
10463 (profile_probability::debug): New member function.
10464 (profile_probability::differs_from_p): New member function.
10465 (profile_probability::differs_lot_from_p): New member function.
10466 (profile_probability::stream_in): New member function.
10467 (profile_probability::stream_out): New member function.
10468 * profile-count.h (profile_count_quality): Rename to ...
10469 (profile_quality): ... this one.
10470 (profile_probability): New.
10471 (profile_count): Update.
10472 * profile.c (compute_branch_probabilities): Update.
10473 * recog.c (peep2_attempt): Update.
10474 * sched-ebb.c (schedule_ebbs): Update.
10475 * sched-rgn.c (find_single_block_region): Update.
10476 (compute_dom_prob_ps): Update.
10477 (schedule_region): Update.
10478 * sel-sched-ir.c (compute_succs_info): Update.
10479 * stmt.c (struct case_node): Update.
10480 (do_jump_if_equal): Update.
10481 (get_outgoing_edge_probs): Update.
10482 (conditional_probability): Update.
10483 (emit_case_dispatch_table): Update.
10484 (expand_case): Update.
10485 (expand_sjlj_dispatch_table): Update.
10486 (emit_case_nodes): Update.
10487 * targhooks.c: Update.
10488 * tracer.c (better_p): Update.
10489 (find_best_successor): Update.
10490 * trans-mem.c (expand_transaction): Update.
10491 * tree-call-cdce.c: Update.
10492 * tree-cfg.c (gimple_split_edge): Upate.
10493 (move_sese_region_to_fn): Upate.
10494 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10495 * tree-eh.c (lower_resx): Upate.
10496 (cleanup_empty_eh_move_lp): Upate.
10497 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10498 * tree-inline.c (copy_edges_for_bb): Update.
10499 (copy_cfg_body): Update.
10500 * tree-parloops.c (gen_parallel_loop): Update.
10501 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10502 (gimple_gen_time_profiler): Update.
10503 * tree-ssa-dce.c (remove_dead_stmt): Update.
10504 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10505 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10506 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10507 (unloop_loops): Update.
10508 (try_peel_loop): Update.
10509 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10510 * tree-ssa-loop-split.c (connect_loops): Update.
10511 (split_loop): Update.
10512 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10513 (hoist_guard): Update.
10514 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10515 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10516 (value_replacement): Update.
10517 * tree-ssa-reassoc.c (branch_fixup): Update.
10518 * tree-ssa-tail-merge.c (replace_block_by): Update.
10519 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10520 (create_edge_and_update_destination_phis): Update.
10521 (compute_path_counts): Update.
10522 (recompute_probabilities): Update.
10523 (update_joiner_offpath_counts): Update.
10524 (freqs_to_counts_path): Update.
10525 (duplicate_thread_path): Update.
10526 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10527 (struct switch_conv_info): Update.
10528 (gen_inbound_check): Update.
10529 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10530 (vect_do_peeling): Update.
10531 (vect_loop_versioning): Update.
10532 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10533 (optimize_mask_stores): Update.
10534 * ubsan.c (ubsan_expand_null_ifn): Update.
10535 * value-prof.c (gimple_divmod_fixed_value): Update.
10536 (gimple_divmod_fixed_value_transform): Update.
10537 (gimple_mod_pow2): Update.
10538 (gimple_mod_pow2_value_transform): Update.
10539 (gimple_mod_subtract): Update.
10540 (gimple_mod_subtract_transform): Update.
10541 (gimple_ic): Update.
10542 (gimple_stringop_fixed_value): Update.
10543 (gimple_stringops_transform): Update.
10544 * value-prof.h: Update.
10546 2017-06-29 Carl Love <cel@us.ibm.com>
10548 * config/rs6000/rs6000-c.c: Add support for built-in functions
10549 vector signed int vec_signed (vector float);
10550 vector signed long long vec_signed (vector double);
10551 vector signed int vec_signed2 (vector double, vector double);
10552 vector signed int vec_signede (vector double);
10553 vector signed int vec_signedo (vector double);
10554 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10555 instruction generator.
10556 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10557 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10558 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10560 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10561 vunsignede_v2df): Add define_expands.
10562 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10563 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10564 VEC_UNSIGNEDO): Add definitions.
10565 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10566 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10567 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10568 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10569 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10570 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10571 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10572 vec_unsignede, vec_unsignedo): Add builtin defines.
10573 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10575 * doc/extend.texi: Update the built-in documentation file for the
10576 new built-in functions.
10578 2017-06-29 Richard Biener <rguenther@suse.de>
10580 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10581 reduction chains to LOOP_VINFO_REDUCTIONS.
10582 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10583 SLP reductions after processing reduction chains.
10585 2017-06-29 Nathan Sidwell <nathan@acm.org>
10587 * builtins.c (fold_builtin_FUNCTION): Use
10588 lang_hooks.decl_printable_name.
10590 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10592 PR middle-end/81194
10593 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10594 with only one label.
10595 * stmt.c (expand_case): Assert NCASES is greater than one.
10597 2017-06-29 Richard Biener <rguenther@suse.de>
10599 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10601 (group_case_labels): Likewise.
10602 (find_taken_edge): Push sanity checking on val to workers...
10603 (find_taken_edge_cond_expr): ... here
10604 (find_taken_edge_switch_expr): ... and here, handle cases
10605 with just a default label.
10606 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10607 (group_case_labels): Likewise.
10608 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10609 group_case_labels does anything cleanup the CFG again.
10611 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10613 PR tree-optimization/81196
10614 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10615 exit condition comparing two IVs.
10617 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10619 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10620 profile to the dummy entry at the end of the list of architectures.
10621 * config/arm/arm-cpu-cdata.h: Regenerated.
10623 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10624 Michael Collison <michael.collison@arm.com>
10627 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10629 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10630 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10631 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10632 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10633 of shift when the shift amount is masked with constant equal to
10634 the size of the mode.
10635 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10638 2017-06-29 Martin Liska <mliska@suse.cz>
10640 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10642 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10645 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10648 (edge_growth_cache_entry::edge_growth_cache_entry): New
10650 (reset_edge_growth_cache): Update to use constructor.
10652 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10654 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10655 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10656 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10658 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10660 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10661 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10663 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10665 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10666 (*-linux-uclibc*): Add t-uclibc tmake_file.
10667 * config/t-musl: New.
10668 * config/t-uclibc: New.
10670 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10672 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10674 (gen_comm_data): Emit architectural setting of arch_prof.
10675 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10677 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10678 (armv8-m.base, armv8-m.main): Likewise.
10679 * arm-protos.h (arm_build_target): Add profile field.
10680 (arch_option): Likewise.
10681 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10683 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10684 arm_active_target.profile.
10686 2017-06-28 Richard Biener <rguenther@suse.de>
10688 PR middle-end/81227
10689 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10690 TYPE_OVERFLOW_WRAPS.
10691 * match.pd (negate_expr_p): Likewise.
10692 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10693 fold_build2, not fold_binary.
10695 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10697 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10698 Convert memory address to Pmode.
10699 (aarch64_print_operand): Assert MEM operands are always Pmode.
10701 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10704 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10705 Remove redundant if.
10706 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10707 * config/arm/aarch-common-protos.h
10708 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10709 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10711 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10714 * multiple_target.c (create_dispatcher_calls): Set the default
10715 clone to be static, not public.
10717 2017-06-28 Richard Biener <rguenther@suse.de>
10719 * tree-vect-loop.c (vectorizable_reduction): Move special
10720 cond reduction IV var creation ...
10721 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10722 parameter. Use STMT_VINFO_VECTYPE.
10723 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10726 2017-06-28 Martin Liska <mliska@suse.cz>
10729 * ipa-visibility.c (non_local_p): Handle visibility.
10731 2017-06-28 Martin Liska <mliska@suse.cz>
10734 * common.opt: Add IntegerRange to various options.
10735 * opt-functions.awk (integer_range_info): New function.
10736 * optc-gen.awk: Add integer_range_info to cl_options struct.
10737 * opts-common.c (decode_cmdline_option): Handle
10738 CL_ERR_INT_RANGE_ARG.
10739 (cmdline_handle_error): Likewise.
10740 * opts.c (print_filtered_help): Show valid interval in
10741 when --help is provided.
10742 * opts.h (struct cl_option): Add range_min and range_max fields.
10743 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10745 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10747 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10748 (x * C EQ/NE y * C): New transformation.
10750 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10752 * genmultilib (combination_space): Accept '+' in option names.
10754 2017-06-28 Martin Liska <mliska@suse.cz>
10757 * asan.c (instrument_derefs): Bail out inner references
10758 that are hard register variables.
10760 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10763 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10764 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10766 2017-06-28 Richard Biener <rguenther@suse.de>
10768 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10769 (vect_get_slp_defs): Adjust.
10770 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10772 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10774 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10775 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10776 (vectorizable_reduction): Adjust.
10777 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10779 (vect_get_slp_defs): Likewise.
10780 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10781 (vectorizable_bswap): Adjust.
10782 (vectorizable_call): Likewise.
10783 (vectorizable_conversion): Likewise.
10784 (vectorizable_assignment): Likewise.
10785 (vectorizable_shift): Likewise.
10786 (vectorizable_operation): Likewise.
10787 (vectorizable_store): Likewise.
10788 (vectorizable_condition): Likewise.
10789 (vectorizable_comparison): Likewise.
10791 2017-06-28 Michael Collison <michael.collison@arm.com>
10794 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10796 (arm_gen_movmemqi): Removed unused variable 'i'.
10797 Convert 'for' loop into 'while' loop.
10798 (arm_expand_prologue): Remove last unnecessary set of insn.
10799 (thumb_pop): Remove unused variable 'pushed_words'.
10800 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10802 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10804 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10805 * config/s390/s390-protos.h: Add prototype of
10806 s390_rel_address_ok_p.
10807 * config/s390/s390.c (s390_got_symbol): New function.
10808 (s390_rel_address_ok_p): New function.
10809 (legitimize_pic_address): Use s390_rel_address_ok_p.
10810 (s390_load_got): Use s390_got_symbol.
10811 (s390_option_override): Issue error if
10812 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10813 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10815 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10817 2017-06-27 Andrew Pinski <apinski@cavium.com>
10819 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10820 (X * copysign (1.0, X)): New pattern.
10821 (X * copysign (1.0, -X)): New pattern.
10822 (copysign (-1.0, CST)): New pattern.
10824 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10826 * genmultilib (combination_space): Remove variable.
10827 Validate reuse rules against regular expression for any sequence
10828 of multilib options in any order.
10830 2017-06-27 Michael Collison <michael.collison@arm.com>
10832 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10833 call aarch64_split_simd_combine.
10834 * (aarch64_combine_internal<mode>): Delete pattern.
10835 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10836 Allow register and subreg operands.
10838 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10840 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10841 specific need, just fallback on defaults.
10842 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10844 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10845 Olivier Hainque <hainque@adacore.com>
10847 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10849 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10850 targets. Pick a default if no particular attempt applied.
10851 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10854 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10856 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10857 (x86_64-wrs-vxworks7): Likewise.
10859 2017-06-27 Marek Polacek <polacek@redhat.com>
10862 * ubsan.c (instrument_null): Check get_base_address's result for null.
10864 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10866 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10868 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10870 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10871 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10872 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10873 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10874 New function types.
10875 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10876 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10877 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10878 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10879 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10880 BUILT_IN_FEUPDATEENV): New builtins.
10881 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10882 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10883 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10884 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10886 (builtin_structptr_types): Adjust size.
10887 * tree.c (builtin_structptr_types): Add four entries.
10889 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10890 Olivier Hainque <hainque@adacore.com>
10892 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10893 (TLS_SYM): New local macro, forcing reference to __tls__ on
10894 link command lines for VxWorks 7 RTPs, triggering initialization
10896 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10897 OS features TLS support, true for RTPs on VxWorks 7.
10898 * config/vxworks.c (vxworks_override_options): Setup emutls
10901 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10903 * predict.c (test_prediction_value_range): Use -1U instead of -1
10904 to avoid narrowing conversion warning.
10905 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10906 to avoid narrowing conversion warning.
10907 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10909 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10911 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10913 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10914 64bit configurations.
10915 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10916 (SIZE_TYPE): Likewise.
10917 * config/vxworks.c (vxworks_emutls_var_fields): Use
10918 long_unsigned_type_node instead of unsigned_type_node as the offset
10919 field type, which is "pointer" mode in emutls.c.
10921 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10924 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10926 PR middle-end/81207
10927 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10928 gimple_vuse copying separately from gimple_vdef copying.
10930 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10932 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10935 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10936 Olivier Hainque <hainque@adacore.com>
10938 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10939 for all vxworks7 targets.
10940 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10941 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10942 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10943 variations for VX6/VX7 and 32/64bits later on in ...
10944 (VXWORKS_LIB_SPEC): Leverage new macros.
10945 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10946 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10948 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10950 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10951 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10953 2017-06-26 Carl Love <cel@us.ibm.com>
10955 * config/rs6000/rs6000-c.c: Add support for built-in functions
10956 vector bool char vec_reve (vector bool char);
10957 vector signed char vec_reve (vector signed char);
10958 vector unsigned char vec_reve (vector unsigned char);
10959 vector bool int vec_reve (vector bool int);
10960 vector signed int vec_reve (vector signed int);
10961 vector unsigned int vec_reve (vector unsigned int);
10962 vector bool long long vec_reve (vector bool long long);
10963 vector signed long long vec_reve (vector signed long long);
10964 vector unsigned long long vec_reve (vector unsigned long long);
10965 vector bool short vec_reve (vector bool short);
10966 vector signed short vec_reve (vector signed short);
10967 vector double vec_reve (vector double);
10968 vector float vec_reve (vector float);
10969 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10970 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10971 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10972 (altivec_vreve): New pattern.
10973 * config/rs6000/altivec.h (vec_reve): New define.
10974 * doc/extend.texi (vec_rev): Update the built-in documentation file
10975 for the new built-in functions.
10977 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10979 PR tree-optimization/71815
10980 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10982 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10984 (slsr_process_phi): Likewise.
10985 (replace_uncond_cands_and_profitable_phis): Don't replace a
10986 multiply candidate with a stride of 1 (copy or cast).
10987 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10989 (lowest_cost_path): Likewise.
10990 (total_savings): Likewise.
10992 2017-06-26 Richard Biener <rguenther@suse.de>
10995 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10996 Use def_builtin_pure for all gather builtins.
10998 2017-06-26 Richard Biener <rguenther@suse.de>
11000 PR tree-optimization/81203
11001 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11002 non-dominating BBs.
11004 2017-06-26 Marek Polacek <polacek@redhat.com>
11007 * doc/invoke.texi: Document -Wmultistatement-macros.
11009 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11011 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11012 arm_neon_ok_no_float_abi effective target.
11014 2017-06-26 Richard Biener <rguenther@suse.de>
11016 PR tree-optimization/80928
11017 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11018 (copy_bbs): Set BB_DUPLICATED flag early.
11019 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11021 (execute_on_shrinking_pred): Likewise.
11022 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11023 BB_DUPLICATED blocks.
11024 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11025 iterate over all PHIs considering removal of *gsi.
11027 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11029 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11032 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11034 * config/rs6000/rs6000-string.c: (expand_block_clear,
11035 do_load_for_compare, select_block_compare_mode,
11036 compute_current_alignment, expand_block_compare,
11037 expand_strncmp_align_check, expand_strn_compare,
11038 expand_block_move, rs6000_output_load_multiple)
11039 Move functions related to string/block move/compare
11040 to a separate file.
11041 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11042 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11043 for this function which is now used in two files.
11044 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11045 * config.gcc: Add rs6000-string.o to extra_objs for
11046 targets powerpc*-*-* and rs6000*-*-*.
11048 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11051 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11052 32-bit, since indexed is not valid for DImode.
11053 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11054 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11055 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11056 (define_peephole2 for Altivec d-form store): Likewise.
11059 * multiple_target.c (create_dispatcher_calls): Only create the
11060 dispatcher call if the function is the default clone of a
11061 versioned function.
11063 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11065 PR middle-end/80902
11066 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11067 a call, force the call to not be a tail call.
11069 2017-06-23 Jeff Law <law@redhat.com>
11071 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11074 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11076 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11077 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11078 handling for early expansion of vector shifts (sl,sr,sra,rl).
11079 (builtin_function_type): Add vector shift right instructions
11080 to the unsigned argument list.
11082 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11084 rtl-optimizatoin/79286
11085 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11086 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11087 trap. PIC register plus a const unspec without offset can never trap.
11089 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11091 * tree.h (builtin_structptr_type): New type.
11092 (builtin_structptr_types): Declare new array.
11093 * tree.c (builtin_structptr_types): New array.
11094 (free_lang_data, build_common_tree_nodes): Use it.
11096 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11099 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11102 2017-06-22 Matt Turner <mattst88@gmail.com>
11104 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11105 Lake models to skylake case. Assume skylake for unknown
11106 models with clflushopt.
11108 2017-06-22 Jeff Law <law@redhat.com>
11110 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11111 frame sizes that do not satisfy aarch64_uimm12_shift.
11113 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11115 * profile-count.h (apply_probability,
11116 apply_scale, probability_in): Fix checks for zero.
11118 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11120 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11121 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11123 2016-06-22 Richard Biener <rguenther@suse.de>
11125 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11126 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11127 REDUC_MAX_EXPR support.
11128 (vectorizable_reduction): Likewise.
11129 (vect_create_epilog_for_reduction): Likewise.
11131 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11133 * match.pd (A / (1 << B) -> A >> B): New.
11134 * generic-match-head.c: Include optabs-tree.h.
11135 * gimple-match-head.c: Likewise.
11136 * optabs-tree.h (target_supports_op_p): New.
11137 * optabs-tree.c (target_supports_op_p): New.
11139 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11141 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11142 $gcc_cv_ld --help output.
11143 (gcc_cv_ld_demangle): Likewise.
11144 (gcc_cv_ld_eh_frame_hdr): Likewise.
11145 (gcc_cv_ld_pie): Likewise.
11146 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11147 (gcc_cv_ld_buildid): Likewise.
11148 (gcc_cv_ld_sysroot): Likewise.
11149 (ld_bndplt_support): Likewise.
11150 (ld_pushpopstate_support): Likewise.
11151 * configure: Regenerate.
11152 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11154 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11157 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11158 operands indexes to avoid gap between operands and match_dups.
11160 2017-06-21 Andrew Pinski <apinski@cavium.com>
11162 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11163 Increment Arith_shift and Arith_shift_reg by 1.
11164 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11166 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11167 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11168 (aarch64_strip_extend): Add new argument and test for it.
11169 (aarch64_cheap_mult_shift_p): New function.
11170 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11171 add a cost if it is true.
11172 Update calls to aarch64_strip_extend.
11173 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11175 2017-06-21 Andrew Pinski <apinski@cavium.com>
11177 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11179 (thunderxt88): Likewise.
11180 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11181 (thunderx_prefetch_tune): New variable.
11182 (thunderx2t99_prefetch_tune): Update for the correct values.
11183 (thunderxt88_tunings): New variable.
11184 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11185 generic_prefetch_tune.
11186 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11188 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11190 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11191 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11192 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11193 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11194 (aarch64_atomic_cas<mode>, GPI): Likewise.
11196 2017-06-21 Martin Liska <mliska@suse.cz>
11198 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11199 statements on cold and hot labels.
11200 * predict.c (tree_estimate_probability_bb): Remove the
11201 prediction from this place.
11203 2017-06-21 Martin Liska <mliska@suse.cz>
11205 PR tree-optimization/79489
11206 * gimplify.c (maybe_add_early_return_predict_stmt): New
11208 (gimplify_return_expr): Call the function.
11209 * predict.c (tree_estimate_probability_bb): Remove handling
11211 * predict.def: Update comment about early return predictor.
11212 * gimple-predict.h (is_gimple_predict): New function.
11213 * predict.def: Change default value of early return to 66.
11214 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11216 * passes.def: Put pass_strip_predict_hints to the beginning of
11219 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11221 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11222 FUNCTION_DECL declarations.
11223 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11225 (dwaf2out_decl): Likewise.
11226 * godump.c (go_early_global_decl): Skip call to the real debug hook
11227 for FUNCTION_DECL declarations.
11228 * passes.c (rest_of_decl_compilation): Skip call to the
11229 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11230 -fdump-go-spec is passed.
11232 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11234 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11235 Reorder for compactness.
11236 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11237 (def_builtin_pure, def_builtin_pure2): New functions.
11238 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11240 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11242 * match.pd (nop_convert): New predicate.
11243 ((A +- CST1) +- CST2): Allow some NOP conversions.
11245 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11248 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11249 with ctors/dtors if GOVD_SHARED is set.
11251 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11253 * config/aarch64/aarch64.md (movti_aarch64):
11254 Emit mov rather than orr.
11255 (movtf_aarch64): Likewise.
11256 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11257 Emit mov rather than orr.
11259 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11261 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11262 Swap alternatives, make integer dup more expensive.
11264 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11266 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11267 Return true for non-tls symbols.
11269 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11271 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11272 (cortex-a75): Likewise.
11273 (cortex-a75.cortex-a55): Likewise.
11274 * config/aarch64/aarch64-tune.md: Regenerate.
11275 * doc/invoke.texi (-mtune): Document new values for -mtune.
11277 2017-06-21 Tom de Vries <tom@codesourcery.com>
11279 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11280 stack_size feature.
11281 (Effective-Target Keywords, Other attributes): Suggest using
11282 dg-add-options stack_size feature to get stack limit in stack_size
11283 effective target documentation.
11285 2017-06-21 Julian Brown <julian@codesourcery.com>
11286 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11288 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11289 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11290 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11292 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11293 attribute type list for neon_multiply.
11294 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11295 attribute type list for neon_multiply.
11296 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11297 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11298 attribute type list for neon_multiply.
11299 * config/arm/types.md (crypto_pmull): Add.
11300 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11301 attribute type list.
11303 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11305 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11308 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11310 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11311 to make sure not to dereference a NULL cost_classes_ptr pointer.
11313 2017-06-20 Carl Love <cel@us.ibm.com>
11315 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11316 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11317 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11318 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11319 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11320 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11321 VMULOSW): New enum "unspec" values.
11322 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11323 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11324 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11325 altivec_vmulosw): New patterns.
11326 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11327 VMULOSW): Add definitions.
11329 2017-06-20 Julia Koval <julia.koval@intel.com>
11331 * config/i386/i386.c: Fix rounding expand for new pattern.
11332 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11334 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11336 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11337 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11339 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11341 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11344 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11346 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11347 architecture, then by implementer ID.
11348 * config/aarch64/aarch64-tune.md: Regenerate.
11350 2017-06-20 Richard Biener <rguenther@suse.de>
11352 PR middle-end/81097
11353 * fold-const.c (split_tree): Fold to type before negating.
11355 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11357 * diagnostic-show-locus.c
11358 (selftest::test_fixit_deletion_affecting_newline): New function.
11359 (selftest::diagnostic_show_locus_c_tests): Call it.
11361 2017-06-20 Andreas Schwab <schwab@suse.de>
11364 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11367 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11369 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11370 __ARM_FEATURE_COPROC according to support.
11372 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11374 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11375 Rewritten to avoid overflow for > 32-bit pointers.
11378 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11379 by removing enum keyword.
11380 (ubsan_type_descriptor): Likewise. Formatting fix.
11383 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11384 splitter): Require TARGET_SSE2 in the condition.
11386 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11389 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11390 for doing vector set of SFmode on ISA 3.0.
11391 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11392 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11394 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11395 SFmode value into a V4SF variable that was extracted from another
11396 V4SF variable without converting the element to double precision
11397 and back to single precision vector format.
11398 (vsx_insert_extract_v4sf_p9_2): Likewise.
11400 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11402 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11403 in UWHI to avoid undefined overflow.
11406 * ubsan.h (enum ubsan_encode_value_phase): New.
11407 (ubsan_encode_value): Change second argument to
11408 enum ubsan_encode_value_phase with default value of
11409 UBSAN_ENCODE_VALUE_GENERIC.
11410 * ubsan.c (ubsan_encode_value): Change second argument to
11411 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11412 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11413 create_tmp_var_raw instead of create_tmp_var and use a
11415 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11416 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11417 ubsan_encode_value callers.
11420 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11421 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11422 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11424 2017-06-19 Richard Biener <rguenther@suse.de>
11426 PR middle-end/81118
11427 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11428 estimates if we changed anything.
11430 2017-06-19 Richard Biener <rguenther@suse.de>
11432 PR tree-optimization/80887
11433 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11434 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11435 simplified lookups, then reset mprts_hook.
11436 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11438 (try_to_simplify): Likewise.
11440 2017-06-19 Martin Liska <mliska@suse.cz>
11443 * gimplify.c (gimplify_switch_expr):
11444 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11446 2017-06-19 Martin Liska <mliska@suse.cz>
11448 * doc/install.texi: Document that PGO runs in 4 stages.
11450 2017-06-19 Martin Liska <mliska@suse.cz>
11453 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11454 to dispatcher function name.
11455 * multiple_target.c (replace_function_decl): New function.
11456 (create_dispatcher_calls): Redirect both edges and references.
11458 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11460 * profile-count.c (profile_count::dump): Dump quality.
11461 (profile_count::differs_from_p): Update for unsigned val.
11462 * profile-count.h (profile_count_quality): New enum.
11463 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11465 2017-06-19 Richard Biener <rguenther@suse.de>
11467 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11468 struct function as arg.
11469 (estimate_numbers_of_iterations): Export overload with loop arg.
11470 (free_numbers_of_iterations_estimates_loop): Use an overload of
11471 free_numbers_of_iterations_estimates instead.
11472 * tree-cfg.c (remove_bb): Adjust.
11473 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11474 * tree-parloops.c (gen_parallel_loop): Likewise.
11475 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11477 (tree_unroll_loops_completely): Likewise.
11478 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11479 Use an overload instead and export.
11480 (estimated_loop_iterations): Adjust.
11481 (max_loop_iterations): Likewise.
11482 (likely_max_loop_iterations): Likewise.
11483 (estimate_numbers_of_iterations): Take struct function as arg
11485 (loop_exits_before_overflow): Adjust.
11486 (free_numbers_of_iterations_estimates_loop): Use an overload.
11487 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11488 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11490 2017-06-19 Richard Biener <rguenther@suse.de>
11493 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11494 RANGE_EXPR conservatively.
11496 2017-06-16 Carl Love <cel@us.ibm.com>
11498 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11499 definitions for vec_float, vec_float2, vec_floato,
11500 vec_floate built-ins.
11501 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11502 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11504 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11505 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11506 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11507 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11508 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11509 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11510 vec_floato): Add builtin defines.
11511 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11512 Update the built-in documentation file for the new built-in
11515 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11517 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11518 (mthumb): Mark as the negative of -marm.
11520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11522 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11524 (ARM Options, -mtune): Document that this accepts the same
11525 extension options as -mcpu.
11526 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11528 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11530 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11531 permitted extensions.
11533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11535 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11536 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11537 (armv8-m.main): Add option +nodsp.
11538 * config/arm/arm-cpu-cdata.h: Regenerated.
11540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11542 * config/arm/t-fuchsia: New file.
11543 * config.gcc (arm*-*-fuchsia*): Use it.
11545 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11547 * config/arm/t-symbian: Rewrite for new option infrastructure.
11549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11551 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11552 (MULTILIB_REQUIRED): Likewise.
11554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11556 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11557 (MULTILIB_RESUE): Likewise.
11558 (MULTILIB_MATCHES): Likewise.
11559 (MULTLIB_REQUIRED): Likewise.
11561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11563 * config/arm/t-rtems: Rewrite for new option framework.
11565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11567 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11568 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11569 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11570 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11571 * config/arm/t-multilib: ... here.
11572 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11573 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11574 armv7-a and armv8*-a when A-profile libraries have not been built.
11575 * config/arm/t-rmprofile: Rewrite.
11577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11579 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11580 with a backslash. Remove the backslash after substituting unescaped
11582 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11586 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11587 the driver to pass through the default setting of -mfloat-abi.
11588 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11590 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11591 (all_feat_combs): New rule.
11592 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11594 * config/arm/t-aprofile: Rewrite.
11596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11598 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11599 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11600 fpu is not specified by the user/command-line.
11601 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11602 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11603 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11604 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11605 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11606 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11610 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11611 * config/arm/t-arm-elf: Rewritten.
11613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11615 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11616 have some floating-point instructions.
11617 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11618 (TARGET_MAYBE_HARD_FLOAT): New macro.
11619 * config/arm/arm-builtins.c (arm_init_builtins): Use
11620 TARGET_MAYBE_HARD_FLOAT.
11621 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11625 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11626 (configargs.h): Include it.
11627 (arm_print_hint_for_fpu_option): New function.
11628 (arm_parse_fpu_option): New function.
11629 (candidate_extension): New class.
11630 (arm_canon_for_multilib): New function.
11631 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11632 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11633 (ARCH_CANONICAL_SPECS): New macro.
11634 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11638 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11639 are set after handling multilib fragments. Set target_cpu_default2
11642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11644 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11646 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11647 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11649 (check_arch): Likewise.
11650 * config/arm/arm.c (arm_configure_build_target): Handle
11651 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11652 options in the default.
11654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11656 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11657 when an option is an alias of another.
11658 * config/arm/parsecpu.awk (optalias): New parser token.
11659 (gen_comm_data): Mark non-alias options as such. Emit entries
11660 for extension aliases.
11661 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11662 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11663 (armv6kz, armv6zk, armv6t2): Likewise.
11664 (armv7): Make vfpv3-d16 an alias.
11665 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11667 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11668 Sort in canonical order.
11669 (armv8-a): Sort in canonical order.
11670 (armv8.1-a, armv8.2-a): Likewise.
11671 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11673 (cortex-a9): Sort in canonical order.
11674 * config/arm/arm.c (selftests.h): Include it.
11675 (arm_test_cpu_arch_data): New function.
11676 (arm_run_self_tests): New function.
11677 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11678 (targetm): Move declaration to the end of the file.
11679 * arm-cpu-cdata.h: Regenerated.
11681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11683 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11684 call to target_mode_check describing the type of option passed.
11685 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11686 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11687 arm_parse_cpu_option_name to match parameters against list of
11689 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11690 arm_arch_core_flags data structure.
11691 * config/arm/arm-cpu_cdata.h: Regenerated.
11693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11695 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11697 (arm_print_hint_for_cpu_option): Likewise.
11698 (arm_print_hint_for_arch_option): Likewise.
11699 (arm_parse_cpu_option_name): Likewise.
11700 (arm_parse_arch_option_name): Likewise.
11701 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11702 of entries in the all_fpus list.
11703 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11704 (arm_parse_cpu_option_name): Declare.
11705 (arm_parse_arch_option_name): Declare.
11706 (arm_parse_option_features): Declare.
11707 (arm_intialize_isa): Declare.
11708 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11710 (gen_comm_data): ... here. Make definitions non-static.
11711 * config/arm/arm-cpu-data.h: Regenerated.
11712 * config/arm/arm-cpu-cdata.h: Regenerated.
11714 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11716 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11717 (cpu_arch_extension): New structure.
11718 (cpu_arch_option, arch_option, cpu_option): New structures.
11719 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11720 architecture types.
11721 (gen_data): Generate new format data tables.
11722 * config/arm/arm.c (cpu_tune): New structure.
11723 (cpu_option, processors): Delete.
11724 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11725 (arm_print_hint_for_cpu_option): ... this and ...
11726 (arm_print_hint_for_arch_option): ... this.
11727 (arm_parse_arch_cpu_name): Delete. Replace with ...
11728 (arm_parse_cpu_option_name): ... this and ...
11729 (arm_parse_arch_option_name): ... this.
11730 (arm_unrecognized_feature): Change type of target parameter to
11732 (arm_parse_arch_cpu_features): Delete. Replace with ...
11733 (arm_parse_option_features): ... this.
11734 (arm_configure_build_target): Rework to use new configuration data
11736 (arm_print_tune_info): Rework for new configuration data tables.
11737 * config/arm/arm-cpu-data.h: Regenerated.
11738 * config/arm/arm-cpu.h: Regenerated.
11740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11742 * Makefile.in (OBJS): Move sbitmap.o from here ...
11743 (OBJS-libcommon): ... to here.
11745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11747 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11748 (ISA_ALL_CRYPTO): New macro.
11749 (ISA_ALL_SIMD): New macro
11750 (ISA_ALL_FP): New macro.
11751 * config/arm/arm.c (fpu_bitlist): Update initializer.
11752 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11754 (arm9e): Add fpu. Add option for nofp
11755 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11756 (arm926ej-s, arm1026ej-s): Likewise.
11757 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11758 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11759 neon-fp16, neon-vfpv4, nofp and nosimd.
11760 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11761 (cortex-a8): Add fpu. Add option for nofp.
11762 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11763 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11764 (cortex-r4f): Add fpu.
11765 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11766 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11768 (cortex-r8): Likewise.
11769 (cortex-m4): Add fpu. Add option for nofp.
11770 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11771 (cortex-a17.cortex-a7): Likewise.
11772 (cortex-a32): Add fpu. Add options for crypto and nofp.
11773 (cortex-a35, cortex-a53): Likewise.
11774 (cortex-a57): Add fpu. Add option for crypto.
11775 (cortex-a72, cortex-a73): Likewise.
11776 (exynos-m1): Likewise.
11777 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11778 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11779 (cortex-m33): Add fpu. Add option for nofp.
11780 * config/arm/arm-cpu-cdata.h: Regenerated
11781 * config/arm/arm-cpu-data.h: Regenerated.
11783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11785 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11786 (armv5te, armv5tej): Likewise.
11787 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11788 (armv7): Add options fp and vfpv3-d16.
11789 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11790 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11792 (armv7ve): Likewise.
11793 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11794 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11795 (armv8-a): Add nocrypto option.
11796 (armv8.1-a, armv8.2-a): Likewise.
11797 (armv8-m.main): add options fp, fp.dp and nofp.
11799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11801 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11803 (armv8-a+crc): Delete.
11804 (armv8.1-a): Add options simd, crypto and nofp.
11805 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11806 (armv8.2-a+fp16): Delete.
11807 (armv8-m.main): Add option dsp.
11808 (armv8-m.main+dsp): Delete.
11809 (cortex-a8): Add fpu. Add nofp option.
11810 (cortex-a9): Add fpu. Add nofp and nosimd options.
11811 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11812 link to main cpu and architecture data structures.
11813 (gen_comm_data): Only put isa attributes from the main architecture
11815 (option): New statement for architecture and CPU entries.
11816 * arm.c (struct cpu_option): New structure.
11817 (struct processors): Add entry for options.
11818 (arm_unrecognized_feature): New function.
11819 (arm_parse_arch_cpu_name): Ignore any characters after the first
11821 (arm_parse_arch_cpu_feature): New function.
11822 (arm_configure_build_target): Separate out any CPU and architecture
11823 features and parse separately. Don't error out if -mfpu=auto is
11824 used with only an architecture string.
11825 (arm_print_asm_arch_directives): New function.
11826 (arm_file_start): Call it.
11827 * config/arm/arm-cpu-cdata.h: Regenerated.
11828 * config/arm/arm-cpu-data.h: Likewise.
11829 * config/arm/arm-tables.opt: Likewise.
11831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11833 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11834 assembler when it is not -mfpu=auto.
11836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11838 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11839 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11840 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11841 (ASM_CPU_SPEC): Rewrite.
11842 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11843 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11844 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11845 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11846 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11847 copied string is NUL-terminated. Also strip any characters prefixed
11849 (arm_rewrite_selected_arch): New function.
11850 (arm_rewrite_march): New function.
11852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11854 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11855 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11856 (march, mcpu, mtune): Convert to string-based options.
11857 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11858 (arm_parse_arch_cpu_name): New function.
11859 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11860 identify selected architecture or CPU.
11861 (arm_option_save): New function.
11862 (TARGET_OPTION_SAVE): Redefine.
11863 (arm_option_restore): Restore string options.
11864 (arm_option_print): Print string options.
11866 2017-06-16 Martin Sebor <msebor@redhat.com>
11868 PR tree-optimization/80933
11869 PR tree-optimization/80934
11870 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11871 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11872 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11873 (gimple_fold_builtin): Call them.
11875 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11877 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11878 as unlikely; update profile.
11880 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11882 * predict.c (force_edge_cold): Handle declaring edges impossible
11885 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11887 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11889 (try_unroll_loop_completely): Fix reporting.
11891 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11893 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11895 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11898 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11899 if given a non-constant argument for an intrinsic which requires a
11902 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11904 * profile.c (compare_freqs): New function.
11905 (branch_prob): Sort edge list.
11906 (find_spanning_tree): Assume that the list is priority sorted.
11908 2017-06-16 Richard Biener <rguenther@suse.de>
11910 PR tree-optimization/81090
11911 * passes.def (pass_record_bounds): Remove.
11912 * tree-pass.h (make_pass_record_bounds): Likewise.
11913 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11914 make_pass_record_bounds): Likewise.
11915 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11916 not free niter estimates at the beginning but at the end.
11917 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11919 2017-06-16 Richard Biener <rguenther@suse.de>
11921 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11922 initializer to workaround ICE in host GCC 4.8.
11924 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11926 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11928 (clone_inlined_nodes): Update.
11930 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11932 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11933 prefetch settings, and enable prefetching by default at -O3.
11935 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11937 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11938 Set flag_prefetch_loop_arrays according to tuning data.
11940 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11942 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11943 New tune structure.
11944 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11945 [Unrelated to main purpose of the patch] Place the pointer field last
11946 to enable type checking errors when tune structure are wrongly merged.
11947 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11948 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11949 (thunderx2t99_prefetch_tune): New tune constants.
11950 (tune_params *_tunings): Update all tunings (no functional change).
11951 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11952 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11953 from tunings structures.
11955 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11958 * ubsan.c (instrument_null): Add T argument, use it instead
11959 of computing it based on IS_LHS.
11960 (instrument_object_size): Likewise.
11961 (pass_ubsan::execute): Adjust instrument_null and
11962 instrument_object_size callers to pass gimple_get_lhs or
11963 gimple_assign_rhs1 result to it. Use instrument_null instead of
11964 calling get_base_address and instrument_mem_ref. Handle
11965 aggregate call arguments for object-size sanitization.
11967 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11969 PR tree-optimization/81089
11970 * tree-vrp.c (is_masked_range_test): Validate operands of
11973 2017-06-15 Martin Sebor <msebor@redhat.com>
11976 * dumpfile.c (dump_register): Avoid calling memset to initialize
11977 a class with a default ctor.
11978 * gcc.c (struct compiler): Remove const qualification.
11979 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11980 * hash-table.h: Ditto.
11981 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11983 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11984 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11986 * params.h (struct param_info): Make struct members non-const.
11987 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11988 with default initialization.
11989 * vec.h (vec_copy_construct, vec_default_construct): New helper
11991 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11992 with vec_copy_construct.
11993 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11994 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11995 * doc/invoke.texi (-Wclass-memaccess): Document.
11997 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11999 * emit-rtl.h (is_leaf): Update comment about local
12000 register allocator.
12002 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12005 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12006 for a variable to have a section before checking if the section has a
12008 Set section to.persistent if persistent attribute is set.
12009 Warn if .persistent attribute is used on an automatic variable.
12011 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12013 PR rtl-optimization/80474
12014 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12016 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12018 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12021 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12023 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12024 (casesi): Emit bounds checking as RTL.
12025 (casesi_internal_mips16_<mode>): Remove bounds checking.
12027 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12029 * config/xtensa/xtensa.c (xtensa_option_override): Append
12030 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12031 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12032 xtensa_doloop_hooks): Define unconditionally.
12033 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12035 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12036 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12037 for it in xtensa_option_override.
12038 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12039 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12041 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12043 * doc/cppopts.texi: Document '-' special value to -MF.
12045 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12047 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12048 (cortex_a53_fconst): Likewise.
12049 (cortex_a53_fpmul): Likewise.
12050 (cortex_a53_f_load_64): Likewise.
12051 (cortex_a53_f_load_many): Likewise.
12052 (cortex_a53_advsimd_alu): Likewise.
12053 (cortex_a53_advsimd_alu_q): Likewise.
12054 (cortex_a53_advsimd_mul): Likewise.
12055 (cortex_a53_advsimd_mul_q): Likewise.
12056 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12057 Add missing fmul, r2f_cvt and fconst cases.
12059 2017-06-14 Richard Biener <rguenther@suse.de>
12061 PR middle-end/81088
12062 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12064 (fold_binary_loc): When associating do not treat pre-existing
12065 TREE_OVERFLOW on literal constants as a reason to allow
12066 TREE_OVERFLOW on associated literal constants.
12068 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12070 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12071 (MASK_FEATURES): New macro.
12072 * config/sparc/sparc.c (sparc_option_override): Remove the special
12073 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12075 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12077 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12078 a division of 0 if non-call exceptions are enabled.
12080 2017-06-14 Andrew Pinski <apinski@cavium.com>
12081 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12084 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12085 Improve vector initialization code gen for only variable case.
12087 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12089 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12091 2017-06-14 Richard Biener <rguenther@suse.de>
12093 PR tree-optimization/81083
12094 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12097 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12099 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12100 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12101 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12102 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12103 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12104 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12106 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12108 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12109 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12111 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12113 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12115 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12117 * config/rs6000/t-rtems: Don't handle SPE.
12119 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12121 * config/rs6000/t-linux: Don't handle SPE.
12123 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12125 * config/rs6000/eabispe.h: Delete file.
12127 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12129 * config/rs6000/t-spe: Delete file.
12131 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12133 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12134 (rs6000_legitimate_offset_address_p): Return false for anything in
12135 V2SImode or V2SFmode.
12137 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12139 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12140 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12141 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12143 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12144 (rs6000_legitimate_offset_address_p): Ditto.
12145 (rs6000_emit_move): Ditto.
12146 (rs6000_init_builtins): Remove V4HI_type_node.
12148 2017-06-13 Martin Liska <mliska@suse.cz>
12151 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12152 (gate_asan): Likewise.
12153 * asan.h (asan_no_sanitize_address_p): Remove the function.
12154 (sanitize_flags_p): New function.
12155 * builtins.def: Fix coding style.
12156 * common.opt: Use renamed enum value.
12157 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12158 * doc/extend.texi: Document no_sanitize attribute.
12159 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12160 to SANITIZE_UNDEFINED_NONDEFAULT.
12161 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12162 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12163 Use sanitize_flags_p.
12164 * gimplify.c (gimplify_function_tree): Likewise.
12165 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12166 * opts.c (parse_no_sanitize_attribute): New function.
12167 (common_handle_option): Use renamed enum value.
12168 * opts.h (parse_no_sanitize_attribute): Declare.
12169 * tree.c (sanitize_flags_p): New function.
12170 * tree.h: Declared here.
12171 * tsan.c: Use sanitize_flags_p.
12172 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12173 (instrument_mem_ref): Likewise.
12174 (instrument_bool_enum_load): Likewise.
12175 (do_ubsan_in_current_function): Remove the function.
12176 (pass_ubsan::execute): Use sanitize_flags_p.
12177 * ubsan.h: Remove do_ubsan_in_current_function
12178 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12179 (dump_function_to_file): Use it here.
12181 2017-06-13 Martin Jambor <mjambor@suse.cz>
12183 PR tree-optimization/80803
12184 PR tree-optimization/81063
12185 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12186 (propagate_subaccesses_across_link): Enqueue subtree whenever
12187 necessary instead of relying on the caller.
12189 2017-06-13 Martin Jambor <mjambor@suse.cz>
12191 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12192 that have a first_link.
12193 (sort_and_splice_var_accesses): Do not check first_link before
12195 (subtree_mark_written_and_enqueue): Likewise.
12196 (propagate_all_subaccesses): Likewise and do not stop at first
12197 parent with a first_link.
12199 2017-06-13 Martin Jambor <mjambor@suse.cz>
12201 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12204 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12206 * match.pd: New pattern.
12208 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12210 * tree-vrp.c (is_masked_range_test): New function.
12211 (register_edge_assert_for): Determine ranges for
12214 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12216 PR tree-optimization/67328
12217 * fold-const.c (maskable_range_p): New function.
12218 (build_range_check): Generate bittests if possible.
12220 2017-06-13 Martin Liska <mliska@suse.cz>
12222 * gimple-pretty-print.c (dump_probability): Add new argument.
12223 (dump_edge_probability): Dump both probability and count.
12224 (dump_gimple_label): Likewise.
12225 (dump_gimple_bb_header): Likewise.
12227 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12230 * config/avr/avr-devices.c: Fix indentation.
12231 * config/avr/gen-avr-mmcu-specs.c: Dito.
12233 2017-06-13 Richard Biener <rguenther@suse.de>
12235 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12236 instead get vector type from stmt_info.
12237 (vectorizable_reduction): Adjust. Remove dead code.
12239 2017-06-13 Richard Biener <rguenther@suse.de>
12241 PR middle-end/81065
12242 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12243 case of C * (x * C2 + C3).
12244 (fold_addr_of_array_ref_difference): Properly fold index difference.
12246 2017-06-12 David S. Miller <davem@davemloft.net>
12249 * config/sparc/sparc.md (return expander): Emit frame blockage if
12250 function uses alloca.
12252 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12254 * combine.c (make_field_assignment): Check len rather than the mode
12255 precision when calling force_to_mode.
12257 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12259 Support multilibs and devices that see flash in RAM address range.
12262 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12263 (avr_mcu_t) <flash_pm_offset>: New field.
12264 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12265 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12266 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12267 (AVR_TINY_PM_OFFSET): Remove macro.
12268 * config/avr/avr.opt (-mshort-calls): New option.
12269 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12270 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12271 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12272 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12273 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12274 instead of avr_arch->have_jmp_call.
12275 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12276 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12277 avr_arch->flash_pm_offset to define.
12278 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12279 new field flash_pm_offset. Add entry for avrxmega3.
12280 (avr_texinfo): Add entry for avrxmega3.
12281 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12282 attiny212, attiny214,
12283 attiny412, attiny414, attiny416, attiny417,
12284 attiny814, attiny816, attiny817,
12285 attiny1614, attiny1616, attiny1617,
12286 attiny3214, attiny3216, attiny3217.
12287 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12288 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12289 (avr_print_operand_address) [AVR_TINY]: Same.
12290 (avr_asm_init_sections) <readonly_data_section>: Only patch
12291 callback if avr_arch->flash_pm_offset = 0.
12292 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12293 for rodata if avr_arch->flash_pm_offset != 0.
12294 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12295 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12296 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12297 (m_options): Append opt_rcall.
12298 (m_dirnames): Append dir_rcall.
12299 * config/avr/t-multilib: Regenerate.
12301 * configure.ac [target=avr]: Check whether avrxmega3 default
12302 linker description file works as needed.
12303 * configure: Regenerate.
12304 * doc/avr-mmcu.texi: Regenerate.
12305 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12306 <__AVR_ARCH__>: Document avrxmega3 and 103.
12307 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12308 <__AVR_SHORT_CALLS__>: Document it.
12309 <__AVR_PM_BASE_ADDRESS__>: Document it.
12310 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12311 (AVR Variable Attributes) <progmem>: Document this is
12312 not needed for avrxmega3.
12313 (AVR Named Address Spaces) <__flash>: Dito.
12315 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12317 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12319 2017-06-12 Doug Rupp <rupp@adacore.com>
12321 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12322 Append vxworks-stdint.h to the tm_file list.
12323 * config/vxworks-stdint.h: New file.
12325 2017-06-12 Martin Liska <mliska@suse.cz>
12327 PR tree-optimization/81041
12328 * tree-profile.c (gimple_gen_ic_func_profiler):
12329 Create an extra BB in profile-generate
12330 (gimple_gen_time_profiler): Likewise.
12332 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12334 PR tree-optimization/81003
12335 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12336 (update_range_test): Use it instead of force_gimple_operand_gsi.
12338 2017-06-12 Richard Biener <rguenther@suse.de>
12340 PR tree-optimization/81053
12341 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12342 with backedge value not defined in loop. Simplify def stmt
12345 2017-06-11 Tom de Vries <tom@codesourcery.com>
12348 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12350 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12351 nvptx_cannot_force_const_mem.
12353 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12355 * opts.c (finish_options): Move test for flag_split_stack after
12356 it has been initialized.
12358 2017-06-11 Jason Merrill <jason@redhat.com>
12360 * tree.h (id_equal): New.
12361 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12362 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12363 instead of strcmp of IDENTIFIER_POINTER.
12365 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12367 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12368 (mark_all_inlined_calls_cdtor): Fix formating.
12369 (inline_transform): Rescale profile before inlining.
12371 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12373 * cgraph.h (cgraph_edge::clone): Update prototype.
12374 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12375 (cgraph_node::create_clone): Update.
12376 (cgraph_node::create_version_clone): Update.
12377 * tree-inline.c (copy_bb): Update.
12378 (expand_call_inline): Update.
12380 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12382 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12383 factored out from ...
12384 (rs6000_emit_prologue): ... here.
12386 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12388 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12389 factored out from ...
12390 (rs6000_emit_prologue): ... here.
12392 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12394 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12396 (handle_missing_profiles): Fix computation of tp_first_run.
12397 (counts_to_freqs): Do not touch freqs when count is 0.
12399 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12401 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12404 2017-06-10 Tom de Vries <tom@codesourcery.com>
12406 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12407 attributes): Document signal effective target.
12409 2017-06-10 Tom de Vries <tom@codesourcery.com>
12411 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12412 Document effective target stack_size.
12414 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12416 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12417 to the edit_context if they can be auto-applied.
12419 2017-06-9 Ian Lance Taylor <iant@golang.org>
12421 * opts.c (finish_options): If -fsplit-stack, disable implicit
12422 -forder-blocks-and-partition.
12423 * doc/invoke.texi (Optimize Options): Document that when using
12424 -fsplit-stack -forder-blocks-and-partition is not implicitly
12427 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12429 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12430 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12431 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12432 * builtins.def (abort, trap, unreachable): Declare cold.
12433 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12434 * tree-core.h (ECF_COLD): New.
12435 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12436 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12438 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12440 * predict.c (unlikely_executed_stmt_p): Cleanup.
12442 2017-06-09 Richard Biener <rguenther@suse.de>
12444 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12445 model if the ref is always written to.
12447 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12449 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12451 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12453 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12456 2017-06-09 Tom de Vries <tom@codesourcery.com>
12459 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12460 "target cannot support label values" when encountering LABEL_REF.
12462 2017-06-09 Martin Liska <mliska@suse.cz>
12464 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12465 (gimple_gen_ic_func_profiler): Emit direct comparison
12466 of __gcov_indirect_call_callee with NULL.
12467 (gimple_gen_time_profiler): Change probability from
12468 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12470 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12472 * profile.c (edge_gcov_counts): Turn to pointer.
12473 (compute_branch_probabilities, compute_branch_probabilities): Update.
12474 (branch_prob): Do not clear edge_gcov_count.
12475 * profile.h (edge_gcov_counts): Turn to pointer.
12476 (edge_gcov_count): Update.
12478 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12480 * gimple.h (gimple_check_failed): Mark cold.
12482 2017-06-09 Richard Biener <rguenther@suse.de>
12484 PR tree-optimization/66623
12485 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12486 refactor check_reduction into two parts, properly computing
12487 whether we have to check reduction validity for outer loop
12490 2017-06-09 Richard Biener <rguenther@suse.de>
12492 PR tree-optimization/79483
12493 * graphite-scop-detection.c (order): New global.
12494 (get_order): Compute bb to order mapping that satisfies code
12495 generation constraints.
12496 (cmp_pbbs): New helper.
12497 (build_scops): Start domwalk at entry block, sort generated
12500 2017-06-09 Richard Biener <rguenther@suse.de>
12502 PR middle-end/81007
12503 * ipa-polymorphic-call.c
12504 (ipa_polymorphic_call_context::restrict_to_inner_class):
12505 Skip FIELD_DECLs with error_mark_node type.
12506 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12509 2017-06-09 Martin Liska <mliska@suse.cz>
12511 * predict.c (struct branch_predictor): New struct.
12512 (test_prediction_value_range): New test.
12513 (predict_c_tests): New function.
12514 * selftest-run-tests.c (selftest::run_tests): Run the function.
12515 * selftest.h: Declare new tests.
12517 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12520 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12521 gen_add3_insn did not fail.
12522 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12523 r0, construct that number in a temporary reg and add that reg to r0.
12524 If asked to put the result in r0 as well, fail.
12526 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12528 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12529 for early expansion of vec_eqv.
12531 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12533 PR middle-end/81005
12534 * ubsan.c (instrument_null): Avoid pointless code temporary.
12535 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12537 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12541 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12544 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12547 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12549 * predict.c (unlikely_executed_edge_p): Move ahead.
12550 (probably_never_executed_edge_p): Use it.
12552 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12554 PR middle-end/79988
12555 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12556 gimple_call_builtin_p call.
12558 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12560 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12561 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12562 rtl_check_failed_type2, rtl_check_failed_code1,
12563 rtl_check_failed_code2, rtl_check_failed_code_mode,
12564 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12565 rtvec_check_failed_bounds, rtl_check_failed_flag,
12566 _fatal_insn_not_found, _fatal_insn): Likewise.
12567 * tree.h (tree_contains_struct_check_failed,
12568 tree_check_failed, tree_not_check_failed,
12569 tree_class_check_failed, tree_range_check_failed,
12570 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12571 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12572 tree_operand_check_failed, omp_clause_check_failed,
12573 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12576 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12578 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12579 flag_branch_probabilities.
12580 * ipa-inline.c (edge_badness): Likewise.
12581 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12582 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12583 * predict.c (maybe_hot_frequency_p): Likewise.
12584 (probably_never_executed): Likewise.
12585 * sched-ebb.c (schedule_ebbs): Likewise.
12586 * sched-rgn.c (find_single_block_region): Likewise.
12587 * tracer.c (tail_duplicate): Likewise.
12589 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12591 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12592 longer requires x_flag_profile_use.
12594 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12596 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12597 instead of flag_reorder_blocks_and_partition.
12598 * dbxout.c (dbxout_function_end): Likewise.
12599 * dwarf2out.c (gen_subprogram_die): Likewise.
12600 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12601 * hw-doloop.c (reorg_loops): Likewise.
12602 * varasm.c (assemble_start_function,
12603 assemble_end_function): Likewise.
12604 (decide_function_section): Do not check for
12605 flag_reorder_blocks_and_partition.
12607 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12609 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12611 (chkp_get_hard_register_fake_addr_expr): Ditto.
12612 (chkp_build_addr_expr): Add check for hard reg case.
12613 (chkp_parse_array_and_component_ref): Ditto.
12614 (chkp_find_bounds_1): Ditto.
12615 (chkp_process_stmt): Don't generate bounds store for
12618 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12620 * predict.c (maybe_hot_bb_p): Do not check profile status.
12621 (maybe_hot_edge_p): Likewise.
12622 (probably_never_executed): Check for zero counts even if profile
12624 (unlikely_executed_edge_p): New function.
12625 (unlikely_executed_stmt_p): New function.
12626 (unlikely_executed_bb_p): New function.
12627 (set_even_probabilities): Use unlikely predicates.
12628 (combine_predictions_for_bb): Likewise.
12629 (predict_paths_for_bb): Likewise.
12630 (predict_paths_leading_to_edge): Likewise.
12631 (determine_unlikely_bbs): New function.
12632 (estimate_bb_frequencies): Use it.
12633 (compute_function_frequency): Use zero counts even if profile is
12635 * profile-count.h: Fix typo.
12637 2017-08-08 Julia Koval <julia.koval@intel.com>
12639 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12640 _mm512_mask_cvtsepi16_storeu_epi8,
12641 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12642 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12643 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12644 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12645 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12646 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12647 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12648 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12649 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12650 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12651 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12652 __builtin_ia32_pmovuswb256mem_mask,
12653 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12654 __builtin_ia32_pmovwb512mem_mask): New builtins.
12656 2017-08-08 Julia Koval <julia.koval@intel.com>
12658 PR target/73350,80862
12659 * config/i386/subst.md (round): Fix round pattern.
12660 * config/i386/i386.c (ix86_erase_embedded_rounding):
12661 Fix erasing rounding for the fixed pattern.
12663 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12665 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12667 2017-06-08 Martin Liska <mliska@suse.cz>
12669 PR gcov-profile/80911
12670 * gcov.c (block_info::block_info): New constructor.
12672 2017-06-07 Carl Love <cel@us.ibm.com>
12674 * config/rs6000/rs6000-c: The return type of the following
12675 built-in functions was implemented as int not long long. Fix sign
12676 of return value for the unsigned version of vec_mulo and vec_mule.
12677 vector unsigned long long vec_bperm (vector unsigned long long,
12678 vector unsigned char)
12679 vector signed long long vec_mule (vector signed int,
12681 vector unsigned long long vec_mule (vector unsigned int,
12682 vector unsigned int)
12683 vector signed long long vec_mulo (vector signed int,
12685 vector unsigned long long vec_mulo (vector unsigned int,
12686 vector unsigned int)
12687 * doc/extend.texi: Fix the documentation for the built-in
12690 2017-06-07 Carl Love <cel@us.ibm.com>
12693 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12696 2017-06-07 Carl Love <cel@us.ibm.com>
12698 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12699 support, Generate doublehv for signed int/float for BE case only.
12701 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12703 * doc/invoke.texi (mcx16): Rewrite.
12705 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12707 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12708 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12709 *mov<mode>_softfloat, and an anonymous splitter): Use
12710 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12714 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12716 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12717 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12718 (rs6000_debug_reg_global): Adjust.
12719 (rs6000_init_hard_regno_mode_ok): Adjust.
12720 (rs6000_dbx_register_number): Adjust.
12721 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12722 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12723 Remove SPE_ACC and SPEFSCR.
12724 (REG_ALLOC_ORDER): Ditto.
12725 (FRAME_POINTER_REGNUM): Change to 111.
12726 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12727 (REG_CLASS_NAMES): Ditto.
12728 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12729 (REGISTER_NAMES): Ditto.
12730 (ADDITIONAL_REG_NAMES): Ditto.
12731 (rs6000_reg_names): Ditto.
12732 * config/rs6000/rs6000.md: Renumber some register number
12735 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12737 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12739 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12740 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12742 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12743 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12744 Delete the SPE high registers.
12745 (REG_ALLOC_ORDER): Ditto.
12746 (enum reg_class): Remove SPE_HIGH_REGS.
12747 (REG_CLASS_NAMES): Ditto.
12748 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12749 (REGISTER_NAMES): Ditto.
12750 (rs6000_reg_names): Ditto.
12751 * doc/tm.texi.in: Remove SPE as example.
12752 * doc/tm.texi: Regenerate.
12754 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12756 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12757 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12758 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12759 * config/rs6000/rs6000.md (type): Remove "brinc".
12761 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12763 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12764 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12765 * config/rs6000/linuxspe.h: Delete file.
12766 * config/rs6000/rs6000.md: Don't include spe.md.
12767 * config/rs6000/spe.h: Delete file.
12768 * config/rs6000/spe.md: Delete file.
12769 * config/rs6000/t-rs6000: Remove spe.md.
12771 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12773 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12774 (reg_or_none500mem_operand): Delete.
12775 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12776 instead of reg_or_none500mem_operand.
12778 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12780 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12781 handling of SPE flags.
12782 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12784 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12786 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12788 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12789 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12790 paired_divv2sf3): Similar.
12791 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12792 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12793 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12795 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12796 Rename the paired_* instruction patterns.
12797 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12799 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12800 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12801 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12802 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12803 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12804 PAIRED_VECTOR_MODE.
12805 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12806 (spe_func_has_64bit_regs_p): Delete.
12807 (spe_expand_predicate_builtin): Delete.
12808 (spe_expand_evsel_builtin): Delete.
12809 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12810 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12811 (invalid_e500_subreg): Delete.
12812 (rs6000_legitimize_address): Always force_reg op2 as well, for
12813 paired single memory accesses.
12814 (rs6000_member_type_forces_blk): Delete.
12815 (rs6000_spe_function_arg): Delete.
12816 (rs6000_expand_unop_builtin): Delete SPE handling.
12817 (rs6000_expand_binop_builtin): Ditto.
12818 (spe_expand_stv_builtin): Delete.
12819 (bdesc_2arg_spe): Delete.
12820 (spe_expand_builtin): Delete.
12821 (spe_expand_predicate_builtin): Delete.
12822 (spe_expand_evsel_builtin): Delete.
12823 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12824 (spe_init_builtins): Delete.
12825 (spe_func_has_64bit_regs_p): Delete.
12826 (savres_routine_name): Delete "info" parameter. Adjust callers.
12827 (rs6000_emit_stack_reset): Ditto.
12828 (rs6000_dwarf_register_span): Delete.
12829 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12830 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12831 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12833 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12835 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12836 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12837 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12838 mulv2sf3, divv2sf3): Delete expanders.
12840 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12842 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12844 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12846 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12847 * config/rs6000/rs6000.c: Ditto.
12849 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12851 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12852 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12853 comparison_operator.
12855 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12857 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12858 * config/rs6000/rs6000.opt: Ditto.
12859 * config/rs6000/t-rtems: Ditto.
12861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12863 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12864 TARGET_E500_SINGLE by 0, simplify.
12865 * config/rs6000/rs6000.c: Ditto.
12866 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12867 (spe_build_register_parallel): Delete.
12868 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12869 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12870 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12871 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12872 (E500_CONVERT): Delete.
12873 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12875 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12877 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12878 * config/rs6000/dfp.md: Ditto.
12879 (negdd2, *negdd2_fpr): Merge.
12880 (absdd2, *absdd2_fpr): Merge.
12881 (negtd2, *negtd2_fpr): Merge.
12882 (abstd2, *abstd2_fpr): Merge.
12883 * config/rs6000/e500.h: Delete file.
12884 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12885 TARGET_FPRS by 1 and simplify.
12886 * config/rs6000/rs6000-c.c: Ditto.
12887 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12888 TARGET_DF_SPE by 0.
12889 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12891 * config/rs6000/rs6000.md: Ditto.
12892 (floatdidf2, *floatdidf2_fpr): Merge.
12893 (move_from_CR_gt_bit): Delete.
12894 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12895 (E500_CR_IOR_COMPARE): Delete.
12896 (All patterns that require !TARGET_FPRS): Delete.
12897 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12899 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12901 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12903 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12905 * graphds.c (add_edge): Intitialize edge's attached data.
12906 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12907 pointer parameter. Call pointed function on each edge during
12908 graph traversing. Skip traversing the edge when the function
12910 (graphds_dfs, graphds_scc): Ditto.
12911 (for_each_edge): New parameter. Pass the new parameter to callback
12913 * graphds.h (skip_edge_callback): New function pointer type.
12914 (graphds_dfs, graphds_scc): New function pointer parameter.
12915 (graphds_edge_callback, for_each_edge): New parameter.
12917 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12919 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12920 out code checking if runtime alias check is possible to below ...
12921 Call the new function.
12922 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12923 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12925 2017-06-07 Marek Polacek <polacek@redhat.com>
12928 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12929 TYPE_OVERFLOW_WRAPS check.
12931 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12933 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12934 if versioning is required.
12935 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12936 peeling with the check for versioning.
12938 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12940 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12941 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12942 Set true to new parameter if new ssa variable is defined.
12943 (vect_gen_vector_loop_niters): Refactor. Set range information
12944 for the new vector loop bound variable.
12945 (vect_do_peeling): Ditto.
12947 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12949 * tree-affine.c (ssa.h): Include header file.
12950 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12951 has wrapping overflow behavior.
12953 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12955 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12957 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12959 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12960 (tree_to_aff_combination): ... here.
12962 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12964 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12965 reg_pressure model function.
12966 (ivopts_global_cost_for_size): Delete.
12967 (determine_set_costs, iv_ca_recount_cost): Call new model function
12968 ivopts_estimate_reg_pressure.
12970 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12972 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12973 expensive than udiv. Remove floating point cases from mod.
12975 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12977 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12978 Increase idiv cost.
12980 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12982 * config/aarch64/aarch64.md
12983 (copysignsf3): Fix mask generation.
12985 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12987 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12989 (class dump_manager): Add register_dumps method.
12990 * dumpfile.c: Include langhooks.h.
12991 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12992 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12993 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12994 (dump_manager::dump_register): Start with 512 entries instead of 32.
12995 (dump_manager::register_dumps): New method.
12996 * toplev.c (general_init): Instead of invoking register_dumps
12997 langhook, invoke register_dumps method on the dump manager.
12998 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13001 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13003 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13004 Say that # requires an associated define_split to exist, and that
13005 the define_split must be suitable for use after register allocation.
13007 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13009 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13010 (compute_outgoing_frequencies): Also initialize zero counts.
13011 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13012 around loops; preserve more of profile when nothing changes.
13014 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13016 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13018 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13019 * config/arm/arm-cpu-cdata.h: Regenerate.
13020 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13021 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13022 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13023 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13024 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13026 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13028 * config/arm/t-rmprofile: Likewise.
13029 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13031 2017-06-06 David S. Miller <davem@davemloft.net>
13034 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13035 blockage if function uses alloca.
13037 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13039 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13040 New "uid" fields to hold pretty-print IDs of group and ref.
13041 Memory references are now identified as <group_id>:<ref_id>
13042 instead of using [random] addresses.
13043 (dump_mem_details): Simplify, no functional change.
13044 (dump_mem_ref): Simplify and make output more concise.
13045 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13046 (find_or_create_group): Initialize group uid.
13047 (record_ref): Initialize ref uid. Improve debug output.
13048 (prune_group_by_reuse, should_issue_prefetch_p,)
13049 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13050 (mark_nontemporal_store, determine_loop_nest_reuse):
13051 Improve debug output.
13053 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13055 * dbgcnt.def (prefetch): New debug counter.
13056 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13057 (schedule_prefetches): Stop issueing prefetches if debug counter
13060 2017-06-06 Tom de Vries <tom@codesourcery.com>
13062 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13063 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13065 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13067 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13068 Use aarch64_reg_or_zero predicate for operand 4.
13069 (aarch64_compare_and_swap<mode> define_insn_and_split):
13070 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13071 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13073 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13075 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13076 (arm_compute_save_core_reg_mask): This.
13077 (thumb1_compute_save_reg_mask): Rename into ...
13078 (thumb1_compute_save_core_reg_mask): This.
13079 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13080 (arm_compute_frame_layout): Likewise.
13082 2017-06-06 Richard Biener <rguenther@suse.de>
13084 PR tree-optimization/80974
13085 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13086 keep or clear leaders SSA info.
13088 2017-06-06 Tom de Vries <tom@codesourcery.com>
13090 * config/nvptx/nvptx.c (split_mode_p): New function.
13091 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13093 2017-06-06 Tom de Vries <tom@codesourcery.com>
13095 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13097 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13100 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13103 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13105 * shrink-wrap.c (handle_simple_exit): Update profile.
13106 (try_shrink_wrapping): Upate profile.
13108 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13110 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13111 (tree_guess_outgoing_edge_probabilities): New.
13112 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13113 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13115 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13117 * ipa-split.c (split_function): Initialize return bb profile.
13119 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13121 * profile.c (compute_branch_probabilities): Also initialize
13122 EXIT_BLOCK profile.
13124 2017-06-06 Richard Biener <rguenther@suse.de>
13126 PR tree-optimization/80928
13127 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13128 (vect_analyze_loop_operations): Properly guard analysis for
13130 (vect_transform_loop): Likewise.
13131 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13132 (vect_model_induction_cost): Do not cost for pure SLP.
13133 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13134 of induction in inner loop vectorization.
13135 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13136 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13137 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13139 (vect_analyze_slp_cost_1): Cost induction.
13140 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13141 (vect_get_slp_vect_defs): Likewise.
13142 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13143 (vect_transform_stmt): Handle SLP reductions.
13144 * tree-vectorizer.h (vectorizable_induction): Adjust.
13146 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13148 * config/rs6000/rs6000.c (make_resolver_func): Update
13149 init_lowered_empty_function call.
13151 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13153 * doc/invoke.texi: Document the -fprofile-abs-path option.
13154 * common.opt (fprofile-abs-path): New option.
13155 * gcov-io.h (gcov_write_filename): Declare.
13156 * gcov-io.c (gcov_write_filename): New function.
13157 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13158 * profile.c (output_location): Likewise.
13160 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13162 * shring-wrap.c: Revert accidental commit.
13164 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13166 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13168 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13170 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13172 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13173 profile in callgraph edge.
13174 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13175 (apply_scale): Likewise.
13176 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13177 Also scale profile when inlining function with zero profile.
13178 (initialize_cfun): Update exit block profile even when it is zero.
13179 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13180 when profile is read.
13182 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13184 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13185 (CLONE_*): New constants to define the processors we can generate
13186 code for with the target_clone attribute.
13187 (rs6000_clone_map): New array to identify which clone processors
13188 the current program is running on.
13189 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13190 target_clone attribute.
13191 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13192 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13193 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13194 (cpu_expand_builtin): Add support for target_clone attribute.
13195 (rs6000_valid_attribute_p): Allow "default" attribute.
13196 (get_decl_name): New debug function to simplify printing the
13197 current function name in debugging statements.
13198 (rs6000_clone_priority): New functions to support the target_clone
13199 attribute, and be able to generate code to switch between ISA 2.05
13200 through ISA 3.0 (power6 through power9).
13201 (rs6000_compare_version_priority): Likewise.
13202 (rs6000_get_function_versions_dispatcher): Likewise.
13203 (make_resolver_func): Likewise.
13204 (add_condition_to_bb): Likewise.
13205 (dispatch_function_versions): Likewise.
13206 (rs6000_generate_version_dispatcher_body): Likewise.
13207 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13208 (fusion_gpr_load_p): Fix a spacing issue.
13209 * doc/extend.texi (Common Function Attributes): Document that the
13210 PowerPC supports the target_clone attribute.
13212 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13214 * config/arm/arm.h: explain F symbol found in description of ARM
13215 register allocation in its legend.
13217 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13219 * config/mips/frame-header-opt.c: Include profile-count.h.
13220 * config/riscv/riscv.c: Include profile-count.h
13222 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13224 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13226 (sm_set_flag_if_changed): Add bbs field.
13227 (execute_sm_if_changed_flag_set): Pass BBS.
13228 (execute_sm): Update.
13230 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13232 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13235 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13237 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13238 (peephole2): New peephole2 to emit the above.
13239 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13241 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13243 * config/aarch64/aarch64.c (define_peephole2 above
13244 *sub_<shift>_<mode>): New peephole.
13246 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13248 * config/i386/i386.c (make_resolver_func): Update.
13249 * Makefile.in: Add profile-count.h and profile-count.o
13250 * auto-profile.c (afdo_indirect_call): Update to new API.
13251 (afdo_set_bb_count): Update.
13252 (afdo_propagate_edge): Update.
13253 (afdo_propagate_circuit): Update.
13254 (afdo_calculate_branch_prob): Update.
13255 (afdo_annotate_cfg): Update.
13256 * basic-block.h: Include profile-count.h
13257 (struct edge_def): Turn count to profile_count.
13258 (struct basic_block_def): Likewie.
13259 (REG_BR_PROB_BASE): Move to profile-count.h
13260 (RDIV): Move to profile-count.h
13261 * bb-reorder.c (max_entry_count): Turn to profile_count.
13262 (find_traces): Update.
13263 (rotate_loop):Update.
13264 (connect_traces):Update.
13265 (sanitize_hot_paths):Update.
13266 * bt-load.c (migrate_btr_defs): Update.
13267 * cfg.c (RDIV): Remove.
13268 (init_flow): Use alloc_block.
13269 (alloc_block): Uninitialize count.
13270 (unchecked_make_edge): Uninitialize count.
13271 (check_bb_profile): Update.
13272 (dump_edge_info): Update.
13273 (dump_bb_info): Update.
13274 (update_bb_profile_for_threading): Update.
13275 (scale_bbs_frequencies_int): Update.
13276 (scale_bbs_frequencies_gcov_type): Update.
13277 (scale_bbs_frequencies_profile_count): New.
13278 * cfg.h (update_bb_profile_for_threading): Update.
13279 (scale_bbs_frequencies_profile_count): Declare.
13280 * cfgbuild.c (compute_outgoing_frequencies): Update.
13281 (find_many_sub_basic_blocks): Update.
13282 * cfgcleanup.c (try_forward_edges): Update.
13283 (try_crossjump_to_edge): Update.
13284 * cfgexpand.c (expand_gimple_tailcall): Update.
13285 (construct_exit_block): Update.
13286 * cfghooks.c (verify_flow_info): Update.
13287 (dump_bb_for_graph): Update.
13288 (split_edge): Update.
13289 (make_forwarder_block): Update.
13290 (duplicate_block): Update.
13291 (account_profile_record): Update.
13292 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13293 (get_estimated_loop_iterations): Update.
13294 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13295 (single_likely_exit): Update.
13296 * cfgloopmanip.c (scale_loop_profile): Update.
13298 (set_zero_probability): Update.
13299 (lv_adjust_loop_entry_edge): Update.
13300 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13301 (purge_dead_edges): Update.
13302 (rtl_account_profile_record): Update.
13303 * cgraph.c (cgraph_node::create): Uninitialize count.
13304 (symbol_table::create_edge): Uninitialize count.
13305 (cgraph_update_edges_for_call_stmt_node): Update.
13306 (cgraph_edge::dump_edge_flags): Update.
13307 (cgraph_node::dump): Update.
13308 (cgraph_edge::maybe_hot_p): Update.
13309 * cgraph.h: Include profile-count.h
13310 (create_clone), create_edge, create_indirect_edge): Update.
13311 (cgraph_node): Turn count to profile_count.
13312 (cgraph_edge0: Likewise.
13313 (make_speculative, clone): Update.
13314 (create_edge): Update.
13315 (init_lowered_empty_function): Update.
13316 * cgraphclones.c (cgraph_edge::clone): Update.
13317 (duplicate_thunk_for_node): Update.
13318 (cgraph_node::create_clone): Update.
13319 * cgraphunit.c (cgraph_node::analyze): Update.
13320 (cgraph_node::expand_thunk): Update.
13321 * final.c (dump_basic_block_info): Update.
13322 * gimple-streamer-in.c (input_bb): Update.
13323 * gimple-streamer-out.c (output_bb): Update.
13324 * graphite.c (print_global_statistics): Update.
13325 (print_graphite_scop_statistics): Update.
13326 * hsa-brig.c: Include basic-block.h.
13327 * hsa-dump.c: Include basic-block.h.
13328 * hsa-gen.c (T sum_slice): Update.
13329 (convert_switch_statements):Update.
13330 * hsa-regalloc.c: Include basic-block.h.
13331 * ipa-chkp.c (chkp_produce_thunks): Update.
13332 * ipa-cp.c (struct caller_statistics): Update.
13333 (init_caller_stats): Update.
13334 (gather_caller_stats): Update.
13335 (ipcp_cloning_candidate_p): Update.
13336 (good_cloning_opportunity_p): Update.
13337 (get_info_about_necessary_edges): Update.
13338 (dump_profile_updates): Update.
13339 (update_profiling_info): Update.
13340 (update_specialized_profile): Update.
13341 (perhaps_add_new_callers): Update.
13342 (decide_about_value): Update.
13343 (ipa_cp_c_finalize): Update.
13344 * ipa-devirt.c (struct odr_type_warn_count): Update.
13345 (struct decl_warn_count): Update.
13346 (struct final_warning_record): Update.
13347 (possible_polymorphic_call_targets): Update.
13348 (ipa_devirt): Update.
13349 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13350 * ipa-icf.c (sem_function::merge): Update.
13351 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13352 * ipa-inline.c (compute_uninlined_call_time): Update.
13353 (compute_inlined_call_time): Update.
13354 (want_inline_small_function_p): Update.
13355 (want_inline_self_recursive_call_p): Update.
13356 (edge_badness): Update.
13357 (lookup_recursive_calls): Update.
13358 (recursive_inlining): Update.
13359 (inline_small_functions): Update.
13360 (dump_overall_stats): Update.
13361 (dump_inline_stats): Update.
13362 * ipa-profile.c (ipa_profile_generate_summary): Update.
13363 (ipa_propagate_frequency): Update.
13364 (ipa_profile): Update.
13365 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13366 * ipa-utils.c (ipa_merge_profiles): Update.
13367 * loop-doloop.c (doloop_modify): Update.
13368 * loop-unroll.c (report_unroll): Update.
13369 (unroll_loop_runtime_iterations): Update.
13370 * lto-cgraph.c (lto_output_edge): Update.
13371 (lto_output_node): Update.
13372 (input_node): Update.
13373 (input_edge): Update.
13374 (merge_profile_summaries): Update.
13375 * lto-streamer-in.c (input_cfg): Update.
13376 * lto-streamer-out.c (output_cfg): Update.
13377 * mcf.c (create_fixup_graph): Update.
13378 (adjust_cfg_counts): Update.
13379 (sum_edge_counts): Update.
13380 * modulo-sched.c (sms_schedule): Update.
13381 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13382 * predict.c (maybe_hot_count_p): Update.
13383 (probably_never_executed): Update.
13384 (dump_prediction): Update.
13385 (combine_predictions_for_bb): Update.
13386 (propagate_freq): Update.
13387 (handle_missing_profiles): Update.
13388 (counts_to_freqs): Update.
13389 (rebuild_frequencies): Update.
13390 (force_edge_cold): Update.
13391 * predict.h: Include profile-count.h
13392 (maybe_hot_count_p, counts_to_freqs): UPdate.
13393 * print-rtl-function.c: Do not include cfg.h
13394 * print-rtl.c: Include basic-block.h
13395 * profile-count.c: New file.
13396 * profile-count.h: New file.
13397 * profile.c (is_edge_inconsistent): Update.
13398 (correct_negative_edge_counts): Update.
13399 (is_inconsistent): Update.
13400 (set_bb_counts): Update.
13401 (read_profile_edge_counts): Update.
13402 (compute_frequency_overlap): Update.
13403 (compute_branch_probabilities): Update; Initialize and deinitialize
13405 (branch_prob): Update.
13406 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13407 (edge_gcov_count): New.
13408 (bb_gcov_count): New.
13409 * shrink-wrap.c (try_shrink_wrapping): Update.
13410 * tracer.c (better_p): Update.
13411 * trans-mem.c (expand_transaction): Update.
13412 (ipa_tm_insert_irr_call): Update.
13413 (ipa_tm_insert_gettmclone_call): Update.
13414 * tree-call-cdce.c: Update.
13415 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13416 (gimple_duplicate_sese_tail): Update.
13417 (gimple_account_profile_record): Update.
13418 (execute_fixup_cfg): Update.
13419 * tree-inline.c (copy_bb): Update.
13420 (copy_edges_for_bb): Update.
13421 (initialize_cfun): Update.
13422 (freqs_to_counts): Update.
13423 (copy_cfg_body): Update.
13424 (expand_call_inline): Update.
13425 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13426 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13427 (try_unroll_loop_completely): Update.
13428 (try_peel_loop): Update.
13429 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13430 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13431 * tree-ssa-loop-split.c (connect_loops): Update.
13432 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13433 * tree-ssa-reassoc.c (branch_fixup): Update.
13434 * tree-ssa-tail-merge.c (replace_block_by): Update.
13435 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13436 (compute_path_counts): Update.
13437 (update_profile): Update.
13438 (recompute_probabilities): Update.
13439 (update_joiner_offpath_counts): Update.
13440 (estimated_freqs_path): Update.
13441 (freqs_to_counts_path): Update.
13442 (clear_counts_path): Update.
13443 (ssa_fix_duplicate_block_edges): Update.
13444 (duplicate_thread_path): Update.
13445 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13446 (struct switch_conv_info): Update.
13447 * tree-tailcall.c (decrease_profile): Update.
13448 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13449 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13450 * value-prof.c (check_counter): Update.
13451 (gimple_divmod_fixed_value): Update.
13452 (gimple_mod_pow2): Update.
13453 (gimple_mod_subtract): Update.
13454 (gimple_ic_transform): Update.
13455 (gimple_stringop_fixed_value): Update.
13456 * value-prof.h (gimple_ic): Update.
13458 2017-06-02 Carl Love <cel@us.ibm.com>
13460 * config/rs6000/rs6000-c: Add support for built-in functions
13461 vector double vec_doublee (vector signed int);
13462 vector double vec_doublee (vector unsigned int);
13463 vector double vec_doublee (vector float);
13464 vector double vec_doubleh (vector signed int);
13465 vector double vec_doubleh (vector unsigned int);
13466 vector double vec_doubleh (vector float);
13467 vector double vec_doublel (vector signed int);
13468 vector double vec_doublel (vector unsigned int);
13469 vector double vec_doublel (vector float);
13470 vector double vec_doubleo (vector signed int);
13471 vector double vec_doubleo (vector unsigned int);
13472 vector double vec_doubleo (vector float);.
13473 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13474 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13476 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13477 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13478 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13480 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13481 vec_doublel, vec_doubleh.
13482 * doc/extend.texi: Update the built-in documentation file for the
13483 new built-in functions.
13485 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13488 * ipa-inline-analysis.c (free_growth_caches): Set
13489 edge_removal_hook_holder to NULL after removing it.
13491 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13493 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13494 comparision with zero.
13496 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13497 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13498 for early expansion of vec_min and vec_max builtins.
13499 (builtin_function_type): Add min/max unsigned variants to those
13500 identified as having unsigned arguments.
13502 2017-06-02 Olivier Hainque <hainque@adacore.com>
13504 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13506 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13508 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13509 Use VALL_F16 iterator rather than VALL.
13511 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13513 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13514 Emit CBNZ inside loop when doing a strong exchange and comparing
13515 against zero. Generate the CC flags after the loop.
13517 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13519 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13520 (dl_section_ref): New.
13521 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13522 On AIX, append an expression to subtract the size of the
13523 section length to dl_section_ref.
13525 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13527 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13528 for early expansion of vector absolute builtins.
13530 2017-06-02 Richard Biener <rguenther@suse.de>
13532 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13533 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13535 2017-06-02 Richard Biener <rguenther@suse.de>
13537 PR tree-optimization/80948
13538 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13539 stmt order as well.
13541 2017-06-02 Richard Biener <rguenther@suse.de>
13543 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13545 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13546 for inductions as relevant.
13548 2017-06-02 Richard Biener <rguenther@suse.de>
13550 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13551 (vectorizable_induction): ... this. Remove dead code.
13553 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13555 * builtins. (expand_builtin_alloca): Remove second parameter and
13556 infer its value from the first parameter instead.
13557 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13559 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13561 PR rtl-optimization/80903
13562 * loop-doloop.c (add_test): Unshare sequence.
13564 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13566 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13568 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13570 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13572 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13573 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13574 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13575 (xlogue_layout::get_stub_rtx): Make static.
13576 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13577 (xlogue_layout::compute_stub_managed_regs): Rename to...
13578 (xlogue_layout::count_stub_managed_regs): ...this.
13579 (xlogue_layout::is_stub_managed_reg): New function.
13580 (xlogue_layout::m_stub_names): Rename to...
13581 (xlogue_layout::s_stub_names): ...this, make static.
13582 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13583 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13584 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13585 xlogue_layout::s_stub_names): Instantiate statics.
13586 (stub_managed_regs): Remove.
13587 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13588 (disable_call_ms2sysv_xlogues): Rename to...
13589 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13590 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13592 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13593 change after reload_completed.
13594 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13596 (ix86_expand_prologue): Likewise.
13597 (ix86_expand_epilogue): Likewise.
13598 (ix86_expand_split_stack_prologue): Likewise.
13599 (ix86_compute_frame_layout): Remove frame parameter ...
13600 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13601 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13603 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13604 (ix86_frame): Move from here ...
13605 * config/i386/i386.h (ix86_frame): ... to here.
13606 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13607 complete ix86_frame data structure instead. Remove some_ld_name.
13609 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13611 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13612 symbols that hold a DECL_VALUE_EXPR.
13614 2017-06-01 Martin Jambor <mjambor@suse.cz>
13616 PR tree-optimization/80898
13617 * tree-sra.c (process_subtree_disqualification): Removed.
13618 (disqualify_candidate): Do not acll
13619 process_subtree_disqualification.
13620 (subtree_mark_written_and_enqueue): New function.
13621 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13622 RHS has been disqualified and re-queue LHS if necessary. Apart
13623 from that, ignore disqualified RHS.
13625 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13627 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13628 address fetch for z10 or later.
13630 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13632 * config/arc/arc.md (tst_movb): Add guard when splitting.
13634 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13636 * config/arc/arc.c (arc_can_eliminate): Test against
13637 arc_frame_pointer_needed.
13639 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13641 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13642 to prevent store reordering.
13643 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13644 (type): Add block type.
13645 (stack_tie): Define special instruction to be used in
13648 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13650 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13651 constraint. It is not valid for the pattern.
13652 (noncommutative_binary_comparison): Likewise.
13654 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13656 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13659 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13661 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13662 be used by the reg-alloc.
13664 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13666 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13667 reg-alloc when having mul64 or mul32x16 instructions.
13668 (mulsidi3): Likewise.
13669 (umulsidi3): Likewise.
13670 (mulsi32x16): New pattern.
13671 (mulsi64): Likewise.
13672 (mulsidi64): Likewise.
13673 (umulsidi64): Likewise.
13674 (MUL32x16_REG): Define.
13675 (mul64_600): Use MUL32x16_REG.
13676 (mac64_600): Likewise.
13677 (umul64_600): Likewise.
13678 (umac64_600): Likewise.
13680 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13682 * config/arc/arc.md (mulsi3_700): Make it commutative.
13684 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13686 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13688 (*sign_extendsidi2_insn): Likewise for movstosw.
13690 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13692 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13693 the type of the input discriminant value. Convert the
13694 discriminant value of signedness vary.
13696 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13698 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13699 Add to -Wall section.
13701 2017-06-01 Richard Biener <rguenther@suse.de>
13703 PR middle-end/66313
13704 * fold-const.c (fold_plusminus_mult_expr): If the factored
13705 factor may be zero use a wrapping type for the inner operation.
13706 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13707 and handle moved defs.
13708 (process_assignment): Properly guard the unary op case. Return a
13709 tri-state indicating that moving the stmt before the call may allow
13710 to continue. Pass through to_move.
13711 (find_tail_calls): Handle moving unrelated defs before
13714 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13717 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13718 splitter result in the canonical way.
13720 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13722 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13723 also for 32bit target. Update insn attributes.
13724 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13726 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13728 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13729 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13730 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13731 (_mm_maskz_min_ss): New intrinsics.
13733 2017-05-31 Martin Liska <mliska@suse.cz>
13735 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13736 Change comment style to one we normally use.
13737 (vectorizable_reduction): Likewise.
13738 (vectorizable_induction): Likewise.
13739 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13740 (vectorizable_call): Likewise.
13741 (vectorizable_simd_clone_call): Likewise.
13742 (vectorizable_conversion): Likewise.
13743 (vectorizable_assignment): Likewise.
13744 (vectorizable_shift): Likewise.
13745 (vectorizable_operation): Likewise.
13746 (vectorizable_store): Likewise.
13747 (vectorizable_load): Likewise.
13748 * tree-vectorizer.h: Likewise.
13750 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13752 * passes.c (emergency_dump_function): New.
13753 * tree-pass.h (emergency_dump_function): Declare.
13754 * plugin.c (plugins_internal_error_function): Remove.
13755 * plugin.h (plugins_internal_error_function): Remove declaration.
13756 * toplev.c (internal_error_function): New static function. Use it...
13757 (general_init): ...here.
13759 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13761 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13762 (arc_rtx_costs): Add costs for new patterns.
13763 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13764 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13766 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13768 * tree-ssa-strlen.c (get_next_strinfo): New function.
13769 (get_stridx_plus_constant): Use it.
13770 (zero_length_string): Likewise.
13771 (adjust_related_strinfos): Likewise.
13772 (adjust_last_stmt): Likewise.
13774 2017-05-31 Richard Biener <rguenther@suse.de>
13777 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13778 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13780 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13782 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13783 loop_vinfo argument and use of dependence distance vectors.
13784 Check instead whether the two references differ only in their
13785 initial value and assume that they have the same alignment if the
13786 difference is a multiple of the vector alignment.
13787 (vect_analyze_data_refs_alignment): Update call accordingly.
13789 2017-05-31 Martin Liska <mliska@suse.cz>
13792 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13794 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13796 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13797 (create_intersect_range_checks): Move from ...
13798 * tree-data-ref.c (create_intersect_range_checks_index)
13799 (create_intersect_range_checks): ... to here.
13800 (create_runtime_alias_checks): New function factored from ...
13801 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13802 here. Call above function.
13803 * tree-data-ref.h (create_runtime_alias_checks): New function.
13805 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13807 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13808 segment length for dr_b and compute it in wide_int.
13810 2017-05-31 Richard Biener <rguenther@suse.de>
13812 PR tree-optimization/80906
13813 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13814 and pass through iv_map.
13815 (copy_bb_and_scalar_dependences): Adjust.
13816 (translate_pending_phi_nodes): Likewise.
13817 (copy_loop_close_phi_args): Handle code-generating IVs instead
13820 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13822 * diagnostic-color.c (color_dict): Add "type-diff".
13823 (parse_gcc_colors): Update comment.
13824 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13825 -fdiagnostics-show-template-tree and -fno-elide-type.
13826 (GCC_COLORS): Add type-diff to example.
13828 (-fdiagnostics-show-template-tree): New.
13829 (-fno-elide-type): New.
13830 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13831 the pp_format_decoder callback. Call any m_format_postprocessor's
13833 (pretty_printer::pretty_printer): Initialize
13834 m_format_postprocessor.
13835 (pretty_printer::~pretty_printer): Delete any
13836 m_format_postprocessor.
13837 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13838 (class format_postprocessor): New class.
13839 (struct pretty_printer::format_decoder): Document the new parameters.
13840 (struct pretty_printer::m_format_postprocessor): New field.
13841 * tree-diagnostic.c (default_tree_printer): Update for new
13842 bool and const char ** params.
13843 * tree-diagnostic.h (default_tree_printer): Likewise.
13845 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13847 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13848 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13850 rs6000_gen_cell_microcode code.
13851 (rs6000_final_prescan_insn): Delete.
13852 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13853 "warn-cell-microcode" entries.
13854 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13855 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13856 throughout. Change cc_reg_not_micro_cr0_operand to
13857 cc_reg_not_cr0_operand throughout.
13858 (*extendhi<mode>2_noload): Delete.
13859 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13860 (mwarn-cell-microcode): Delete.
13861 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13862 -mgen-cell-microcode and -mwarn-cell-microcode.
13864 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13867 * config/i386/constraints.md (Yd): New constraint.
13869 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13870 and (?Yd, r) alternatives. Update insn attributes.
13871 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13872 and (?*Yd, r) alternatives. Update insn attributes.
13873 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13875 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13877 * gimplify.c (gimplify_modify_expr): Don't create a
13878 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13881 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13883 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13885 2017-05-30 Richard Biener <rguenther@suse.de>
13887 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13888 and reduc_def fields.
13889 (STMT_VINFO_REDUC_TYPE): New define.
13890 (STMT_VINFO_REDUC_DEF): Likewise.
13891 (vect_force_simple_reduction): Adjust prototype.
13892 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13893 (vect_is_simple_reduction): Remove check_reduction argument.
13894 (vect_force_simple_reduction): Adjust and set
13895 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13896 (vectorizable_reduction): Do not re-do reduction analysis
13897 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13898 * tree-parloops.c (gather_scalar_reductions): Adjust.
13900 2017-05-30 Richard Biener <rguenther@suse.de>
13902 PR middle-end/80901
13903 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13906 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13908 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13909 Introduce unknown_misalignment parameter and remove vf.
13910 (vect_peeling_hash_get_lowest_cost):
13911 Pass unknown_misalignment parameter.
13912 (vect_enhance_data_refs_alignment):
13913 Fix unsupportable data ref treatment.
13915 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13917 * tree-vect-data-refs.c (vect_get_data_access_cost):
13918 Workaround for SLP handling.
13919 (vect_enhance_data_refs_alignment):
13920 Compute costs for doing no peeling at all, compare to the best
13921 peeling costs so far and avoid peeling if cheaper.
13923 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13925 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13926 Return peeling info and set costs to zero for unlimited cost
13928 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13929 with unknown misalignment. Compute and costs for unknown
13930 misalignment, compare them to the costs for known misalignment
13931 and choose the cheapest for peeling.
13933 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13935 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13936 (vect_get_peeling_costs_all_drs): Create function.
13937 (vect_peeling_hash_get_lowest_cost):
13938 Use vect_get_peeling_costs_all_drs.
13939 (vect_peeling_supportable): Create function.
13940 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13942 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13944 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13945 DR_HAS_NEGATIVE_STEP.
13946 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13947 (vect_enhance_data_refs_alignment): Use.
13948 (vect_duplicate_ssa_name_ptr_info): Use.
13949 * tree-vectorizer.h (dr_misalignment): Use.
13950 (known_alignment_for_access_p): Use.
13952 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13955 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13957 (has_section_name): New function.
13959 2017-05-30 Martin Liska <mliska@suse.cz>
13962 * auto-profile.c (get_function_decl_from_block): Fix
13965 2017-05-30 Richard Biener <rguenther@suse.de>
13967 PR middle-end/80876
13968 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13970 2017-05-30 Martin Liska <mliska@suse.cz>
13972 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13973 * dumpfile.h (struct dump_file_info): Remove ctors.
13975 2017-05-30 Martin Liska <mliska@suse.cz>
13977 * predict.def: Fix GNU coding style.
13979 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13981 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13982 Mark 'to' argument with ATTRIBUTE_UNUSED.
13984 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13986 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13987 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13988 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13991 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13993 * doc/install.texi (Options specification): Restore entry of
13994 --enable-sjlj-exceptions.
13996 2017-05-27 Michael Eager <eager@eagercon.com>
13999 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14001 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14003 * config/microblaze/microblaze.h
14004 (FIXED_REGISTERS): Update in macro.
14005 (CALL_USED_REGISTERS): Update in macro.
14007 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14009 * doc/install.texi: Add links to macOS binary distributions.
14011 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14015 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14017 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14019 2017-05-26 Martin Liska <mliska@suse.cz>
14021 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14023 2017-05-26 Martin Liska <mliska@suse.cz>
14025 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14026 always leading ';; '.
14027 (dump_bb_info): Likewise.
14028 (brief_dump_cfg): Likewise.
14029 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14030 * dumpfile.c: Remove usage of TDF_VERBOSE.
14031 * dumpfile.h (enum dump_kind): Likewise.
14032 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14033 * print-tree.c (debug_verbose): Remove.
14034 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14035 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14036 * tree-diagnostic.c (default_tree_printer): Replace
14037 TDF_DIAGNOSTIC with TDF_SLIM.
14039 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14041 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14042 in parameter loop, rather than loop_vinfo.
14043 (create_intersect_range_checks): Ditto.
14044 (vect_create_cond_for_alias_checks): Update call to above functions.
14046 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14048 PR tree-optimization/80815
14049 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14050 for merging runtime alias checks. Handle negative DR_STEPs.
14052 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14054 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14056 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14057 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14058 out code pruning runtime alias checks.
14059 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14060 factored out from above.
14061 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14063 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14065 (prune_runtime_alias_test_list): New decalaration.
14067 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14069 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14070 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14071 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14072 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14073 (operator==, comp_dr_with_seg_len_pair): Ditto.
14074 (vect_prune_runtime_alias_test_list): Ditto.
14076 2017-05-26 Martin Liska <mliska@suse.cz>
14079 * params.def: Bound partial-inlining-entry-probability param.
14081 2017-05-26 Marek Polacek <polacek@redhat.com>
14084 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14087 2017-05-26 Richard Biener <rguenther@suse.de>
14089 PR tree-optimization/80842
14090 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14093 2017-05-26 Richard Biener <rguenther@suse.de>
14095 PR tree-optimization/80844
14096 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14098 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14100 * doc/md.texi (Machine Constraints): Update x86 family
14101 machine constraints section to match 'config/i386/constraints.md'.
14103 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14105 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14107 2017-05-25 Nathan Sidwell <nathan@acm.org>
14109 * doc/invoke.texi (--enable-languages): Update documentation.
14111 2017-05-25 Martin Liska <mliska@suse.cz>
14113 * dumpfile.c: Add TDF_FOLDING.
14114 * dumpfile.h (enum dump_kind): Likewise.
14115 * genmatch.c (dt_simplify::gen_1): Use it.
14117 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14119 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14121 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14123 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14124 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14126 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14128 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14129 * match.pd (X == C): Rewrite it here.
14130 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14131 with_certain_nonzero_bits2): New predicates.
14132 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14134 2017-05-24 Nathan Sidwell <nathan@acm.org>
14136 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14139 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14142 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14144 * config/powerpcspe: New port. Files are copied from the rs6000
14145 port, with "rs6000" in filenames replaced by "powerpcspe".
14147 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14149 PR rtl-optimization/80754
14150 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14152 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14154 * config/sparc/sparc.md (length): Return the correct value for -mflat
14155 sibcalls to match output_sibcall.
14157 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14161 * config/rs6000/rs6000.c (struct machine_function): Add new field
14163 (rs6000_get_separate_components): Init that field, use it.
14164 (rs6000_components_for_bb): Use the field.
14166 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14168 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14170 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14172 PR middle-end/80823
14173 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14175 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14178 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14179 address against address_operand predicate.
14180 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14182 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14184 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14185 parameters passed indirectly.
14187 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14189 * config/i386/i386.md (*movdi_internal): Remove SSE4
14190 alternative 18 (?r, *v). Update insn attributes.
14191 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14192 Update insn attributes.
14193 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14194 Update insn attributes.
14195 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14196 alternative 1 (r, v). Remove isa attribute.
14197 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14198 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14199 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14201 2017-05-23 Tom de Vries <tom@codesourcery.com>
14203 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14206 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14208 * cgraphunit.c (symbol_table::process_new_functions): Update.
14209 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14210 (inline_generate_summary): Rename to ...
14211 (ipa_fn_summary_generate): ... this one.
14212 (inline_read_summary): Rename to ...
14213 (ipa_fn_summary_read): ... this one.
14214 (inline_write_summary): Rename to ...
14215 (ipa_fn_summary_write): ... this one.
14216 (inline_free_summary): Rename to ...
14217 (ipa_free_fn_summary): ... this one.
14218 (pass_data_local_fn_summary, pass_local_fn_summary,
14219 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14220 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14221 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14222 make_pass_ipa_fn_summary): New.
14223 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14224 inline_write_summary, inline_free_summary): Remove.
14225 (ipa_free_fn_summary) : New.
14226 * ipa-inline.c (ipa_inline): Update.
14227 (pass_ipa_inline): Do not generate summaries.
14228 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14230 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14231 and add pass_ipa_fn_summary.
14232 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14234 (make_pass_inline_parameters): Remove.
14236 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14238 * omp-low.c (struct omp_context): Remove "default_kind" member.
14241 * omp-offload.c (execute_oacc_device_lower): Remove the
14242 parallelism dimensions function attributes for unparallelized
14243 OpenACC kernels constructs.
14245 2017-05-23 Martin Liska <mliska@suse.cz>
14247 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14249 (cgraph_edge::make_speculative): Likewise.
14250 (cgraph_edge::resolve_speculation): Likewise.
14251 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14252 (cgraph_node::dump): Likewise.
14253 * cgraph.h: Likewise.
14254 * cgraphunit.c (analyze_functions): Likewise.
14255 (symbol_table::compile): Likewise.
14256 * ipa-cp.c (print_all_lattices): Likewise.
14257 (determine_versionability): Likewise.
14258 (initialize_node_lattices): Likewise.
14259 (ipcp_verify_propagated_values): Likewise.
14260 (estimate_local_effects): Likewise.
14261 (update_profiling_info): Likewise.
14262 (create_specialized_node): Likewise.
14263 (perhaps_add_new_callers): Likewise.
14264 (decide_about_value): Likewise.
14265 (decide_whether_version_node): Likewise.
14266 (identify_dead_nodes): Likewise.
14267 (ipcp_store_bits_results): Likewise.
14268 * ipa-devirt.c (dump_targets): Likewise.
14269 (ipa_devirt): Likewise.
14270 * ipa-icf.c (sem_item::dump): Likewise.
14271 (sem_function::equals): Likewise.
14272 (sem_variable::equals): Likewise.
14273 (sem_item_optimizer::read_section): Likewise.
14274 (sem_item_optimizer::execute): Likewise.
14275 (congruence_class::dump): Likewise.
14276 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14277 (dump_inline_summary): Likewise.
14278 (estimate_node_size_and_time): Likewise.
14279 (inline_analyze_function): Likewise.
14280 * ipa-inline-transform.c (inline_call): Likewise.
14281 * ipa-inline.c (report_inline_failed_reason): Likewise.
14282 (want_early_inline_function_p): Likewise.
14283 (edge_badness): Likewise.
14284 (update_edge_key): Likewise.
14285 (inline_small_functions): Likewise.
14286 * ipa-profile.c (ipa_profile): Likewise.
14287 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14288 (ipa_make_edge_direct_to_target): Likewise.
14289 (remove_described_reference): Likewise.
14290 (ipa_impossible_devirt_target): Likewise.
14291 (propagate_controlled_uses): Likewise.
14292 (ipa_print_node_params): Likewise.
14293 (ipcp_transform_function): Likewise.
14294 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14295 (propagate_pure_const): Likewise.
14296 * ipa-reference.c (generate_summary): Likewise.
14297 (read_write_all_from_decl): Likewise.
14298 (propagate): Likewise.
14299 (ipa_reference_read_optimization_summary): Likewise.
14300 * ipa-utils.c (ipa_merge_profiles): Likewise.
14301 * ipa.c (walk_polymorphic_call_targets): Likewise.
14302 (symbol_table::remove_unreachable_nodes): Likewise.
14303 (ipa_single_use): Likewise.
14304 * passes.c (execute_todo): Likewise.
14305 * predict.c (drop_profile): Likewise.
14306 * symtab.c (symtab_node::get_dump_name): New function.
14307 (symtab_node::dump_name): Likewise.
14308 (symtab_node::dump_asm_name): Likewise.
14309 (symtab_node::dump_references): Likewise.
14310 (symtab_node::dump_referring): Likewise.
14311 (symtab_node::dump_base): Likewise.
14312 (symtab_node::debug_symtab): Likewise.
14313 * tree-sra.c (convert_callers_for_node): Likewise.
14314 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14315 * value-prof.c (init_node_map): Likewise.
14317 2017-05-23 Martin Liska <mliska@suse.cz>
14319 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14320 and symtab_node::debug_symtab to symbol_table::debug.
14321 * cgraphunit.c (analyze_functions): Use the renamed function.
14322 (symbol_table::compile): Likewise.
14323 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14324 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14325 * passes.c (execute_todo): Likewise.
14326 * symtab.c (symbol_table::dump): New function.
14327 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14329 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14331 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14332 that nonconst implies exec.
14334 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14336 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14337 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14338 (inline_edge_summary_vec): Turn into ...
14339 (ipa_call_summaries): ... this one.
14340 (redirect_to_unreachable, edge_set_predicate,
14341 evaluate_properties_for_edge, inline_summary_alloc,
14342 reset_ipa_call_summary, reset_inline_summary,
14343 inline_summary_t::duplicate): Update.
14344 (inline_edge_duplication_hook): Turn to ...
14345 (ipa_call_summary_t::duplicate): ... this one.
14346 (inline_edge_removal_hook): Turn to ...
14347 (ipa_call_summary_t::remove): ... this one.
14348 (dump_inline_edge_summary): Turn to ...
14349 (dump_ipa_call_summary): ... this one.
14350 (estimate_function_body_sizes): Update.
14351 (inline_update_callee_summaries): Update.
14352 (remap_edge_change_prob): Update.
14353 (remap_edge_summaries): Update.
14354 (inline_merge_summary): Update.
14355 (do_estimate_edge_time): Update.
14356 (inline_generate_summary): Update.
14357 (inline_read_section): Update.
14358 (inline_read_summary): Update.
14359 (inline_free_summary): Update.
14360 * ipa-inline.c (can_inline_edge_p): Update.
14361 (compute_inlined_call_time): Update.
14362 (want_inline_small_function_p): Update.
14363 (edge_badness): Update.
14364 (early_inliner): Update.
14365 * ipa-inline.h (inline_edge_summary): Turn to ...
14366 (ipa_call_summary): ... this one.
14367 (ipa_call_summary_t): New class.
14368 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14369 (ipa_call_summaries): New.
14370 (inline_edge_summary): Remove.
14371 (estimate_edge_growth): Update.
14372 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14373 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14374 * ipa-split.c (execute_split_functions): Update.
14375 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14377 2017-05-23 Tom de Vries <tom@codesourcery.com>
14379 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14380 attributes): Document rdrand effective target.
14382 2017-05-23 Tom de Vries <tom@codesourcery.com>
14384 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14385 attributes): Sort alphabetically.
14387 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14389 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14391 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14394 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14395 V2DF/V2DI splat into two separate patterns, one that handles
14396 registers, and the other that only handles memory. Drop support
14397 for splatting from a GPR on ISA 2.07 and then splitting the
14398 splat into direct move and splat.
14399 (vsx_splat_<mode>_reg): Likewise.
14400 (vsx_splat_<mode>_mem): Likewise.
14402 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14404 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14406 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14408 PR middle-end/80809
14409 * omp-low.c (finish_taskreg_remap): New function.
14410 (finish_taskreg_scan): If unit size of ctx->record_type
14411 is non-constant, unshare the size expression and replace
14412 decls in it with possible outer var refs.
14414 PR middle-end/80809
14415 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14416 GOVD_SHARED rather than GOVD_PRIVATE with it.
14417 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14418 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14420 PR middle-end/80853
14421 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14422 as last argument to build_outer_var_ref for pointer bases of array
14423 section reductions.
14425 2017-05-19 Martin Sebor <msebor@redhat.com>
14427 * print-tree.c (print_node): Print DECL_READ_P flag.
14429 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14431 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14432 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14433 * cgraph.c: Likewise.
14434 * cgraphunit.c: Likewise.
14435 * gengtype.c: Likewise.
14436 * ipa-cp.c: Likewise.
14437 * ipa-devirt.c: Likewise.
14438 * ipa-icf.c: Likewise.
14439 * ipa-predicate.c: Likewise.
14440 * ipa-profile.c: Likewise.
14441 * ipa-prop.c: Likewise.
14442 * ipa-split.c: Likewise.
14444 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14445 edge_predicate_pool, dump_inline_hints,
14446 inline_summary::account_size_time, redirect_to_unreachable,
14447 edge_set_predicate, set_hint_predicate,
14448 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14449 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14450 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14451 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14452 ipa_call_summary_t::remove, initialize_growth_caches,
14453 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14454 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14455 mark_modified, unmodified_parm_1, unmodified_parm,
14456 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14457 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14458 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14459 will_be_nonconstant_predicate, record_modified_bb_info,
14460 get_minimal_bb, record_modified, param_change_prob,
14461 phi_result_unknown_predicate, predicate_for_phi_result,
14462 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14463 estimate_function_body_sizes, compute_inline_parameters,
14464 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14465 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14466 inline_update_callee_summaries, remap_edge_change_prob,
14467 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14468 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14469 inline_analyze_function, inline_summary_t::insert,
14470 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14471 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14472 inline_free_summary): Move to ipa-fnsummary.h
14473 (predicate_t): Remove.
14474 * ipa-fnsummary.c: New file.
14475 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14476 (enum inline_hints_vals, inline_hints, agg_position_info,
14477 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14478 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14479 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14480 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14481 inline_read_summary, inline_write_summary, inline_free_summary,
14482 inline_analyze_function, initialize_inline_failed,
14483 inline_merge_summary, inline_update_overall_summary,
14484 compute_inline_parameters): Move to ipa-fnsummary.h
14485 * ipa-fnsummary.h: New file.
14486 * ipa-inline-transform.h: Include ipa-inline.h.
14487 * ipa-inline.c: LIkewise.
14489 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14491 * ipa-inline.c (edge_badness): Use inlined_time instead of
14492 inline_summaries->get.
14494 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14496 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14498 2017-05-22 Nathan Sidwell <nathan@acm.org>
14500 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14501 (fdump-lang): Document 'raw' option.
14502 * dumpfile.h (TDI_tu): Delete.
14503 * dumpfile.c (dump_files): Remove translation-unit.
14504 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14506 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14508 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14509 command option from $(AWK) call.
14510 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14511 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14512 [FORMAT]: Remove handling of variable.
14513 * config/avr/t-multilib: Regenerate.
14515 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14517 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14519 (dump_inline_summary): Do not print self_time.
14520 (estimate_function_body_sizes): Do not set self_time.
14521 (compute_inline_parameters): Likewise.
14522 (inline_read_section, inline_write_summary): Do not stream self_time.
14523 * ipa-inline.h (inline_summary): Drop self_time.
14525 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14527 * ipa-inline-analysis.c (account_size_time): Rename to ...
14528 (inline_summary::account_size_time): ... this one.
14529 (reset_ipa_call_summary): Turn to ...
14530 (ipa_call_summary::reset): ... this one.
14531 (reset_inline_summary): Turn to ...
14532 (inline_summary::reset): ... this one.
14533 (inline_summary_t::remove): Update.
14534 (inline_summary_t::duplicate): Update.
14535 (ipa_call_summary_t::remove): Update.
14536 (dump_inline_summary): Update.
14537 (estimate_function_body_sizes): Update.
14538 (compute_inline_parameters): Update.
14539 (estimate_node_size_and_time): Update.
14540 (inline_merge_summary): Update.
14541 (inline_update_overall_summary): Update.
14542 (inline_read_section): Update.
14543 (inline_write_summary): Update.
14544 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14545 add account_size_time and reset member functions.
14546 (ipa_call_summary): Add reset function.
14547 * ipa-predicate.h (predicate::operator &): Constify.
14549 2017-05-22 Richard Biener <rguenther@suse.de>
14551 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14553 2017-05-19 Jason Merrill <jason@redhat.com>
14555 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14557 2017-05-19 Marek Polacek <polacek@redhat.com>
14560 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14561 TYPE_OVERFLOW_WRAPS checks.
14563 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14565 * tree-core.h (enum omp_clause_default_kind): Add
14566 "OMP_CLAUSE_DEFAULT_PRESENT".
14567 * tree-pretty-print.c (dump_omp_clause): Handle it.
14568 * gimplify.c (enum gimplify_omp_var_data): Add
14569 "GOVD_MAP_FORCE_PRESENT".
14570 (gimplify_adjust_omp_clauses_1): Map it to
14571 "GOMP_MAP_FORCE_PRESENT".
14572 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14574 * gimplify.c (oacc_default_clause): Clarify.
14576 2017-05-19 Nathan Sidwell <nathan@acm.org>
14578 LANG_HOOK_REGISTER_DUMPS
14579 * toplev.c (general_init): Call register dump lang hook.
14580 * doc/invoke.texi: Document -fdump-lang option family.
14581 * dumpfile.c (dump_files): Remove class dump here.
14582 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14583 * dumpfile.h (tree_dump_index): Remove TDI_class.
14584 * langhooks-def.h (lhd_register_dumps): Declare.
14585 (LANG_HOOKS_REGISTER_DUMPS): Define.
14586 (LANG_HOOKS_INITIALIZER): Add it.
14587 * langhooks.c (lhd_register_dumps): Define.
14588 * langhooks.h (struct lang_hooks): Add register_dumps.
14590 2017-05-19 Nathan Sidwell <nathan@acm.org>
14592 * context.h (context::set_passes): New.
14593 * context.c (context::context): Do not create pass manager.
14594 * toplev.c (general_init): Create pass manager here.
14596 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14598 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14599 use this splitter if two add or or instructions would also work for
14600 the constant we want to generate.
14602 2017-05-19 Richard Biener <rguenther@suse.de>
14605 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14606 predicate evaluation.
14608 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14610 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14612 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14615 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14617 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14618 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14619 (inline_edge_summary_vec): Turn into ...
14620 (ipa_call_summaries): ... this one.
14621 (redirect_to_unreachable, edge_set_predicate,
14622 evaluate_properties_for_edge, inline_summary_alloc,
14623 reset_ipa_call_summary, reset_inline_summary,
14624 inline_summary_t::duplicate): Update.
14625 (inline_edge_duplication_hook): Turn to ...
14626 (ipa_call_summary_t::duplicate): ... this one.
14627 (inline_edge_removal_hook): Turn to ...
14628 (ipa_call_summary_t::remove): ... this one.
14629 (dump_inline_edge_summary): Turn to ...
14630 (dump_ipa_call_summary): ... this one.
14631 (estimate_function_body_sizes): Update.
14632 (inline_update_callee_summaries): Update.
14633 (remap_edge_change_prob): Update.
14634 (remap_edge_summaries): Update.
14635 (inline_merge_summary): Update.
14636 (do_estimate_edge_time): Update.
14637 (inline_generate_summary): Update.
14638 (inline_read_section): Update.
14639 (inline_read_summary): Update.
14640 (inline_free_summary): Update.
14641 * ipa-inline.c (can_inline_edge_p): Update.
14642 (compute_inlined_call_time): Update.
14643 (want_inline_small_function_p): Update.
14644 (edge_badness): Update.
14645 (early_inliner): Update.
14646 * ipa-inline.h (inline_edge_summary): Turn to ...
14647 (ipa_call_summary): ... this one.
14648 (ipa_call_summary_t): New class.
14649 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14650 (ipa_call_summaries): New.
14651 (inline_edge_summary): Remove.
14652 (estimate_edge_growth): Update.
14653 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14654 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14655 * ipa-split.c (execute_split_functions): Update.
14656 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14658 2017-05-19 Richard Biener <rguenther@suse.de>
14660 PR middle-end/80764
14661 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14663 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14665 * config/rs6000/rs6000.c (struct machine_function): Add field
14666 fpr_is_wrapped_separately.
14667 (rs6000_get_separate_components): Use 64 components. Handle the
14668 new FPR components.
14669 (rs6000_components_for_bb): Handle the FPR components.
14670 (rs6000_emit_prologue_components): Handle the FPR components.
14671 (rs6000_emit_epilogue_components): Handle the FPR components.
14672 (rs6000_set_handled_components): Handle the FPR components.
14673 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14674 that are already separately shrink-wrapped.
14675 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14676 that are already separately shrink-wrapped.
14678 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14681 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14684 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14685 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14686 cases where the register allocator uses a move and an offsettable
14687 memory operation to/from a FPR register on ISA 2.06/2.07.
14688 (define_peephole2 for Altivec d-form store): Likewise.
14690 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14693 * config/i386/mmx.md (*mov<mode>_internal): Enable
14694 alternatives 11, 12, 13 and 14 also for 32bit targets.
14695 Remove alternatives 15, 16, 17 and 18.
14696 * config/i386/sse.md (vec_concatv2di): Change
14697 alternative (!x, *y) to (x, ?!*Yn).
14699 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14701 * dumpfile.h (enum dump_kind): Remove stray comma.
14703 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14705 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14706 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14707 predicate::num_conditions
14708 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14709 (CHANGED): turn into predicate::changed.
14710 (agg_position_info): Move to ipa-predicate.h
14711 (add_condition, predicate::add_clause, predicate::operator &=,
14712 predicate::or_with, predicate::evaluate, predicate::probability,
14713 dump_condition, dump_clause, predicate::dump,
14714 predicate::remap_after_duplication, predicate::remap_after_inlining,
14715 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14716 (evaluate_conditions_for_known_args): Update.
14717 (set_cond_stmt_execution_predicate): Update.
14718 * ipa-inline.h: Include ipa-predicate.h
14719 (condition, inline_param_summary, conditions, agg_position_info,
14720 predicate): Move to ipa-predicate.h
14721 * ipa-predicate.c: New file.
14722 * ipa-predicate.h: New file.
14724 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14726 * final.c (leaf_function_p): Check we are not in a sequence.
14728 2017-05-18 Martin Liska <mliska@suse.cz>
14730 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14731 * dumpfile.c (dump_register): Use new enum dump_kind.
14732 (get_dump_file_name): Likewise.
14733 (dump_enable_all): Likewise.
14734 (dump_switch_p_1): Likewise.
14735 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14736 * dumpfile.h (enum dump_kind): New enum type.
14737 (struct dump_file_info): Create constructor and
14738 format fields and comments.
14739 * passes.c (pass_manager::register_one_dump_file):
14741 * statistics.c (statistics_early_init): Likewise.
14742 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14743 TDF_TREE with TDF_SLIM.
14744 (gather_memory_references_ref): Likewise.
14746 2017-05-18 Martin Liska <mliska@suse.cz>
14748 * vec.h (struct vnull): Use it.
14750 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14752 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14753 (true_predicate, false_predicate, true_predicate_p,
14754 false_predicate_p): Remove.
14755 (single_cond_predicate, not_inlined_predicate): Turn to member function
14757 (add_condition): Update.
14758 (add_clause): Turn to...
14759 (predicate::add_clause): ... this one; update; allow passing NULL
14761 (and_predicates): Turn to ...
14762 (predicate::operator &=): ... this one.
14763 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14764 (or_predicates): Turn to ...
14765 (predicate::or_with): ... this one.
14766 (evaluate_predicate): Turn to ...
14767 (predicate::evaluate): ... this one.
14768 (predicate_probability): Turn to ...
14769 (predicate::probability): ... this one.
14770 (dump_condition): Update.
14771 (dump_predicate): Turn to ...
14772 (predicate::dump): ... this one.
14773 (account_size_time): Update.
14774 (edge_set_predicate): Update.
14775 (set_hint_predicate): UPdate.
14776 (evaluate_conditions_for_known_args): Update.
14777 (evaluate_properties_for_edge): Update.
14778 (remap_predicate_after_duplication): Turn to...
14779 (predicate::remap_after_duplication): ... this one.
14780 (remap_hint_predicate_after_duplication): Update.
14781 (inline_summary_t::duplicate): UPdate.
14782 (dump_inline_edge_summary): Update.
14783 (dump_inline_summary): Update.
14784 (set_cond_stmt_execution_predicate): Update.
14785 (set_switch_stmt_execution_predicate): Update.
14786 (compute_bb_predicates): Update.
14787 (will_be_nonconstant_expr_predicate): Update.
14788 (will_be_nonconstant_predicate): Update.
14789 (phi_result_unknown_predicate): Update.
14790 (predicate_for_phi_result): Update.
14791 (array_index_predicate): Update.
14792 (estimate_function_body_sizes): Update.
14793 (estimate_node_size_and_time): Update.
14794 (estimate_ipcp_clone_size_and_time): Update.
14795 (remap_predicate): Rename to ...
14796 (predicate::remap_after_inlining): ... this one.
14797 (remap_hint_predicate): Update.
14798 (inline_merge_summary): Update.
14799 (inline_update_overall_summary): Update.
14800 (estimate_size_after_inlining): Update.
14801 (read_predicate): Rename to ...
14802 (predicate::stream_in): ... this one.
14803 (read_inline_edge_summary): Update.
14804 (write_predicate): Rename to ...
14805 (predicate::stream_out): ... this one.
14806 (write_inline_edge_summary): Update.
14807 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14808 (clause_t): Turn to uint32_t
14809 (predicate): Turn to class; implement constructor and operators
14811 (size_time_entry): Update.
14812 (inline_summary): Update.
14813 (inline_edge_summary): Update.
14815 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14817 * fold-const.c (fold_binary_loc): Move transformation...
14818 * match.pd (C - X CMP X): ... here.
14820 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14822 * config/sparc/sparc.c (sparc_option_override): Set function
14823 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14824 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14826 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14828 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14830 2017-05-18 Marek Polacek <polacek@redhat.com>
14833 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14834 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14835 gimple_assign_load_p.
14837 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14839 PR middle-end/80692
14840 * real.c (do_compare): Give decimal_do_compare preference over
14841 comparing just the signs.
14843 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14845 * doc/md.texi (Canonicalization of Instructions): Describe the
14846 canonical form of instructions that inherently set a condition
14849 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14851 PR middle-end/80775
14852 * tree-cfg.c: Move deletion of unreachable case statements to after
14853 the merging of consecutive case labels.
14855 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14857 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14858 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14859 restoring of callee-saved registers.
14861 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14863 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14864 * config/visium/visium.c (single_set_and_flags): Likewise.
14865 * config/visium/visium.md (Substitutions): Likewise.
14867 2017-05-17 Martin Liska <mliska@suse.cz>
14869 * cfg.c: Introduce dump_flags_t type and
14870 use it instead of int type.
14872 * cfghooks.c: Likewise.
14873 * cfghooks.h (struct cfg_hooks): Likewise.
14874 * cfgrtl.c: Likewise.
14875 * cfgrtl.h: Likewise.
14876 * cgraph.c (cgraph_node::get_body): Likewise.
14877 * coretypes.h: Likewise.
14878 * domwalk.c: Likewise.
14879 * domwalk.h: Likewise.
14880 * dumpfile.c (struct dump_option_value_info): Likewise.
14881 (dump_enable_all): Likewise.
14882 (dump_switch_p_1): Likewise.
14883 (opt_info_switch_p): Likewise.
14884 * dumpfile.h (enum tree_dump_index): Likewise.
14885 (struct dump_file_info): Likewise.
14886 * genemit.c: Likewise.
14887 * generic-match-head.c: Likewise.
14888 * gengtype.c (open_base_files): Likewise.
14889 * gimple-pretty-print.c: Likewise.
14890 * gimple-pretty-print.h: Likewise.
14891 * graph.c (print_graph_cfg): Likewise.
14892 * graphite-scop-detection.c (dot_all_sese): Likewise.
14893 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14894 * loop-unroll.c (report_unroll): Likewise.
14895 * passes.c (pass_manager::register_one_dump_file): Likewise.
14896 * print-tree.c: Likewise.
14897 * statistics.c: Likewise.
14898 * tree-cfg.c: Likewise.
14899 * tree-cfg.h: Likewise.
14900 * tree-dfa.c: Likewise.
14901 * tree-dfa.h: Likewise.
14902 * tree-dump.c (dump_function): Likewise.
14903 * tree-dump.h (struct dump_info): Likewise.
14904 * tree-pretty-print.c: Likewise.
14905 * tree-pretty-print.h: Likewise.
14906 * tree-ssa-live.c: Likewise.
14907 * tree-ssa-live.h: Likewise.
14908 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14909 * tree-vect-loop.c: Likewise.
14910 * tree-vect-slp.c: Likewise.
14912 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14913 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14915 PR tree-optimization/80457
14916 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14917 of all arguments to a statement as scalar_to_vec operations.
14918 (vectorizable_call): Adjust call to vect_model_simple_cost for
14920 (vectorizable_conversion): Likewise.
14921 (vectorizable_assignment): Likewise.
14922 (vectorizable_shift): Likewise.
14923 (vectorizable_operation): Likewise.
14924 (vectorizable_comparison): Likewise.
14925 (vect_is_simple_cond): Record the def types for operands.
14926 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14927 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14928 for statement argument count.
14930 2017-05-16 Carl Love <cel@us.ibm.com>
14932 * config/rs6000/rs6000-c: Add support for built-in functions
14933 vector unsigned long long vec_bperm (vector unsigned long long,
14934 vector unsigned char)
14935 vector signed long long vec_mule (vector signed int,
14937 vector unsigned long long vec_mule (vector unsigned int,
14938 vector unsigned int)
14939 vector signed long long vec_mulo (vector signed int,
14941 vector unsigned long long vec_mulo (vector unsigned int,
14942 vector unsigned int)
14943 vector signed char vec_sldw (vector signed char,
14944 vector signed char,
14946 vector unsigned char vec_sldw (vector unsigned char,
14947 vector unsigned char,
14949 vector signed short vec_sldw (vector signed short,
14950 vector signed short,
14952 vector unsigned short vec_sldw (vector unsigned short,
14953 vector unsigned short,
14955 vector signed int vec_sldw (vector signed int,
14958 vector unsigned int vec_sldw (vector unsigned int,
14959 vector unsigned int,
14961 vector signed long long vec_sldw (vector signed long long,
14962 vector signed long long,
14964 vector unsigned long long vec_sldw (vector unsigned long long,
14965 vector unsigned long long,
14967 * config/rs6000/rs6000-c: Add support for built-in functions
14968 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14969 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14970 * doc/extend.texi: Update the built-in documentation for the
14971 new built-in functions.
14973 2017-05-16 Marek Polacek <polacek@redhat.com>
14977 * tree.c (save_expr): Don't fold the expression.
14979 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14981 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14982 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14983 and (?*y,m). Update insn attributes.
14985 2017-05-16 Martin Liska <mliska@suse.cz>
14987 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14988 flags argument of print_gimple_stmt, print_gimple_expr,
14989 print_generic_stmt and print_generic_expr.
14990 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14991 * coretypes.h: Likewise.
14992 * except.c (dump_eh_tree): Likewise.
14993 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14994 * gimple-pretty-print.h: Likewise.
14995 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14996 (backprop::push_to_worklist): Likewise.
14997 (backprop::pop_from_worklist): Likewise.
14998 (backprop::process_use): Likewise.
14999 (backprop::intersect_uses): Likewise.
15000 (note_replacement): Likewise.
15001 * gimple-ssa-store-merging.c
15002 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15003 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15004 (pass_store_merging::execute): Likewise.
15005 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15006 (ssa_base_cand_dump_callback): Likewise.
15007 (dump_incr_vec): Likewise.
15008 (replace_refs): Likewise.
15009 (replace_mult_candidate): Likewise.
15010 (create_add_on_incoming_edge): Likewise.
15011 (create_phi_basis): Likewise.
15012 (insert_initializers): Likewise.
15013 (all_phi_incrs_profitable): Likewise.
15014 (introduce_cast_before_cand): Likewise.
15015 (replace_one_candidate): Likewise.
15016 * gimplify.c (gimplify_expr): Likewise.
15017 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15018 (set_rename): Likewise.
15019 (rename_uses): Likewise.
15020 (copy_loop_phi_nodes): Likewise.
15021 (add_close_phis_to_merge_points): Likewise.
15022 (copy_loop_close_phi_args): Likewise.
15023 (copy_cond_phi_args): Likewise.
15024 (graphite_copy_stmts_from_block): Likewise.
15025 (translate_pending_phi_nodes): Likewise.
15026 * graphite-poly.c (print_pdr): Likewise.
15027 (dump_gbb_cases): Likewise.
15028 (dump_gbb_conditions): Likewise.
15029 (print_scop_params): Likewise.
15030 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15031 (build_cross_bb_scalars_use): Likewise.
15032 (gather_bbs::before_dom_children): Likewise.
15033 * hsa-dump.c (dump_hsa_immed): Likewise.
15034 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15035 (get_replacement_map): Likewise.
15036 * ipa-inline-analysis.c (dump_condition): Likewise.
15037 (estimate_function_body_sizes): Likewise.
15038 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15039 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15040 * ipa-prop.c (ipa_dump_param): Likewise.
15041 (ipa_print_node_jump_functions_for_edge): Likewise.
15042 (ipa_modify_call_arguments): Likewise.
15043 (ipa_modify_expr): Likewise.
15044 (ipa_dump_param_adjustments): Likewise.
15045 (ipa_dump_agg_replacement_values): Likewise.
15046 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15047 * ipa-pure-const.c (check_stmt): Likewise.
15048 (pass_nothrow::execute): Likewise.
15049 * ipa-split.c (execute_split_functions): Likewise.
15050 * omp-offload.c (dump_oacc_loop_part): Likewise.
15051 (dump_oacc_loop): Likewise.
15052 * trans-mem.c (tm_log_emit): Likewise.
15053 (tm_memopt_accumulate_memops): Likewise.
15054 (dump_tm_memopt_set): Likewise.
15055 (dump_tm_memopt_transform): Likewise.
15056 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15057 (print_loop): Likewise.
15058 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15059 (chkp_gather_checks_info): Likewise.
15060 (chkp_get_check_result): Likewise.
15061 (chkp_remove_check_if_pass): Likewise.
15062 (chkp_use_outer_bounds_if_possible): Likewise.
15063 (chkp_reduce_bounds_lifetime): Likewise.
15064 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15065 (chkp_mark_completed_bounds): Likewise.
15066 (chkp_register_incomplete_bounds): Likewise.
15067 (chkp_mark_invalid_bounds): Likewise.
15068 (chkp_maybe_copy_and_register_bounds): Likewise.
15069 (chkp_build_returned_bound): Likewise.
15070 (chkp_get_bound_for_parm): Likewise.
15071 (chkp_build_bndldx): Likewise.
15072 (chkp_get_bounds_by_definition): Likewise.
15073 (chkp_generate_extern_var_bounds): Likewise.
15074 (chkp_get_bounds_for_decl_addr): Likewise.
15075 * tree-chrec.c (chrec_apply): Likewise.
15076 * tree-data-ref.c (dump_data_reference): Likewise.
15077 (dump_subscript): Likewise.
15078 (dump_data_dependence_relation): Likewise.
15079 (analyze_overlapping_iterations): Likewise.
15080 * tree-inline.c (expand_call_inline): Likewise.
15081 (tree_function_versioning): Likewise.
15082 * tree-into-ssa.c (dump_defs_stack): Likewise.
15083 (dump_currdefs): Likewise.
15084 (dump_names_replaced_by): Likewise.
15085 (dump_update_ssa): Likewise.
15086 (update_ssa): Likewise.
15087 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15088 * tree-parloops.c (build_new_reduction): Likewise.
15089 (try_create_reduction_list): Likewise.
15090 (ref_conflicts_with_region): Likewise.
15091 (oacc_entry_exit_ok_1): Likewise.
15092 (oacc_entry_exit_single_gang): Likewise.
15093 * tree-pretty-print.h: Likewise.
15094 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15095 (get_scalar_evolution): Likewise.
15096 (add_to_evolution): Likewise.
15097 (get_loop_exit_condition): Likewise.
15098 (analyze_evolution_in_loop): Likewise.
15099 (analyze_initial_condition): Likewise.
15100 (analyze_scalar_evolution): Likewise.
15101 (instantiate_scev): Likewise.
15102 (number_of_latch_executions): Likewise.
15103 (gather_chrec_stats): Likewise.
15104 (final_value_replacement_loop): Likewise.
15105 (scev_const_prop): Likewise.
15106 * tree-sra.c (dump_access): Likewise.
15107 (disqualify_candidate): Likewise.
15108 (create_access): Likewise.
15109 (reject): Likewise.
15110 (maybe_add_sra_candidate): Likewise.
15111 (create_access_replacement): Likewise.
15112 (analyze_access_subtree): Likewise.
15113 (analyze_all_variable_accesses): Likewise.
15114 (sra_modify_assign): Likewise.
15115 (initialize_constant_pool_replacements): Likewise.
15116 (find_param_candidates): Likewise.
15117 (decide_one_param_reduction): Likewise.
15118 (replace_removed_params_ssa_names): Likewise.
15119 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15120 * tree-ssa-copy.c (dump_copy_of): Likewise.
15121 (copy_prop_visit_cond_stmt): Likewise.
15122 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15123 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15124 (record_equivalences_from_stmt): Likewise.
15125 * tree-ssa-dse.c (compute_trims): Likewise.
15126 (delete_dead_call): Likewise.
15127 (delete_dead_assignment): Likewise.
15128 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15129 (forward_propagate_into_cond): Likewise.
15130 (pass_forwprop::execute): Likewise.
15131 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15132 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15134 (move_computations_worker): Likewise.
15135 (execute_sm): Likewise.
15136 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15137 (remove_exits_and_undefined_stmts): Likewise.
15138 (remove_redundant_iv_tests): Likewise.
15139 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15140 (adjust_iv_update_pos): Likewise.
15141 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15142 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15143 (value_replacement): Likewise.
15144 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15145 * tree-ssa-pre.c (print_pre_expr): Likewise.
15146 (get_representative_for): Likewise.
15147 (create_expression_by_pieces): Likewise.
15148 (insert_into_preds_of_block): Likewise.
15149 (eliminate_insert): Likewise.
15150 (eliminate_dom_walker::before_dom_children): Likewise.
15151 (eliminate): Likewise.
15152 (remove_dead_inserted_code): Likewise.
15153 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15154 * tree-ssa-reassoc.c (get_rank): Likewise.
15155 (eliminate_duplicate_pair): Likewise.
15156 (eliminate_plus_minus_pair): Likewise.
15157 (eliminate_not_pairs): Likewise.
15158 (undistribute_ops_list): Likewise.
15159 (eliminate_redundant_comparison): Likewise.
15160 (update_range_test): Likewise.
15161 (optimize_range_tests_var_bound): Likewise.
15162 (optimize_vec_cond_expr): Likewise.
15163 (rewrite_expr_tree): Likewise.
15164 (rewrite_expr_tree_parallel): Likewise.
15165 (linearize_expr): Likewise.
15166 (break_up_subtract): Likewise.
15167 (linearize_expr_tree): Likewise.
15168 (attempt_builtin_powi): Likewise.
15169 (attempt_builtin_copysign): Likewise.
15170 (transform_stmt_to_copy): Likewise.
15171 (transform_stmt_to_multiply): Likewise.
15172 (dump_ops_vector): Likewise.
15173 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15174 (print_scc): Likewise.
15175 (set_ssa_val_to): Likewise.
15176 (visit_reference_op_store): Likewise.
15177 (visit_use): Likewise.
15178 (sccvn_dom_walker::before_dom_children): Likewise.
15179 (run_scc_vn): Likewise.
15180 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15182 (expr_hash_elt::print): Likewise.
15183 (const_and_copies::pop_to_marker): Likewise.
15184 (const_and_copies::record_const_or_copy_raw): Likewise.
15185 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15186 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15187 (dump_predicates): Likewise.
15188 (find_uninit_use): Likewise.
15189 (warn_uninitialized_phi): Likewise.
15190 (pass_late_warn_uninitialized::execute): Likewise.
15191 * tree-ssa.c (verify_vssa): Likewise.
15192 (verify_ssa): Likewise.
15193 (maybe_optimize_var): Likewise.
15194 * tree-vrp.c (dump_value_range): Likewise.
15195 (dump_all_value_ranges): Likewise.
15196 (dump_asserts_for): Likewise.
15197 (register_edge_assert_for_2): Likewise.
15198 (vrp_visit_cond_stmt): Likewise.
15199 (vrp_visit_switch_stmt): Likewise.
15200 (vrp_visit_stmt): Likewise.
15201 (vrp_visit_phi_node): Likewise.
15202 (simplify_cond_using_ranges_1): Likewise.
15203 (fold_predicate_in): Likewise.
15204 (evrp_dom_walker::before_dom_children): Likewise.
15205 (evrp_dom_walker::push_value_range): Likewise.
15206 (evrp_dom_walker::pop_value_range): Likewise.
15207 (execute_early_vrp): Likewise.
15209 2017-05-16 Richard Biener <rguenther@suse.de>
15211 * dwarf2out.c (loc_list_from_tree_1): Do not create
15212 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15214 2017-05-16 Richard Biener <rguenther@suse.de>
15216 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15218 (note_variable_value_in_expr): If we resolved the decl ref
15219 do not push to the stack.
15221 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15223 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15224 operations in fast-math mode.
15225 (vaddq_f16): Likewise.
15226 (vmul_f16): Likewise.
15227 (vmulq_f16): Likewise.
15228 (vsub_f16): Likewise.
15229 (vsubq_f16): Likewise.
15230 * config/arm/neon.md (add<mode>3): New.
15232 (fma:<VH:mode>3): New. Also remove outdated comment.
15235 2017-05-16 Martin Liska <mliska@suse.cz>
15239 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15240 (odr_types_equivalent_p): Likewise.
15242 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15244 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15246 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15249 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15250 non-interunit SSE move alternatives with '?'.
15251 (zero-extendsidi peephole2): New peephole to skip intermediate
15252 general register in SSE zero-extend sequence.
15254 2017-05-15 Jeff Law <law@redhat.com>
15256 * reorg.c (relax_delay_slots): Create a new variable to hold
15257 the temporary target rather than clobbering TARGET_LABEL.
15259 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15260 missing argument to extract_bit_field call.
15261 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15263 2017-05-15 Martin Liska <mliska@suse.cz>
15266 * gcc.c (process_command): Do not allow empty argument of -o option.
15268 2017-05-15 Renlin Li <renlin.li@arm.com>
15270 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15271 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15272 * config/aarch64/constraints.md (Usf): Add long call check.
15273 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15274 (call_value): Likewise.
15275 (sibcall): Likewise.
15276 (sibcall_value): Likewise.
15278 (call_value_insn): New.
15279 (sibcall_insn): Update rtx pattern.
15280 (sibcall_value_insn): Likewise.
15281 (call_internal): Remove.
15282 (call_value_internal): Likewise.
15283 (sibcall_internal): Likewise.
15284 (sibcall_value_internal): Likewise.
15285 (call_reg): Likewise.
15286 (call_symbol): Likewise.
15287 (call_value_reg): Likewise.
15288 (call_value_symbol): Likewise.
15290 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15293 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15295 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15297 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15298 compatible with CCGOCmode and with CCZmode.
15300 2017-05-14 Martin Sebor <msebor@redhat.com>
15302 PR middle-end/77671
15303 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15304 (gimple_fold_builtin_snprintf): Same.
15305 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15306 (gimple_fold_builtin_snprintf): Same.
15307 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15308 of character types.
15309 (is_call_safe): New function.
15310 (try_substitute_return_value): Call it.
15311 (try_simplify_call): New function.
15312 (pass_sprintf_length::handle_gimple_call): Call it.
15314 2017-05-14 Martin Sebor <msebor@redhat.com>
15316 PR middle-end/80669
15317 * builtins.c (expand_builtin_stpncpy): Simplify.
15319 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15321 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15322 * config/i386/i386.h
15323 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15324 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15325 (struct machine_function): Add new members call_ms2sysv,
15326 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15327 (struct machine_frame_state): New fields sp_realigned and
15328 sp_realigned_offset.
15329 * config/i386/i386.c
15330 (enum xlogue_stub): New enum.
15331 (enum xlogue_stub_sets): New enum.
15332 (class xlogue_layout): New class.
15333 (struct ix86_frame): New fields stack_realign_allocate_offset,
15334 stack_realign_offset and outlined_save_offset. Modify comments to
15335 detail stack layout when using out-of-line stubs.
15336 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15337 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15338 -mcall-ms2sysv-xlogues.
15339 (stub_managed_regs): New static variable.
15340 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15341 registers managed by out-of-line stub.
15342 (disable_call_ms2sysv_xlogues): New function.
15343 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15344 m->call_ms2sysv when appropriate and compute frame layout for
15346 (sp_valid_at, fp_valid_at): New inline functions.
15347 (choose_basereg): New function.
15348 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15350 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15351 Use align parameter of choose_baseaddr to generated aligned SSE movs
15353 (pro_epilogue_adjust_stack): Modify to track
15354 machine_frame_state::sp_realigned.
15355 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15356 (ix86_nsaved_sseregs): Likewise.
15357 (ix86_emit_save_regs): Likewise.
15358 (ix86_emit_save_regs_using_mov): Likewise.
15359 (ix86_emit_save_sse_regs_using_mov): Likewise.
15360 (get_scratch_register_on_entry): Likewise.
15361 (gen_frame_set): New function.
15362 (gen_frame_load): Likewise.
15363 (gen_frame_store): Likewise.
15364 (emit_outlined_ms2sysv_save): Likewise.
15365 (emit_outlined_ms2sysv_restore): Likewise.
15366 (ix86_expand_prologue): Modify stack re-alignment code and call
15367 emit_outlined_ms2sysv_save when appropriate.
15368 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15369 parameter rtx_insn *insn, which allows the function to be used to only
15370 generate the notes.
15371 (ix86_expand_epilogue): Modify validity checks of frame and stack
15372 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15373 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15374 * config/i386/predicates.md
15375 (save_multiple): New predicate.
15376 (restore_multiple): Likewise.
15377 * config/i386/sse.md
15378 (save_multiple<mode>): New pattern.
15379 (save_multiple_realign<mode>): Likewise.
15380 (restore_multiple<mode>): Likewise.
15381 (restore_multiple_and_return<mode>): Likewise.
15382 (restore_multiple_leave_return<mode>): Likewise.
15383 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15385 2017-05-14 Julia Koval <julia.koval@intel.com>
15387 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15388 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15389 (__builtin_ia32_xsetbv): New builtins.
15390 * config/i386/i386.c (ix86_expand_special_args_builtin):
15392 (ix86_expand_builtin): Special expand for new intrinsics.
15393 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15394 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15395 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15397 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15399 * cfganal.c (inverted_post_order_compute): Change argument type
15401 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15402 * df-core.c (rest_of_handle_df_initialize): Adjust.
15403 (rest_of_handle_df_finish): Likewise.
15404 (df_analyze_1): Likewise.
15405 (df_analyze): Likewise.
15406 (loop_inverted_post_order_compute): Change argument to be a vec *.
15407 (df_analyze_loop): Adjust.
15408 (df_get_n_blocks): Likewise.
15409 (df_get_postorder): Likewise.
15410 * df.h (struct df_d): Change field to be a vec.
15411 * lcm.c (compute_laterin): Adjust.
15412 (compute_available): Likewise.
15413 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15414 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15415 * tree-ssa-pre.c (compute_antic): Likewise.
15417 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15419 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15420 (depth_first_search::depth_first_search): Change structure init
15421 function to this constructor.
15422 (depth_first_search::add_bb): Rename function to this member.
15423 (depth_first_search::execute): Likewise.
15424 (flow_dfs_compute_reverse_finish): Adjust.
15426 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15428 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15429 (longest_simple_path): Likewise.
15430 * shrink-wrap.c (spread_components): Likewise.
15431 (disqualify_problematic_components): Likewise.
15432 (emit_common_heads_for_components): Likewise.
15433 (emit_common_tails_for_components): Likewise.
15434 (insert_prologue_epilogue_for_components): Likewise.
15436 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15438 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15441 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15443 * df-core.c (df_set_blocks): Start using auto_bitmap.
15444 (df_compact_blocks): Likewise.
15445 * df-problems.c (df_rd_confluence_n): Likewise.
15446 * df-scan.c (df_insn_rescan_all): Likewise.
15447 (df_process_deferred_rescans): Likewise.
15448 (df_update_entry_block_defs): Likewise.
15449 (df_update_exit_block_uses): Likewise.
15450 (df_entry_block_bitmap_verify): Likewise.
15451 (df_exit_block_bitmap_verify): Likewise.
15452 (df_scan_verify): Likewise.
15453 * lra-constraints.c (lra_constraints): Likewise.
15454 (undo_optional_reloads): Likewise.
15455 (lra_undo_inheritance): Likewise.
15456 * lra-remat.c (calculate_gen_cands): Likewise.
15457 (do_remat): Likewise.
15458 * lra-spills.c (assign_spill_hard_regs): Likewise.
15459 (spill_pseudos): Likewise.
15460 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15461 (bitmap_set_subtract_values): Likewise.
15463 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15465 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15466 management with auto_bitmap.
15467 (fix_inter_tick): Likewise.
15468 (fix_recovery_deps): Likewise.
15469 * ira.c (add_store_equivs): Likewise.
15470 (find_moveable_pseudos): Likewise.
15471 (split_live_ranges_for_shrink_wrap): Likewise.
15472 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15473 (rtx_reuse_manager::seen_def_p): Likewise.
15474 (rtx_reuse_manager::set_seen_def): Likewise.
15475 * print-rtl.h (class rtx_reuse_manager): Likewise.
15477 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15479 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15481 (migrate_btr_def): Likewise.
15482 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15483 * df-core.c (loop_post_order_compute): Likewise.
15484 (loop_inverted_post_order_compute): Likewise.
15485 * hsa-common.h: Likewise.
15486 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15487 * init-regs.c (initialize_uninitialized_regs): Likewise.
15488 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15489 (inline_small_functions): Likewise.
15490 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15491 * ira.c (combine_and_move_insns): Likewise.
15492 (build_insn_chain): Likewise.
15493 * loop-invariant.c (find_invariants): Likewise.
15494 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15495 * predict.c (tree_predict_by_opcode): Likewise.
15496 (predict_paths_leading_to): Likewise.
15497 (predict_paths_leading_to_edge): Likewise.
15498 (estimate_loops_at_level): Likewise.
15499 (estimate_loops): Likewise.
15500 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15501 (spread_components): Likewise.
15502 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15503 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15504 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15505 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15506 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15507 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15508 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15509 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15510 (mark_threaded_blocks): Likewise.
15511 (thread_through_all_blocks): Likewise.
15512 * tree-ssa.c (verify_ssa): Likewise.
15513 (execute_update_addresses_taken): Likewise.
15514 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15516 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15518 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15520 (post_order_compute): Likewise.
15521 (inverted_post_order_compute): Likewise.
15522 (pre_and_rev_post_order_compute_fn): Likewise.
15524 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15526 * genrecog.c (int_set::int_set): Explicitly construct our
15527 auto_vec base class.
15528 * vec.h (auto_vec::auto_vec): New constructor.
15530 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15532 * bitmap.h (class auto_bitmap): New constructor taking
15533 bitmap_obstack * argument.
15535 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15537 * bitmap.h (class auto_bitmap): Change type of m_bits to
15538 bitmap_head, and adjust ctor / dtor and member operators.
15540 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15542 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15543 when returned register mode doesn't match original mode.
15545 2017-05-12 Jeff Law <law@redhat.com>
15546 Jakub Jelinek <jakub@redhat.com>
15548 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15549 we look for cc setter after the compare-elim changes.
15550 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15551 within the vector to match what compare-elim now expects.
15552 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15553 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15555 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15556 after the compare-elim changes.
15557 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15558 the vector to match what compare-elim now expects.
15559 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15560 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15561 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15562 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15563 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15565 * config/visium/visium.c (single_set_and_flags): Fix where
15566 we look for cc setter after the compare-elim changes.
15567 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15568 with the vector to match what compare-elim now expects.
15569 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15570 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15571 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15572 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15573 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15574 (neg<mode>2_insn_set_overflow): Likewise.
15576 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15578 PR middle-end/79794
15579 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15580 maybe_expand_insn call, set ops[0].target. If still set after call,
15581 set alt_rtl. Add extra arg to recursive calls.
15582 (extract_bit_field): Add alt_rtl argument. Pass to
15584 * expmed.h (extract_bit_field): Fix prototype.
15585 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15586 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15587 to extract_bit_field_calls.
15588 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15589 Pass alt_rtl to extract_bit_field calls.
15590 * calls.c (store_unaligned_arguments_into_psuedos)
15591 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15592 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15594 * optabs.h (struct expand_operand): Add target bitfield.
15596 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15598 * compare-elim.c (try_eliminate_compare): Canonicalize
15599 operation with embedded compare to
15600 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15601 (set (reg) (operation)].
15603 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15605 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15608 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15609 cost of adding a carry flag for ADC instruction.
15610 [case MINUS]: Ignore the cost of subtracting a carry flag
15611 for SBB instruction.
15613 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15615 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15617 * config/rs6000/bmiintrin.h: New file.
15618 * config/rs6000/bmi2intrin.h: New file.
15619 * config/rs6000/x86intrin.h: New file.
15621 2017-05-12 Jeff Law <law@redhat.com>
15623 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15626 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15628 PR middle-end/80707
15629 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15631 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15633 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15634 early expansion of vector divide builtins.
15635 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15636 builtins identified as having unsigned arguments.
15638 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15640 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15641 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15642 expansion of vector logical operations (and, andc, or, xor,
15645 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15647 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15648 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15650 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15652 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15653 early GIMPLE expansion of vector multiplies.
15655 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15657 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15658 TARGET_HAVE_MOVT conditional.
15659 (movt splitter): Likewise.
15661 2017-05-12 Richard Biener <rguenther@suse.de>
15663 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15664 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15665 Fold all stmts not inplace.
15667 2017-05-12 Richard Biener <rguenther@suse.de>
15669 PR tree-optimization/80713
15670 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15671 inserted_exprs bit for not removed stmts.
15673 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15675 PR middle-end/69921
15676 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15677 parallelized" attribute for parallelized OpenACC kernels.
15678 * omp-offload.c (execute_oacc_device_lower): Use it.
15680 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15681 Set "oacc kernels" attribute.
15682 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15683 parameter. Adjust all users.
15684 (oacc_fn_attrib_kernels_p): Remove function.
15685 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15686 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15687 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15688 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15689 assert "oacc kernels" attribute is set.
15691 2017-05-11 Carl Love <cel@us.ibm.com>
15693 * config/rs6000/rs6000-c: Add support for built-in functions
15694 vector unsigned char vec_popcnt (vector signed char)
15695 vector unsigned char vec_popcnt (vector unsigned char)
15696 vector unsigned short vec_popcnt (vector signed short)
15697 vector unsigned short vec_popcnt (vector unsigned short)
15698 vector unsigned int vec_popcnt (vector signed int)
15699 vector unsigned int vec_popcnt (vector unsigned int)
15700 vector unsigned long long vec_popcnt (vector signed long long)
15701 vector unsigned long long vec_popcnt (vector unsigned long long)
15702 vector signed long long vec_slo (vector signed long long,
15703 vector signed char)
15704 vector signed long long vec_slo (vector signed long long,
15705 vector unsigned char)
15706 vector unsigned long long vec_slo (vector unsigned long long,
15707 vector signed char)
15708 vector unsigned long long vec_slo (vector unsigned long long,
15709 vector unsigned char)
15710 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15711 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15712 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15713 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15714 * doc/extend.texi: Update the built-in documentation file for the
15715 new built-in functions.
15717 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15719 * attribs.h (sorted_attr_string): Move machine independent
15720 functions for target clone support from the i386 port to common
15721 code. Rename ix86_function_versions to common_function_versions.
15722 Rename make_name to make_unique_name.
15723 (common_function_versions): Likewise.
15724 (make_unique_name): Likewise.
15725 (make_dispatcher_decl): Likewise.
15726 (is_function_default_version): Likewise.
15727 * attribs.c (attr_strcmp): Likewise.
15728 (sorted_attr_string): Likewise.
15729 (common_function_versions): Likewise.
15730 (make_unique_name): Likewise.
15731 (make_dispatcher_decl): Likewise.
15732 (is_function_default_version): Likewise.
15733 * config/i386/i386.c (attr_strcmp): Likewise.
15734 (sorted_attr_string): Likewise.
15735 (ix86_function_versions): Likewise.
15736 (make_name): Likewise.
15737 (make_dispatcher_decl): Likewise.
15738 (is_function_default_version): Likewise.
15739 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15741 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15744 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15745 Account for direct move costs for vec_construct of integer
15748 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15751 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15752 (UNSPEC_STX_ATOMIC): Ditto.
15753 (loaddi_via_sse): New insn.
15754 (storedi_via_sse): Ditto.
15755 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15756 Update corresponding peephole2 patterns.
15757 (atomic_storedi_fpu): Ditto.
15759 2017-05-11 Julia Koval <julia.koval@intel.com>
15761 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15762 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15764 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15765 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15766 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15768 2017-05-11 Nathan Sidwell <nathan@acm.org>
15770 * graphite-poly.c: Include dumpfile.h.
15772 * dumpfle.h (dump_function): Declare here ...
15773 * tree-dump.h (dump_function): ... not here.
15774 * dumpfile.c: #include tree-cfg.h.
15775 (dump_function): Move here from ...
15776 * tree-dump.c (dump_function): ... here.
15777 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15778 * graphite-poly.c: Don't include tree-dump.h.
15779 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15780 * print-tree.c: Likewise.
15781 * stor-layout.c: Likewise.
15782 * tree-nested.c: Likewise.
15784 * dumpfile.c (dump_start): Use TDF_FLAGS.
15785 (dump_enable_all): Fix TDF_KIND check thinko.
15787 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15789 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15790 array entries to represent two legal parameterizations of the
15791 overloaded __builtin_cmpb function, as represented by the
15792 P6_OV_BUILTIN_CMPB constant.
15793 (altivec_resolve_overloaded_builtin): Add special case handling
15794 for the __builtin_cmpb function, as represented by the
15795 P6_OV_BUILTIN_CMPB constant.
15796 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15797 (BU_P6_64BIT_2): New macro.
15798 (BU_P6_OVERLOAD_2): New macro
15799 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15800 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15801 (CMPB): Add overload support to represent both 32-bit and 64-bit
15802 compare-bytes function.
15803 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15804 support for TARGET_CMPB.
15805 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15806 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15807 documentation of the __builtin_cmpb overloaded built-in function.
15809 2017-05-11 Richard Biener <rguenther@suse.de>
15811 PR tree-optimization/80705
15812 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15813 bases are not vectorizable.
15815 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15817 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15818 when counting register pressure.
15820 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15822 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15823 (iv_ca_more_deps): Renamed to ...
15824 (iv_ca_compare_deps): ... this.
15825 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15829 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15831 (determine_group_iv_costs): ... here.
15832 (find_inv_vars_cb): Record inv var if it's not recorded before.
15834 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15836 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15837 (get_shiftadd_cost): Ditto.
15839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15841 * tree-ssa-address.c: Include header file.
15842 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15844 (add_to_parts): Refactor.
15845 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15846 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15849 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15851 PR tree-optimization/53090
15852 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15854 (extract_cond_operands): Detect condition with IV on both sides
15855 and return COMP_IV_EXPR_2.
15856 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15857 (rewrite_use_compare): Simplify by removing call to function
15858 extract_cond_operands.
15860 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15862 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15863 (extract_cond_operands): Detect condition comparing against non-
15864 invariant bound and return appropriate enum value.
15865 (find_interesting_uses_cond): Update use of extract_cond_operands.
15866 Handle its return value accordingly.
15867 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15869 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15871 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15872 nonlinear iv_use computation in loop invariant sensitive way.
15874 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15876 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15877 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15881 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15882 (dump_cand): Support iv_cand.inv_exprs.
15883 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15885 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15888 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15890 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15892 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15893 as local function. Include necessary header files.
15894 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15896 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15898 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15900 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15902 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15903 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15904 RSHIFT_EXPR and BIT_NOT_EXPR.
15906 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15908 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15909 (adjust_setup_cost): New parameter supporting round up adjustment.
15910 (struct address_cost_data): Delete.
15911 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15912 (split_address_cost, ptr_difference_cost): Delete.
15913 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15914 (struct ainc_cost_data): New struct.
15915 (get_address_cost_ainc): New function.
15916 (get_address_cost, get_computation_cost): Reimplement.
15917 (determine_group_iv_cost_address): Record inv_expr for all uses of
15919 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15920 (iv_ca_has_deps): Reimplemented to ...
15921 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15923 (iv_ca_extend): Call iv_ca_more_deps.
15925 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15927 * tree-ssa-address.c (struct mem_address): Move to header file.
15928 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15929 * tree-ssa-address.h (struct mem_address): Move from C file.
15930 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15932 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15934 * tree-affine.h (aff_combination_type): New interface.
15935 (aff_combination_zero_p): Remove static.
15936 (aff_combination_const_p): New interface.
15937 (aff_combination_singleton_var_p): New interfaces.
15939 2017-05-11 Richard Biener <rguenther@suse.de>
15941 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15942 Skip unreachable blocks and destinations.
15943 (eliminate): Move stmt removal and fixup ...
15944 (fini_eliminate): ... here. Skip inserted exprs.
15945 (pass_pre::execute): Move fini_pre after fini_eliminate.
15946 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15947 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15948 PRE to get rid of dead code that has invalid SSA form and
15949 split critical edges again.
15951 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15953 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15955 2017-05-11 Richard Biener <rguenther@suse.de>
15957 * passes.c (execute_function_todo): Verify loops if they are
15958 said to be up-to-date.
15959 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15960 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15962 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15965 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15966 handle calling assemble_external ourself.
15969 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15970 modes with zero size. Enhance comment.
15972 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15974 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15975 built-ins for vec_xl and vec_xst with short and char pointer
15978 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15980 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15981 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15982 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15983 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15984 (_mm_maskz_min_round_ss): New intrinsics.
15985 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15986 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15987 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15988 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15989 (__builtin_ia32_minss_mask_round): New builtins.
15990 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15991 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15992 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15994 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15995 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15997 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16000 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16002 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16003 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16004 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16005 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16006 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16007 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16008 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16009 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16010 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16011 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16012 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16013 (__builtin_ia32_mulss_mask_round): New builtins.
16014 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16015 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16016 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16018 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16019 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16021 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16024 2017-05-10 Julia Koval <julia.koval@intel.com>
16026 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16027 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16028 (_mm256_setr_m128i): New intrinsics.
16030 2017-05-10 Julia Koval <julia.koval@intel.com>
16032 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16033 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16034 (_mm_maskz_rcp14_ss): New intrinsics.
16035 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16036 (__builtin_ia32_rcp14ss_mask): New builtins.
16037 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16039 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16041 PR tree-optimization/51513
16042 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16043 (assert_unreachable_fallthru_edge_p): Use it.
16044 (group_case_labels_stmt): Likewise.
16045 * tree-cfg.h: Prototype it.
16046 * stmt.c: Include cfghooks.h and tree-cfg.h.
16047 (emit_case_dispatch_table) <gap_label>: New local variable.
16048 Use it to fill dispatch table gaps.
16049 Test for default_label before updating probabilities.
16050 (expand_case) <default_label>: Remove unneeded initialization.
16051 Test for unreachable default case statement and remove its edge.
16052 Set default_label accordingly.
16053 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16055 2017-05-10 Carl Love <cel@us.ibm.com>
16057 * config/rs6000/rs6000-c: Add support for built-in functions
16058 vector signed char vec_neg (vector signed char)
16059 vector signed short int vec_neg (vector short int)
16060 vector signed int vec_neg (vector signed int)
16061 vector signed long long vec_neg (vector signed long long)
16062 vector float vec_neg (vector float)
16063 vector double vec_neg (vector double)
16064 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16066 * config/rs6000/altivec.h: Add define for vec_neg
16067 * doc/extend.texi: Update the built-in documentation for the
16068 new built-in functions.
16070 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16072 PR tree-optimization/77644
16073 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16075 2017-05-10 Nathan Sidwell <nathan@acm.org>
16077 * dumpfile.h (TDI_lang_all): New.
16078 (TDF_KIND): New. Renumber others
16079 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16081 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16083 (get_dump_file_name): Adjust suffix generation.
16084 (dump_enable_all): Use TDF_KIND.
16085 * doc/invoke.texi (-fdump-lang-all): Document.
16087 * dumpfile.h: Tabify.
16089 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16092 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16093 Move member access before delete.
16095 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16097 * tree-inline.c (expand_call_inline): Split block at stmt
16100 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16103 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16104 are now unused after splitting mov{sf,sd}_hardfloat.
16105 (f32_lr2): Likewise.
16106 (f32_lm): Likewise.
16107 (f32_lm2): Likewise.
16108 (f32_li): Likewise.
16109 (f32_li2): Likewise.
16110 (f32_lv): Likewise.
16111 (f32_sr): Likewise.
16112 (f32_sr2): Likewise.
16113 (f32_sm): Likewise.
16114 (f32_sm2): Likewise.
16115 (f32_si): Likewise.
16116 (f32_si2): Likewise.
16117 (f32_sv): Likewise.
16118 (f32_dm): Likewise.
16119 (f32_vsx): Likewise.
16120 (f32_av): Likewise.
16121 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16122 For movsf, order stores so the VSX stores occur before the GPR
16123 store which encourages the register allocator to use a traditional
16124 FPR instead of a GPR. For movsd, order the stores so that the GPR
16125 store comes before the VSX stores to allow the power6 to work.
16126 This is due to the power6 not having a 32-bit integer store
16127 instruction from a FPR.
16128 (movsf_hardfloat): Likewise.
16129 (movsd_hardfloat): Likewise.
16131 2017-05-09 Martin Sebor <msebor@redhat.com>
16133 PR translation/80280
16134 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16137 PR translation/80280
16138 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16139 data member added in r247778.
16140 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16142 2017-05-09 Nathan Sidwell <nathan@acm.org>
16144 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16146 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16149 2017-05-09 Marek Polacek <polacek@redhat.com>
16151 * doc/invoke.texi: Fix typo.
16153 2017-05-09 Richard Biener <rguenther@suse.de>
16155 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16156 (vrp_val_is_min): Likewise.
16157 (set_value_range_to_value): Likewise.
16158 (set_value_range_to_nonnegative): Likewise.
16159 (gimple_assign_nonzero_p): Likewise.
16160 (gimple_stmt_nonzero_p): Likewise.
16161 (vrp_int_const_binop): Likewise. Remove unreachable case.
16162 (adjust_range_with_scev): Adjust comments.
16163 (compare_range_with_value): Likewise.
16164 (extract_range_from_phi_node): Likewise.
16165 (test_for_singularity): Likewise.
16167 2017-05-09 Richard Biener <rguenther@suse.de>
16169 * tree-vrp.c (get_single_symbol): Add assert that we don't
16170 get overflowed constants as invariant part.
16171 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16172 checks. Use wi::cmp instead of recursing for integer constants.
16173 (compare_values): Just ignore whether we assumed undefined
16174 overflow instead of failing the compare.
16175 (extract_range_for_var_from_comparison_expr): Add comment before the
16176 TREE_NO_WARNING sets.
16177 (test_for_singularity): Likewise.
16178 (extract_range_from_comparison): Do not disable optimization
16179 when we assumed undefined overflow.
16180 (extract_range_basic): Remove init of unused var.
16182 2017-05-09 Richard Biener <rguenther@suse.de>
16184 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16185 (extract_range_from_multiplicative_op_1): Adjust.
16186 (extract_range_from_binary_expr_1): Use int_const_binop.
16188 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16191 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16192 rs6000_store_data_bypass_p in seven define_bypass directives and
16193 in several comments.
16194 * config/rs6000/rs6000-protos.h: Add prototype for
16195 rs6000_store_data_bypass_p function.
16196 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16197 function implements slightly different (rs6000-specific) semantics
16198 than store_data_bypass_p, returning false rather than aborting
16199 with assertion error when arguments do not satisfy the
16200 requirements of store data bypass.
16201 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16202 rs6000_store_data_bypass_p.
16204 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16206 * config/xtensa/xtensa-protos.h
16207 (xtensa_initial_elimination_offset): New declaration.
16208 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16209 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16210 macro definition, add case for FRAME_POINTER_REGNUM when
16211 FRAME_GROWS_DOWNWARD.
16212 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16213 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16214 xtensa_initial_elimination_offset.
16216 2017-05-08 Nathan Sidwell <nathan@acm.org>
16218 * doc/invoke.texi: Alphabetize -fdump options.
16220 2017-05-08 Martin Sebor <msebor@redhat.com>
16222 PR translation/80280
16223 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16225 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16227 * target.def (compute_frame_layout): New optional target hook.
16228 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16229 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16230 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16232 * reload1.c (verify_initial_elim_offsets): Likewise.
16233 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16234 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16235 (arm_get_frame_offsets): Split up into this ...
16236 (arm_compute_frame_layout): ... and this function.
16238 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16240 * config/aarch64/constraints.md (Usa): New constraint.
16241 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16243 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16245 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16246 with_multilib_list after it has been checked.
16248 2017-05-08 Richard Biener <rguenther@suse.de>
16250 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16251 (bitmap_set_subtract_values): Likewise.
16253 2017-05-08 Richard Biener <rguenther@suse.de>
16255 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16256 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16258 (gimple_stmt_nonzero_warnv_p): Rename to ...
16259 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16261 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16262 (extract_range_basic): Adjust, do not disable propagation on
16263 strict overflow sensitive simplification.
16264 (vrp_visit_cond_stmt): Likewise.
16266 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16268 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16269 body size unconditionally.
16271 2017-05-07 Jeff Law <law@redhat.com>
16274 2017-05-06 Jeff Law <law@redhat.com>
16275 PR tree-optimization/78496
16276 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16279 PR tree-optimization/78496
16280 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16281 (simplify_stmt_using_ranges): Call it.
16282 (vrp_dom_walker::before_dom_children): Extract equivalences
16283 from an ASSERT_EXPR with an equality comparison against a
16286 2017-05-06 Jeff Law <law@redhat.com>
16288 PR tree-optimization/78496
16289 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16292 PR tree-optimization/78496
16293 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16294 (simplify_stmt_using_ranges): Call it.
16295 (vrp_dom_walker::before_dom_children): Extract equivalences
16296 from an ASSERT_EXPR with an equality comparison against a
16299 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16301 * lra-constraints.c (lra_copy_reg_equiv): New function.
16302 (split_reg): Use it to copy equivalence information from the
16303 original register to the spill register.
16305 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16307 PR rtl-optimization/75964
16308 * simplify-rtx.c (simplify_const_relational_operation): Remove
16309 invalid handling of comparisons of integer ABS.
16311 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16313 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16314 initialize to zero.
16315 (init_regs): Remove declaration.
16316 (function_arg_advance_32): Initialize error_p as boolean variable.
16318 2017-05-05 Nathan Sidwell <nathan@acm.org>
16320 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16321 lines. Use for (;;).
16323 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16325 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16326 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16327 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16328 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16329 VF=2 that require versioning.
16331 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16333 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16336 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16338 * diagnostic.h (diagnostic_override_option_index): Convert from
16339 macro to inline function.
16341 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16343 * diagnostic.c (last_module_changed_p): New function.
16344 (set_last_module): New function.
16345 (diagnostic_report_current_module): Convert macro usage to
16346 the above functions.
16347 * diagnostic.h (diagnostic_context::last_module): Strengthen
16348 from const line_map * to const line_map_ordinary *.
16349 (diagnostic_last_module_changed): Delete macro.
16350 (diagnostic_set_last_module): Delete macro.
16352 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16354 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16355 with diagnostic_report_diagnostic.
16356 (diagnostic_n_impl_richloc): Likewise.
16357 * diagnostic.h (report_diagnostic): Delete macro.
16358 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16359 with diagnostic_report_diagnostic.
16360 * substring-locations.c (format_warning_va): Likewise.
16362 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16364 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16365 save/restor of format_spec. Move option-printing code to...
16366 (print_option_information): ...this new function, and
16367 reimplement by simply printing to the pretty_printer,
16368 rather than appending to the format string.
16370 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16372 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16373 handling logic into...
16374 (update_effective_level_from_pragmas): ...this new function.
16376 2017-05-04 Andrew Waterman <andrew@sifive.com>
16378 * config/riscv/riscv.opt (mstrict-align): New option.
16379 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16380 (SLOW_UNALIGNED_ACCESS): Define.
16381 (riscv_slow_unaligned_access): Declare.
16382 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16384 (riscv_slow_unaligned_access): New variable.
16385 (rocket_tune_info): Set slow_unaligned_access to true.
16386 (optimize_size_tune_info): Set slow_unaligned_access to false.
16387 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16388 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16389 (riscv_option_override): Set riscv_slow_unaligned_access.
16390 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16392 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16394 * config/riscv/riscv.md: Unify indentation.
16396 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16401 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16403 (extendsi<mode>2): Add support for doing sign extension via
16404 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16405 don't have ISA 3.0 instructions.
16406 (extendsi<mode>2 splitter): Likewise.
16407 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16408 generate the normal insns since SImode can now go in vector
16409 registers. Disallow the special UNSPECs needed for previous
16410 machines to hide SImode being used. Add new insns
16411 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16412 (fix_trunc<mode>si2_stfiwx): Likewise.
16413 (fix_trunc<mode>si2_internal): Likewise.
16414 (fixuns_trunc<mode>si2): Likewise.
16415 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16416 (fctiw<u>z_<mode>_smallint): Likewise.
16417 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16418 of floating point to 32-bit integer from doing a direct move to
16419 the GPR registers to do a store.
16420 (fctiwz_<mode>): Break long line.
16422 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16424 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16425 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16426 (addr_list, addr_offset_valid_p): New.
16427 (split_address_groups): Check offset validity with above function.
16428 (gt-tree-ssa-loop-ivopts.h): Include header file.
16430 2017-05-05 Nathan Sidwell <nathan@acm.org>
16432 * config.gcc (arm*-*-*): Add missing 'fi'.
16434 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16436 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16437 included in -fopt-info does not matter.
16438 * doc/optinfo.texi (-fopt-info): Fix description of default
16439 behavour. Explicitly say order of options included in -fopt-info
16442 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16444 * config.gcc: Allow combinations of aprofile and rmprofile values for
16445 --with-multilib-list.
16446 * config/arm/t-multilib: New file.
16447 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16448 variables. Remove setting of ISA and floating-point ABI in
16449 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16450 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16451 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16452 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16454 * config/arm/t-rmprofile: Likewise except for the matches changes.
16455 * doc/install.texi (--with-multilib-list): Document the combination of
16456 aprofile and rmprofile values and warn about pitfalls in doing that.
16458 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16460 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16461 (movdi_aarch64): Likewise.
16463 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16465 PR tree-optimization/80632
16466 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16468 (build_arrays): Initialize it for virtual phis.
16469 (fix_phi_nodes): Use it for virtual phis.
16471 PR tree-optimization/80558
16472 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16473 [x, y] op z into [x op, y op z] for op & or | if conditions
16476 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16477 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16480 * config/arm/arm.md (use_literal_pool): Remove.
16481 (64-bit immediate split): No longer takes cost into consideration
16482 if arm_disable_literal_pool is enabled.
16483 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16484 used when arm_disable_literal_pool is enabled.
16485 (arm_max_const_double_inline_cost): Remove use of
16486 arm_disable_literal_pool.
16487 (push_minipool_fix): Add assert.
16488 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16489 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16490 (no_literal_pool_sf_immediate): New.
16492 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16494 PR tree-optimization/80613
16495 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16496 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16498 2017-05-05 Richard Biener <rguenther@suse.de>
16500 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16502 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16504 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16505 of this flag from insn conditions due to removal from r247495.
16507 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16509 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16511 (arm_early_store_addr_dep_ptr): Likewise.
16512 * config/arm/aarch-common-protos.h
16513 (arm_early_load_addr_dep_ptr): Add prototype.
16514 (arm_early_store_addr_dep_ptr): Likewise.
16515 * config/arm/cortex-a53.md: Add new bypasses.
16517 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16519 * tree.c (next_type_uid): Change type to unsigned.
16520 (type_hash_canon): Decrement back next_type_uid if
16521 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16522 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16525 2017-05-04 Martin Sebor <msebor@redhat.com>
16527 * builtins.c: Fix a trivial typo in a comment.
16529 PR middle-end/79234
16530 * builtins.c (check_sizes): Adjust to handle reading past the end.
16531 Avoid printing excessive upper bound of ranges. Use %E to print
16532 tree nodes instead of converting them to %wu.
16533 (expand_builtin_memchr): New function.
16534 (compute_dest_size): Rename...
16535 (compute_objsize): ...to this.
16536 (expand_builtin_memcpy): Adjust.
16537 (expand_builtin_mempcpy): Adjust.
16538 (expand_builtin_strcat): Adjust.
16539 (expand_builtin_strcpy): Adjust.
16540 (check_strncat_sizes): Adjust.
16541 (expand_builtin_strncat): Adjust.
16542 (expand_builtin_strncpy): Adjust and simplify.
16543 (expand_builtin_memset): Adjust.
16544 (expand_builtin_bzero): Adjust.
16545 (expand_builtin_memcmp): Adjust.
16546 (expand_builtin): Handle memcmp.
16547 (maybe_emit_chk_warning): Check strncat just once.
16549 2017-05-04 Martin Sebor <msebor@redhat.com>
16551 PR preprocessor/79214
16552 PR middle-end/79222
16553 PR middle-end/79223
16554 * builtins.c (check_sizes): Add inlining context and issue
16555 warnings even when -Wno-system-headers is set.
16556 (check_strncat_sizes): Same.
16557 (expand_builtin_strncat): Same.
16558 (expand_builtin_memmove): New function.
16559 (expand_builtin_stpncpy): Same.
16560 (expand_builtin): Handle memmove and stpncpy.
16562 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16564 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16565 which is not used any more.
16567 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16569 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16571 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16573 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16574 (cortexa53_tunings): Likewise.
16575 (cortexa57_tunings): Likewise.
16576 (cortexa72_tunings): Likewise.
16577 (cortexa73_tunings): Likewise.
16579 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16581 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16582 Set loop alignment to 8.
16584 2017-05-04 Martin Sebor <msebor@redhat.com>
16586 PR translation/80280
16587 * builtins.c (expand_builtin_object_size): Add missing quoting to
16588 %D and like directives.
16589 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16590 (hsa_type_for_tree_type): Same.
16591 (verify_function_arguments): Same.
16592 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16593 * varasm.c (get_section): Same.
16596 2017-05-04 Martin Sebor <msebor@redhat.com>
16598 PR translation/80280
16599 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16601 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16603 * config/aarch64/aarch64.c (generic_addrcost_table):
16604 Change HI/TI mode setting.
16606 2017-05-04 Martin Jambor <mjambor@suse.cz>
16608 PR tree-optimization/80622
16609 * tree-sra.c (comes_initialized_p): New function.
16610 (build_accesses_from_assign): Only set write lazily when
16611 comes_initialized_p is false.
16612 (analyze_access_subtree): Use comes_initialized_p.
16613 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16614 instead of testing for PARM_DECL.
16616 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16618 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16619 constraint on operand 0 to allow more general addressing modes.
16620 Adjust output template.
16621 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16623 * config/aarch64/aarch64-protos.h
16624 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16625 * config/aarch64/constraints.md (Dp): New address constraint.
16626 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16629 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16631 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16632 update use of estimate_ipcp_clone_size_and_time.
16633 (estimate_local_effects): Update use of
16634 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16635 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16636 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16637 Return nonspecialized time.
16639 2017-05-04 Richard Biener <rguenther@suse.de>
16641 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16642 for the last VUSE which def dominates the PHI. Directly call
16644 (get_continuation_for_phi_1): Remove.
16646 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16648 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16649 to explain the use of truncating division. Cap the number of
16650 iterations to the maximum given by nb_iterations_upper_bound,
16653 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16655 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16656 * configure: Regenerate.
16657 * config.in: Regenerate.
16658 * config/i386/driver-mingw32.c: new file.
16659 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16660 * config.host: Link driver-mingw32.o on MinGW host.
16661 * doc/install.texi: Document new --enable-mingw-wildcard configure
16664 2017-05-04 Marek Polacek <polacek@redhat.com>
16666 PR tree-optimization/80612
16667 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16669 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16670 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16672 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16673 (movt splitter): Likewise.
16674 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16675 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16676 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16677 block for Thumb-1 with MOVT.
16678 (thumb2_legitimate_address_p): Move code block ...
16679 (can_avoid_literal_pool_for_label_p): ... into this new function.
16680 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16682 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16683 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16684 "M-profile targets with the MOVT instruction".
16686 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16688 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16689 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16690 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16692 2017-05-04 Martin Liska <mliska@suse.cz>
16694 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16695 variable cond_code.
16697 2017-05-04 Richard Biener <rguenther@suse.de>
16699 * tree.c (array_at_struct_end_p): Handle arrays at struct
16700 end with flexarrays more conservatively. Refactor and treat
16701 arrays of arrays or aggregates more strict. Fix
16702 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16703 * tree.c (array_at_struct_end_p): Adjust prototype.
16704 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16705 * gimple-fold.c (get_range_strlen): Likewise.
16706 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16708 2017-05-04 Richard Biener <rguenther@suse.de>
16710 PR tree-optimization/31130
16711 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16713 (supports_overflow_infinity): Likewise.
16714 (is_negative_overflow_infinity): Likewise.
16715 (is_positive_overflow_infinity): Likewise.
16716 (is_overflow_infinity): Likewise.
16717 (stmt_overflow_infinity): Likewise.
16718 (overflow_infinity_range_p): Likewise.
16719 (usable_range_p): Remove as always returning true.
16720 (make_overflow_infinity): Remove.
16721 (negative_overflow_infinity): Likewise.
16722 (positive_overflow_infinity): Likewise.
16723 (avoid_overflow_infinity): Likewise.
16724 (set_value_range): Adjust accordingly.
16725 (set_value_range_to_nonnegative): Likewise, remove now unused
16726 overflow_infinity arg.
16727 (vrp_operand_equal_p): Adjust.
16728 (update_value_range): Likewise.
16729 (range_int_cst_singleton_p): Likewise.
16730 (operand_less_p): Likewise.
16731 (compare_values_warnv): Likewise.
16732 (extract_range_for_var_from_comparison_expr): Likewise.
16733 (vrp_int_const_binop): Likewise.
16734 (zero_nonzero_bits_from_vr): Likewise.
16735 (extract_range_from_multiplicative_op_1): Likewise.
16736 (extract_range_from_binary_expr_1): Likewise.
16737 (extract_range_from_unary_expr): Likewise.
16738 (extract_range_from_comparison): Likewise.
16739 (extract_range_basic): Likewise.
16740 (adjust_range_with_scev): Likewise.
16741 (compare_ranges): Likewise.
16742 (compare_range_with_value): Likewise.
16743 (dump_value_range): Likewise.
16744 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16746 (simplify_cond_using_ranges): Adjust.
16748 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16750 * brig-builtins.def: Added a builtin for class_f64.
16751 * builtin-types.def: Added a builtin type needed by class_f64.
16753 2017-05-03 Jason Merrill <jason@redhat.com>
16755 * timevar.def: Add TV_CONSTEXPR.
16757 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16759 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16761 2017-05-03 Martin Jambor <mjambor@suse.cz>
16763 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16764 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16765 * ipa-cp.c (ipcp_propagate_stage): Do not call
16766 ipa_update_after_lto_read.
16767 * ipa-inline.c (ipa_inline): Likewise.
16769 2017-05-03 Martin Jambor <mjambor@suse.cz>
16771 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16772 tag. Added a default constructor and a destructor.
16773 (ipa_edge_args_sum_t): New class;
16774 (ipa_edge_args_sum): Declare.
16775 (ipa_edge_args_vector): Remove declaration.
16776 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16777 (ipa_free_edge_args_substructures): Remove declaration.
16778 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16779 (ipa_edge_args_info_available_for_edge_p): Likewise.
16780 * ipa-prop.c (ipa_edge_args_vector): Removed.
16781 (edge_removal_hook_holder): Likewise.
16782 (edge_duplication_hook_holder): Likewise.
16783 (ipa_edge_args_sum): New variable.
16784 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16785 ipa_edge_args_vector.
16786 (ipa_free_edge_args_substructures): Likewise.
16787 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16788 ipa_edge_args_vector.
16789 (ipa_edge_removal_hook): Turned into method
16790 ipa_edge_args_sum_t::remove.
16791 (ipa_edge_duplication_hook): Turned into method
16792 ipa_edge_args_sum_t::duplicate.
16793 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16794 registering edge hooks.
16795 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16796 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16797 ipa_edge_args_sum instead of ipa_edge_args_vector.
16798 * ipa-profile.c (ipa_profile): Likewise.
16800 2017-05-03 Martin Jambor <mjambor@suse.cz>
16802 * symbol-summary.h (function_summary): New method exists.
16803 (function_summary::symtab_removal): Deallocate through release.
16804 (call_summary): New class.
16805 (gt_ggc_mx): New overload.
16806 (gt_pch_nx): Likewise.
16807 (gt_pch_nx): Likewise.
16809 2017-05-03 Jeff Law <law@redhat.com>
16811 PR tree-optimization/78496
16812 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16813 from simplify_cond_using_ranges. Split off code to walk
16814 backwards through casts into ...
16815 (simplify_cond_using_ranges_2): New function.
16816 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16817 (execute_vrp): After identifying jump threads, call
16818 simplify_cond_using_ranges_2.
16820 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16823 * ipa-inline.h (inline_summary): Add ctor.
16824 (create_ggc): Do not use ggc_cleared_alloc.
16826 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16827 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16829 * gcc.c (handle_braces): Support escaping in switch matching
16831 * doc/invoke.texi (Spec Files): Document it.
16832 Remove superfluous @code markup in items.
16834 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16836 * diagnostic-show-locus.c (struct column_range): New struct.
16837 (get_affected_columns): New function.
16838 (get_printed_columns): New function.
16839 (struct correction): New struct.
16840 (correction::ensure_capacity): New function.
16841 (correction::ensure_terminated): New function.
16842 (struct line_corrections): New struct.
16843 (line_corrections::~line_corrections): New dtor.
16844 (line_corrections::add_hint): New function.
16845 (layout::print_trailing_fixits): Reimplement in terms of the new
16847 (selftest::test_overlapped_fixit_printing): New function.
16848 (selftest::diagnostic_show_locus_c_tests): Call it.
16850 2017-05-03 Nathan Sidwell <nathan@acm.org>
16852 Canonicalize canonical type hashing
16853 * tree.h (type_hash_canon_hash): Declare.
16854 * tree.c (type_hash_list, attribute_hash_list): Move into
16855 type_hash_canon_hash.
16856 (build_type_attribute_qual_variant): Break out hash code calc into
16857 type_hash_canon_hash.
16858 (type_hash_canon_hash): New. Generic type hash computation.
16859 (build_range_type_1, build_array_type_1, build_function_type,
16860 build_method_type_directly, build_offset_type, build_complex_type,
16861 make_vector_type): Call it.
16863 2017-05-03 Richard Biener <rguenther@suse.de>
16865 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16866 When all DRs have unknown misaligned do not always peel
16867 when there is a store but apply the same costing model as if
16868 there were only loads.
16870 2017-05-03 Richard Biener <rguenther@suse.de>
16873 PR tree-optimization/80492
16874 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16875 compare_base_decls returning dont-know properly.
16877 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16879 * config/arm/iterators.md (CCSI): New mode iterator.
16880 (arch): New mode attribute.
16881 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16882 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16883 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16884 code iterator for success result mode.
16885 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16886 the corresponding new insn generators.
16888 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16891 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16892 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16894 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16896 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16897 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16898 (DDR_A): Wrap DDR argument in brackets.
16899 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16900 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16901 (DDR_REVERSED_P): Likewise.
16903 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16905 PR tree-optimization/79472
16906 * tree-switch-conversion.c (struct switch_conv_info): Add
16907 contiguous_range and default_case_nonstandard fields.
16908 (collect_switch_conv_info): Compute contiguous_range and
16909 default_case_nonstandard fields, don't clear final_bb if
16910 contiguous_range and only the default case doesn't have the required
16912 (check_all_empty_except_final): Set default_case_nonstandard instead
16913 of failing if contiguous_range and the default case doesn't have empty
16915 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16916 and only the default case doesn't have the required constants. Skip
16918 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16919 if default_case_nonstandard.
16920 (build_constructors): Build constant 1 just once. Assert that default
16921 values aren't inserted in between cases if contiguous_range. Skip
16923 (build_arrays): Skip virtual phis.
16924 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16925 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16926 Handle virtual phis.
16927 (gen_inbound_check): Handle default_case_nonstandard case.
16928 (process_switch): Adjust check_final_bb caller. Call
16929 gather_default_values with the first non-default case instead of
16930 default case if default_case_nonstandard.
16932 2017-05-02 Nathan Sidwell <nathan@acm.org>
16934 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16935 check. Fix formatting.
16937 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16939 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16940 errors when comparing specialized and unspecialized times.
16942 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16944 * diagnostic-show-locus.c
16945 (layout::should_print_annotation_line_p): Make private.
16946 (layout::print_annotation_line): Make private.
16947 (layout::annotation_line_showed_range_p): Make private.
16948 (layout::show_ruler): Make private.
16949 (layout::print_source_line): Make private. Pass in line and
16950 line_width, rather than calling location_get_source_line. Drop
16952 (layout::print_leading_fixits): New method.
16953 (layout::print_any_fixits): Rename to...
16954 (layout::print_trailing_fixits): ...this, and make private.
16955 Don't print newline fixits.
16956 (diagnostic_show_locus): Move logic for printing one row into...
16957 (layout::print_line): ...this new function. Move the
16958 location_get_source_line call and error-handling from
16959 print_source_line to here. Call print_leading_fixits, and rename
16960 print_any_fixits to print_trailing_fixits.
16961 (selftest::test_fixit_insert_containing_newline): Update now that
16962 newlines are partially supported.
16963 (selftest::test_fixit_insert_containing_newline_2): New test.
16964 (selftest::test_fixit_replace_containing_newline): Update comments.
16965 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16966 * edit-context.c (class added_line): New class.
16967 (class edited_line): Describe newline handling in comment.
16968 (edited_line::actually_edited_p): New method.
16969 (edited_line::print_content): Delete redundant decl.
16970 (edited_line::m_predecessors): New field.
16971 (edited_file::print_content): Call edited_line::print_content.
16972 (edited_file::print_diff): Update to support newlines.
16973 (edited_file::print_diff_hunk): Likewise.
16974 (edited_file::print_run_of_changed_lines): New function.
16975 (edited_file::print_diff_line): Convert to...
16976 (print_diff_line): ...this.
16977 (edited_file::get_effective_line_count): New function.
16978 (edited_line::edited_line): Initialize new field m_predecessors.
16979 (edited_line::~edited_line): Clean up m_predecessors.
16980 (edited_line::apply_fixit): Handle newlines.
16981 (edited_line::get_effective_line_count): New function.
16982 (edited_line::print_content): New function.
16983 (edited_line::print_diff_lines): New function.
16984 (selftest::test_applying_fixits_insert_containing_newline): New
16986 (selftest::test_applying_fixits_replace_containing_newline): New
16988 (selftest::insert_line): New function.
16989 (selftest::test_applying_fixits_multiple_lines): Add example of
16991 (selftest::edit_context_c_tests): Call the new tests.
16993 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16995 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16996 parameter cand. Update dump information.
16997 (get_computation_cost): Update uses.
16999 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17001 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17002 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17003 (get_computation_at, rewrite_use_address): Update use of
17004 get_computation_aff.
17006 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17008 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17009 (get_computation): Delete.
17010 (get_computation_cost): Implement like get_computation_cost_at.
17011 Use get_computation_at.
17012 (get_computation_cost_at): Delete.
17013 (rewrite_use_nonlinear_expr): Use get_computation_at.
17014 (rewrite_use_compare, remove_unused_ivs): Ditto.
17016 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17018 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17020 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17022 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17023 (ivopts_global_cost_for_size): Rename parameter and update uses.
17024 (iv_ca_recount_cost): Update uses.
17025 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17026 candidates seperately in n_invs and n_cands.
17027 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17029 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17031 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17032 (find_inv_vars_cb): New.
17033 (find_depends): Renamed to ...
17034 (find_inv_vars): ... this.
17035 (add_candidate_1, force_var_cost): Call find_inv_vars.
17036 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17038 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17040 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17041 inv_vars. Add inv_exprs.
17042 (struct iv_cand): Rename depends_on to inv_vars.
17043 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17044 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17045 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17046 (dump_cand): Dump inv_vars.
17047 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17048 (record_invariant, find_depends, add_candidate_1): Ditto.
17049 (set_group_iv_cost, force_var_cost): Ditto.
17050 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17051 (get_computation_cost_at, get_computation_cost): Ditto.
17052 (determine_group_iv_cost_generic): Ditto.
17053 (determine_group_iv_cost_address): Ditto.
17054 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17055 (determine_group_iv_costs): Ditto.
17056 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17057 (iv_ca_set_remove_invariants): Renamed to ...
17058 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17059 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17060 (iv_ca_set_add_invariants): Renamed to ...
17061 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17062 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17063 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17064 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17065 (create_new_ivs): Remove useless dump.
17067 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17069 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17071 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17072 (iv_ca_set_no_cp, create_new_iv): Ditto.
17074 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17076 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17078 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17080 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17081 function tree_check2.
17083 2017-05-02 Martin Liska <mliska@suse.cz>
17085 * doc/gcov.texi: Add missing preposition.
17086 * gcov.c (function_info::function_info): Properly fill up
17087 all member variables.
17089 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17091 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17093 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17095 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17097 2017-05-02 Martin Liska <mliska@suse.cz>
17100 * lto-streamer-in.c (lto_read_tree_1): Remove
17101 LTO_STREAMER_DEBUG.
17102 * lto-streamer.c (struct tree_hash_entry): Likewise.
17103 (struct tree_entry_hasher): Likewise.
17104 (tree_entry_hasher::hash): Likewise.
17105 (tree_entry_hasher::equal): Likewise.
17106 (lto_streamer_init): Likewise.
17107 (lto_orig_address_map): Likewise.
17108 (lto_orig_address_get): Likewise.
17109 (lto_orig_address_remove): Likewise.
17110 * lto-streamer.h: Likewise.
17111 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17112 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17114 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17116 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17117 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17118 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17119 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17120 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17121 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17122 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17123 (mm_maskz_sub_ss): New intrinsics.
17124 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17125 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17126 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17127 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17128 (__builtin_ia32_subss_mask_round): New builtins.
17129 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17130 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17131 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17133 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17134 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17136 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17139 2017-05-02 Martin Jambor <mjambor@suse.cz>
17141 PR tree-optimization/78687
17142 * tree-sra.c (access): New field parent.
17143 (process_subtree_disqualification): New function.
17144 (disqualify_candidate): Call it.
17145 (build_accesses_from_assign): Reset write flag if creating an
17147 (build_access_subtree): Fill in parent field and also prpagate
17148 down grp_write flag.
17149 (create_artificial_child_access): New parameter set_grp_write, set
17150 grp_write to its value.
17151 (propagate_subaccesses_across_link): Also propagate grp_write flag
17153 (propagate_all_subaccesses): Push the closest parent back to work
17154 queue if add_access_to_work_queue returned true.
17156 2017-05-02 Richard Biener <rguenther@suse.de>
17158 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17159 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17160 -fstrict-overflow documentation.
17161 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17162 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17163 flag_strict_overflow.
17164 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17165 * lto-opts.c (lto_write_options): Do not stream it.
17166 * lto-wrapper.c (merge_and_complain): Do not handle it.
17167 * opts.c (default_options_table): Do not set -fstrict-overflow.
17168 (finish_options): Likewise do not clear it when sanitizing.
17169 * simplify-rtx.c (simplify_const_relational_operation): Do not
17170 test flag_strict_overflow.
17172 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17174 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17175 using enabled attribute.
17176 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17177 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17178 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17179 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17180 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17181 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17182 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17183 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17184 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17185 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17187 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17189 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17191 2017-05-02 Richard Biener <rguenther@suse.de>
17193 PR tree-optimization/80591
17195 2017-04-10 Richard Biener <rguenther@suse.de>
17197 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17200 2017-05-02 Richard Biener <rguenther@suse.de>
17202 PR tree-optimization/80549
17203 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17204 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17205 headers if they do not have a preheader.
17207 2017-05-02 Martin Liska <mliska@suse.cz>
17210 * common.opt: Fix typo.
17211 * doc/invoke.texi: Likewise.
17213 2017-05-01 Jan Beulich <jbeulich@suse.com>
17215 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17216 swapping, add (x,x,m,x,n) alternative.
17218 2017-05-01 Nathan Sidwell <nathan@acm.org>
17220 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17221 unnecessary unadjusted_alignment check.
17223 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17226 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17228 * gimplify.c (gimplify_cilk_detach): New function.
17229 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17230 * tree-core.h: Document EXPR_CILK_SPAWN.
17231 * tree.h (EXPR_CILK_SPAWN): Define.
17233 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17235 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17236 to use new fixit_hint representation, using the "replace" logic.
17237 (get_line_span_for_fixit_hint): Likewise.
17238 (layout::print_any_fixits): Likewise.
17239 (selftest::test_one_liner_many_fixits): Rename to...
17240 (selftest::test_one_liner_many_fixits_1): ...this, and update
17241 comment and expected output to reflect that the multiple fix-it
17242 hints are now consolidated into one insertion.
17243 (selftest::test_one_liner_many_fixits_2): New test.
17244 (selftest::test_diagnostic_show_locus_one_liner): Update for
17246 (selftest::test_fixit_consolidation): Update for fix-it API
17248 * diagnostic.c (print_parseable_fixits): Likewise.
17249 * edit-context.c (edited_line::m_line_events): Convert from
17250 auto_vec <line_event *> to auto_vec <line_event>.
17251 (class line_event): Convert from abstract base class to a concrete
17252 class, taking over the role of replace_event.
17253 (class insert_event): Delete.
17254 (class replace_event): Rename to class line_event. Convert to
17256 (edit_context::add_fixits): Reimplement.
17257 (edit_context::apply_insert): Delete.
17258 (edit_context::apply_replace): Rename to...
17259 (edit_context::apply_fixit): ...this. Convert to half-open range.
17260 (edited_file::apply_insert): Delete.
17261 (edited_file::apply_replace): Rename to...
17262 (edited_file::apply_fixit): ...this.
17263 (edited_line::~edited_line): Drop deletion of events.
17264 (edited_line::apply_insert): Delete.
17265 (edited_line::apply_replace): Rename to...
17266 (edited_line::apply_fixit): ...this. Convert to half-open range.
17267 Update for change to type of m_line_events.
17268 * edit-context.h (edit_context::apply_insert): Delete.
17269 (edit_context::apply_replace): Rename to...
17270 (edit_context::apply_fixit): ...this.
17272 2017-05-01 Martin Sebor <msebor@redhat.com>
17274 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17277 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17280 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17281 __get_cpuid_max returns 0.
17282 (__get_cpuid_count): Ditto.
17284 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17286 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17287 replacement expression is another instance of one of its arguments.
17289 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17292 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17293 check for stack push/pop autoinc.
17294 * config/i386/i386.c (ix86_agi_dependent): Return false
17295 if the only reason why modified_in_p returned true is that
17296 addr is SP based and set_insn is a push or pop.
17298 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17300 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17303 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17306 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17307 (account_size_time): Use two predicates - exec_pred and
17309 (evaluate_conditions_for_known_args): Compute both clause and
17311 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17312 (inline_summary_t::duplicate): Update.
17313 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17315 (compute_inline_parameters): Likewise.
17316 (estimate_edge_size_and_time): Update caluclation of time.
17317 (estimate_node_size_and_time): Compute both time and nonspecialized
17319 (estimate_ipcp_clone_size_and_time): Update.
17320 (inline_merge_summary): Update.
17321 (do_estimate_edge_time): Update.
17322 (do_estimate_edge_size): Update.
17323 (do_estimate_edge_hints): Update.
17324 (inline_read_section, inline_write_summary): Stream both new predicates.
17325 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17327 (compute_inlined_call_time): Cleanup.
17328 (big_speedup_p): Update.
17329 (edge_badness): Update.
17330 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17331 (size_time_entry): Replace predicate by exec_predicate and
17332 nonconst_predicate.
17333 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17334 (estimate_edge_time): Return also nonspec_time.
17335 (reset_edge_growth_cache): Update.
17337 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17339 PR rtl-optimization/80491
17340 * ifcvt.c (noce_process_if_block): When looking for x setter
17341 with missing else_bb, don't check only the insn right before
17342 cond_earliest, but look for the last insn that x is modified in
17343 within the same bb.
17345 PR rtl-optimization/80491
17346 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17348 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17350 PR tree-optimization/80487
17351 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17353 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17355 PR tree-optimization/79697
17356 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17357 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17358 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17360 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17361 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17363 2017-04-28 Martin Sebor <msebor@redhat.com>
17365 PR tree-optimization/80523
17366 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17367 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17369 (maybe_warn, format_directive, parse_directive): Use new functions.
17370 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17372 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17374 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17376 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17378 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17379 target_header_dir): Set correctly.
17380 * configure: Regenerated.
17381 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17382 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17383 instead of SYSTEM_HEADER_DIR.
17385 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17387 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17388 (estimate_local_effects): Likewise.
17389 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17390 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17391 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17392 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17393 do_estimate_edge_time, estimate_edge_time): Likewise.
17394 * ipa-inline-analysis.c (estimate_node_size_and_time,
17395 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17396 (estimate_time_after_inlining): Remove.
17398 2017-04-28 Martin Liska <mliska@suse.cz>
17400 * doc/gcov.texi: Enhance documentation of gcov.
17402 2017-04-28 Martin Liska <mliska@suse.cz>
17404 * doc/gcov.texi: Sort options in alphabetic order.
17405 * doc/gcov-dump.texi: Likewise.
17406 * doc/gcov-tool.texi: Likewise.
17407 * gcov.c (print_usage): Likewise.
17408 * gcov-dump.c (print_usage): Likewise.
17409 * gcov-tool.c (print_merge_usage_message): Likewise.
17410 (print_rewrite_usage_message): Likewise.
17411 (print_overlap_usage_message): Likewise.
17413 2017-04-28 Martin Liska <mliska@suse.cz>
17415 PR gcov-profile/53915
17416 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17418 2017-04-28 Martin Liska <mliska@suse.cz>
17420 PR gcov-profile/79891
17421 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17422 is marked by compiler as living on a line.
17423 (get_cycles_count): Remove usage of the union.
17424 (output_intermediate_file): Likewise.
17425 (find_source): Fix GNU coding style.
17426 (accumulate_line_counts): Remove old non-all block mode.
17427 (output_lines): Remove usage of the union.
17428 * profile.c (output_location): Include all BBs, even if
17429 belonging to a same line (and file) as a previous BB.
17431 2017-04-28 Martin Liska <mliska@suse.cz>
17433 * gcov.c (process_args): Handle new argument 'w'.
17434 (read_graph_file): Assign ID to BBs.
17435 (output_branch_count): Display BB # if verbose flag is set.
17436 (output_lines): Likewise for arcs.
17437 (print_usage): Add '--verbose' option help.
17438 * doc/gcov.texi: Document --verbose (-w) option.
17440 2017-04-28 Martin Liska <mliska@suse.cz>
17442 * gcov.c (struct block_location_info): New struct.
17443 (process_file): Fill up the new structure.
17444 (read_graph_file): Replace usage of encoding by the newly added
17446 (add_line_counts): Likewise.
17447 (accumulate_line_counts): Remove usage of the union.
17448 (function_info::function_info): New function.
17449 (function_info::~function_info): Likewise.
17450 (process_file): Call delete instead of release_function.
17451 (release_function): Release the function.
17452 (release_structures): Call delete instead of release_function.
17453 (solve_flow_graph): Replace usage of num_blocks.
17454 (find_exception_blocks): Likewise.
17455 (output_lines): Fix GNU coding style.
17457 2017-04-28 Martin Liska <mliska@suse.cz>
17460 * coverage.c (coverage_remove_note_file): New function.
17461 * coverage.h: Declare the function.
17462 * toplev.c (finalize): Clean if an error has been seen.
17464 2017-04-28 Martin Liska <mliska@suse.cz>
17466 PR gcov-profile/80031
17467 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17468 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17469 * gcov.c (read_graph_file): Read just number of blocks.
17470 * profile.c (branch_prob): Do not stream 0 flags per a basic
17473 2017-04-28 Martin Liska <mliska@suse.cz>
17475 * gcov-dump.c (tag_*): Add new argument to declarations.
17476 (dump_gcov_file): Likewise.
17477 (tag_blocks): Add and use new argument depth.
17478 (tag_arcs): Likewise.
17479 (tag_lines): Likewise.
17480 (tag_counters): Likewise.
17481 (tag_summary): Likewise.
17482 (dump_working_sets): Use depth to do a proper indentation.
17484 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17487 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17488 * symtab.c (symtab_node::debug_symtab): Move definition here.
17490 2017-04-28 Richard Biener <rguenther@suse.de>
17492 * lto-streamer.h (LTO_major_version): Bump to 7.
17494 2017-04-28 Richard Biener <rguenther@suse.de>
17496 * tree-vrp.c (assert_info): New struct.
17497 (add_assert_info): New helper.
17498 (register_edge_assert_for_2): Refactor to add asserts to a vector
17500 (register_edge_assert_for_1): Likewise.
17501 (register_edge_assert_for): Likewise.
17502 (finish_register_edge_assert_for): New helper actually registering
17503 asserts where live on edge.
17504 (find_conditional_asserts): Adjust.
17505 (find_switch_asserts): Likewise.
17506 (evrp_dom_walker::try_find_new_range): Generalize.
17507 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17509 2017-04-27 Marek Polacek <polacek@redhat.com>
17512 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17513 arg10 and arg11 to itype.
17515 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17517 * doc/extend.texi (Object Size Checking): Improve grammar.
17519 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17522 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17523 that the logic for permitting reciprocal estimates matches that
17526 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17529 * tree.c (type_cache_hasher::equal): Only compare
17530 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17531 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17532 non-aggregate element types.
17533 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17534 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17536 2017-04-27 Richard Biener <rguenther@suse.de>
17538 PR middle-end/80533
17539 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17540 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17541 keeping a reference to a trailing array.
17543 2017-04-27 Richard Biener <rguenther@suse.de>
17545 PR middle-end/80539
17546 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17547 being in loop-closed SSA form conservatively.
17548 (chrec_fold_multiply_poly_poly): Likewise.
17550 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17552 PR middle-end/79665
17553 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17554 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17556 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17559 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17560 (aarch64_function_arg_alignment): Return unsigned int again, but still
17561 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17562 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17563 Don't emit -Wpsabi note.
17564 (aarch64_function_arg_boundary): Likewise.
17565 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17568 2017-04-26 Nathan Sidwell <nathan@acm.org>
17570 * tree.h (crc32_unsigned_n): Declare.
17571 (crc32_unsigned, crc32_unsigned): Make inline.
17572 * tree.c (crc32_unsigned_bits): Replace with ...
17573 (crc32_unsigned_n): ... this.
17574 (crc32_unsigned, crc32_byte): Remove.
17575 (crc32_string): Remove unnecessary braces.
17577 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17579 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17580 * ipa-inline-analysis.c (MAX_TIME): Remove.
17581 (account_size_time): Use sreal for time.
17582 (dump_inline_summary): Update.
17583 (estimate_function_body_sizes): Update.
17584 (estimate_edge_size_and_time): Update.
17585 (estimate_calls_size_and_time): Update.
17586 (estimate_node_size_and_time): Update.
17587 (inline_merge_summary): Update.
17588 (inline_update_overall_summary): Update.
17589 (estimate_time_after_inlining): Update.
17590 (inline_read_section): Update.
17591 (inline_write_summary): Update.
17592 * ipa-inline.c (compute_uninlined_call_time): Update.
17593 (compute_inlined_call_time): Update.
17594 (recursive_inlining): Update.
17595 (inline_small_functions): Update.
17596 (dump_overall_stats): Update.
17597 * ipa-inline.h: Include sreal.h.
17598 (size_time_entry): Turn time to sreal.
17599 (inline_summary): Turn self_time nad time to sreal.
17601 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17603 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17605 (sreal::stream_out, sreal::stream_in): New.
17606 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17608 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17610 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17613 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17616 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17618 (dimode_scalar_chain::compute_convert_gain): Ditto.
17619 (dimode_scalar_chain::make_vector_copies): Ditto.
17620 (dimode_scalar_chain::convert_reg): Ditto.
17621 (dimode_scalar_chain::convert_insn): Ditto.
17622 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17623 (VI248_AVX512BW_1): New mode iterator.
17624 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17625 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17628 2017-04-25 Martin Sebor <msebor@redhat.com>
17630 PR tree-optimization/80497
17631 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17632 constants are representable in HOST_WIDE_INT.
17633 (parse_directive): Ditto.
17635 2017-04-25 Martin Sebor <msebor@redhat.com>
17638 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17639 (new_zero_array): Adjust signature.
17640 (dom_info::dom_init): Used unsigned rather that size_t.
17641 (dom_info::dom_info): Same.
17643 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17644 Jakub Jelinek <jakub@redhat.com>
17647 * config/arm/arm.c: Include gimple.h.
17648 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17649 returns negative, increment ncrn only if it returned positive.
17650 (arm_needs_doubleword_align): Return int instead of bool,
17651 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17652 members, but if there is any such non-FIELD_DECL
17653 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17654 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17655 returns negative, increment nregs only if it returned positive.
17656 (arm_setup_incoming_varargs): Likewise.
17657 (arm_function_arg_boundary): Emit -Wpsabi note if
17658 arm_needs_doubleword_align returns negative, return
17659 DOUBLEWORD_ALIGNMENT only if it returned positive.
17661 2017-04-25 Marek Polacek <polacek@redhat.com>
17664 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17665 first argument to type.
17667 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17670 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17671 type checks to test for compatibility instead of equality.
17673 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17674 Jakub Jelinek <jakub@redhat.com>
17677 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17679 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17680 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17681 the alignment computation, but return their maximum in warn_alignment.
17682 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17683 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17685 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17686 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17689 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17691 * config/arc/simdext.md (dmpyh): Fix typo.
17693 2017-04-25 Richard Biener <rguenther@suse.de>
17695 PR tree-optimization/80492
17696 * alias.c (compare_base_decls): Handle registers with asm
17697 specification conservatively.
17698 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17699 compare_base_decls returning dont-know properly.
17701 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17703 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17704 (legitimate_offset_address_p): New function.
17705 (arc_legitimate_address_p): Use above function.
17707 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17709 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17711 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17713 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17714 ACCH registers whenever they are available.
17716 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17718 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17719 double regs fix when not used.
17721 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17723 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17725 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17726 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17730 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17731 when emitting short ADD instructions.
17733 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17735 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17737 (cmpsi_cc_c_insn): Likewise.
17738 (cbranchsi4_scratch): Compute proper instruction length using
17739 compact_hreg_operand.
17740 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17742 2017-04-25 Richard Biener <rguenther@suse.de>
17744 PR middle-end/80509
17745 * passes.c (pass_manager::pass_manager): Initialize
17746 m_name_to_pass_map.
17748 2017-04-25 Richard Biener <rguenther@suse.de>
17750 PR tree-optimization/79201
17751 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17753 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17756 * config/s390/vector.md: Split MEM->GPR vector moves for
17757 non-s_operand addresses.
17759 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17762 * config/s390/predicates.md (reload_const_wide_int_operand): New
17764 * config/s390/s390.md ("movti"): Remove d/P alternative.
17765 ("movti_bigconst"): New pattern definition.
17767 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17770 * s390-protos.h (s390_expand_cs_hqi): Removed.
17771 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17772 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17773 modes as well as CCZ1mode and CCZmode.
17774 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17775 signature of s390_emit_compare_and_swap.
17776 (s390_expand_cs_hqi): Likewise, make static.
17777 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17778 compare-and-swap, in some cases.
17779 (s390_expand_cs): Wrapper function.
17780 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17782 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17783 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17784 patterns for small and large integers. Forbid symref memory operands.
17785 Move expander to s390.c. Require cc register.
17786 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17787 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17788 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17789 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17790 symref memory operands. Remove CC mode and call s390_match_ccmode
17792 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17794 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17796 * config/s390/s390.md (define_peephole2): New peephole to help
17797 combining the load-and-test pattern with volatile memory.
17799 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17801 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17802 with CCZmode for TARGET_Z196.
17804 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17806 PR rtl-optimization/80501
17807 * combine.c (make_compound_operation_int): Set subreg_code to SET
17808 even for AND with mask of the sign bit of mode.
17810 PR rtl-optimization/80500
17811 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17812 sum's initial value.
17814 2017-04-25 Julian Brown <julian@codesourcery.com>
17815 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17817 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17819 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17821 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17823 2017-04-25 Julian Brown <julian@codesourcery.com>
17824 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17826 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17827 (thunderx2t99_sha): New Reservation.
17829 2017-04-25 Julian Brown <julian@codesourcery.com>
17830 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17832 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17833 type for 1-element load.
17835 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17837 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17839 2017-04-24 Martin Jambor <mjambor@suse.cz>
17841 PR tree-optimization/80293
17842 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17843 char arrays not totally scalarizable if it is false.
17844 (analyze_all_variable_accesses): Pass correct value in the new
17845 parameter. Add a statistics counter.
17847 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17849 PR middle-end/79931
17850 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17852 2017-04-24 Richard Biener <rguenther@suse.de>
17854 PR tree-optimization/80494
17855 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17856 out for complex types.
17858 2017-04-24 Richard Biener <rguenther@suse.de>
17860 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17861 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17862 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17863 (DFS): Adjust and never fail.
17864 (sccvn_dom_walker::fail): Remove.
17865 (sccvn_dom_walker::before_dom_children): Adjust.
17866 (run_scc_vn): Likewise and never fail.
17867 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17868 (pass_fre::execute): Likewise.
17870 2017-04-24 Richard Biener <rguenther@suse.de>
17872 PR tree-optimization/79725
17873 * tree-ssa-sink.c (statement_sink_location): Return whether
17874 failure reason was zero uses. Move that check later.
17875 (sink_code_in_bb): Deal with zero uses by removing the stmt
17878 2017-04-24 Richard Biener <rguenther@suse.de>
17881 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17882 pointer-based references.
17884 2017-04-24 Richard Biener <rguenther@suse.de>
17887 * pass_manager.h (pass_manager::operator new): Remove.
17888 (pass_manager::operator delete): Likewise.
17889 * passes.c (pass_manager::operator new): Remove.
17890 (pass_manager::operator delete): Likewise.
17891 (pass_manager::pass_manager): Zero individual pass members.
17893 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17896 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17897 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17898 Check "XEXP (src, 1)" operand here.
17899 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17900 Check "XEXP (src, 1)" operand here.
17901 (dimode_scalar_chain::make_vector_copies): Detect count register
17902 of a shift instruction. Zero extend count register from QImode
17903 to DImode to satisfy vector shift pattern count operand predicate.
17904 Substitute vector shift count operand with a DImode copy.
17905 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17908 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17910 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17911 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17912 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17913 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17914 (UNSPEC_NOREX_MEM): Remove definition.
17916 2017-04-21 Richard Biener <rguenther@suse.de>
17918 PR tree-optimization/79547
17919 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17920 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17921 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17922 without any constraints.
17924 2017-04-21 Richard Biener <rguenther@suse.de>
17926 PR tree-optimization/78847
17927 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17929 2017-04-21 Richard Biener <rguenther@suse.de>
17931 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17932 (build_distinct_type_copy): Likewise.
17933 (build_variant_type_copy): Likewise.
17934 * tree.c (build_qualified_type): Pass down mem-stat info.
17935 (build_distinct_type_copy): Likewise.
17936 (build_variant_type_copy): Likewise.
17938 2017-04-21 Richard Biener <rguenther@suse.de>
17940 PR tree-optimization/80237
17941 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17943 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17944 for a simplified result.
17946 2016-04-21 Richard Biener <rguenther@suse.de>
17948 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17949 sth as strict as a simple_iv but a chrec without symbols and an
17950 operand defined in the loop we are peeling (and not some subloop).
17951 (propagate_constants_for_unrolling): Propagate all constants.
17953 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17956 * config/i386/i386.c (print_reg): Remove assert for disalowed
17957 regno values, call output_operand_lossage instead.
17959 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17962 * config/i386/constraints.md (Yc): New register constraint.
17963 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17964 Use Yc constraint for alternative 2 of operand 0. Remove
17965 preferred_for_speed attribute.
17967 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17969 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17970 lastprivate clauses in SIMT case.
17972 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17974 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17976 2017-04-20 Richard Biener <rguenther@suse.de>
17978 PR tree-optimization/57796
17979 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17980 as N scalar stores.
17981 (vect_model_load_cost): Cost gathers as N scalar loads.
17983 2017-04-20 Richard Biener <rguenther@suse.de>
17985 * ggc-page.c (ggc_allocated_p): Rename to ...
17986 (safe_lookup_page_table_entry): ... this and return the lookup
17988 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17990 2017-04-20 Richard Biener <rguenther@suse.de>
17992 PR tree-optimization/80453
17993 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17994 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17995 from the conditions.
17996 (vn_phi_eq): Pass them down.
17997 (vn_phi_lookup): Record them.
17998 (vn_phi_insert): Likewise.
18000 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18002 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18003 uninitialized variable warning to avoid buffer overrun.
18005 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18008 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18009 is suppressed for '{ 0 }' in C.
18011 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18013 * BASE-VER: Set to 8.0.0.
18015 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18017 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18018 priority .init_array and .fini_array section with SECTION_NOTYPE
18021 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18023 PR middle-end/80423
18024 * tree.h (build_array_type): Add typeless_storage default argument.
18025 * tree.c (type_cache_hasher::equal): Also compare
18026 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18027 (build_array_type): Add typeless_storage argument, set
18028 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18030 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18031 (build_array_type): Likewise. Add typeless_storage argument.
18033 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18034 Jakub Jelinek <jakub@redhat.com>
18036 PR tree-optimization/80426
18037 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18038 operation on symbolic operands, also compute the overflow for the
18039 invariant part when the operation degenerates into a negation.
18041 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18044 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18045 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18048 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18050 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18053 * config/avr/avr.c (tree.h): Include it.
18054 (cgraph.h): Include it.
18055 (avr_encode_section_info): Don't warn for uninitialized progmem
18056 variable if it's just an alias.
18058 2017-04-19 Richard Biener <rguenther@suse.de>
18061 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18062 when needed by AutoPGO.
18064 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18067 * doc/lto.texi: Remove an extra 'that'.
18069 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18071 PR rtl-optimization/80429
18072 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18073 are only used in debug insns.
18075 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18076 Vladimir Makarov <vmakarov@redhat.com>
18078 * config/sparc/predicates.md (input_operand): Add comment. Return
18079 true for any memory operand when LRA is in progress.
18080 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18082 2017-04-18 Jeff Law <law@redhat.com>
18085 * mips.md ({return,simple_return}_internal): Do not overwrite
18088 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18090 PR tree-optimization/80443
18091 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18092 instead of adding 1, subtract -1 and similarly instead of subtracting
18095 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18097 PR rtl-optimization/80357
18098 * haifa-sched.c (tmp_bitmap): New variable.
18099 (model_recompute): Handle duplicate use records.
18100 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18101 (free_global_sched_pressure_data): Free it.
18103 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18106 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18107 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18108 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18109 instead of SYSTEM_HEADER_DIR.
18111 2017-04-18 Jeff Law <law@redhat.com>
18113 PR middle-end/80422
18114 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18115 predecessors after walking up the insn chain.
18117 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18120 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18121 sizetype type into debug info.
18123 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18126 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18127 unneeded test for TARGET_UPPER_REGS_SF.
18128 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18130 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18133 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18134 instead of gsi_after_labels.
18136 2017-04-18 Jeff Law <law@redhat.com>
18138 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18142 2017-04-13 Jeff Law <law@redhat.com>
18143 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18144 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18146 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18149 * config/avr/avr.c (intl.h): Include it.
18150 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18152 2017-04-18 Martin Liska <mliska@suse.cz>
18154 PR gcov-profile/78783
18155 * gcov-tool.c (gcov_output_files): Validate that destination
18156 file is either removed by the tool or by a user.
18158 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18159 Guy Benyei <guybe@mellanox.com>
18161 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18162 block, and do not negate it, the stored id is already negative.
18164 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18166 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18168 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18171 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18172 masks of options that should be turned off if the VSX vector
18173 options are turned off.
18174 (OTHER_P8_VECTOR_MASKS): Likewise.
18175 (OTHER_VSX_VECTOR_MASKS): Likewise.
18176 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18177 rs6000_disable_incompatible_switches to validate no type switches
18179 (rs6000_incompatible_switch): New function to disallow turning on
18180 other vector options if -mno-vsx, -mno-power8-vector, or
18181 -mno-power9-vector are specified.
18183 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18185 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18187 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18189 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18190 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18191 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18192 (ARG_POINTER_CFA_OFFSET): Likewise.
18194 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18196 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18197 conditions to take advantage of various optimizations.
18199 2017-04-13 Jeff Law <law@redhat.com>
18201 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18202 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18203 (zero_extendsidi2_dext): Likewise.
18205 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18208 * fold-const.c (fold_ternary_loc): Revert
18209 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18212 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18214 PR rtl-optimization/80343
18215 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18216 new scratch pseudo.
18218 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18221 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18222 to ubsan_encode_value.
18224 2017-04-13 Jeff Law <law@redhat.com>
18226 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18227 appearing in DEBUG_INSNs.
18229 2017-04-13 Martin Liska <mliska@suse.cz>
18231 PR gcov-profile/80413
18232 * gcov-io.c (gcov_write_string): Copy to buffer just when
18233 allocated size is greater than zero.
18235 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18238 * dwarf2out.c (decls_for_scope): Ignore declarations of
18239 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18241 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18244 * ipa-visibility.c (non_local_p): Fix typos.
18245 (localize_node): When localizing symbol in same comdat group,
18246 dissolve the group only when we know external symbols are going
18248 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18250 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18252 PR tree-optimization/79390
18253 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18254 order does not result in usable sequence, retry with reversed operand
18260 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18261 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18262 op0 instead of fold_convert_loc (loc, type, arg0).
18264 2017-04-12 Jeff Law <law@redhat.com>
18266 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18267 has a delay slot in the generated code.
18269 * config/cris/cris.md (cris_preferred_reload_class): Return
18270 GENNONACR_REGS rather than GENERAL_REGS.
18272 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18275 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18276 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18277 signedness of the result type.
18279 2017-04-12 Richard Biener <rguenther@suse.de>
18280 Jeff Law <law@redhat.com>
18282 PR tree-optimization/80359
18283 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18284 trim stores to TARGET_MEM_REFs.
18286 2017-04-12 Richard Biener <rguenther@suse.de>
18288 PR tree-optimization/79390
18289 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18290 threading case even more.
18292 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18295 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18296 for quad_address_p for TImode, instead of just not indexed_address.
18298 2017-04-12 Richard Biener <rguenther@suse.de>
18299 Bernd Edlinger <bernd.edlinger@hotmail.de>
18301 PR middle-end/79671
18302 * alias.c (component_uses_parent_alias_set_from): Handle
18303 TYPE_TYPELESS_STORAGE.
18304 (get_alias_set): Likewise.
18305 * tree-core.h (tree_type_common): Add typeless_storage flag.
18306 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18307 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18308 for types containing members with TYPE_TYPELESS_STORAGE.
18309 (place_field): Likewise.
18310 (layout_type): Likewise for ARRAY_TYPE.
18311 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18312 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18313 TYPE_TYPELESS_STORAGE.
18314 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18316 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18319 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18320 first argument to type.
18322 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18326 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18327 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18328 (rs6000_expand_binop_builtin): Likewise.
18329 (rs6000_expand_ternop_builtin): Likewise; also add missing
18330 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18331 vshasigma built-ins.
18332 * doc/extend.texi: Document that vec_xxpermdi's third argument
18333 must be a constant.
18335 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18337 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18338 Use shift_const cost parameter when calculating gain of STV shifts.
18340 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18342 PR rtl-optimization/70478
18343 * lra-constraints.c (process_alt_operands): Check memory for
18344 disfavoring memory insn operand.
18346 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18348 PR middle-end/80100
18349 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18350 left shift in unsigned HOST_WIDE_INT type.
18352 PR rtl-optimization/80385
18353 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18354 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18357 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18358 if they have any depend clauses.
18360 2017-04-11 Martin Liska <mliska@suse.cz>
18363 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18364 * ipa-split.c (split_function): Create a local comdat symbol
18365 if caller is in a comdat group.
18367 2017-04-11 Martin Liska <mliska@suse.cz>
18370 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18373 2017-04-11 Martin Sebor <msebor@redhat.com>
18375 PR middle-end/80364
18376 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18377 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18379 (directive::set_width, directive::set_precision, format_character):
18381 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18384 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18387 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18388 conflict, set target->arch_name instead of target->cpu_name.
18390 2017-04-11 Richard Biener <rguenther@suse.de>
18392 PR tree-optimization/80374
18393 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18394 build_zero_cst, remove fold_convertible_p check again.
18396 2017-04-11 Martin Liska <mliska@suse.cz>
18399 * ubsan.c (instrument_object_size): Do not instrument register
18402 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18405 * config/i386/i386-builtin-types.def
18406 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18407 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18408 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18409 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18410 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18411 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18412 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18413 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18414 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18415 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18416 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18417 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18418 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18419 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18420 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18421 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18422 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18423 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18424 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18425 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18426 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18427 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18428 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18429 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18430 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18431 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18432 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18433 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18434 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18435 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18436 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18437 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18438 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18439 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18440 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18441 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18442 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18443 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18444 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18445 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18446 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18447 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18448 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18449 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18450 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18451 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18453 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18454 flag to second_arg_count, handle 4 argument function type _COUNT
18455 aliases, handle second_arg_count on second argument rather than last.
18457 2017-04-10 Jeff Law <law@redhat.com>
18459 PR tree-optimization/80374
18460 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18461 record anything if we can not convert integer_zero_node to the
18464 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18467 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18468 Enhance special handling given to the TARGET_P9_MINMAX option in
18469 relation to certain other options.
18471 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18473 PR tree-optimization/80153
18474 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18475 remove POINTER_PLUS_EXPR's base part directly, rather than through
18478 2017-04-10 Richard Biener <rguenther@suse.de>
18479 Bin Cheng <bin.cheng@arm.com>
18481 PR tree-optimization/80153
18482 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18483 the first element of pointer type aff_tree. Build result expr in
18485 (add_elt_to_tree): Convert to type unconditionally. Remove other
18486 fold_convert calls.
18487 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18488 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18490 2017-04-10 Richard Biener <rguenther@suse.de>
18492 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18495 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18497 PR rtl-optimization/70478
18498 * lra-constraints.c (curr_small_class_check): New.
18499 (update_and_check_small_class_inputs): New.
18500 (process_alt_operands): Update curr_small_class_check. Disfavor
18501 alternative insn memory operands. Check available regs for small
18504 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18507 * config/mips/mips.opt (-mvirt): Update description.
18508 * doc/invoke.texi (-mvirt): Likewise.
18510 2017-04-10 Richard Biener <rguenther@suse.de>
18512 PR middle-end/80362
18513 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18514 looking for NEGATE_EXPR in -A / -B to A / B folding.
18516 2017-04-10 Martin Liska <mliska@suse.cz>
18518 PR gcov-profile/80224
18519 * gcov.c (print_usage): Fix usage string.
18520 (get_gcov_intermediate_filename): Remove.
18521 (output_gcov_file): Use both for normal and intermediate format.
18522 (generate_results): Do not initialize special file for
18523 intermediate format.
18525 2017-04-10 Richard Biener <rguenther@suse.de>
18527 PR tree-optimization/80304
18528 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18531 2017-04-10 Nathan Sidwell <nathan@acm.org>
18534 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18535 (rs6000_init_builtins): Use it.
18537 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18539 * config/arm/arm.md (<mrc>): Add mode to SET source.
18540 (<mrrc>): Likewise.
18542 2017-04-10 Richard Biener <rguenther@suse.de>
18544 PR middle-end/80344
18545 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18547 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18550 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18551 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18552 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18553 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18554 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18555 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18556 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18557 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18558 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18559 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18560 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18561 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18562 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18563 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18564 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18565 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18566 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18567 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18568 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18569 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18570 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18571 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18572 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18574 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18576 PR rtl-optimization/70478
18577 * lra-constraints.c: Reverse the last patch.
18579 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18581 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18582 Add comment for WCHAR_T.
18584 2017-04-08 Martin Liska <mliska@suse.cz>
18587 2017-04-07 Martin Liska <mliska@suse.cz>
18590 * ipa-split.c (split_function): Add function part to a same comdat
18593 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18596 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18598 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18600 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18601 (vec_pairing): ...to this.
18602 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18603 (rs6000_sched_init): Adjust for name change.
18604 (struct rs6000_sched_context): Likewise.
18605 (rs6000_init_sched_context): Likewise.
18606 (rs6000_set_sched_context): Likewise.
18608 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18614 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18616 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18617 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18618 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18620 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18622 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18624 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18626 PR rtl-optimization/70703
18627 * ira-color.c (update_conflict_hard_regno_costs): Use
18628 int64_t instead of HOST_WIDE_INT.
18630 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18632 PR rtl-optimization/70478
18633 * lra-constraints.c (process_alt_operands): Disfavor alternative
18634 insn memory operands.
18636 2017-04-07 Jeff Law <law@redhat.com>
18638 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18639 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18641 2017-04-07 Martin Liska <mliska@suse.cz>
18644 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18645 Show error message instead of an ICE.
18647 2017-04-07 Martin Liska <mliska@suse.cz>
18650 * ipa-split.c (split_function): Add function part to a same comdat
18653 2017-04-07 Richard Biener <rguenther@suse.de>
18655 PR middle-end/80341
18656 * tree.c (get_unwidened): Also handle ! for_type case for
18658 * convert.c (do_narrow): Split out from ...
18659 (convert_to_integer_1): ... here. Do not pass final truncation
18660 type to get_unwidened for TRUNC_DIV_EXPR.
18662 2017-04-07 Richard Biener <rguenther@suse.de>
18664 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18666 (aff_combination_const): Adjust.
18667 (aff_combination_scale): Likewise.
18668 (aff_combination_add_elt): Likewise.
18669 (aff_combination_add_cst): Likewise.
18670 (aff_combination_convert): Likewise.
18671 (add_elt_to_tree): Likewise. Remove unused argument.
18672 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18674 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18676 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18678 * config/arm/arm.c (arm_default_short_enums): Use
18679 ARM_DEFAULT_SHORT_ENUMS.
18680 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18682 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18685 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18686 members with redundant out-of-class redeclaration.
18688 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18691 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18692 * config/i386/i386.md (*zero_extendsidi2):
18693 Add (?*x,*x) and (?*v,*v) alternatives.
18695 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18698 * config/i386/i386.c (ix86_expand_builtin)
18699 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18700 mode from insn data. Convert operands to insn operand mode.
18701 Copy operands that don't satisfy insn predicate to a register.
18703 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18705 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18708 2017-04-06 Richard Biener <rguenther@suse.de>
18710 PR tree-optimization/80334
18711 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18712 preserve alignment of accesses.
18714 2017-04-06 Richard Biener <rguenther@suse.de>
18716 PR tree-optimization/80262
18717 * tree-sra.c (build_ref_for_offset): Preserve address-space
18719 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18720 Drop useless address-space information on MEM_REF offsets.
18722 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18724 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18726 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18728 PR rtl-optimization/70703
18729 * ira-color.c (update_conflict_hard_regno_costs): Use
18730 HOST_WIDE_INT instead of long.
18732 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18735 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18736 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18738 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18739 for x86_64 target. Handle -m3dnowa option.
18741 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18743 PR rtl-optimization/70703
18744 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18745 (update_conflict_hard_regno_costs): Use long instead of unsigned
18746 arithmetic for cost calculation.
18748 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18749 Bernd Edlinger <bernd.edlinger@hotmail.de>
18752 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18755 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18758 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18759 ptr_mode with Pmode throughout.
18760 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18761 into probe_stack_range and use DImode.
18763 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18766 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18767 call_eh_return is true.
18769 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18771 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18772 Initialize last_match_fntype_index.
18774 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18777 * tree-nvr.c: Include internal-fn.h.
18778 (pass_return_slot::execute): Ignore internal calls without
18781 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18782 Richard Biener <rguenther@suse.de>
18785 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18786 captures used multiple times, except for the last use.
18787 * generic-match-head.c: Include gimplify.h.
18789 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18791 PR tree-optimization/79390
18792 * target.h (struct noce_if_info): Declare.
18793 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18794 * target.def (noce_conversion_profitable_p): New target hook.
18795 * ifcvt.h (struct noce_if_info): New type, moved from ...
18796 * ifcvt.c (struct noce_if_info): ... here.
18797 (noce_conversion_profitable_p): Renamed to ...
18798 (default_noce_conversion_profitable_p): ... this. No longer
18800 (noce_try_store_flag_constants, noce_try_addcc,
18801 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18802 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18803 instead of noce_conversion_profitable_p.
18804 * config/i386/i386.c: Include ifcvt.h.
18805 (ix86_option_override_internal): Don't override
18806 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18807 (ix86_noce_conversion_profitable_p): New function.
18808 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18809 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18810 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18811 * doc/tm.texi: Regenerated.
18813 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18815 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18818 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18821 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18822 instructions for small multiply cores.
18824 2017-04-04 Jeff Law <law@redhat.com>
18826 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18828 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18829 that are not set by the loop over the elements.
18831 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18834 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18835 int mode, convert_modes it to mode as unsigned, otherwise use
18836 lowpart_subreg to mode rather than SImode.
18837 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18838 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18839 Use DImode instead of SImode for the shift count operand.
18840 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18843 2017-04-04 Richard Biener <rguenther@suse.de>
18845 PR middle-end/80281
18846 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18847 arithmetic done for the negate or the plus. Simplify.
18848 (A - (-B) -> A + B): Likewise.
18849 * fold-const.c (split_tree): Make sure to not negate pointers.
18851 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18853 PR rtl-optimization/60818
18854 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18855 a compare of comparisons with the thing compared if this results
18856 in a different machine mode.
18858 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18860 * alias.c (base_alias_check): Fix typo in comment.
18861 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18862 * cgraphunit.c (symbol_table::compile): Likewise.
18863 * collect2.c (maybe_run_lto_and_relink): Likewise.
18864 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18865 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18866 * config/avr/avr.c (avr_map_op_t): Likewise.
18867 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18868 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18869 * config/epiphany/epiphany.md (movcc): Likewise.
18870 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18871 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18873 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18874 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18875 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18876 * config/sh/sh.c (sh_rtx_costs): Likewise.
18877 * fold-const.c (fold_truth_andor): Likewise.
18878 * genautomata.c (collapse_flag): Likewise.
18879 * gengtype.h (struct type::u::s): Likewise.
18880 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18881 * input.c (FORMAT_AMOUNT): Likewise.
18882 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18883 (known_aggs_to_agg_replacement_list): Likewise.
18884 * ipa-inline-analysis.c: Likewise.
18885 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18886 * ipa-polymorphic-call.c
18887 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18888 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18889 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18891 * modulo-sched.c (apply_reg_moves): Likewise.
18892 * omp-expand.c (build_omp_regions_1): Likewise.
18893 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18894 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18895 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18896 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18897 * value-prof.c: Likewise.
18898 * var-tracking.c (val_reset): Likewise.
18900 2017-04-03 Richard Biener <rguenther@suse.de>
18902 PR tree-optimization/80275
18903 * fold-const.c (split_address_to_core_and_offset): Handle
18906 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18908 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18909 descriptors is at least equal to that of functions.
18911 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18913 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18915 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18918 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18919 (mov<IMOD4:mode>): New expander.
18920 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18922 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18924 PR rtl-optimization/79405
18925 * fwprop.c (propagations_left): New variable.
18926 (forward_propagate_into): Decrement it.
18927 (fwprop_init): Initialize it.
18928 (fw_prop): If the variable has reached zero, stop propagating.
18929 (fwprop_addr): Ditto.
18931 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18934 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18935 a FUNCTION_DECL, pass it as decl instead of origin to
18938 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18940 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18943 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18946 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18947 TARGET_VSX_SMALL_INTEGER.
18949 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18951 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18952 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18954 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18956 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18957 extraction from odd-numbered MSA register.
18959 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18961 PR middle-end/80173
18962 * expmed.c (store_bit_field_1): Don't attempt to create
18963 a word subreg out of hard registers wider than word if they
18964 have HARD_REGNO_NREGS of 1 for their mode.
18966 PR middle-end/80163
18967 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18968 conversions to integer types wider than word and pointer.
18971 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18972 (rtx_equal_for_cselib_p): Pass 0 to it.
18973 * cselib.c (cselib_hasher::equal): Likewise.
18974 (rtx_equal_for_cselib_1): Add depth argument. If depth
18975 is 128, don't look up VALUE locs and punt. Increment
18976 depth in recursive calls when walking VALUE locs.
18978 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18980 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18981 (make_gcov_file_name): Use the canonical path name for generating
18983 (read_line): Fix handling of files with ascii null bytes.
18985 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18987 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18988 to initialise a vector register instead
18989 of using a const_int.
18991 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18993 PR translation/80189
18994 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18995 diagnostic messages.
18997 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19000 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19001 (dfp_diex_<mode>): Update mode of operand 1.
19002 * doc/extend.texi (dxex, dxexq): Document change to return type.
19003 (diex, diexq): Document change to argument type.
19005 2017-03-30 Martin Jambor <mjambor@suse.cz>
19008 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19009 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19010 it reflects the signature changes performed at the callee side.
19011 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19012 to cgraph_build_function_type_skip_args.
19013 (build_function_decl_skip_args): Adjust call to the above function.
19015 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19018 * config/i386/sse.md
19019 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19020 register as dest whenever it is a MEM not rtx_equal_p to the
19021 corresponding dup operand, and when forcing into reg move the
19022 reg into the memory afterwards.
19023 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19024 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19025 for the force_reg mode.
19026 (avx512vl_vextractf128<mode>): Use register as dest either
19027 always when a MEM, or when it is a MEM not rtx_equal_p to the
19028 corresponding dup operand, or even not when it is a CONST_VECTOR
19029 depending on the mode and lo vs. hi.
19030 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19032 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19033 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19034 Likewise. Require that operands[2] is even.
19035 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19036 Remove extraneous parens. Require that operands[2] is a multiple
19038 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19039 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19040 disallow memory then.
19042 2017-03-30 Richard Biener <rguenther@suse.de>
19044 PR tree-optimization/77498
19045 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19046 to non-constants over backedges.
19048 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19050 PR rtl-optimization/80233
19051 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19052 as last_combined_insn. Do not test for BARRIER_P separately.
19054 2017-03-29 Andreas Schwab <schwab@suse.de>
19057 * calls.c (prepare_call_address): Convert funexp to Pmode before
19058 copying to temp reg.
19060 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19062 PR tree-optimization/80158
19063 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19064 Handle possible future case of more than one alternate
19066 (replace_rhs_if_not_dup): Likewise.
19067 (replace_one_candidate): Likewise.
19069 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19071 PR rtl-optimization/80193
19072 * ira.c (ira): Do not check allocation for LRA.
19074 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19076 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19077 (nvptx_output_simt_exit): Declare.
19078 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19079 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19080 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19081 (nvptx_declare_function_name): ...here. Emit declaration of local
19082 memory space buffer for omp_simt_enter insn.
19083 (nvptx_output_unisimt_switch): New.
19084 (nvptx_output_softstack_switch): New.
19085 (nvptx_output_simt_enter): New.
19086 (nvptx_output_simt_exit): New.
19087 * config/nvptx/nvptx.h (struct machine_function): New fields
19088 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19089 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19090 (UNSPECV_SIMT_EXIT): Ditto.
19091 (omp_simt_enter_insn): New insn.
19092 (omp_simt_enter): New expansion.
19093 (omp_simt_exit): New insn.
19094 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19096 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19097 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19098 (expand_GOMP_SIMT_EXIT): New.
19099 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19100 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19101 (GOMP_SIMT_EXIT): Ditto.
19102 * target-insns.def (omp_simt_enter): New insn.
19103 (omp_simt_exit): Ditto.
19104 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19106 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19107 (lower_rec_input_clauses): Likewise.
19108 (lower_lastprivate_clauses): Handle SIMT privatization.
19110 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19111 (ompdevlow_adjust_simt_enter): New.
19112 (find_simtpriv_var_op): New.
19113 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19114 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19116 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19117 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19118 (copy_decl_for_dup_finish): Ditto.
19120 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19122 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19125 * config/i386/i386.c (ix86_option_override_internal): Always
19126 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19128 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19130 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19132 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19134 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19135 mark new edge's irreducible flag accordign to it.
19136 (vect_do_peeling): Check loop preheader edge's irreducible flag
19137 and pass it to function slpeel_add_loop_guard.
19139 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19141 PR tree-optimization/80218
19142 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19143 Update block frequencies and counts.
19145 2017-03-28 Richard Biener <rguenther@suse.de>
19147 PR tree-optimization/78644
19148 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19149 of a simplification result we may not use it at all.
19151 2017-03-28 Richard Biener <rguenther@suse.de>
19154 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19155 without arguments, generate default definition of a SSA name.
19157 2017-03-28 Richard Biener <rguenther@suse.de>
19159 PR middle-end/80222
19160 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19161 TYPE_REF_CAN_ALIAS_ALL references.
19162 * fold-const.c (fold_indirect_ref_1): Likewise.
19164 2017-03-28 Martin Liska <mliska@suse.cz>
19167 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19168 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19170 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19171 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19173 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19174 (EXTRA_SPECS): Define.
19175 (SUBTARGET_EXTRA_SPECS): Likewise.
19176 (SUBTARGET_CPP_SPEC): Likewise.
19177 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19178 SUBTARGET_EXTRA_SPECS.
19179 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19181 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19183 * config/arc/simdext.md (vst64_insn): Update pattern.
19184 (vld32wh_insn): Likewise.
19185 (vld32wl_insn): Likewise.
19186 (vld64_insn): Likewise.
19187 (vld32_insn): Likewise.
19189 2017-03-28 Marek Polacek <polacek@redhat.com>
19192 * fold-const.c (fold_comparison): Use protected_set_expr_location
19193 instead of SET_EXPR_LOCATION.
19195 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19197 * tree.c (add_expr): Avoid name lookup warning.
19199 2017-03-27 Jeff Law <law@redhat.com>
19201 PR tree-optimization/80216
19202 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19203 function name. Limit recursion depth.
19204 (record_temporary_equivalences): Corresponding changes.
19206 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19208 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19211 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19214 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19216 * cfgcleanup.c (reg_note_cfa_p): New array.
19217 (insns_have_identical_cfa_notes): New function.
19218 (old_insns_match_p): Don't cross-jump in between /f
19219 and non-/f instructions. If both i1 and i2 are frame related,
19220 verify all CFA notes, their order and content.
19222 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19225 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19226 HImode and SImode with zero extend to DImode to one insn.
19227 (bswap<mode>2_extenddi): Likewise.
19228 (bswapsi2_extenddi): Likewise.
19229 (bswaphi2_extendsi): Likewise.
19230 (bswaphi2): Combine bswap HImode and SImode into one insn.
19231 Separate memory insns from swapping register.
19232 (bswapsi2): Likewise.
19233 (bswap<mode>2): Likewise.
19234 (bswaphi2_internal): Delete, no longer used.
19235 (bswapsi2_internal): Likewise.
19236 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19237 store, and gpr<-gpr swap insns.
19238 (bswap<mode>2_store): Likewise.
19239 (bswaphi2_reg): Register only splitter, combine with the splitter.
19240 (bswaphi2 splitter): Likewise.
19241 (bswapsi2_reg): Likewise.
19242 (bswapsi2 splitter): Likewise.
19243 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19244 the insns into load, store, and register/register insns.
19245 (bswapdi2_ldbrx): Likewise.
19246 (bswapdi2_load): Likewise.
19247 (bswapdi2_store): Likewise.
19248 (bswapdi2_reg): Likewise.
19250 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19252 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19253 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19255 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19258 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19260 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19261 special handling for target option conflicts between dform
19262 options (-mpower9-dform, -mpower9-dform-vector,
19263 -mpower9-dform-scalar) and -mno-direct-move.
19265 2017-03-27 Richard Biener <rguenther@suse.de>
19267 PR tree-optimization/80181
19268 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19270 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19272 * config/arc/predicates.md (move_double_src_operand): Replace the
19273 call to move_double_src_operand with a call to address_operand.
19275 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19277 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19278 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19279 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19281 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19283 * config/arc/predicates.md (long_immediate_loadstore_operand):
19284 Consider scaled addresses cases.
19286 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19288 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19289 restored when in interrupt.
19290 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19291 doesn't have delay slot.
19293 2017-03-27 Richard Biener <rguenther@suse.de>
19296 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19297 inlined thunk clones.
19299 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19302 * asan.c (instrument_derefs): Copy over last operand from
19303 original COMPONENT_REF to the new COMPONENT_REF with
19304 DECL_BIT_FIELD_REPRESENTATIVE.
19305 * ubsan.c (instrument_object_size): Likewise.
19307 2017-03-27 Richard Biener <rguenther@suse.de>
19309 PR tree-optimization/80170
19310 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19311 sure DR/SCEV didnt fold in constants we do not see when looking
19312 at the reference base alignment.
19314 2017-03-27 Richard Biener <rguenther@suse.de>
19316 PR middle-end/80171
19317 * gimple-fold.c (fold_ctor_reference): Properly guard against
19318 NULL return value from canonicalize_constructor_val.
19320 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19323 * config/i386/i386.c (ix86_expand_builtin)
19324 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19325 flags reg setting and flags reg using instructions.
19326 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19327 clobbering instructions to zero extend op2.
19329 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19331 * doc/install.texi (Configuration) <--with-aix-soname>:
19332 Update link to AIX ld.
19334 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19336 PR rtl-optimization/80160
19337 PR rtl-optimization/80159
19338 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19339 reg_alternate_class into account.
19341 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19344 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19345 to consider in curr_insn_transform.
19347 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19349 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19350 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19351 and emit_mode_inner.
19353 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19355 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19356 argument to the overloaded builtin variants. Use the new flag to
19357 deprecate certain builtin variants.
19358 * config/s390/s390-builtin-types.def: Add new builtin types.
19359 * config/s390/s390-builtins.h: Support new flags field for
19360 overloaded builtins.
19361 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19362 (s390_macro_to_expand): Enable vector float data type.
19363 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19364 builtins by incrementing the __VEC__ version number.
19365 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19367 (s390_resolve_overloaded_builtin): Emit error messages depending
19368 on the builtin flags.
19369 * config/s390/s390.c (s390_expand_builtin): Support additional
19370 flags argument. Change error message to match the messages
19371 emitted in s390-c.c.
19372 * config/s390/s390.md: New UNSPEC_* constants.
19373 (op_type): Add new instruction types.
19374 * config/s390/vecintrin.h: Add new builtins and test data class
19376 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19377 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19378 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19379 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19380 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19381 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19383 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19384 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19385 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19386 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19388 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19389 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19390 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19391 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19393 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19394 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19395 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19396 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19397 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19398 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19399 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19401 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19402 ("vec_scatter_element<V_HW_4:mode>_DI")
19403 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19404 ("vec_fpint<mode>", "vflls")
19405 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19406 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19407 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19408 ("*vec_cmphe<mode>_cc"): ... these.
19410 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19411 mode constant instead of magic value.
19413 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19415 * config/s390/s390.c (s390_expand_vec_compare): Support other
19416 vector floating point modes than just V2DF.
19417 (s390_expand_vcond): Likewise.
19418 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19419 (s390_cannot_change_mode_class): Prevent mode changes between TF
19420 and V1TF in vector registers.
19421 * config/s390/s390.md (DF, SF): New mode attributes.
19422 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19423 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19424 SFmode support for VRs.
19425 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19427 (VFT, VF_HW): New mode iterators.
19428 (vw, sdx): New mode attributes.
19429 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19430 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19431 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19432 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19433 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19434 also the new vector floating point modes. Renaming to ...
19436 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19437 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19438 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19439 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19440 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19441 ("vec_unordered<mode>"): ... these.
19443 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19444 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19445 ("*vec_extendv2df"): New insn definitions.
19447 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19449 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19450 ("mulditi3_2", "*muldi3_sign"): New patterns.
19451 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19452 rename the pattern definition.
19454 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19456 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19458 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19460 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19462 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19463 instruction if possible.
19464 * config/s390/vector.md (vec_halfnumelts): New mode
19466 ("*vec_vllezlf<mode>"): New pattern.
19468 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19470 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19471 ("popcountv4si2", "popcountv2di2"): Rename to ...
19472 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19473 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19475 ("popcount<mode>2_vxe"): New pattern.
19477 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19479 * common/config/s390/s390-common.c (processor_flags_table): Add
19481 * config.gcc: Add arch12.
19482 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19483 Default to arch12 for unknown CPU model numbers.
19484 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19485 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19486 PROCESSOR_max sanity check.
19487 * config/s390/s390-opts.h (enum processor_type): Add
19489 * config/s390/s390.c (processor_table): Add arch12.
19490 (s390_expand_builtin): Add check for B_VXE flag.
19491 (s390_issue_rate): Add PROCESSOR_ARCH12.
19492 (s390_get_sched_attrmask): Likewise.
19493 (s390_get_unit_mask): Likewise.
19494 (s390_sched_score): Enable z13 scheduling for arch12.
19495 (s390_sched_reorder): Likewise.
19496 (s390_sched_variable_issue): Likewise.
19497 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19499 (s390_tune_attr): Use z13 scheduling also for arch12.
19500 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19501 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19502 (TARGET_VXE_P): New macros.
19503 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19504 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19505 * config/s390/s390.opt: Add arch12 as processor_type.
19507 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19509 * config/s390/s390.md
19510 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19511 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19512 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19514 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19515 Rename expanders to ...
19517 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19518 ("fixuns_truncdddi2_emu"): ... these.
19520 ("fixuns_trunc<mode>si2_emu"): New expander.
19522 ("*fixuns_truncdfdi2_z13"): Rename to ...
19523 ("*fixuns_truncdfdi2_vx"): ... this.
19525 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19527 * config/s390/2964.md: Remove the single element vector compare
19528 instructions which are no longer used.
19529 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19531 (s390_canonicalize_comparison): Remove handling of DFmode
19533 (s390_expand_vec_compare_scalar): Remove function.
19534 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19535 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19537 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19539 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19541 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19543 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19544 will anyway by matched by mov<mode>_64dfp.
19546 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19548 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19549 vlef/vstef. Add missing operand to vleif.
19551 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19553 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19554 pair for all vector types with 64 bit elements.
19555 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19556 * config/s390/vector.md (V_HW_64): ... here.
19557 (V_128_NOSINGLE): New mode iterator.
19558 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19559 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19560 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19561 ("*vec_load_pairv2di"): Change to ...
19562 ("*vec_load_pair<mode>"): ... this one.
19564 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19566 * config/s390/constraints.md: Add comments.
19567 (jKK): Reject element sizes > 8 bytes.
19568 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19570 * config/s390/s390.md: Add the s_operand checks formerly in
19571 s390_split_ok_p to various splitters where they are still
19573 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19574 for 128 bit vectors. Plus two splitters.
19576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19578 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19581 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19584 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19585 error if the boundary argument is not constant.
19587 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19589 PR rtl-optimization/80112
19590 * loop-doloop.c (doloop_condition_get): Don't check condition
19591 if cmp isn't SET with IF_THEN_ELSE src.
19593 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19595 PR tree-optimization/80158
19596 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19597 replacing a candidate statement, also replace it for the
19598 candidate's alternate interpretation.
19599 (replace_rhs_if_not_dup): Likewise.
19600 (replace_one_candidate): Likewise.
19602 2017-03-24 Richard Biener <rguenther@suse.de>
19604 PR tree-optimization/80167
19605 * graphite-isl-ast-to-gimple.c
19606 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19608 (translate_isl_ast_to_gimple::get_rename): Likewise.
19610 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19612 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19613 handling of certain combinations of target options, including the
19614 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19615 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19617 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19620 * config/arm/arm.md (*load_multiple): Add reload_completed to
19621 matching condition.
19623 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19624 Richard Biener <rguenth@suse.de>
19626 PR tree-optimization/79908
19627 PR tree-optimization/80136
19628 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19629 been cast away, gimplify_and_add suffices.
19631 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19633 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19635 2017-03-23 Richard Biener <rguenther@suse.de>
19637 PR tree-optimization/80032
19638 * gimplify.c (gimple_push_cleanup): Forced unconditional
19639 cleanups still have to go to the conditional_cleanups
19642 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19644 PR tree-optimization/80072
19645 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19647 (next_operand_entry_id): Change type to unsigned int.
19648 (sort_by_operand_rank): Make sure to return the right return value
19649 even if unsigned fields are bigger than INT_MAX.
19650 (struct oecount): Change cnt and id type to unsigned int.
19651 (oecount_hasher::equal): Formatting fix.
19652 (oecount_cmp): Make sure to return the right return value
19653 even if unsigned fields are bigger than INT_MAX.
19654 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19657 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19658 TREE_READONLY on result if writing it more than once.
19661 * doc/invoke.texi (-fsanitize=thread): Document that with
19662 -fnon-call-exceptions atomics are not able to throw
19666 * tsan.c: Include tree-eh.h.
19667 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19668 maybe_clean_or_replace_eh_stmt where needed.
19669 (instrument_memory_accesses): Add cfg_changed argument.
19670 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19671 if it returned true.
19672 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19674 PR rtl-optimization/63191
19675 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19676 wrapper function, moved the whole old content into ...
19677 (ix86_delegitimize_address_1): ... this. New inline function.
19678 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19679 true as last argument instead of ix86_delegitimize_address.
19681 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19683 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19684 cortexa57_branch_cost.
19686 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19688 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19690 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19693 * doc/md.texi (Constraints): Document wA constraint.
19694 * config/rs6000/constraints.md (wA): New.
19695 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19696 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19697 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19698 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19700 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19703 * gimplify.c (is_oacc_declared): New function.
19704 (oacc_default_clause): Use it to set default flags for acc declared
19705 variables inside parallel regions.
19706 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19707 declared variables.
19708 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19709 declare attribute to any decl as necessary.
19711 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19714 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19715 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19716 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19717 (arm_arch_lpae): This.
19718 * config/arm/arm.c (arm_arch7ve): Rename into ...
19719 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19720 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19723 2017-03-22 Martin Liska <mliska@suse.cz>
19726 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19727 error message instead of an ICE.
19729 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19731 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19733 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19735 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19737 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19740 2017-03-21 Martin Sebor <msebor@redhat.com>
19742 * doc/extend.texi: Use "cannot" instead of "can't."
19743 * doc/hostconfig.texi: Same.
19744 * doc/install.texi: Same.
19745 * doc/invoke.texi: Same.
19746 * doc/loop.texi: Same.
19747 * doc/md.texi: Same.
19748 * doc/objc.texi: Same.
19749 * doc/rtl.texi: Same.
19750 * doc/tm.texi: Same.
19751 * doc/tm.texi.in: Same.
19752 * doc/trouble.texi: Same.
19754 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19757 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19758 (collect_checksum_attributes): Set it.
19759 (die_checksum_ordered): Use it.
19761 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19763 PR tree-optimization/79908
19764 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19765 change: For a VA_ARG whose LHS has been cast away, use
19766 force_gimple_operand to construct the side effects.
19768 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19770 PR translation/80001
19771 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19772 more amenable to translation.
19773 (oacc_loop_auto_partitions): Likewise.
19775 2017-03-21 Marek Polacek <polacek@redhat.com>
19776 Martin Sebor <msebor@redhat.com>
19778 PR tree-optimization/80109
19779 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19780 on INTEGRAL_TYPE_P.
19782 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19783 Segher Boessenkool <segher@kernel.crashing.org>
19786 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19787 check reg_used_between_p between insn and one of succ or succ2
19788 depending on if succ is artificial insn not inserted into insn
19791 2017-03-21 Martin Liska <mliska@suse.cz>
19793 PR gcov-profile/80081
19794 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19795 * doc/gcc.texi: Include gcov-dump stuff.
19796 * doc/gcov-dump.texi: New file.
19798 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19800 PR rtl-optimization/79150
19801 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19802 conditional jump, if the jump is the last insn of the loop.
19804 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19805 Richard Biener <rguenth@suse.de>
19807 PR tree-optimization/79908
19808 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19809 been cast away, use force_gimple_operand to construct the side
19812 2017-03-21 Martin Liska <mliska@suse.cz>
19814 PR libfortran/79956
19815 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19818 2017-03-21 Brad Spengler <spender@grsecurity.net>
19821 * plugin.c (htab_hash_plugin): New function.
19822 (add_new_plugin): Use it and adjust.
19823 (parse_plugin_arg_opt): Adjust.
19824 (init_one_plugin): Likewise.
19826 2017-03-21 Richard Biener <rguenther@suse.de>
19828 PR tree-optimization/80032
19829 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19830 if set force the cleanup to happen unconditionally.
19831 (gimplify_target_expr): Push inserted clobbers with force_uncond
19832 to avoid them being removed by control-dependent DCE.
19834 2017-03-21 Richard Biener <rguenther@suse.de>
19836 PR tree-optimization/80122
19837 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19838 va_arg_pack_len when the inlined call stmt requires pack
19840 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19842 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19845 * tsan.c (instrument_builtin_call): If the memory model argument
19846 is not a constant, assume it is valid.
19849 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19852 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19854 PR rtl-optimization/79910
19855 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19856 if its dest is used by an insn before I2 (other than the combined
19857 insns themselves, which are properly handled already).
19859 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19862 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19864 * combine.c (record_used_regs): New static function.
19865 (try_combine): Handle situations where there is an additional
19866 instruction between I2 and I3 which needs to have a LOG_LINK
19870 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19872 * combine.c (try_combine): Delete redundant i1 test. Call
19873 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19875 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19878 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19879 alternatives 13/14.
19881 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19883 PR tree-optimization/80054
19884 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19885 the optimization if a PHI or any of its arguments is not dominated
19886 by the candidate's basis. Use gphi* rather than gimple* as
19888 (replace_profitable_candidates): Clean up a gimple* variable that
19889 should be a gphi* variable.
19891 2017-03-20 Martin Sebor <msebor@redhat.com>
19894 * doc/extend.texi (attribute constructor): Document present limitation.
19896 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19899 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19900 __POWER9_VECTOR__ #ifdef control, change template definition to
19901 use Power9-specific built-in function.
19902 (vec_any_eq): Likewise.
19903 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19904 to control outcomes from this test.
19905 (vector_ae_<mode>p): For VEC_F modes, likewise.
19907 2017-03-20 Ian Lance Taylor <iant@google.com>
19909 * config/i386/i386.c (ix86_function_regparm): Save an extra
19910 register for -fsplit-stack with DECL_STATIC_CHAIN.
19912 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19915 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19916 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19918 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19920 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19922 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19925 2017-03-20 Marek Polacek <polacek@redhat.com>
19928 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19930 2017-03-20 Richard Biener <rguenther@suse.de>
19932 PR tree-optimization/80113
19933 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19934 allocate extra SSA name for PHI def.
19935 (add_close_phis_to_outer_loops): Likewise.
19936 (add_close_phis_to_merge_points): Likewise.
19937 (copy_loop_close_phi_args): Likewise.
19938 (copy_cond_phi_nodes): Likewise.
19940 2017-03-20 Martin Liska <mliska@suse.cz>
19942 PR middle-end/79753
19943 * tree-chkp.c (chkp_build_returned_bound): Do not build
19944 returned bounds for a LHS that's not a BOUNDED_P type.
19946 2017-03-20 Martin Liska <mliska@suse.cz>
19950 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19951 COMPLEX_CST and VECTOR_CST.
19953 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19956 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19957 target operand. A new splitter adds the clobber statement in case
19958 the target operand is dead anyway.
19960 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19962 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19963 to age-old versions of binutils and glibc.
19965 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19967 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19969 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19971 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19973 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19975 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19976 requirement for binutils 2.13.
19978 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19980 * combine.c (try_combine): Delete redundant i1 test. Call
19981 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19983 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19985 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19986 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19988 <riscv64-*-elf>: Re-arrange section
19989 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19990 <riscv32-*-linux>: Likewise.
19991 <riscv64-*-elf>: Likewise
19992 <riscv64-*-linux>: Likewise.
19994 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19997 * aarch64.opt(verbose-cost-dump): Fix typo.
19999 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20002 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20003 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20005 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20007 * reload.c (find_reloads): When reloading a nonoffsettable address,
20008 use RELOAD_OTHER for it and its address reloads.
20010 PR rtl-optimization/79910
20011 * combine.c (record_used_regs): New static function.
20012 (try_combine): Handle situations where there is an additional
20013 instruction between I2 and I3 which needs to have a LOG_LINK
20016 2017-03-17 Jeff Law <law@redhat.com>
20018 PR tree-optimization/71437
20019 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20020 conditional in the hash table first.
20021 (vrp_dom_walker::before_dom_children): Extract condition from
20022 ASSERT_EXPR. Record condition, its inverion and any implied
20023 conditions as well.
20025 2017-03-17 Marek Polacek <polacek@redhat.com>
20026 Markus Trippelsdorf <markus@trippelsdorf.de>
20028 PR tree-optimization/80079
20029 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20032 2017-03-17 Richard Biener <rguenther@suse.de>
20034 PR middle-end/80075
20035 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20036 Properly verify the LHS before the RHS possibly claims to be
20038 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20041 2017-03-17 Martin Jambor <mjambor@suse.cz>
20043 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20044 (List of -O2 options): Likewise.
20045 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20048 2017-03-17 Tom de Vries <tom@codesourcery.com>
20050 * gcov-dump.c (print_usage): Print bug_report_url.
20052 2017-03-17 Richard Biener <rguenther@suse.de>
20054 PR middle-end/80050
20055 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20056 (parser::peek): Likewise.
20058 2017-03-17 Richard Biener <rguenther@suse.de>
20060 PR tree-optimization/80048
20061 * sese.c (free_sese_info): Properly release rename_map and
20062 copied_bb_map elements.
20064 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20066 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20067 Add linked-list forward and backlinks. Insert on
20068 construction, remove on destruction.
20069 (class pass_store_merging): Add m_stores_head field.
20070 (pass_store_merging::terminate_and_process_all_chains):
20071 Iterate over m_stores_head list.
20072 (pass_store_merging::terminate_all_aliasing_chains):
20074 (pass_store_merging::execute): Check for debug stmts first.
20075 Push new chains onto the m_stores_head stack.
20077 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20080 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20081 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20082 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20084 2017-03-16 Jeff Law <law@redhat.com>
20086 PR tree-optimization/71437
20087 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20088 member function. Implementation moved into after_dom_children
20089 member function and into the threader's thread_outgoing_edges
20091 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20092 some code into new thread_outgoing_edges.
20093 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20094 definition. Simplify marker handling (do it here). Assume we always
20095 have the available expression and the const/copies tables.
20096 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20098 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20099 * tree-vrp.c (equiv_stack): No longer file scoped.
20100 (vrp_dom_walker): New class.
20101 (vrp_dom_walker::before_dom_children): New member function.
20102 (vrp_dom_walker::after_dom_children): Likewise.
20103 (identify_jump_threads): Setup domwalker. Use it rather than
20104 walking edges in a random order by hand. Simplify setup/finalization.
20105 (finalize_jump_threads): Remove.
20106 (vrp_finalize): Do not call identify_jump_threads here.
20107 (execute_vrp): Do it here instead and call thread_through_all_blocks
20110 PR tree-optimization/71437
20111 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20113 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20115 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20116 (dom_opt_dom_walker::thread_across_edge): Remove
20117 handle_dominating_asserts argument. All callers changed.
20118 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20119 changes. Remove calls to lhs_of_dominating_assert. Other
20120 uses of handle_dominating_asserts turn into unconditional code
20121 (simplify_control_stmt_condition_1): Likewise.
20122 (simplify_control_stmt_condition): Likewise.
20123 (thread_through_normal_block, thread_across_edge): Likewise.
20124 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20125 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20126 object if it is not an SSA_NAME.
20127 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20128 before calling into the VRP specific simplifiers.
20129 (identify_jump_threads): Remove handle_dominating_asserts
20132 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20135 * tree-diagnostic.c (default_tree_printer): No longer static.
20136 * tree-diagnostic.h (default_tree_printer): New prototype.
20138 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20140 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20141 Change ins into fmov.
20143 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20145 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20146 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20147 Use h_con constraint for operand 1.
20148 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20149 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20151 2017-03-15 Jeff Law <law@redhat.com>
20153 PR tree-optimization/71437
20154 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20155 (record_temporary_equivalences): Use it.
20157 PR tree-optimization/71437
20158 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20159 tree-ssa-scopedtables.
20160 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20161 (record_conditions, record_cond, vuse_eq): Likewise.
20162 (record_edge_info): Adjust to API tweak of record_conditions.
20163 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20164 (record_temporary_equivalences, optimize_stmt): Likewise.
20165 (eliminate_redundant_computations): Likewise.
20166 (record_equivalences_from_stmt): Likewise.
20167 * tree-ssa-scopedtables.c: Include options.h and params.h.
20168 (vuse_eq): New function, moved from tree-ssa-dom.c
20169 (build_and_record_new_cond): Likewise.
20170 (record_conditions): Likewise. Accept vector of conditions rather
20171 than edge_equivalence structure for first argument.
20172 for the first argument.
20173 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20174 from tree-ssa-dom.c.
20175 (avail_exprs_stack::record_cond): Likewise.
20176 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20177 from tree-ssa-dom.c.
20178 (avail_exprs_stack): Add new member functions lookup_avail_expr
20180 (record_conditions): Declare.
20182 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20185 * lra-constraints.c (process_alt_operands): Increase reject for
20186 reloading an input/output operand.
20188 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20191 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20192 insns to convert from signed/unsigned char/short to IEEE 128-bit
20194 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20196 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20199 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20200 subreg of inner mode for values already in registers.
20202 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20204 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20205 iteration reg is used after the loop.
20207 2017-03-14 Martin Sebor <msebor@redhat.com>
20209 PR tree-optimization/79800
20210 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20211 precision in negative-positive range.
20212 (format_floating): Call non-const overload with adjusted precision.
20214 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20217 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20220 2017-03-14 Martin Sebor <msebor@redhat.com>
20222 PR middle-end/80020
20223 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20224 * builtins.def (aligned_alloc): Use it.
20227 * Makefile.in (GTFILES): Add calls.c.
20228 * calls.c: Include "gt-calls.h".
20230 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20232 PR rtl-optimization/79728
20233 * regs.h (struct target_regs): New field
20234 x_contains_allocatable_regs_of_mode.
20235 (contains_allocatable_regs_of_mode): New macro.
20236 * reginfo.c (init_reg_sets_1): Initialize it, and change
20237 contains_reg_of_mode so it includes global regs as well.
20238 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20239 rather than contains_regs_of_mode.
20241 2017-03-14 Martin Liska <mliska@suse.cz>
20243 * doc/invoke.texi: Document options that can't be combined with
20244 -fcheck-pointer-bounds.
20246 2017-03-14 Martin Liska <mliska@suse.cz>
20248 PR middle-end/79831
20249 * doc/invoke.texi (-Wchkp): Document the option.
20251 2017-03-14 Martin Liska <mliska@suse.cz>
20253 * Makefile.in: Install gcov-dump.
20255 2017-03-14 Martin Liska <mliska@suse.cz>
20257 * multiple_target.c (expand_target_clones): Bail out for
20258 an invalid attribute.
20260 2017-03-14 Richard Biener <rguenther@suse.de>
20262 * alias.c (struct alias_set_entry): Pack properly.
20263 * cfgloop.h (struct loop): Likewise.
20264 * cse.c (struct set): Likewise.
20265 * ipa-utils.c (struct searchc_env): Likewise.
20266 * loop-invariant.c (struct invariant): Likewise.
20267 * lra-remat.c (struct cand): Likewise.
20268 * recog.c (struct change_t): Likewise.
20269 * rtl.h (struct address_info): Likewise.
20270 * symbol-summary.h (function_summary): Likewise.
20271 * tree-loop-distribution.c (struct partition): Likewise.
20272 * tree-object-size.c (struct object_size_info): Likewise.
20273 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20274 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20275 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20276 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20277 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20278 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20279 (struct _stmt_vec_info): Likewise.
20281 2017-03-14 Martin Liska <mliska@suse.cz>
20284 * multiple_target.c (create_dispatcher_calls): Check that
20285 a target can create a function dispatcher.
20287 2017-03-14 Martin Liska <mliska@suse.cz>
20290 * multiple_target.c (expand_target_clones): Drop local.local
20291 flag for default implementation.
20293 2017-03-14 Richard Biener <rguenther@suse.de>
20295 PR tree-optimization/80030
20296 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20298 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20300 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20301 gcc_fallthrough() instead of __attribute__((fallthrough));
20303 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20305 * doc/gcc.texi: Remove "up" link to (DIR).
20306 * doc/gccint.texi: Ditto.
20308 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20310 * doc/install.texi (Specific) <avr>: Remove reference to
20313 2017-03-13 Jeff Law <law@redhat.com>
20315 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20316 attribute rather than comments.
20318 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20319 match_scratch operand is highest.
20321 2017-03-13 Martin Liska <mliska@suse.cz>
20323 PR middle-end/78339
20324 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20325 is a CHKP clone, use original declaration.
20327 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20329 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20330 (arc_conditional_register_usage): Use a different allocation order
20331 when optimizing for size.
20332 * common/config/arc/arc-common.c (arc_option_optimization_table):
20333 Section anchors default on when optimizing for size.
20335 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20337 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20339 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20341 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20342 * config/arc/arc.md (cpu_facility): Add cd variant.
20343 (*movqi_insn): Add code density variant.
20344 (*movhi_insn): Likewise.
20345 (*movqi_insn): Likewise.
20346 (*addsi3_mixed): Likewise.
20347 (subsi3_insn): Likewise.
20349 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20351 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20353 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20355 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20356 expressions with MINUS and UNARY ops.
20358 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20361 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20363 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20364 between vec_select and vector argument.
20365 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20366 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20367 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20368 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20369 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20370 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20372 2017-03-13 Richard Biener <rguenther@suse.de>
20375 * params.def (vect-max-peeling-for-alignment): Fix typo.
20377 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20379 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20380 issue that only occurred with binutils below 2.18.
20382 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20384 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20385 refer to binutils 2.11/2.12 minimum.
20387 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20389 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20390 ftp.kernel.org and simplify binutils requirement.
20392 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20394 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20396 (Optimize Options): Ditto. Also remove redundancy.
20398 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20400 PR translation/79848
20401 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20403 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20404 to G_ to avoid double translation.
20406 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20408 PR translation/79923
20409 * auto-profile.c (get_combined_location): Convert leading
20410 character of diagnostics to lower case and remove trailing period.
20411 (read_profile): Likewise for various diagnostics.
20412 * config/arm/arm.c (arm_option_override): Remove trailing period
20413 from various diagnostics.
20414 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20415 (msp430_expand_delay_cycles): Likewise.
20417 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20420 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20421 full command-line argument, rather than just "str".
20422 (aarch64_validate_march): Likewise.
20423 (aarch64_validate_mtune): Likewise.
20425 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20427 PR rtl-optimization/78911
20428 * lra-assigns.c (must_not_spill_p): New function.
20429 (spill_for): Use it.
20431 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20433 PR tree-optimization/79981
20434 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20435 ATOMIC_COMPARE_EXCHANGE ifn result.
20436 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20437 IFN_ATOMIC_COMPARE_EXCHANGE.
20439 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20442 * opts.c (parse_sanitizer_options): Add missing question mark to
20443 "did you mean" message.
20445 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20447 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20449 (VMULEUH_UNS): Likewise.
20450 (VMULOUB_UNS): Likewise.
20451 (VMULOUH_UNS): Likewise.
20452 * config/rs6000/rs6000.c (builtin_function_type): Remove
20453 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20455 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20458 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20459 x with result of extra_parsing_for_operand_code_0.
20460 (function_reader::extra_parsing_for_operand_code_0): Convert
20461 return type from void to rtx, returning x. When reading
20462 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20463 larger size containing struct block_symbol.
20465 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20467 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20468 -mfloat128-hardware without -m64.
20470 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20473 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20474 entries to the case statement that marks unsigned arguments to
20475 overloaded functions.
20477 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20480 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20482 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20485 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20486 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20488 2017-03-10 Martin Liska <mliska@suse.cz>
20492 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20493 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20496 2017-03-10 Olivier Hainque <hainque@adacore.com>
20498 * tree-switch-conversion (array_value_type): Start by resetting
20499 candidate type to it's main variant.
20501 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20503 PR rtl-optimization/79909
20504 * combine.c (try_combine): Use simplify_replace_rtx on individual
20505 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20506 of the whole CALL_INSN_FUNCTION_USAGE.
20508 PR tree-optimization/79972
20509 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20510 get_range_info on SSA_NAMEs. Formatting fixes.
20512 2017-03-10 Richard Biener <rguenther@suse.de>
20513 Jakub Jelinek <jakub@redhat.com>
20515 PR tree-optimization/77975
20516 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20517 edge to be constant.
20518 (get_val_for): For constant x return it. Formatting fix.
20519 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20520 would use the same bases as the current one.
20522 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20524 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20525 instead of vec_select for V1TImode.
20526 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20528 (VSX_LE_128): Add V1TI to this mode iterator.
20529 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20530 (*vsx_le_perm_store_<mode>): Likewise.
20531 (pre-reload splitter for VSX stores): Likewise.
20532 (post-reload splitter for VSX stores): Likewise.
20533 (*vsx_xxpermdi2_le_<mode>): Likewise.
20534 (*vsx_lxvd2x2_le_<mode>): Likewise.
20535 (*vsx_stxvd2x2_le_<mode>): Likewise.
20537 2017-03-09 Michael Eager <eager@eagercon.com>
20539 Correct failures with --enable-checking=yes,rtl.
20541 * config/microblaze/microblaze.c (microblaze_expand_shift):
20542 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20543 test for const0_rtx.
20544 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20545 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20547 2017-03-09 Richard Biener <rguenther@suse.de>
20549 PR tree-optimization/79977
20550 * graphite-scop-detection.c (scop_detection::merge_sese):
20551 Handle the case of extra exits to blocks dominating the entry.
20553 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20555 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20558 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20560 PR rtl-optimization/79949
20561 * lra-constraints.c (process_alt_operands): Check memory when
20562 trying to predict a cycle. Print about the overall increase.
20564 2017-03-09 Richard Biener <rguenther@suse.de>
20566 PR middle-end/79971
20567 * gimple-expr.c (useless_type_conversion_p): Preserve
20568 TYPE_SATURATING for fixed-point types.
20570 2017-03-09 Richard Biener <rguenther@suse.de>
20573 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20574 alignment of BLKmode params.
20576 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20579 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20580 (VALL_NO_V2Q): Likewise.
20582 * config/aarch64/aarch64-simd.md
20583 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20585 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20586 VALL_NO_V2Q mode iterator.
20587 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20589 2017-03-09 Martin Liska <mliska@suse.cz>
20591 PR tree-optimization/79631
20592 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20593 tree_int_cst_sign_bit just for INTEGER constants.
20595 2017-03-09 Martin Liska <mliska@suse.cz>
20599 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20602 2017-03-09 Marek Polacek <polacek@redhat.com>
20605 * tree.c (inchash::add_expr): Handle TREE_VEC.
20607 2017-03-09 Martin Liska <mliska@suse.cz>
20610 (chkp_narrow_size_and_offset): New function.
20611 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20612 (void chkp_parse_bit_field_ref): New function.
20613 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20614 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20616 2017-03-09 Martin Liska <mliska@suse.cz>
20619 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20620 (chkp_find_bounds_1): Remove gcc_unreachable.
20622 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20625 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20626 BUILT_IN_SYNC*, determine the access type from the size suffix and
20627 always build a MEM_REF with that type. Handle forgotten
20628 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20631 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20632 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20633 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20634 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20635 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20636 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20637 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20638 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20639 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20640 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20641 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20642 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20643 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20644 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20645 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20646 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20647 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20648 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20649 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20650 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20651 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20652 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20653 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20654 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20655 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20656 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20657 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20658 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20659 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20660 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20661 definitions outside of __OPTIMIZE__ guarded section.
20664 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20665 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20666 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20667 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20670 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20672 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20673 ("vfenez<mode>"): Add missing constraints.
20675 2017-03-08 Martin Sebor <msebor@redhat.com>
20678 * config/nds32/nds32.c (nds32_option_override):
20679 Fix misspelled diagnostic.
20681 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20684 * gimplify.c (gimplify_omp_for): Replace index var in outer
20685 taskloop statement with an artificial variable and add
20686 OMP_CLAUSE_PRIVATE clause for it.
20688 2017-03-08 Richard Biener <rguenther@suse.de>
20690 PR tree-optimization/79955
20691 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20692 for accesses that are completely outside of the variable.
20694 2017-03-08 Andrew Haley <aph@redhat.com>
20696 PR tree-optimization/79943
20697 * tree-ssa-loop-split.c (compute_new_first_bound): When
20698 calculating the new upper bound, (END-BEG) should be added, not
20701 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20703 * config/avr/avr.md (setmemhi): Make sure match_dup
20704 operand number comes before match_scratch.
20706 2017-03-08 Richard Biener <rguenther@suse.de>
20708 PR tree-optimization/79920
20709 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20710 with ncopies == 1 to ...
20711 (vect_transform_slp_perm_load): ... here. Properly compute
20712 all element loads by iterating VF times over the group. Do
20713 not handle ncopies (computed in a broken way) in
20714 vect_create_mask_and_perm.
20716 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20719 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20720 is a uniform vector, use uniform_vector_p return value instead of
20721 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20723 2017-03-07 Marek Polacek <polacek@redhat.com>
20725 PR middle-end/79809
20726 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20727 (alloca_call_type): Likewise.
20729 2017-03-07 Martin Liska <mliska@suse.cz>
20731 * gcov.c (process_args): Put comment to correct location.
20733 2017-03-07 Martin Liska <mliska@suse.cz>
20735 PR middle-end/68270
20736 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20737 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20738 (chkp_narrow_bounds_for_field): Likewise.
20739 (chkp_parse_array_and_component_ref): Pass one more argument to
20742 2017-03-07 Richard Biener <rguenther@suse.de>
20744 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20747 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20749 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20750 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20752 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20755 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20756 to end of description.
20757 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20759 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20761 PR rtl-optimization/79901
20762 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20764 (*avx512f_<code><mode>3<mask_name>): ... this.
20765 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20766 iterator instead of VI8_AVX2_AVX512BW.
20768 PR rtl-optimization/79901
20769 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20770 min/max expander, expand it using expand_vec_cond_expr.
20773 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20776 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20779 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20780 to void * for PCH reasons.
20781 * dwarf2out.c (output_loc_operands, output_die): Cast
20782 v.val_vec.array to unsigned char *.
20784 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20787 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20790 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20792 PR rtl-optimization/79571
20793 * lra-constraints.c (process_alt_operands): Calculate static
20794 reject and subtract it from overall when only addresses will be
20797 2017-03-06 Julia Koval <julia.koval@intel.com>
20800 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20801 incoming stack boundary to 128 for 64-bit targets.
20803 2017-03-06 Richard Biener <rguenther@suse.de>
20805 PR tree-optimization/79894
20806 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20807 to NULL after folding it.
20809 2017-03-06 Richard Biener <rguenther@suse.de>
20811 PR tree-optimization/79824
20812 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20813 check disabling peeling for gaps.
20815 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20817 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20818 attributes): Document gettimeofday.
20820 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20822 * config/s390/s390.c (s390_option_override_internal): Set
20823 PARAM_MIN_VECT_LOOP_BOUND
20825 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20827 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20828 * config/s390/s390.md: Likewise.
20830 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20833 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20834 (<avx2_avx512>_perm<mode>): Rename to ...
20835 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20837 (<avx512>_perm<mode>_mask): Rename to ...
20838 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20840 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20841 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20842 instead of VI8F_256_512.
20843 (avx512f_perm<mode>): New define_expand.
20844 (avx512f_perm<mode>_mask): Likewise.
20845 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20846 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20848 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20850 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20851 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20853 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20855 2017-03-06 Martin Liska <mliska@suse.cz>
20858 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20859 when having a SSA NAME w/o VAR_DECL assigned to it.
20861 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20863 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20864 msa_dpsub_<su>_d): Fix MODE for vec_select.
20866 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20868 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20870 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20872 2017-03-06 Richard Biener <rguenther@suse.de>
20874 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20875 * plugin.c (register_plugin_info): Likewise.
20876 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20878 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20880 * config/i386/sse.md (sse_storehps, sse_storelps,
20881 avx_<castmode><avxsizesuffix>_<castmode>,
20882 avx512f_<castmode><avxsizesuffix>_<castmode>,
20883 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20884 in condition that at least one operand is not a MEM.
20886 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20888 PR middle-end/79805
20889 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20890 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20892 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20893 gimple_call_nothrow_p flag based on whether original builtin can throw.
20894 If it can, emit following stmts on the fallthrough edge.
20895 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20896 don't create new bb if inserting just debug stmts on the edge, try to
20897 insert them on the fallthru bb or just reset debug stmts.
20899 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20902 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20903 restore recog_data (including the operand rtxes inside it) around
20904 the call to get_insn_template.
20906 2017-03-03 Martin Sebor <msebor@redhat.com>
20908 PR tree-optimization/79699
20909 * context.c (context::~context): Free MPFR caches to avoid
20910 a memory leak on program exit.
20912 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20914 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20915 Use wide_int::ulow () instead of .elt (0).
20917 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20919 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20920 (*pushxf): Limit oF constraint to 32bit targets and add oC
20921 constraint for 64bit targets.
20922 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20923 (*pushdf): Change rmF constraint to rmC.
20925 2017-03-03 Martin Liska <mliska@suse.cz>
20927 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20928 Remove unused variable.
20930 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20933 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20934 is a memory operand, increase num_memory.
20935 (ix86_expand_args_builtin): Likewise.
20937 2017-03-03 Jan Hubicka <jh@suse.cz>
20940 * ipa-devirt.c (maybe_record_node): Properly handle
20941 __cxa_pure_virtual visibility.
20943 2017-03-03 Martin Liska <mliska@suse.cz>
20945 PR tree-optimization/79803
20946 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20948 (pass_loop_prefetch::execute): Disabled optimization if an
20949 assumption about L1 cache size is not met.
20951 2017-03-03 Martin Liska <mliska@suse.cz>
20953 PR rtl-optimization/79574
20954 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20955 (hash_scan_set): Likewise.
20956 (dump_hash_table): Likewise.
20957 (hoist_code): Likewise.
20959 2017-03-03 Richard Biener <rguenther@suse.de>
20961 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20962 in place of uhigh ().
20963 (fixed_convert_from_real): Likewise.
20965 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20968 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20970 2017-03-03 Richard Biener <rguenther@suse.de>
20972 PR middle-end/79818
20973 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20974 TYPE_OVERFLOW_UNDEFINED check.
20976 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20978 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20980 (vector_ae_<mode>_p): Likewise.
20981 (vector_nez_<mode>_p): Likewise.
20982 (vector_ne_v2di_p): Likewise.
20983 (vector_ae_v2di_p): Likewise.
20984 (vector_ne_<mode>_p): Likewise.
20985 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20987 (vsx_tsqrt<mode>2_fe): Likewise.
20989 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20992 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20994 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20996 PR rtl-optimization/79780
20997 * cprop.c (one_cprop_pass): When second and further conditional trap
20998 in a single basic block is turned into an unconditional trap, turn it
20999 into a deleted note to avoid RTL verification failures.
21001 2017-03-02 Richard Biener <rguenther@suse.de>
21003 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21005 2017-03-02 Richard Biener <rguenther@suse.de>
21007 PR tree-optimization/79345
21009 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21010 param and abort the walk, returning -1 if it is hit.
21011 (walk_aliased_vdefs): Take a limit param and pass it on.
21012 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21013 defaulting to 0 and return a signed int.
21014 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21015 (check_defs): New helper.
21016 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21017 about uninitialized memory.
21018 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21019 bogus uninitialized warning.
21020 (fixed_convert_from_real): Likewise.
21022 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21024 PR tree-optimization/66768
21025 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21026 iv_use if base object can't be determined.
21028 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21030 PR tree-optimization/79345
21031 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21032 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21033 (get_pattern_stats): Initialize it.
21034 * genemit.c (gen_expand): Verify match_scratch numbers come after
21035 match_operand/match_dup numbers.
21036 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21037 match_scratch numbers.
21038 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21040 * config/s390/s390.md (trunctdsd2): Likewise.
21042 2017-03-02 Richard Biener <rguenther@suse.de>
21044 * wide-int.h (wide_int_storage::operator=): Implement in terms
21047 2017-03-02 Richard Biener <rguenther@suse.de>
21049 PR tree-optimization/79777
21050 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21051 the to insert expression to sth existing.
21053 2017-03-01 Martin Sebor <msebor@redhat.com>
21055 PR middle-end/79692
21056 * gimple-ssa-sprintf.c
21057 (directive::known_width_and_precision): New function.
21058 (format_integer): Use it.
21059 (get_mpfr_format_length): Consider the full range of precision
21060 when computing %g output with the # flag. Set the likely byte
21061 count to 3 rather than 1 when precision is indeterminate.
21062 (format_floating): Correct the lower bound of precision.
21064 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21066 * doc/invoke.texi: Document default code model for 64-bit Linux.
21068 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21071 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21072 udiv rather than div since input pattern is unsigned.
21074 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21076 * config/i386/i386.c (print_reg): Warn for values of
21077 unsupported size in integer register.
21079 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21082 * config/rs6000/predicates.md (current_file_function_operand): Do
21083 not allow self calls to be local if the function is replaceable.
21085 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21088 * config/rs6000/altivec.h (vec_ctz and others): Change the
21089 preprocessor macro that controls conditional compilation from
21090 _ARCH_PWR9 to __POWER9_VECTOR__.
21091 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21092 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21093 control (instead of _ARCH_PWR9 control) so that template
21094 definition uses power9-specific function.
21095 (vec_any_eq): Likewise.
21096 (vec_all_ne): Change macro definition to use a power9-specific
21097 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21098 _ARCH_PWR9 control).
21099 (vec_any_eq) Likewise.
21100 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21101 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21102 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21103 support for xvcmpnedp instruction.
21104 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21105 macro expansion so that Power9 implementation of vec_all_ne does
21106 not use the AltiVec predicate framework.
21107 (VCMPNEH_P): Likewise.
21108 (VCMPNEW_P): Likewise.
21109 (VCMPNED_P): Likewise.
21110 (VCMPNEFP_P): Likewise.
21111 (VCMPNEDP_P): Likewise.
21112 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21113 implementation of vec_any_eq to not use AltiVec predicate
21115 (VCMPAEH_P): Likewise.
21116 (VCMPAEW_P): Likewise.
21117 (VCMPAED_P): Likewise.
21118 (VCMPAEFP_P): Likewise.
21119 (VCMPAEDP_P): Likewise.
21120 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21121 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21122 not use the AltiVec predicate framework.
21123 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21124 of vec_any_eq to not use AltiVec predicate framework.
21125 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21126 support for predefined __POWER9_VECTOR__ macro to indicate that
21127 Power9 instruction selection is enabled.
21128 (altivec_overloaded_builtins): Remove extraneous
21129 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21130 function argument types RS6000_BTI_bool_V16QI and
21131 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21132 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21133 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21134 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21135 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21136 Power9 for implementations of vec_cmpne. Change the signature for
21137 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21138 (representing vec_all_ne) to remove the previously described first
21139 argument of type RS6000_BTI_INTSI, as this was an artifact of
21140 reliance on the AltiVec predicate framework, which is no longer
21141 used in the implementation of these functions. Add
21142 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21143 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21144 since, unlike the AltiVec predicate framework implementation, we
21145 do not share function descriptors between vec_alle and vec_anyeq.
21146 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21147 set of modes that receive special treatment even when
21148 TARGET_P9_VECTOR is true. The special treatment emits code that
21149 does not depend on Power9 instructions.
21150 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21151 define_expand to not rely on AltiVec predicate framework.
21152 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21154 (vector_ne_v2di_p): Change this define_expand to not rely on
21155 AltiVec predicate framework.
21156 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21158 (vector_ne_<mode>_p): Change this define_expand to not rely on
21159 AltiVec predicate framework.
21160 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21162 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21163 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21164 define_insn pattern.
21165 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21166 define_insn pattern because the xvcmpne<VSs>. instruction is not
21168 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21169 instruction is not supported.
21171 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21173 * config/nvptx/nvptx.c: Include intl.h.
21175 2017-03-01 Martin Jambor <mjambor@suse.cz>
21178 * ipa-prop.h (ipa_bits): Removed field known.
21179 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21180 to pointers. Adjusted their comments to warn about their sharing.
21181 (ipcp_transformation_summary): Change bits to a vector of pointers.
21182 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21183 (ipa_get_ipa_bits_for_value): Declare.
21184 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21185 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21186 (ipa_bits_hash_table): Likewise.
21187 (ipa_vr_ggc_hash_traits): Likewise.
21188 (ipa_vr_hash_table): Likewise.
21189 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21190 being pointers and vr_known being removed.
21191 (ipa_set_jf_unknown): Likewise.
21192 (ipa_get_ipa_bits_for_value): New function.
21193 (ipa_set_jfunc_bits): Likewise.
21194 (ipa_get_value_range): New overloaded functions.
21195 (ipa_set_jfunc_vr): Likewise.
21196 (ipa_compute_jump_functions_for_edge): Use the above functions to
21197 construct bits and vr parts of jump functions.
21198 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21199 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21201 (ipcp_grow_transformations_if_necessary): Also allocate
21202 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21204 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21205 them. Fix too long lines.
21206 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21207 vr_known being removed.
21208 (ipa_read_jump_function): Use new setter functions to construct bits
21209 and vr parts of jump functions or set them to NULL.
21210 (write_ipcp_transformation_info): Adjust for bits being pointers.
21211 (read_ipcp_transformation_info): Likewise.
21212 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21214 Include gt-ipa-prop.h.
21215 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21217 (ipcp_store_bits_results): Likewise.
21218 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21219 Do not write to existing jump functions but use a temporary instead.
21221 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21224 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21225 attempt to use its first operand as BIT_FIELD_REF base.
21227 2017-03-01 Richard Biener <rguenther@suse.de>
21229 PR middle-end/79721
21230 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21231 interpolating formula in wrapping arithmetic.
21232 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21234 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21236 PR tree-optimization/79734
21237 * tree-vect-generic.c (expand_vector_condition): Optimize
21238 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21239 Handle VEC_COND_EXPR where comparison has different inner width from
21240 type's inner width.
21242 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21244 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21245 markup, and similar issues. Remove @opindex entries for things
21246 that aren't options. Add missing -mmpy-option entries.
21248 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21250 PR tree-optimization/79737
21251 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21252 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21253 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21254 instead of byte_size. Formatting fix.
21255 (shift_bytes_in_array_right): Formatting fix.
21257 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21260 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21261 condition on optimize for the leaf function test.
21263 2017-02-28 Martin Liska <mliska@suse.cz>
21266 * read-rtl-function.c (function_reader::handle_unknown_directive):
21267 Bail out when one uses -flto.
21269 2017-02-28 Martin Liska <mliska@suse.cz>
21271 * common.opt: Replace space with tabular for options of <number>
21273 * config/i386/i386.opt: Show <number> value for
21274 -mlarge-data-threshold.
21275 * opts.c (print_filtered_help): Do not display number in hexadecimal
21278 2017-02-28 Martin Liska <mliska@suse.cz>
21280 * common.opt: Fix --help=option -Q for options which are of
21283 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21285 * config/i386/i386.c (print_reg): Error out for values
21286 of 8-bit size in invalid integer register.
21288 2017-02-28 Martin Sebor <msebor@redhat.com>
21290 PR tree-optimization/79691
21291 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21293 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21296 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21297 gcc_unreachable with output_operand_lossage.
21299 2017-02-28 Richard Biener <rguenther@suse.de>
21301 PR tree-optimization/79740
21302 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21304 (visit_nary_op): Insert the nary into the hashtable if we
21305 pattern-matched sth.
21306 * tree-ssa-pre.c (eliminate_insert): Robustify.
21308 2017-02-28 Richard Biener <rguenther@suse.de>
21310 PR middle-end/79731
21311 * fold-const.c (decode_field_reference): Reject out-of-bound
21314 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21316 * config/i386/i386.c: Include intl.h.
21317 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21318 instead of just cond ? "..." : "...".
21319 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21320 * coverage.c (read_counts_file): Likewise.
21321 * omp-offload.c: Include intl.h.
21322 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21323 of just cond ? "..." : "...".
21324 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21325 of just cond ? "..." : "...".
21327 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21330 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21332 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21333 'tune for' CPU name.
21334 * config/arm/arm-cpu-data.h: Regenerated.
21336 2017-02-28 Richard Biener <rguenther@suse.de>
21338 PR tree-optimization/79732
21339 * tree-inline.c (expand_call_inline): Do not shadow var.
21341 2017-02-28 Richard Biener <rguenther@suse.de>
21343 PR tree-optimization/79723
21344 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21345 address-space properly.
21347 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21349 * doc/optinfo.texi (Optimization groups): Fix option used for
21351 * doc/invoke.texi (-fopt-info): Document "omp".
21352 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21353 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21354 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21355 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21356 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21358 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21360 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21361 OPTGROUP_OMP with "omp".
21363 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21366 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21367 for arithmetic shift of unsigned V2DI.
21369 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21371 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21372 arc/linux.h headers.
21373 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21374 (LINK_SPEC): Likewise.
21375 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21376 (EXTRA_SPECS): Likewise.
21377 (STARTFILE_SPEC): Likewise.
21378 (ENDFILE_SPEC): Likewise.
21379 (LIB_SPEC): Likewise.
21380 (TARGET_SDATA_DEFAULT): Likewise.
21381 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21382 (MULTILIB_DEFAULTS): Likewise.
21383 (DWARF2_UNWIND_INFO): Likewise.
21384 * config/arc/big.h: New file.
21385 * config/arc/elf.h: Likewise.
21386 * config/arc/linux.h: Likewise.
21387 * config/arc/t-uClibc: Remove.
21389 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21391 PR tree-optimization/77536
21392 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21393 (tree_transform_and_unroll_loop): Use above function to compute the
21394 estimated niter of unrolled loop and use it when scaling profile.
21395 Also use count info rather than frequency if it's non-zero.
21396 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21397 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21398 (vect_transform_loop): Call above function.
21400 2017-02-27 Richard Biener <rguenther@suse.de>
21402 PR tree-optimization/45397
21403 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21404 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21405 (visit_nary_op): Add pattern matching for CSEing sign-changed
21406 or truncated operations with wider ones.
21408 2017-02-27 Richard Biener <rguenther@suse.de>
21410 PR tree-optimization/79690
21411 * tree-vect-stmts.c (vectorizable_store): Use vector type
21412 built from the DR with address-space.
21414 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21416 * doc/invoke.texi (Optimize Options): Refine the description
21417 of asan-use-after-return.
21419 2017-02-25 Alan Modra <amodra@gmail.com>
21421 PR rtl-optimization/79584
21422 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21423 base, not ad->base_term, the reg within base. Remove assertion
21424 that ad->base == ad->base_term. Replace gen_int_mode using
21425 bogus mode with const0_rtx.
21427 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21429 PR middle-end/79396
21430 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21431 FMA_EXPR like tcc_binary or tcc_unary.
21433 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21436 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21438 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21439 (output_loc_operands): Handle DW_OP_call_ref and
21440 DW_OP_GNU_variable_value.
21441 (struct variable_value_struct): New type.
21442 (struct variable_value_hasher): Likewise.
21443 (variable_value_hash): New variable.
21444 (string_types): Remove.
21445 (copy_loc_descr): New function.
21446 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21447 (prepend_loc_descr_to_each): New function.
21448 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21449 instead of add_loc_descr_to_each if the first argument is single
21450 location list and the second has multiple.
21451 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21452 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21453 when looking for variable value which doesn't have other location info.
21454 (loc_list_from_tree): Formatting fix.
21455 (gen_array_type_die): Simplify DW_AT_string_length handling.
21456 (adjust_string_types): Remove.
21457 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21458 string_types. Call resolve_variable_values.
21459 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21460 (resolve_addr_in_expr): Likewise. Add A argument.
21461 (copy_deref_exprloc): Remove deref argument. Adjust for the
21462 original expression being DW_OP_GNU_variable_value with optionally
21463 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21464 optionally after it.
21465 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21466 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21467 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21468 (variable_value_hasher::hash, variable_value_hasher::equal): New
21470 (resolve_variable_value_in_expr, resolve_variable_value,
21471 resolve_variable_values, note_variable_value_in_expr,
21472 note_variable_value): New functions.
21473 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21476 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21479 * opts.h (handle_generated_option): Add GENERATED_P argument.
21480 * opts-common.c (handle_option): Adjust function comment.
21481 (handle_generated_option): Add GENERATED_P argument, pass it to
21483 (control_warning_option): Pass false to handle_generated_option
21485 * opts.c (maybe_default_option): Pass true to handle_generated_option
21487 * optc-gen.awk: Likewise.
21489 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21491 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21492 a REG, look at the REG it is a SUBREG of.
21493 (splitter for cmpeqsi_t): Ditto.
21495 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21497 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21498 the special USEs with the pattern of the insn, not the insn itself.
21500 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21503 * doc/invoke.texi: Document -mload-store-pairs.
21505 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21506 Sandra Loosemore <sandra@codesourcery.com>
21508 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21509 argument isn't a CONST_INT.
21510 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21511 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21512 (nios2_validate_compare): Bypass alternate compare logic if *op2
21513 is not a CONST_INT.
21514 (ldstwm_operation_p): Return false if first_base is not a REG or
21515 if first_offset is not a CONST_INT.
21517 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21519 * config/cris/cris.md: Use correct operand in a define_peephole2.
21521 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21523 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21525 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21527 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21528 this_insn if it is an INSN or JUMP_INSN.
21529 (force_offsettable): Look at base, not at addr.
21530 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21531 on things that aren't necessarily CONST_INTs.
21533 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21535 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21536 -mfpmath=sse is the default also for x86-32 targets with SSE2
21537 instruction set when @option{-ffast-math} is enabled
21539 2017-02-24 Jeff Law <law@redhat.com>
21541 PR rtl-optimizatoin/79286
21542 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21545 2017-02-24 Richard Biener <rguenther@suse.de>
21547 PR tree-optimization/79389
21548 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21551 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21553 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21554 function comment to reflect reality.
21555 (loop_exits_before_overflow): Fix typo in function description.
21557 2017-02-24 Richard Biener <rguenther@suse.de>
21559 PR tree-optimization/79389
21560 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21561 properly that a threading opportunity exists. Detect conditional
21562 copy/constant propagation opportunities.
21564 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21566 * config/visium/visium.md (type): Add trap.
21567 (b): New mode attribute.
21568 (*btst): Rename into...
21569 (*btst<mode>): ...this and adjust.
21570 (*cbranchsi4_btst_insn): Rename into...
21571 (*cbranch<mode>4_btst_insn): ...this and adjust.
21572 (trap): New define_insn.
21574 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21576 PR tree-optimization/79389
21577 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21578 (noce_reversed_cond_code): New function.
21579 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21580 reversed_comparison_code. Formatting fix.
21581 (noce_try_store_flag): Test rev_cond != NULL in addition to
21582 reversed_comparison_code.
21583 (noce_try_store_flag_constants): Likewise.
21584 (noce_try_store_flag_mask): Likewise.
21585 (noce_try_addcc): Use rev_cond if non-NULL instead of
21586 reversed_comparison_code.
21587 (noce_try_cmove_arith): Likewise. Formatting fixes.
21588 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21589 (noce_find_if_block): Initialize rev_cond.
21590 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21591 instead of false as last argument never attempt to reverse it
21594 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21596 PR tree-optimization/79663
21597 * tree-predcom.c (combine_chains): Process refs in reverse order
21598 only for ZERO length chains, and add explaining comment.
21600 2017-02-23 Jeff Law <law@redhat.com>
21602 PR tree-optimization/79578
21603 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21604 in call to operand_equal_p.
21606 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21609 * config/i386/cpuid.h: Fix another undefined behavior.
21611 2017-02-23 Richard Biener <rguenther@suse.de>
21613 PR tree-optimization/79683
21614 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21615 vector types for data-refs.
21617 2017-02-23 Martin Liska <mliska@suse.cz>
21619 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21621 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21623 PR middle-end/79665
21624 * internal-fn.c (get_range_pos_neg): Moved to ...
21625 * tree.c (get_range_pos_neg): ... here. No longer static.
21626 * tree.h (get_range_pos_neg): New prototype.
21627 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21628 are known to be in between 0 and signed maximum inclusive, try to
21629 expand both unsigned and signed divmod and use the cheaper one from
21632 2017-02-22 Jeff Law <law@redhat.com>
21634 PR tree-optimization/79578
21635 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21636 to compare base operands.
21638 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21641 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21642 gpc_reg_operand instead of fpr_reg_operand.
21644 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21646 * config/mips/mips.c (mips_return_in_memory): Force FP
21647 vector types to be returned in memory for o32 ABI.
21649 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21651 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21652 instead of DW_TAG_member for static data member declarations and don't
21653 set no_linkage_name for static inline data members.
21654 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21657 2017-02-22 Martin Liska <mliska@suse.cz>
21659 * doc/invoke.texi: Replace inequality signs with square brackets
21662 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21664 PR tree-optimization/68644
21665 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21667 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21670 * lra-constraints.c (simplify_operand_subreg): Handle
21671 WORD_REGISTER_OPERATIONS targets.
21673 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21676 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21677 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21678 elimination by swapping fld*.
21680 2017-02-22 Richard Biener <rguenther@suse.de>
21682 PR tree-optimization/79673
21683 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21684 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21685 irrelevant address-space qualifiers and avoiding a
21686 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21688 2017-02-22 Richard Biener <rguenther@suse.de>
21690 PR tree-optimization/79666
21691 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21692 to not symbolically negate if that may introduce undefined
21695 2017-02-22 Martin Liska <mliska@suse.cz>
21698 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21699 * data-streamer-out.c (streamer_write_gcov_count_stream):
21701 * value-prof.c (stream_out_histogram_value): Make assert more
21702 precise based on type of counter.
21704 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21707 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21708 Use nonimmediate_operand instead of memory_operand for operand 1.
21709 (float-extend standard_x87sse_constant_load splitter): Ditto.
21711 2017-02-21 Jeff Law <law@redhat.com>
21713 PR tree-optimization/79621
21714 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21715 blocks with edges to themselves.
21717 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21720 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21721 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21722 Use gimple_call_builtin_p.
21725 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21726 on temporarily removed DEBUG_INSNs.
21728 PR tree-optimization/79649
21729 * tree-loop-distribution.c (classify_partition): Give up on
21730 non-generic address space loads/stores.
21732 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21734 * doc/loop.texi (Loop manipulation): Remove nonexistent
21735 tree_ssa_loop_version from the documentation.
21736 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21738 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21741 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21742 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21743 * config/rs6000/rs6000.c: Include except.h.
21744 (rs6000_expand_split_stack_prologue): Call
21745 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21747 2017-02-21 Martin Jambor <mjambor@suse.cz>
21750 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21751 have been analyzed.
21753 2017-02-21 Martin Jambor <mjambor@suse.cz>
21755 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21756 for backward compatibility only.
21757 * doc/invoke.texi (Option Summary): Remove all references to
21758 -fipa-cp-alignment.
21760 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21764 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21766 * lra-constraints.c (curr_insn_transform): Handle
21767 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21769 2017-02-21 Martin Liska <mliska@suse.cz>
21771 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21773 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21775 PR translation/79638
21776 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21778 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21781 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21782 (arm_function_ok_for_sibcall): Return false for an indirect call by
21783 descriptor if all the argument registers are used.
21784 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21785 alignment of the function.
21787 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21789 PR tree-optimization/61441
21790 * simplify-rtx.c (simplify_const_unary_operation): For
21791 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21792 the sNaN unmodified.
21794 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21796 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21797 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21798 instead of SYSTEM_HEADER_DIR.
21800 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21801 Martin Liška <mliska@suse.cz>
21803 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21804 Fix typos and grammar, use active voice, and clarify.
21806 2017-02-20 Marek Polacek <polacek@redhat.com>
21808 PR middle-end/79537
21809 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21812 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21814 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21817 * config/i386/i386.c (ix86_expand_builtin): Handle
21818 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21819 ix86_builtins_isa[fcode].isa as a requirement of those
21820 flags and any other flag in the bitmask.
21821 (ix86_init_mmx_sse_builtins): Use 0 instead of
21822 ~OPTION_MASK_ISA_64BIT as mask.
21823 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21824 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21825 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21826 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21828 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21831 * lra-constraints.c (split_reg): Check requested split mode
21832 is supported by the register.
21834 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21836 * lra-constraints.c (simplify_operand_subreg): Remove early
21839 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21842 * lra-constraints.c (curr_insn_transform): Tighten condition
21843 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21845 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21848 * lra-constraints.c (curr_insn_transform): Handle
21849 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21851 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21854 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21856 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21858 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21861 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21864 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21866 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21869 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21871 * final.c (last_columnnum, override_columnnum): New variables.
21872 (final_start_function): Set last_columnnum, pass it to begin_prologue
21873 hook and pass 0 to dwarf2out_begin_prologue.
21874 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21875 to source_line debug hook.
21876 (notice_source_line): Compute last_columnnum and for debug_column_info
21877 return true on column changes.
21878 * debug.h (struct gcc_debug_hooks): Add column argument to
21879 source_line and begin_prologue hooks.
21880 (debug_nothing_int_charstar_int_bool): Remove prototype.
21881 (debug_nothing_int_int_charstar,
21882 debug_nothing_int_int_charstar_int_bool): New prototypes.
21883 (dwarf2out_begin_prologue): Add column argument.
21884 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21885 begin_prologue hooks.
21886 (debug_nothing_int_charstar_int_bool): Remove.
21887 (debug_nothing_int_int_charstar,
21888 debug_nothing_int_int_charstar_int_bool): New functions.
21889 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21890 through to dwarf2out_source_line.
21891 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21892 (dwarf2out_source_line): Add column argument, emit it if requested.
21893 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21895 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21896 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21897 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21898 through to dwarf2out_begin_prologue.
21899 (vmsdbgout_source_line): Add column argument, pass it through to
21900 dwarf2out_source_line.
21901 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21902 dbxout_source_line caller.
21903 (dbxout_source_line): Add column argument.
21905 * common.opt (gno-column-info, gcolumn-info): New options.
21906 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21907 (check_die): Also test for multiple DW_AT_decl_column attributes.
21908 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21909 DW_AT_decl_column if requested.
21910 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21912 (gen_variable_die): Likewise.
21913 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21914 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21917 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21918 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21919 (ix86_handle_option): Handle OPT_m3dnowa.
21920 * doc/invoke.texi (-m3dnowa): Document.
21921 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21922 -m3dnowa instead of -m3dnow -march=athlon.
21925 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21926 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21928 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21931 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21932 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21933 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21934 generator for vsx_xxpermdi_<mode>_be.
21935 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21936 force big-endian semantics.
21937 (vsx_xxpermdi_<mode>_be): New define_expand with same
21938 implementation as previous version of vsx_xxpermdi_<mode>.
21940 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21942 PR tree-optimization/79327
21943 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21944 variable, its initialization and use.
21946 2017-02-17 Julia Koval <julia.koval@intel.com>
21948 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21949 (OPTION_MASK_ISA_PKU_UNSET): New.
21950 (ix86_handle_option): Handle -mrdpid.
21951 * config/i386/cpuid.h (bit_RDPID): New.
21952 * config/i386/driver-i386.c (host_detect_local_cpu):
21953 Detect RDPID feature.
21954 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21955 * config/i386/i386-c.c (ix86_target_macros_internal):
21957 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21958 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21959 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21960 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21961 * config/i386/i386.md (define_insn "rdpid"): New.
21962 * config/i386/i386.opt Add -mrdpid.
21963 * config/i386/immintrin.h (_rdpid_u32): New.
21965 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21967 PR rtl-optimization/79541
21968 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21969 instead of transforming it into USE.
21971 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21973 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21974 If HONOR_SNANS (SFmode) force the input to a register.
21975 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21976 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21977 an frsp or similar insn.
21979 2017-02-17 Martin Liska <mliska@suse.cz>
21981 PR rtl-optimization/79577
21982 * params.def (selsched-max-sched-times): Increase minimum to 1.
21984 2017-02-17 Martin Liska <mliska@suse.cz>
21986 PR rtl-optimization/79574
21987 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21989 2017-02-17 Martin Liska <mliska@suse.cz>
21991 PR tree-optimization/79529
21992 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21993 ssa_defined_default_def_p to handle cases which are implicitly
21995 * tree-ssa.c (ssa_defined_default_def_p): New function.
21996 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21997 which are implicitly defined.
21998 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22000 2017-02-17 Richard Biener <rguenther@suse.de>
22002 PR middle-end/79576
22003 * params.def (max-ssa-name-query-depth): Limit to 10.
22005 2017-02-17 Richard Biener <rguenther@suse.de>
22007 PR tree-optimization/79552
22008 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22011 2017-02-17 Richard Biener <rguenther@suse.de>
22014 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22016 2017-02-17 Marek Polacek <polacek@redhat.com>
22018 PR middle-end/79536
22019 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22020 (fold_negate_expr): New wrapper.
22022 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22024 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22025 Correct terminology and de-emphasize pre-standard behavior.
22027 2017-02-16 Alan Modra <amodra@gmail.com>
22029 PR rtl-optimization/79286
22030 * ira.c (def_dominates_uses): New function.
22031 (update_equiv_regs): Don't create an equivalence for insns that
22032 may trap where the register def does not dominate the use.
22034 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22036 PR rtl-optimization/78127
22037 * lra.c (lra): Call lra_eliminate before finish the loop after
22040 2017-02-16 Richard Biener <rguenther@suse.de>
22042 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22044 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22045 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22046 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22047 (isl_val_int_from_wi): New function.
22048 (extract_affine_gmp): Rename to ...
22049 (extract_affine_wi): ... this, take a widest_int.
22050 (extract_affine_int): Just wrap extract_affine_wi.
22051 (add_param_constraints): Use isl_val_int_from_wi.
22052 (add_loop_constraints): Likewise, and extract_affine_wi.
22054 2017-02-15 Jeff Law <law@redhat.com>
22056 PR middle-end/79521
22057 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22058 ira_init_register_move_cost_if_necessary.
22060 2017-02-15 Martin Sebor <msebor@redhat.com>
22062 PR middle-end/32003
22063 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22064 removed in a prior commit.
22066 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22068 PR tree-optimization/79347
22069 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22070 counters during peeling.
22072 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22074 * Makefile.in (site.exp): Remove "set ISLVER".
22076 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22079 * real.c (real_from_integer): Call real_convert even for decimal.
22081 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22084 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22086 2017-02-14 Andrew Pinski <apinski@cavium.com>
22088 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22089 cores and change the partno/implementer to be correct.
22090 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22091 the 'B" as the implementer.
22092 * config/aarch64/aarch64-tune.md: Regenerate.
22094 2017-02-14 Carl Love <cel@us.ibm.com>
22096 * config/rs6000/rs6000.c: Add case statement entry to make the
22097 xvcvuxdsp built-in argument unsigned.
22098 * config/rs6000/vsx.md: Fix the source and return operand types so they
22099 match the instruction definitions from the ISA document. Fix typo
22100 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22103 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22106 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22107 member early_clobber_alts.
22108 * lra-lives.c (reg_early_clobber_p): New.
22109 (process_bb_lives): Use it.
22110 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22111 (debug_operand_data): Initialize early_clobber_alts.
22112 (setup_operand_alternative): Set up early_clobber_alts.
22113 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22114 alternatives to new_insn_reg.
22115 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22117 (lra_update_insn_regno_info): Pass the new arg.
22119 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22121 PR middle-end/79505
22122 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22123 (new_oacc_loop_raw): Don't clear already cleared fields.
22126 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22127 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22128 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22130 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22133 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22135 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22138 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22139 the extra instruction to the right place to store 128-bit constant
22142 2017-02-14 Martin Sebor <msebor@redhat.com>
22144 PR middle-end/79448
22145 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22146 warning for strings of unknown length.
22148 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22150 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22152 2017-02-14 Jeff Law <law@redhat.com>
22155 * ira-costs.c (scan_one_insn): Initialize register move costs
22156 for pseudos seen in USE/CLOBBER insns.
22158 PR tree-optimization/79095
22159 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22160 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22161 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22162 if the operands are known to be not equal, then the resulting range
22164 (intersect_ranges): If the new range is ~[0,0] and the old range is
22165 wide, then prefer ~[0,0].
22166 * tree-vrp.c (overflow_comparison_p_1): New function.
22167 (overflow_comparison_p): New function.
22168 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22169 if NAME is used in an overflow test.
22170 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22171 overflow check that can be expressed as an equality test, then adjust
22172 ops to be that equality test.
22174 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22176 * config/s390/s390-builtin-types.def: Remove flags argument.
22177 * config/s390/s390.c (s390_init_builtins): Likewise.
22179 2017-02-14 Martin Liska <mliska@suse.cz>
22181 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22182 vector. Fix trailing white spaces.
22184 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22186 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22189 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22191 PR rtl-optimization/68664
22192 * config/arm/arm.c (arm_sched_can_speculate_insn):
22193 New function. Declare prototype.
22194 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22196 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22198 PR rtl-optimization/68664
22199 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22201 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22203 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22205 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22206 max skip bytes for function, loop and jump.
22208 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22210 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22211 ABS_EXPR for gimple dump.
22213 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22216 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22218 PR tree-optimization/79408
22219 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22220 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22221 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22222 also if rhs1 is INTEGER_CST.
22224 2017-02-14 Richard Biener <rguenther@suse.de>
22226 PR middle-end/79432
22227 * tree-into-ssa.c (insert_phi_nodes): When the function can
22228 have abnormal edges rewrite SSA names with broken use-def
22229 dominance out of SSA and register them for PHI insertion.
22231 2017-02-13 Martin Sebor <msebor@redhat.com>
22233 PR middle-end/79496
22234 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22235 clearing info.nowrite flag when snprintf size argument is a range.
22237 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22239 * cprop.c (cprop_jump): Add missing space in string literal.
22240 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22241 (get_constraint_for_component_ref): Likewise.
22242 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22243 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22244 * lra-constraints.c (process_alt_operands): Likewise.
22245 * ipa-inline.c (inline_small_functions): Likewise.
22246 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22247 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22248 * trans-mem.c (diagnose_tm_1_op): Likewise.
22249 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22250 (grid_parallel_clauses_gridifiable): Likewise.
22252 * config/nvptx/mkoffload.c (process): Add space in between
22255 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22256 "MOD4_SSE_REGS" and "ALL_REGS".
22258 * spellcheck.c (test_data): Add , in between "foo" and "food".
22260 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22263 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22264 boundary crossing check and subsequent code generation agree.
22266 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22268 * config/aarch64/aarch64.c (has_memory_op): Delete.
22269 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22272 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22274 PR rtl-optimization/79388
22275 PR rtl-optimization/79450
22276 * combine.c (distribute_notes): When removing TEM_INSN for which
22277 corresponding dest has last value recorded, invalidate that last
22280 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22282 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22283 of explicit '@'. Add missing assembly comment marker on branch costs
22286 2017-02-13 Nathan Sidwell <nathan@acm.org>
22288 * gengtype-lex.l (<in_struct>): Add '/'.
22290 2017-02-13 Martin Liska <mliska@suse.cz>
22293 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22295 2017-02-13 Richard Biener <rguenther@suse.de>
22297 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22299 * configure: Re-generate.
22300 * config.in: Likewise.
22301 * graphite-dependences.c: Simplify as if
22302 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22303 * graphite-isl-ast-to-gimple.c: Likewise.
22304 * graphite-optimize-isl.c: Likewise.
22305 * graphite-poly.c: Likewise.
22306 * graphite-sese-to-poly.c: Likewise.
22307 * graphite.h: Likewise.
22308 * toplev.c: Include isl/version.h and use isl_version () for
22309 printing the ISL version.
22310 * doc/install.texi: Update ISL requirement.
22312 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22314 * doc/standards.texi (Standards): Update reference to
22317 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22319 * doc/extend.texi (Named Address Spaces): sourceware.org now
22321 * doc/install.texi (Binaries): Ditto.
22324 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22326 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22327 standard terminology "stringize"/"stringizing" throughout.
22328 * doc/cppinternals.texi: Likewise.
22330 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22332 * doc/extend.texi: Fix some spelling mistakes and typos.
22333 * doc/invoke.texi: Likewise.
22335 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22338 * params.def (inline-min-speedup) Change from 10 to 8.
22340 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22342 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22345 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22348 * ipa-inline-analysis.c (get_minimal_bb): New function.
22349 (record_modified): Use it.
22350 (remap_edge_change_prob): Handle also ancestor functions.
22352 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22354 * doc/contrib.texi (Contributors): Remove broken link into
22355 the Mauve CVS repository.
22357 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22359 PR middle-end/79454
22360 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22361 result computation whenever lhs doesn't have vector mode, not
22362 just when it has BLKmode.
22364 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22366 * doc/makefile.texi (profiledbootstrap): Refer to the
22367 installation instructions only in textual form.
22369 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22372 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22374 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22376 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22377 (Specific): Update mingw-w64 reference.
22379 (Specific): Remove broken link to Renesas RX processor.
22381 2017-02-10 Richard Biener <rguenther@suse.de>
22383 * toplev.c (process_options): Do not mention obsolete graphite
22384 options when printing sorry message about missing graphite support.
22385 Mention -floop-nest-optimize.
22387 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22389 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22390 (vtst_p16): Likewise.
22391 (vtstq_p8): Likewise.
22392 (vtstq_p16): Likewise.
22394 (vtstq_p64): Likewise.
22395 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22396 (vset_lane_p64): New.
22397 (vsetq_lane_p64): New.
22399 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22401 PR tree-optimization/79411
22402 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22403 stmt operands are SSA_NAMEs used in abnormal phis.
22404 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22407 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22410 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22413 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22415 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22417 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22419 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22423 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22424 not arbitrary TREE_CONSTANT.
22427 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22428 "omp declare target link" attribute unless is_global_var.
22429 * omp-offload.c (find_link_var_op): Likewise.
22431 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22432 Chung-Lin Tang <cltang@codesourcery.com>
22434 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22436 (gimplify_adjust_omp_clauses): Don't delete TILE.
22437 (gimplify_omp_for): Deal with TILE.
22438 * internal-fn.c (expand_GOACC_TILE): New function.
22439 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22441 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22442 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22444 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22445 avoid DIV for outermost collapse var.
22446 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22447 Remove out of date comments, fix whitespace.
22448 * omp-general.c (omp_extract_for_data): Deal with tiling.
22449 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22450 adjust OLF_DIM_BASE value.
22451 (struct omp_for_data): Add tiling field.
22452 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22453 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22454 for auto loops. Remove default auto determining, moved to
22455 oacc_loop_fixed_partitions.
22456 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22457 stmts, add e_mask field.
22458 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22459 (oacc_thread_numbers): Use oacc_dim_call.
22460 (oacc_xform_tile): New.
22461 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22462 (finish_oacc_loop): Adjust for ifns vector.
22463 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22464 add case for GOACC_TILE fns.
22465 (oacc_loop_xform_loop): Delete.
22466 (oacc_loop_process): Iterate over call list directly, and add
22467 handling for GOACC_TILE fns.
22468 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22470 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22471 vector partitioning to outer loops. Assign 2 partitions to loops
22472 when available. Add TILE handling.
22473 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22474 (execite_oacc_device_lower): Process GOACC_TILE fns,
22475 ignore unknown specs.
22476 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22477 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22478 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22480 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22482 * configure.ac (ACX_BUGURL): Update.
22483 * configure: Regenerate.
22485 2017-02-09 Richard Biener <rguenther@suse.de>
22487 PR tree-optimization/69823
22488 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22489 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22491 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22493 * config/arc/arc-c.def: Add __NPS400__ definition.
22494 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22495 (TARGET_NPS400): Define.
22497 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22499 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22501 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22502 pointer, arch_info.
22503 (arc_cpu_types): Fill the arch_info field with a pointer into the
22504 arc_arch_types table.
22505 (arc_selected_cpu): Declare.
22506 * config/arc/arc.c (arc_selected_cpu): Make global.
22507 (arc_selected_arch): Delete.
22508 (arc_base_cpu): Delete.
22509 (arc_override_options): Remove references to deleted variables,
22510 update access to arch information.
22511 (ARC_OPT): Update access to arch information.
22512 (ARC_OPTX): Likewise.
22513 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22514 (TARGET_ARC600): Update access to arch information.
22515 (TARGET_ARC601): Likewise.
22516 (TARGET_ARC700): Likewise.
22517 (TARGET_EM): Likewise.
22518 (TARGET_HS): Likewise.
22519 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22522 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22525 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22526 condition/operands for integer GE/LE/GEU/LEU operations.
22528 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22530 PR translation/79397
22531 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22534 2017-02-08 Martin Jambor <mjambor@suse.cz>
22537 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22538 whether allocation happened.
22539 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22540 nothing was allocated.
22542 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22544 PR tree-optimization/79408
22545 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22546 constant, but SSA_NAME with a known integer range, use the minimum
22547 of that range instead of op1 to determine if modulo can be replaced
22548 with its first operand.
22550 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22552 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22554 2017-02-08 Richard Biener <rguenther@suse.de>
22556 PR tree-optimization/71824
22557 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22558 Check all loops contained in the merged region.
22560 2017-02-07 Andrew Pinski <apinski@cavium.com>
22562 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22564 2017-02-07 Andrew Pinski <apinski@cavium.com>
22566 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22567 (thunderxt88): Likewise.
22568 (thunderxt81): Disable LSE and change v8.1 to v8.
22569 (thunderxt83): Likewise.
22571 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22572 Richard Biener <rguenther@suse.de>
22574 PR middle-end/79399
22575 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22576 type from int to size_t.
22577 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22579 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22581 PR rtl-optimization/79386
22582 * cprop.c (bypass_conditional_jumps): Initialize
22583 bypass_last_basic_block already before splitting bbs after
22584 unconditional traps...
22585 (bypass_conditional_jumps): ... rather than here.
22588 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22589 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22590 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22591 fix -masm=intel patterns.
22593 2017-02-07 Richard Biener <rguenther@suse.de>
22595 PR tree-optimization/79256
22596 PR middle-end/79278
22597 * builtins.c (get_object_alignment_2): Use min_align_of_type
22598 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22599 and ADJUST_FIELD_ALIGN.
22601 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22603 * doc/tm.texi: Regenerate.
22604 * stor-layout.c (layout_decl): Adjust.
22605 (update_alignment_for_field): Likewise.
22606 (place_field): Likewise.
22607 (min_align_of_type): Likewise.
22608 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22609 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22610 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22611 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22612 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22613 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22614 * config/i386/i386.c (x86_field_alignment): Likewise.
22615 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22616 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22617 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22618 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22619 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22620 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22624 2017-01-30 Richard Biener <rguenther@suse.de>
22626 PR tree-optimization/79256
22627 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22628 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22631 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22633 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22634 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22635 builtins to SImode and emit a zero-extend, if necessary.
22637 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22639 * docs/invoke.texi (RISC-V Options): Alphabetize.
22641 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22643 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22646 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22648 * config/riscv/riscv.c: New file.
22649 * common/config/riscv/riscv-common.c: Likewise.
22650 * config.gcc: Likewise.
22651 * config/riscv/constraints.md: Likewise.
22652 * config/riscv/elf.h: Likewise.
22653 * config/riscv/generic.md: Likewise.
22654 * config/riscv/linux.h: Likewise.
22655 * config/riscv/multilib-generator: Likewise.
22656 * config/riscv/peephole.md: Likewise.
22657 * config/riscv/pic.md: Likewise.
22658 * config/riscv/predicates.md: Likewise.
22659 * config/riscv/riscv-builtins.c: Likewise.
22660 * config/riscv/riscv-c.c: Likewise.
22661 * config/riscv/riscv-ftypes.def: Likewise.
22662 * config/riscv/riscv-modes.def: Likewise.
22663 * config/riscv/riscv-opts.h: Likewise.
22664 * config/riscv/riscv-protos.h: Likewise.
22665 * config/riscv/riscv.h: Likewise.
22666 * config/riscv/riscv.md: Likewise.
22667 * config/riscv/riscv.opt: Likewise.
22668 * config/riscv/sync.md: Likewise.
22669 * config/riscv/t-elf-multilib: Likewise.
22670 * config/riscv/t-linux: Likewise.
22671 * config/riscv/t-linux-multilib: Likewise.
22672 * config/riscv/t-riscv: Likewise.
22673 * configure.ac: Likewise.
22674 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22675 Waterman as RISC-V maintainers.
22676 * doc/install.texi: Add RISC-V entries.
22677 * doc/invoke.texi: Add RISC-V options section.
22678 * doc/md.texi: Add RISC-V constraints section.
22679 * configure: Regenerated.
22681 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22684 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22685 false values to be constant vectors with all 0 or all 1 bits set.
22686 (vcondu<mode><mode>): Likewise.
22687 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22689 (fpmask_comparison_operator): Update comment.
22690 (vecint_comparison_operator): New predicate.
22691 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22692 vector conditionals when the true and false values are constant
22693 vectors with all 0 bits or all 1 bits set.
22695 2017-02-06 Martin Sebor <msebor@redhat.com>
22697 PR tree-optimization/79376
22698 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22700 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22702 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22703 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22704 to simplify split condition.
22706 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22708 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22709 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22712 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22714 PR rtl-optimization/68664
22715 * target.def (can_speculate_insn): New hook.
22716 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22717 * doc/tm.texi: Regenerate.
22718 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22719 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22720 (rs6000_sched_can_speculate_insn): New function.
22722 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22724 PR tree-optimization/79284
22725 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22726 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22727 vectorizable_mask_load_store, vectorizable_operation,
22728 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22729 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22730 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22731 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22732 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22733 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22734 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22735 is_gimple_assign (stmt). Replace another such test with
22736 is_gimple_assign (stmt).
22738 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22741 * config/avr/avr.c (rtl-iter.h): Include it.
22742 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22743 (avr_legitimate_combined_insn): ...and implementation.
22745 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22747 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22748 * config/s390/s390.c (s390_const_operand_ok)
22749 (s390_canonicalize_comparison, s390_extract_part)
22750 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22751 (s390_contiguous_bitmask_p, s390_rtx_costs)
22752 (legitimize_pic_address): Likewise.
22753 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22754 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22755 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22756 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22757 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22759 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22761 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22762 REGNO($0) == REGNO($1).
22764 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22766 * config/s390/linux.h(SIZE_TYPE): Add comment.
22768 2017-02-06 Julian Brown <julian@codesourcery.com>
22769 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22770 Virendra Pathak <virendra.pathak@broadcom.com>
22772 * config/aarch64/aarch64-cores.def: Change the scheduler
22774 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22775 * config/aarch64/thunderx2t99.md: New file.
22777 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22779 * doc/standards.texi (Go Language): Update link to language
22782 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22784 * tree-eh.c (lower_resx): Sanitize profile.
22785 (cleanup_empty_eh_move_lp): Likewise.
22787 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22790 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22792 * cfgloopmanip.h (loop_version): Update prototype.
22793 * modulo-sched.c (sms_schedule): Update call of loop_version.
22794 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22795 * tree-parloops.c (gen_parallel_loop): Likewise.
22796 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22797 * tree-ssa-loop-split.c (split_loop): Likewise.
22798 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22799 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22801 2017-02-05 Martin Liska <mliska@suse.cz>
22804 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22806 (print_operand_address): Initialize a struct to zero.
22808 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22810 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22811 garbage collector only in textual form.
22813 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22815 * doc/extend.texi (x86 specific memory model extensions for
22816 transactional memory): Simplify a phrase.
22818 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22821 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22822 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22823 (atomic_storedi_1): Likewise.
22825 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22827 PR tree-optimization/79338
22828 * tree-parloops.c (gather_scalar_reductions): Don't call
22829 vect_analyze_loop_form for loop->inner before destroying loop's
22832 2017-02-03 Martin Sebor <msebor@redhat.com>
22834 PR tree-optimization/79327
22835 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22836 when precision has resulted in leading zeros.
22837 (format_integer): Adjust the likely counter to assume an unknown
22838 argument that may be zero is non-zero.
22840 2017-02-03 Jason Merrill <jason@redhat.com>
22843 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22844 avoid copying non-taken branch.
22846 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22848 PR tree-optimization/79340
22849 * tree-vect-loop.c (vectorizable_reduction): Release
22850 vec_defs elements after safe_splicing them into other vectors.
22853 PR tree-optimization/79327
22854 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22855 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22857 (format_integer): Use wide_int_to_tree instead of build_int_cst
22858 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22859 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22860 of shortest and longest sequence.
22862 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22864 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22865 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22867 2017-02-03 Walter Lee <walt@tilera.com>
22870 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22871 after initial stackframe link reg save.
22872 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22874 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22877 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22878 wu for stxssp alternative.
22880 2017-02-03 Martin Sebor <msebor@redhat.com>
22882 PR tree-optimization/79352
22883 * gimple-fold.c (get_range_strlen): Add argument.
22884 (get_range_strlen): Change return type to bool.
22885 (get_maxval_strlen): Pass in a dummy argument.
22886 * gimple-fold.h (get_range_strlen): Change return type to bool.
22887 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22888 * tree.h (array_at_struct_end_p): Add argument.
22889 * tree.c (array_at_struct_end_p): Handle it.
22891 2017-02-03 Martin Liska <mliska@suse.cz>
22894 * multiple_target.c (create_dispatcher_calls): Redirect edge
22895 from a caller of a dispatcher.
22896 (expand_target_clones): Make the clones local.
22897 (ipa_target_clone): Do both target clones and resolvers.
22898 (ipa_dispatcher_calls): Remove the pass.
22899 (pass_dispatcher_calls::gate): Likewise.
22900 (make_pass_dispatcher_calls): Likewise.
22901 * passes.def (pass_target_clone): Put as very first IPA early
22904 2017-02-03 Martin Liska <mliska@suse.cz>
22906 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22907 in case of a function with ifunc attribute.
22909 2017-02-03 Martin Liska <mliska@suse.cz>
22911 * cgraph.c (cgraph_node::dump): Dump function version info.
22912 * symtab.c (symtab_node::dump_base): Add missing new line.
22914 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22916 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22917 (ifcombine_ifandif): Use it.
22919 2017-02-03 Martin Liska <mliska@suse.cz>
22921 * doc/invoke.texi: Document default value for
22922 use-after-scope-direct-emission-threshold.
22924 2017-02-03 Martin Liska <mliska@suse.cz>
22926 PR tree-optimization/79339
22927 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22928 (format_floating): Likewise.
22930 2017-02-03 Martin Liska <mliska@suse.cz>
22933 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22935 (ipa_node_params_t::remove): Likewise.
22936 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22937 initialization from removed ipa_node_params_t::insert.
22938 (ipa_node_params::~ipa_node_params): Move from removed
22939 ipa_node_params_t::release.
22940 * symbol-summary.h (symbol_summary::m_released): New member.
22941 Do not release a summary twice. Do not allow to call finalizer
22942 for types of a summary that live in GGC memory.
22944 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22946 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22949 2017-02-02 Martin Sebor <msebor@redhat.com>
22951 PR middle-end/79275
22952 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22953 (format_string): Tighten up the range of output for non-constant
22954 strings and correct the expected range for wide non-constant strings.
22956 2017-02-02 Martin Sebor <msebor@redhat.com>
22958 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22960 PR middle-end/32003
22961 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22963 (-fdump-tree-@var): Add to index and document how to come up
22964 with pass-specific option and dump file names.
22965 (-fdump-passes): Clarify where to look for output.
22967 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22969 PR middle-end/77445
22970 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22971 statistics of the analyzed path; allow threading for speed when
22972 any of BBs along the path are optimized for speed.
22974 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22976 PR middle-end/78468
22977 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22978 settings of the virtual registers.
22981 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22983 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22984 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22987 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22989 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22990 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22992 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22994 * config/s390/s390.md: Add missing comments with the expanded
22996 * config/s390/vector.md: Likewise.
22997 * config/s390/vx-builtins.md: Likewise.
22999 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23002 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23003 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23004 conditions on a single line.
23006 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23008 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23009 __S390_VX__ to __VX__.
23011 2017-02-01 Andrew Pinski <apinski@cavium.com>
23013 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23014 stmt_info to record_stmt_cost.
23015 (vect_get_known_peeling_cost): Pass stmt_info if known to
23017 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23018 cpu_vector_cost field into
23019 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23020 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23021 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23022 splitting of scalar_stmt_cost and vec_stmt_cost.
23023 (thunderx_vector_cost): Likewise.
23024 (cortexa57_vector_cost): LIkewise.
23025 (exynosm1_vector_cost): Likewise.
23026 (xgene1_vector_cost): Likewise.
23027 (thunderx2t99_vector_cost): Improve after the splitting of the two
23029 (aarch64_builtin_vectorization_cost): Update for the splitting of
23030 scalar_stmt_cost and vec_stmt_cost.
23032 2017-02-01 Torvald Riegel <triegel@redhat.com>
23033 Richard Henderson <rth@redhat.com>
23035 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23036 conditional on existance of a fast atomic load.
23037 * optabs-query.c (can_atomic_load_p): New function.
23038 * optabs-query.h (can_atomic_load_p): Declare it.
23039 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23040 no fast atomic load is available for the particular size of access.
23041 (expand_atomic_compare_and_swap): Likewise.
23042 (expand_atomic_load): Likewise.
23043 (expand_atomic_store): Likewise.
23044 (expand_atomic_fetch_op): Likewise.
23045 * testsuite/lib/target-supports.exp
23046 (check_effective_target_sync_int_128): Remove x86 because it provides
23047 no fast atomic load.
23048 (check_effective_target_sync_int_128_runtime): Likewise.
23050 2017-02-01 Richard Biener <rguenther@suse.de>
23052 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23053 (graphite_transform_loops): Provide opt-info for optimized nests.
23054 * tree-parloop.c (parallelize_loops): Provide opt-info for
23055 parallelized loops.
23057 2017-02-01 Richard Biener <rguenther@suse.de>
23059 PR middle-end/79315
23060 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23061 was not set before.
23063 2017-02-01 Richard Biener <rguenther@suse.de>
23065 PR tree-optimization/71824
23066 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23067 Verify the loops are valid in the merged SESE region.
23068 (scop_detection::can_represent_loop_1): Check analyzing the
23069 evolution of the number of iterations in the region succeeds.
23071 2017-01-31 Ian Lance Taylor <iant@golang.org>
23073 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23074 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23076 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23078 PR preprocessor/79210
23079 * input.c (get_substring_ranges_for_loc): Replace line_width
23080 assertion with error-handling.
23082 2017-01-31 Richard Biener <rguenther@suse.de>
23084 PR tree-optimization/77318
23085 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23086 (create_pw_aff_from_tree): Take loop parameter.
23087 (add_condition_to_pbb): Pass loop of the condition to
23088 create_pw_aff_from_tree.
23090 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23092 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23093 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23095 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23099 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23101 (convert_int_to_float128): Likewise.
23102 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23103 (convert_int_to_float128): Likewise.
23104 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23105 (UNSPEC_IEEE128_CONVERT): Likewise.
23106 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23107 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23108 Use local variables for IBM extended format.
23109 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23110 (fix_trunc<mode>si2_fprs): Likewise.
23111 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23112 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23113 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23114 to know that we can now have integers of all sizes in vector
23116 (fix<uns>_<mode>di2_hw): Likewise.
23117 (float<uns>_<mode>si2_hw): Likewise.
23118 (fix_<mode>si2_hw): Likewise.
23119 (fixuns_<mode>si2_hw): Likewise.
23120 (float<uns>_<mode>di2_hw): Likewise.
23121 (float_<mode>di2_hw): Likewise.
23122 (float_<mode>si2_hw): Likewise.
23123 (floatuns_<mode>di2_hw): Likewise.
23124 (floatuns_<mode>si2_hw): Likewise.
23125 (xscvqp<su>wz_<mode>): Delete, no longer used.
23126 (xscvqp<su>dz_<mode>): Likewise.
23127 (xscv<su>dqp_<mode>): Likewise.
23128 (ieee128_mfvsrd_64bit): Likewise.
23129 (ieee128_mfvsrd_32bit): Likewise.
23130 (ieee128_mfvsrwz): Likewise.
23131 (ieee128_mtvsrw): Likewise.
23132 (ieee128_mtvsrd_64bit): Likewise.
23133 (ieee128_mtvsrd_32bit): Likewise.
23135 2017-01-31 Martin Liska <mliska@suse.cz>
23138 * ipa-prop.c (ipa_free_all_node_params): Call release method
23139 instead of ~sumbol_summary to not to trigger double times
23142 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23144 PR tree-optimization/71691
23145 * bitmap.h (class auto_bitmap): New.
23146 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23147 is_maybe_undefined instead of ssa_undefined_value_p.
23149 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23151 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23152 __S390_ARCH_LEVEL__ to __ARCH__.
23154 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23156 PR tree-optimization/79267
23157 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23158 if should_remove_lhs_p is true.
23160 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23163 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23164 (add_alignment_attribute): New.
23165 (base_type_die): Add alignment attribute.
23166 (subrange_type_die): Likewise.
23167 (modified_type_die): Likewise.
23168 (gen_array_type_die): Likewise.
23169 (gen_descr_array_type_die: Likewise.
23170 (gen_enumeration_type_die): Likewise.
23171 (gen_subprogram_die): Likewise.
23172 (gen_variable_die): Likewise.
23173 (gen_field_die): Likewise.
23174 (gen_ptr_to_mbr_type_die): Likewise.
23175 (gen_struct_or_union_type_die): Likewise.
23176 (gen_subroutine_type_die): Likewise.
23177 (gen_typedef_die): Likewise.
23178 (base_type_cmp): Compare alignment attribute.
23180 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23183 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23184 (setb_unsigned) New pattern for setb with CCUNS.
23185 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23186 subfc./subfe sequence to avoid overflow problems. Generate a
23187 shorter sequence with cmpld/setb for power9.
23188 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23189 for generating subfc. instruction.
23190 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23191 now uses this instruction.
23193 2017-01-30 Ian Lance Taylor <iant@google.com>
23196 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23197 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23199 2017-01-30 Martin Sebor <msebor@redhat.com>
23201 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23202 Move constant to the right of a relational operator.
23203 (get_mpfr_format_length, format_character, format_string): Ditto.
23204 (should_warn_p, maybe_warn): Same.
23206 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23208 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23211 * asan.c (get_translation_unit_decl): Remove function.
23212 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23214 2017-01-30 Martin Liska <mliska@suse.cz>
23216 PR gcov-profile/79259
23217 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23218 -fprofile-generate.
23220 2017-01-30 Martin Liska <mliska@suse.cz>
23223 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23224 Initialize variables with NULL value.
23226 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23229 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23231 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23233 2017-01-30 Richard Biener <rguenther@suse.de>
23235 PR tree-optimization/79276
23236 * tree-vrp.c (process_assert_insertions): Properly adjust common
23237 when removing a duplicate.
23239 * gcc.dg/torture/pr79276.c: New testcase.
23241 2017-01-30 Richard Biener <rguenther@suse.de>
23243 PR tree-optimization/79256
23244 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23245 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23247 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23249 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23252 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23253 ("*r<noxa>sbg_<mode>_sll_bitmask")
23254 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23255 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23256 Use contiguous_bitmask_nowrap_operand.
23258 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23261 * config/rs6000/altivec.h (vec_xl): Revise #define.
23262 (vec_xst): Likewise.
23264 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23266 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23268 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23270 PR rtl-optimization/79194
23271 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23272 traps before call to bypass_conditional_jumps.
23274 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23276 PR tree-optimization/71374
23277 * lra-constraints.c (check_conflict_input_operands): New.
23278 (match_reload): Use it.
23280 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23283 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23284 account to calculate conflict_set.
23286 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23288 PR rtl-optimization/78559
23289 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23290 other_insn in combine.
23292 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23294 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23295 uint16_type_node for BT_UINT16.
23297 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23299 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23300 "RTL Tests" to menu.
23301 (GIMPLE Tests): New node.
23302 (RTL Tests): New node.
23304 2017-01-27 Richard Biener <rguenther@suse.de>
23306 PR tree-optimization/79245
23307 * tree-loop-distribution.c (distribute_loop): Apply cost
23308 modeling also to detected patterns.
23310 2017-01-27 Richard Biener <rguenther@suse.de>
23312 PR tree-optimization/71433
23313 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23314 (compare_assert_loc): New function.
23315 (process_assert_insertions): Sort and optimize assert locations
23316 to remove duplicates and push down identical assertions on
23317 edges to their destination block.
23319 2017-01-27 Richard Biener <rguenther@suse.de>
23321 PR tree-optimization/79244
23322 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23323 out SSA names even if abnormal.
23325 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23327 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23328 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23329 instead of MPFR_RNDN.
23331 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23334 * arm.c (arm_option_override): Don't call build_target_option_node
23335 until after doing all option overrides.
23336 (arm_valid_target_attribute_tree): Likewise.
23338 2017-01-27 Martin Liska <mliska@suse.cz>
23340 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23341 for {cd}tors and C++ {cd}tors.
23343 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23345 * config/s390/s390.md ("*setmem_long_and")
23346 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23348 2017-01-26 Martin Sebor <msebor@redhat.com>
23350 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23353 2017-01-26 Martin Sebor <msebor@redhat.com>
23355 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23356 HAVE_DFmode before using XFmode or DFmode.
23357 (parse_directive): Avoid using the z length modifier to avoid
23358 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23360 PR middle-end/78703
23361 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23362 to accept adjustment as an array.
23363 (get_int_range): New function.
23364 (struct directive): Make width and prec arrays.
23365 (directive::set_width, directive::set_precision): Call get_int_range.
23366 (format_integer, format_floating): Handle width and precision ranges.
23367 (format_string, parse_directive): Same.
23369 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23372 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23373 comdat_type_p set, just clone them, but keep the children in the
23377 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23378 which have direct callers with -fvar-tracking-assignments enabled
23380 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23381 inside of type units.
23383 2017-01-26 Martin Sebor <msebor@redhat.com>
23385 PR middle-end/78703
23386 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23388 (struct format_result): Replace number_chars, number_chars_min,
23389 and number_chars_max with a single member of struct result_range.
23391 (format_result::operator+=): Adjust.
23392 (struct fmtresult): Remove bounded. Handle likely and unlikely
23394 (fmtresult::adjust_for_width_or_precision): New function.
23395 (fmtresult:type_max_digits): New function.
23396 (bytes_remaining): Handle likely and unlikely counters.
23397 (min_bytes_remaining): Remove.
23398 (format_percent): Simplify.
23399 (format_integer, format_floating): Set likely and unlikely counters.
23400 (get_string_length, format_character, format_string): Same.
23401 (format_plain, should_warn_p): New function.
23402 (maybe_warn): Call should_warn_p. Update diagnostic messages
23403 and handle those for all directives, including plain strings.
23404 (format_directive): Handle likely and unlikely counters.
23405 Remove unnecessary quoting from diagnostics. Add an informational
23407 (add_bytes): Remove.
23408 (pass_sprintf_length::compute_format_length): Simplify.
23409 (try_substitute_return_value): Handle likely and unlikely counters.
23411 2017-01-26 Carl Love <cel@us.ibm.com>
23413 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23414 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23416 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23419 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23420 endianess for subregs into account.
23421 * lra-constraints.c (lra_constraints): Do risky transformations
23422 always on the first iteration.
23423 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23424 last_call_used_reg_set.
23425 (process_bb_lives): Define and use last_call_used_reg_set.
23426 * lra.c (lra): Always continue after lra_constraints on the first
23429 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23431 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23433 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23435 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23437 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23438 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23439 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23440 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23441 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23442 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23443 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23444 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23445 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23447 2017-01-26 Marek Polacek <polacek@redhat.com>
23450 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23451 for the third operand.
23453 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23455 PR middle-end/79236
23456 * omp-low.c (struct omp_context): Add simt_stmt field.
23457 (scan_omp_for): Return omp_context *.
23458 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23459 context to the _simt_ SIMD stmt.
23460 (lower_omp_for): For combined SIMD with sibling _simt_
23461 SIMD, make sure to use the same decls in _looptemp_
23462 clauses as in the sibling.
23464 2017-01-26 David Sherwood <david.sherwood@arm.com>
23466 PR middle-end/79212
23467 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23470 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23473 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23474 emit fld b; fld a; if possible.
23476 * brig-builtins.def: Update copyright years.
23477 * config/arm/arm_acle_builtins.def: Update copyright years.
23479 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23482 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23483 constraint instead of o for the stxsd instruction.
23485 2017-01-25 Carl Love <cel@us.ibm.com>
23487 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23488 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23490 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23492 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23494 2017-01-25 Richard Biener <rguenther@suse.de>
23496 PR tree-optimization/69264
23497 * target.def (vector_alignment_reachable): Improve documentation.
23498 * doc/tm.texi: Regenerate.
23499 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23501 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23502 earlier changes with respect to TYPE_USER_ALIGN.
23503 (vector_alignment_reachable_p): Likewise. Improve dumping.
23505 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23508 * config/arm/arm.md (xordi3): Force constant operand into a register
23511 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23513 * doc/invoke.texi (-fstore-merging): Correct default optimization
23514 levels at which it is enabled.
23515 (-O): Move -fstore-merging from list to...
23518 2017-01-25 Richard Biener <rguenther@suse.de>
23521 * omp-expand.c: Include debug.h.
23522 (expand_omp_taskreg): Make sure to generate early debug before
23523 outlining anything from a function.
23524 (expand_omp_target): Likewise.
23525 (grid_expand_target_grid_body): Likewise.
23527 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23530 * asan.c (get_translation_unit_decl): New function.
23531 (asan_add_global): Extract modules file name from globals
23532 TRANSLATION_UNIT_DECL name.
23534 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23537 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23538 for long calls with APCS frame and VFP.
23540 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23542 * cfg.c (original_copy_tables_initialized_p): New function.
23543 * cfg.h (original_copy_tables_initialized_p): New decl.
23544 * cfgrtl.c (relink_block_chain): Guard the call to
23545 free_original_copy_tables with a call to
23546 original_copy_tables_initialized_p.
23547 * cgraph.h (symtab_node::native_rtl_p): New decl.
23548 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23549 (symtab_node::needed_p): Don't assert for early assembly output
23550 for __RTL functions.
23551 (cgraph_node::finalize_function): Set "force_output" for __RTL
23553 (cgraph_node::analyze): Bail out early for __RTL functions.
23554 (analyze_functions): Update assertion to support __RTL functions.
23555 (cgraph_node::expand): Bail out early for __RTL functions.
23556 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23558 * function.h (struct function): Update comment for field
23560 * gimple-expr.c: Include "tree-pass.h".
23561 (gimple_has_body_p): Return false for __RTL functions.
23562 * Makefile.in (OBJS): Add run-rtl-passes.o.
23563 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23565 (gcc::pass_manager::get_clean_slate): New accessor.
23566 * passes.c: Include "insn-addr.h".
23567 (should_skip_pass_p): Add logging. Update logic for running
23568 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23569 property-provider override so it is only done for gimple passes.
23571 (skip_pass): New function.
23572 (execute_one_pass): Call skip_pass when skipping passes.
23573 * read-md.c (md_reader::read_char): Support filtering
23574 the input to a subset of line numbers.
23575 (md_reader::md_reader): Initialize fields
23576 m_first_line and m_last_line.
23577 (md_reader::read_file_fragment): New function.
23578 * read-md.h (md_reader::read_file_fragment): New decl.
23579 (md_reader::m_first_line): New field.
23580 (md_reader::m_last_line): New field.
23581 * read-rtl-function.c (function_reader::create_function): Only
23582 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23583 curr_properties. Set DECL_INITIAL to a dummy block.
23584 (read_rtl_function_body_from_file_range): New function.
23585 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23587 * run-rtl-passes.c: New file.
23588 * run-rtl-passes.h: New file.
23590 2017-01-24 Jeff Law <law@redhat.com>
23592 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23595 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23597 PR tree-optimization/79159
23598 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23599 (record_nonwrapping_iv): Improve boundary using above function if no
23600 value range information.
23602 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23603 Martin Jambor <mjambor@suse.cz>
23605 * brig-builtins.def: New file.
23606 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23607 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23608 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23609 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23610 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23611 * builtin-types.def (BT_INT8): New.
23612 (BT_INT16): Likewise.
23613 (BT_UINT8): Likewise.
23614 (BT_UINT16): Likewise.
23615 (BT_FN_ULONG): Likewise.
23616 (BT_FN_UINT_INT): Likewise.
23617 (BT_FN_UINT_ULONG): Likewise.
23618 (BT_FN_UINT_LONG): Likewise.
23619 (BT_FN_UINT_PTR): Likewise.
23620 (BT_FN_ULONG_PTR): Likewise.
23621 (BT_FN_INT8_FLOAT): Likewise.
23622 (BT_FN_INT16_FLOAT): Likewise.
23623 (BT_FN_UINT32_FLOAT): Likewise.
23624 (BT_FN_UINT16_FLOAT): Likewise.
23625 (BT_FN_UINT8_FLOAT): Likewise.
23626 (BT_FN_UINT64_FLOAT): Likewise.
23627 (BT_FN_UINT16_UINT32): Likewise.
23628 (BT_FN_UINT32_UINT16): Likewise.
23629 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23630 (BT_FN_INT_PTR_INT): Likewise.
23631 (BT_FN_UINT_PTR_UINT): Likewise.
23632 (BT_FN_LONG_PTR_LONG): Likewise.
23633 (BT_FN_ULONG_PTR_ULONG): Likewise.
23634 (BT_FN_VOID_UINT64_UINT64): Likewise.
23635 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23636 (BT_FN_INT8_INT8_INT8): Likewise.
23637 (BT_FN_INT16_INT16_INT16): Likewise.
23638 (BT_FN_INT_INT_INT): Likewise.
23639 (BT_FN_UINT_FLOAT_UINT): Likewise.
23640 (BT_FN_FLOAT_UINT_UINT): Likewise.
23641 (BT_FN_ULONG_UINT_UINT): Likewise.
23642 (BT_FN_ULONG_UINT_PTR): Likewise.
23643 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23644 (BT_FN_UINT_UINT_UINT): Likewise.
23645 (BT_FN_VOID_UINT_PTR): Likewise.
23646 (BT_FN_UINT_UINT_PTR: Likewise.
23647 (BT_FN_UINT32_UINT64_PTR): Likewise.
23648 (BT_FN_INT_INT_UINT_UINT): Likewise.
23649 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23650 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23651 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23652 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23653 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23654 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23655 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23656 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23657 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23658 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23659 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23660 * doc/frontends.texi: List BRIG FE.
23661 * doc/install.texi (Testing): Add BRIG tesring requirements.
23662 * doc/invoke.texi (Overall Options): Mention BRIG.
23663 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23665 2017-01-24 Richard Biener <rguenther@suse.de>
23667 PR translation/79208
23668 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23670 2017-01-24 Martin Jambor <mjambor@suse.cz>
23673 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23674 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23675 and known_contexts.
23677 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23679 PR middle-end/79123
23680 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23681 casts from signed to unsigned really don't have a range.
23683 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23685 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23686 GMP_RNDx for compatiblity.
23688 2017-01-24 Martin Liska <mliska@suse.cz>
23691 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23692 that would prevent us to call alloca with -1 as argument.
23694 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23696 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23697 Avoid -Wformat-security warning.
23699 2017-01-23 Andrew Pinski <apinski@cavium.com>
23701 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23704 2017-01-23 Martin Sebor <msebor@redhat.com>
23706 PR middle-end/78703
23707 * gimple-ssa-sprintf.c (warn_level): New global.
23708 (format_integer): Use it here and throughout the rest of the file.
23709 Use the same switch to compute sign as base.
23710 (maybe_warn): New function.
23711 (format_directive): Factor out warnings into maybe_warn.
23712 Add debugging output. Use warn_level.
23713 (add_bytes): Use warn_level.
23714 (pass_sprintf_length::compute_format_length): Add debugging output.
23715 (try_substitute_return_value): Same.
23716 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23718 PR middle-end/78703
23719 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23720 (struct fmtresult, format_integer, format_floating): Adjust.
23721 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23722 (get_string_length, format_string,format_directive): Same.
23723 (pass_sprintf_length::compute_format_length): Same.
23724 (try_substitute_return_value): Simplify slightly.
23726 PR middle-end/78703
23727 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23728 (fmtresult::operator+=): Outlined.
23729 (struct fmtresult): Add ctors.
23730 (struct conversion_spec): Rename...
23731 (struct directive): ...to this. Add and remove data members.
23732 (directive::set_width, directive::set_precision): New functions.
23733 (format_percent): Use fmtresult ctor.
23734 (get_width_and_precision): Remove.
23735 (format_integer): Make naming changes. Avoid computing width and
23737 (format_floating): Same. Adjust indentation.
23738 (format_character, format_none): New functions.
23739 (format_string): Moved character handling to format_character.
23740 (format_directive): Remove arguments, change return type.
23741 (parse_directive): New function.
23742 (pass_sprintf_length::compute_format_length): Move directive
23743 parsing to parse_directive.
23745 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23747 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23748 (assign_assembler_name_if_needed): ... this.
23749 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23750 (assign_assembler_name_if_needed): ... this.
23751 (free_lang_data_in_cgraph): Adjust callers.
23752 * cgraphunit.c (cgraph_node::analyze): Likewise.
23753 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23756 2017-01-23 Richard Biener <rguenther@suse.de>
23758 PR tree-optimization/79088
23759 PR tree-optimization/79188
23760 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23761 resetting loop bounds after last path deletion. Reset loop
23762 bounds of the target loop, make code match the comments.
23763 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23764 Make sure loops need no fixups.
23766 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23768 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23769 exponent support with double type for first argument.
23770 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23771 type returned by __builtin_vec_extract_sig,
23772 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23773 functions from "vector int" to "vector unsigned int" or from
23774 "vector long long int" to "vector unsigned long long int".
23775 Changed type returned by __builtin_vec_extract_exp,
23776 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23777 functions from "vector int" to "vector unsigned int" or from
23778 "vector long long int" to "vector unsigned long long int".
23779 Changed return type of __builtin_vec_test_data_class,
23780 __builtin_vec_test_data_class_sp, and
23781 __builtin_vec_test_data_class_dp from "vector int" to
23782 "vector bool int" or from "vector long long int" to "vector bool
23783 long long int" and changed second argument type from "unsigned
23784 int" to "int". Added new overloaded function forms "vector float
23785 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23786 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23787 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23788 double, vector unsigned long long int)" and "vector double
23789 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23790 long int)". Changed return type of
23791 __builtin_scalar_test_data_class and
23792 __builtin_scalar_test_data_class_sp and
23793 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23794 int" and changed second argument from "unsigned int" to "int".
23795 Changed type returned by __builtin_scalar_test_neg,
23796 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23797 from "int" to "bool int". Added new overloaded function form
23798 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23799 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23800 exponent double-precision with floating point first argument.
23801 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23802 documentation of scalar_test_data_class, scalar_test_neg,
23803 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23804 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23805 vec_test_data_class built-in functions to reflect refinements in
23806 their type signatures.
23808 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23810 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23812 (aarch64_elf_asm_destructor): Likewise.
23814 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23816 PR rtl-optimization/78634
23817 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23818 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23819 * ifcvt.c (noce_try_cmove): Add missing cost check.
23821 PR rtl-optimization/71724
23822 * combine.c (if_then_else_cond): Look for situations where it is
23823 beneficial to undo the work of one of the recursive calls.
23825 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23827 PR tree-optimization/70754
23828 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23829 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23830 combined stmt before it if not NULL.
23831 (combine_chains): Process refs reversely and compute dominance point
23834 2017-01-23 Martin Liska <mliska@suse.cz>
23836 PR tree-optimization/79196
23837 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23838 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23840 (strlen_optimize_stmt): Call the renamed function.
23842 2017-01-23 Michael Matz <matz@suse.de>
23844 PR tree-optimization/78384
23845 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23847 2017-01-23 Richard Biener <rguenther@suse.de>
23849 PR tree-optimization/79186
23850 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23851 both incoming edges before moving an assert.
23853 2017-01-23 Martin Jambor <mjambor@suse.cz>
23855 * ipa-prop.c (load_from_param_1): Removed.
23856 (load_from_unmodified_param): Bits from load_from_param_1 put back
23858 (load_from_param): Removed.
23859 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23860 with stmt. Reverted back to use of load_from_unmodified_param.
23862 2017-01-23 Martin Jambor <mjambor@suse.cz>
23865 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23866 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23867 field a pointer to garbage collected vector, mark lattices and
23868 ipcp_orig_node with GTY((skip)).
23869 (ipa_get_param_count): Adjust to descriptors being a pointer.
23870 (ipa_get_param): Likewise.
23871 (ipa_get_type): Likewise.
23872 (ipa_get_param_move_cost): Likewise.
23873 (ipa_set_param_used): Likewise.
23874 (ipa_get_controlled_uses): Likewise.
23875 (ipa_set_controlled_uses): Likewise.
23876 (ipa_is_param_used): Likewise.
23877 (ipa_node_params_t): Move into garbage collector. New methods insert
23879 (ipa_node_params_sum): Annotate wth GTY(()).
23880 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23882 (ipa_load_from_parm_agg): Adjust declaration.
23883 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23884 * ipa-profile.c (ipa_profile): Likewise.
23885 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23886 (ipa_populate_param_decls): Make descriptors parameter garbage
23888 (ipa_dump_param): Adjust to descriptors being a pointer.
23889 (ipa_alloc_node_params): Likewise.
23890 (ipa_initialize_node_params): Likewise.
23891 (load_from_param_1): Make descriptors parameter garbage collected.
23892 (load_from_unmodified_param): Likewise.
23893 (load_from_param): Likewise.
23894 (ipa_load_from_parm_agg): Likewise.
23895 (ipa_node_params::~ipa_node_params): Removed.
23896 (ipa_free_all_node_params): Remove call to delete operator.
23897 (ipa_node_params_t::insert): New.
23898 (ipa_node_params_t::remove): Likewise.
23899 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23900 copy known_csts and known_contexts vectors.
23901 (ipa_read_node_info): Adjust to descriptors being a pointer.
23902 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23904 (ipcp_transform_function): Make descriptors variable garbage
23907 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23909 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23910 * config/i386/avx512dqintrin.h: Ditto.
23911 * config/i386/avx512fintrin.h: Ditto.
23912 * config/i386/i386.c: Handle new builtins.
23913 * config/i386/i386-builtin.def: Add new builtins.
23914 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23915 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23917 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23918 Martin Liska <mliska@suse.cz>
23920 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23921 * asan.c (asan_expand_poison_ifn): Support stores and use
23922 appropriate ASAN report function.
23923 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23924 * internal-fn.def (ASAN_POISON_USE): Declare.
23925 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23926 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23927 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23928 ASAN_POISON calls w/o LHS.
23929 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23930 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23931 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23932 * gimplify.c (asan_poison_variables): Add attribute
23933 use_after_scope_memory to variables that really needs to live
23935 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23936 having the attribute.
23938 2017-01-23 Martin Liska <mliska@suse.cz>
23940 * asan.c (create_asan_shadow_var): New function.
23941 (asan_expand_poison_ifn): Likewise.
23942 * asan.h (asan_expand_poison_ifn): New declaration.
23943 * internal-fn.c (expand_ASAN_POISON): Likewise.
23944 * internal-fn.def (ASAN_POISON): New builtin.
23945 * sanopt.c (pass_sanopt::execute): Expand
23946 asan_expand_poison_ifn.
23947 * tree-inline.c (copy_decl_for_dup_finish): Make function
23949 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23950 * tree-ssa.c (is_asan_mark_p): New function.
23951 (execute_update_addresses_taken): Rewrite local variables
23952 (identified just by use-after-scope as addressable) into SSA.
23954 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23956 * doc/install.texi (Specific): opensource.apple.com uses https
23957 now. Remove trailing slash.
23959 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23961 * README.Portability: Remove note on an Irix compatibility issue.
23963 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23965 * gcov.c (INCLUDE_ALGORITHM): Define.
23966 (INCLUDE_VECTOR): Define.
23967 No longer include <vector> and <algorithm> directly.
23969 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23971 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23973 * doc/invoke.texi (Code Gen Options): Ditto.
23975 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23978 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23980 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23982 rtl-optimization/79125
23983 * cprop.c (local_cprop_pass): Handle cases where we make an
23984 unconditional trap.
23986 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23990 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23991 read from, for big endian.
23993 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23995 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23996 register pauth builtins for LP64 only.
23998 2017-01-20 Marek Polacek <polacek@redhat.com>
24001 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24004 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24006 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24008 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24009 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24010 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24012 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24015 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24016 in big-endian mode when they are not a single duplicated value.
24018 2017-01-20 Richard Biener <rguenther@suse.de>
24020 * BASE-VER: Bump to 7.0.1.
24022 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24024 * omp-low.c (omplow_simd_context): New struct. Use it...
24025 (lower_rec_simd_input_clauses): ...here and...
24026 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24027 references to idx, lane, max_vf, is_simt.
24029 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24031 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24034 2017-01-20 Martin Jambor <mjambor@suse.cz>
24036 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24037 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24039 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24040 (GTFILES): Rename hsa.c to hsa-common.c.
24041 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24042 * hsa-dump.c: Likewise.
24043 * hsa-gen.c: Likewise.
24044 * hsa-regalloc.c: Likewise.
24045 * ipa-hsa.c: Likewise.
24046 * omp-expand.c: Likewise.
24047 * omp-low.c: Likewise.
24048 * toplev.c: Likewise.
24050 2017-01-20 Marek Polacek <polacek@redhat.com>
24053 * doc/invoke.texi: Document -Wduplicated-branches.
24054 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24055 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24056 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24057 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24058 return 0 only when not OEP_LEXICOGRAPHIC.
24059 (fold_build_cleanup_point_expr): Use the expression
24060 location when building CLEANUP_POINT_EXPR.
24061 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24062 * tree.c (add_expr): Handle error_mark_node.
24064 2017-01-20 Martin Liska <mliska@suse.cz>
24067 * tree-profile.c (init_ic_make_global_vars): Do not call
24069 (gimple_init_gcov_profiler): Likewise.
24071 2017-01-20 Martin Liska <mliska@suse.cz>
24074 * cgraph.h (maybe_create_reference): Remove argument and
24076 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24078 * ipa-cp.c (create_specialized_node): Likewise.
24079 * symtab.c (symtab_node::maybe_create_reference): Handle
24080 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24082 2017-01-20 Martin Liska <mliska@suse.cz>
24084 * read-rtl-function.c (function_reader::create_function): Use
24085 build_decl instread of build_decl_stat.
24087 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24089 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24090 * config/i386/avx512dqintrin.h: Ditto.
24091 * config/i386/avx512fintrin.h: Ditto.
24092 * config/i386/i386-builtin-types.def: Add new types.
24093 * config/i386/i386.c: Handle new types.
24094 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24095 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24096 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24097 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24098 (__builtin_ia32_kshiftridi): New.
24099 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24101 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24105 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24106 define to rs6000_init_stack_protect_guard.
24107 (rs6000_init_stack_protect_guard): New function.
24109 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24110 Yunqiang Su <yunqiang.su@imgtec.com>
24112 * config.gcc (supported_defaults): Add madd4.
24113 (with_madd4): Add validation.
24114 (all_defaults): Add madd4.
24115 * config/mips/mips.opt (mmadd4): New option.
24116 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24118 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24120 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24121 (ISA_HAS_FUSED_MADD4): Likewise.
24122 * doc/invoke.texi (-mmadd4): Document the new option.
24123 * doc/install.texi (--with-madd4): Document the new option.
24125 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24127 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24128 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24129 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24130 (aarch64_init_pauth_hint_builtins): New.
24131 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24132 (aarch64_expand_builtin): Expand new builtins.
24134 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24136 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24137 * combine-stack-adj.c (no_unhandled_cfa): Handle
24138 REG_CFA_TOGGLE_RA_MANGLE.
24139 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24140 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24141 info for return address signing.
24142 (aarch64_expand_epilogue): Likewise.
24144 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24146 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24147 * config/aarch64/aarch64-protos.h
24148 (aarch64_return_address_signing_enabled): New declaration.
24149 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24151 (aarch64_expand_prologue): Sign return address before it's pushed onto
24153 (aarch64_expand_epilogue): Authenticate return address fetched from
24155 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24156 (aarch64_attributes): New function attributes for "sign-return-address".
24157 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24158 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24159 ("*do_return"): Generate combined instructions according to key index.
24160 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24161 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24163 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24164 * config/aarch64/aarch64.opt (msign-return-address=): New.
24165 * doc/extend.texi (AArch64 Function Attributes): Documents
24166 "sign-return-address=".
24167 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24169 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24171 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24172 overall option summary.
24174 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24176 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24177 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24178 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24179 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24181 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24183 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24184 -mpower9-minmax by default for -mcpu=power9.
24185 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24186 128-bit floating point.
24188 2017-01-20 Alan Modra <amodra@gmail.com>
24190 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24191 optimizing for size.
24193 2017-01-20 Alan Modra <amodra@gmail.com>
24196 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24197 for strcmp and strncmp from corresponding builtin decl.
24199 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24201 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24202 instead of i386/rtems-64.h.
24203 * config/i386/rtems-64.h: Remove.
24205 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24209 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24211 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24213 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24215 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24216 Change int to HOST_WIDE_INT.
24217 * config/aarch64/aarch64-protos.h
24218 (aarch64_simd_gen_const_vector_dup): Likewise.
24219 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24221 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24223 * langhooks-def.h (lhd_type_for_size): New decl.
24224 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24225 * langhooks.c (lhd_type_for_size): New function, taken from
24228 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24230 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24231 define_bypass for CR latency.
24232 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24233 (power9-alu2): Add define_bypass for CR latency.
24235 (power9-mul): Update insn latency.
24236 (power9-mul-compare): Update insn latency, bypass latency and remove
24239 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24241 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24243 * config/aarch64/aarch64.md
24244 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24245 aarch64_nopcrelative_literal_loads.
24246 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24248 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24250 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24251 TARGET_LOONGSON_3A.
24252 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24254 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24257 * config.gcc (supported_defaults): Add lxc1-sxc1.
24258 (with_lxc1_sxc1): Add validation.
24259 (all_defaults): Add lxc1-sxc1.
24260 * config/mips/mips.opt (mlxc1-sxc1): New option.
24261 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24263 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24264 __mips_no_lxc1_sxc1.
24265 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24266 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24267 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24269 2017-01-19 Richard Biener <rguenther@suse.de>
24271 PR tree-optimization/72488
24272 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24273 sure to restore SSA info.
24274 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24276 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24278 PR rtl-optimization/79121
24279 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24280 of the inner type when shifting an extended value.
24282 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24285 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24286 interposable aliases.
24288 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24291 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24292 Use the evmergelohi instruction.
24293 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24294 (mov_sitf_e500_subreg8_2_be): Likewise.
24295 (mov_sitf_e500_subreg12_2_le): Likewise.
24296 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24297 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24298 (mov_sitf_e500_subreg8_2_le): Likewise.
24299 (mov_sitf_e500_subreg12_2_be): Likewise.
24301 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24303 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24304 attribute from vecsimple to vecperm.
24305 (altivec_vbpermq2): Likewise.
24307 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24310 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24312 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24313 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24314 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24315 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24316 case where N arg is SIZE_MAX.
24317 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24318 (cmpstrsi): Add pattern.
24320 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24322 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24323 __builtin_vec_revb builtins.
24324 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24325 built-in functions to support generation of the ISA 3.0 XXBR<x>
24326 vector byte reverse instructions.
24327 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24328 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24329 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24330 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24331 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24332 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24333 (P9V_BUILTIN_VEC_REVB): Likewise.
24334 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24335 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24336 (p9_xxbrq_v16qi): Likewise.
24337 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24338 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24339 (p9_xxbrh_v8hi): Likewise.
24340 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24341 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24342 vec_revb built-in functions.
24344 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24346 PR rtl-optimization/78952
24347 * config/i386/i386.md (any_extract): New code iterator.
24348 (*insvqi_2): Use any_extract for source operand.
24349 (*insvqi_3): Use any_shiftrt for source operand.
24351 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24353 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24355 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24357 2017-01-18 Matthias Klose <doko@ubuntu.com>
24359 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24361 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24363 * config/rs6000/altivec.h (vec_bperm): Change #define.
24364 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24365 (altivec_vbpermq2): New define_insn.
24366 (altivec_vbpermd): Likewise.
24367 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24368 function interface.
24369 (VBPERMD): Likewise.
24370 (VBPERM): New polymorphic function interface.
24371 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24372 Add entries for P9V_BUILTIN_VEC_VBPERM.
24373 * doc/extend.texi: Add interfaces for vec_bperm.
24375 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24377 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24378 first letter of error messages.
24379 (s390_resolve_overloaded_builtin): Likewise.
24380 * config/s390/s390.c (s390_expand_builtin): Likewise.
24381 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24382 (s390_valid_target_attribute_inner_p): Likewise.
24383 * config/s390/s390.md ("tabort"): Likewise.
24385 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24387 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24388 (ISA_AVOID_DIV_HILO): New macro.
24389 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24390 (ISA_HAS_DDIV): Likewise.
24392 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24394 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24396 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24398 * doc/invoke.texi (fabi-version): Spelling fix.
24400 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24403 * doc/invoke.texi (fabi-version): Mention mangling fix for
24406 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24409 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24411 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24414 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24415 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24417 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24419 (stack_protect_test): Ditto.
24420 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24421 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24423 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24424 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24425 -mstack-protector-guard-offset=.
24426 (RS/6000 and PowerPC Options): Ditto.
24428 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24430 * config/i386/i386.h (MASK_CLASS_P): New define.
24431 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24432 there are no registers from different register sets also when
24433 mask registers are used. Update function comment.
24434 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24435 to (*k/*r) and (*k/*km) alternatives.
24437 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24439 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24440 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24441 (EH_RETURN_HANDLER_RTX): New define.
24442 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24443 Force frame pointer in EH return functions.
24444 (aarch64_expand_epilogue): Add barrier for eh_return.
24445 (aarch64_final_eh_return_addr): Remove.
24446 (aarch64_eh_return_handler_rtx): New function.
24447 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24449 (aarch64_eh_return_handler_rtx): New prototype.
24451 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24453 * config/rs6000/altivec.h (vec_rlmi): New #define.
24454 (vec_vrlnm): Likewise.
24455 (vec_rlnm): Likewise.
24456 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24457 (UNSPEC_VRLNM): Likewise.
24458 (VIlong): New mode iterator.
24459 (altivec_vrl<VI_char>mi): New define_insn.
24460 (altivec_vrl<VI_char>nm): Likewise.
24461 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24463 (VRLDNM): Likewise.
24464 (RLNM): New polymorphic function entry.
24465 (VRLWMI): New monomorphic function entry.
24466 (VRLDMI): Likewise.
24467 (RLMI): New polymorphic function entry.
24468 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24469 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24470 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24473 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24476 * dwarf2out.c (field_byte_offset): Restore the
24477 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24478 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24479 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24482 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24485 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24487 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24490 * dwarf2out.c (add_data_member_location_attribute): For constant
24491 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24492 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24493 DW_AT_byte_size attributes.
24495 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24497 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24498 after forcing to constant memory when the code model is medium.
24500 2017-01-17 Julia Koval <julia.koval@intel.com>
24503 * config/i386/avx512fintrin.h
24504 (_mm512_i32gather_ps): Change __addr type to void const*.
24505 (_mm512_mask_i32gather_ps): Ditto.
24506 (_mm512_i32gather_pd): Ditto.
24507 (_mm512_mask_i32gather_pd): Ditto.
24508 (_mm512_i64gather_ps): Ditto.
24509 (_mm512_mask_i64gather_ps): Ditto.
24510 (_mm512_i64gather_pd): Ditto.
24511 (_mm512_mask_i64gather_pd): Ditto.
24512 (_mm512_i32gather_epi32): Ditto.
24513 (_mm512_mask_i32gather_epi32): Ditto.
24514 (_mm512_i32gather_epi64): Ditto.
24515 (_mm512_mask_i32gather_epi64): Ditto.
24516 (_mm512_i64gather_epi32): Ditto.
24517 (_mm512_mask_i64gather_epi32): Ditto.
24518 (_mm512_i64gather_epi64): Ditto.
24519 (_mm512_mask_i64gather_epi64): Ditto.
24520 (_mm512_i32scatter_ps): Change __addr type to void*.
24521 (_mm512_mask_i32scatter_ps): Ditto.
24522 (_mm512_i32scatter_pd): Ditto.
24523 (_mm512_mask_i32scatter_pd): Ditto.
24524 (_mm512_i64scatter_ps): Ditto.
24525 (_mm512_mask_i64scatter_ps): Ditto.
24526 (_mm512_i64scatter_pd): Ditto.
24527 (_mm512_mask_i64scatter_pd): Ditto.
24528 (_mm512_i32scatter_epi32): Ditto.
24529 (_mm512_mask_i32scatter_epi32): Ditto.
24530 (_mm512_i32scatter_epi64): Ditto.
24531 (_mm512_mask_i32scatter_epi64): Ditto.
24532 (_mm512_i64scatter_epi32): Ditto.
24533 (_mm512_mask_i64scatter_epi32): Ditto.
24534 (_mm512_i64scatter_epi64): Ditto.
24535 (_mm512_mask_i64scatter_epi64): Ditto.
24536 * config/i386/avx512pfintrin.h
24537 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24538 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24539 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24540 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24541 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24542 (_mm512_prefetch_i32scatter_ps): Ditto.
24543 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24544 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24545 (_mm512_prefetch_i64scatter_pd): Ditto.
24546 (_mm512_prefetch_i64scatter_ps): Ditto.
24547 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24548 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24549 * config/i386/avx512vlintrin.h
24550 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24551 (_mm_mmask_i32gather_ps): Ditto.
24552 (_mm256_mmask_i32gather_pd): Ditto.
24553 (_mm_mmask_i32gather_pd): Ditto.
24554 (_mm256_mmask_i64gather_ps): Ditto.
24555 (_mm_mmask_i64gather_ps): Ditto.
24556 (_mm256_mmask_i64gather_pd): Ditto.
24557 (_mm_mmask_i64gather_pd): Ditto.
24558 (_mm256_mmask_i32gather_epi32): Ditto.
24559 (_mm_mmask_i32gather_epi32): Ditto.
24560 (_mm256_mmask_i32gather_epi64): Ditto.
24561 (_mm_mmask_i32gather_epi64): Ditto.
24562 (_mm256_mmask_i64gather_epi32): Ditto.
24563 (_mm_mmask_i64gather_epi32): Ditto.
24564 (_mm256_mmask_i64gather_epi64): Ditto.
24565 (_mm_mmask_i64gather_epi64): Ditto.
24566 (_mm256_i32scatter_ps): Change __addr type to void*.
24567 (_mm256_mask_i32scatter_ps): Ditto.
24568 (_mm_i32scatter_ps): Ditto.
24569 (_mm_mask_i32scatter_ps): Ditto.
24570 (_mm256_i32scatter_pd): Ditto.
24571 (_mm256_mask_i32scatter_pd): Ditto.
24572 (_mm_i32scatter_pd): Ditto.
24573 (_mm_mask_i32scatter_pd): Ditto.
24574 (_mm256_i64scatter_ps): Ditto.
24575 (_mm256_mask_i64scatter_ps): Ditto.
24576 (_mm_i64scatter_ps): Ditto.
24577 (_mm_mask_i64scatter_ps): Ditto.
24578 (_mm256_i64scatter_pd): Ditto.
24579 (_mm256_mask_i64scatter_pd): Ditto.
24580 (_mm_i64scatter_pd): Ditto.
24581 (_mm_mask_i64scatter_pd): Ditto.
24582 (_mm256_i32scatter_epi32): Ditto.
24583 (_mm256_mask_i32scatter_epi32): Ditto.
24584 (_mm_i32scatter_epi32): Ditto.
24585 (_mm_mask_i32scatter_epi32): Ditto.
24586 (_mm256_i32scatter_epi64): Ditto.
24587 (_mm256_mask_i32scatter_epi64): Ditto.
24588 (_mm_i32scatter_epi64): Ditto.
24589 (_mm_mask_i32scatter_epi64): Ditto.
24590 (_mm256_i64scatter_epi32): Ditto.
24591 (_mm256_mask_i64scatter_epi32): Ditto.
24592 (_mm_i64scatter_epi32): Ditto.
24593 (_mm_mask_i64scatter_epi32): Ditto.
24594 (_mm256_i64scatter_epi64): Ditto.
24595 (_mm256_mask_i64scatter_epi64): Ditto.
24596 (_mm_i64scatter_epi64): Ditto.
24597 (_mm_mask_i64scatter_epi64): Ditto.
24598 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24599 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24600 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24601 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24602 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24603 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24604 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24605 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24606 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24607 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24608 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24609 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24610 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24611 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24612 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24613 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24614 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24615 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24616 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24617 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24618 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24619 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24620 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24621 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24622 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24623 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24624 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24625 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24626 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24627 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24628 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24629 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24630 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24631 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24632 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24633 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24634 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24635 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24636 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24637 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24638 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24639 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24640 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24641 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24642 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24643 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24644 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24645 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24646 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24647 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24648 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24649 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24650 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24651 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24652 definitions accordingly.
24654 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24655 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24658 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24661 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24664 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24665 hard regs for inner regno.
24667 2017-01-17 Martin Liska <mliska@suse.cz>
24670 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24671 assumption and add comment.
24673 2017-01-17 Nathan Sidwell <nathan@acm.org>
24675 * ipa-visibility.c (localize_node): New function, broken out of ...
24676 (function_and_variable_visibility): ... here. Call it.
24678 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24680 PR middle-end/77445
24681 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24682 correctly set frequency of oudgoing edge.
24683 (duplicate_thread_path): Fix profile updating.
24685 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24688 * configure.ac: Add GCC_BASE_VER.
24689 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24690 version from BASE-VER file.
24691 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24692 (gcc.o): Depend on $(BASEVER).
24693 * common.opt (dumpfullversion): New option.
24694 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24695 * doc/invoke.texi: Document -dumpfullversion.
24696 * doc/install.texi: Document --with-gcc-major-version-only.
24697 * configure: Regenerated.
24699 2017-01-17 Richard Biener <rguenther@suse.de>
24701 PR tree-optimization/71433
24702 * tree-vrp.c (register_new_assert_for): Merge same asserts
24703 on all incoming edges.
24704 (process_assert_insertions_for): Handle insertions at the
24707 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24709 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24710 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24712 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24715 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24718 2017-01-17 Alan Modra <amodra@gmail.com>
24721 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24722 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24723 symbolic stack limit when pic.
24725 2017-01-16 Martin Sebor <msebor@redhat.com>
24727 PR tree-optimization/78608
24728 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24730 2017-01-16 Jeff Law <law@redhat.com>
24733 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24734 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24735 for several include directories that may be relative to sysroot.
24736 * config/i386/x-mingw32 (gplus_includedir): Define.
24737 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24738 (native_system_includedir): Likewise.
24739 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24740 override if TARGET_SYSTEM_ROOT is defined.
24741 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24743 PR tree-optimization/79090
24744 PR tree-optimization/33562
24745 PR tree-optimization/61912
24746 PR tree-optimization/77485
24747 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24748 and computed trims into the dump file.
24750 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24752 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24754 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24757 * gimplify.c (gimplify_init_constructor): If want_value and
24758 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24762 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24763 sequence. Formatting fixes.
24764 (doloop_optimize): Formatting fixes.
24767 * gcc.c (debug_level_greater_than_spec_func): New function.
24768 (static_spec_functions): Add debug-level-gt spec function.
24769 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24771 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24772 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24773 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24774 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24775 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24776 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24778 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24780 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24781 QImode fixups to general and mask registers only.
24783 2017-01-16 Carl Love <cel@us.ibm.com>
24785 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24786 for built-in functions
24787 vector signed char vec_nabs (vector signed char)
24788 vector signed short vec_nabs (vector signed short)
24789 vector signed int vec_nabs (vector signed int)
24790 vector signed long long vec_nabs (vector signed long long)
24791 vector float vec_nabs (vector float)
24792 vector double vec_nabs (vector double)
24793 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24795 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24796 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24797 * doc/extend.texi: Update the documentation file for the new built-in
24800 2017-01-16 Martin Sebor <msebor@redhat.com>
24802 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24805 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24807 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24808 UNSPEC_VSX__XXSPLTD to require special splat handling.
24810 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24813 * system.h: Poison strndup.
24815 2017-01-16 Alan Modra <amodra@gmail.com>
24818 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24821 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24823 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24825 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24827 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24828 call recog here. Assert that INSN_CODE (insn) is non-negative.
24830 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24833 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24835 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24836 in the currently scheduled RTL fragment.
24838 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24840 PR rtl-optimization/78751
24841 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24844 2017-01-14 Jeff Law <law@redhat.com>
24846 PR tree-optimization/79090
24847 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24848 variable length stores.
24849 (compute_trims): Delete dead assignment to *trim_tail.
24850 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24853 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24855 PR rtl-optimization/78626
24856 PR rtl-optimization/78727
24857 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24858 of a block, and split such blocks after everything else is finished.
24860 2017-01-14 Alan Modra <amodra@gmail.com>
24863 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24864 target legitimate_combined_insn.
24865 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24866 (rs6000_legitimate_combined_insn): New function.
24867 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24869 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24870 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24871 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24873 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24875 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24877 2017-01-13 Jeff Law <law@redhat.com>
24879 PR tree-optimization/33562
24880 PR tree-optimization/61912
24881 PR tree-optimization/77485
24882 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24884 (delete_dead_assignment): Likewise.
24885 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24886 statement to delete_dead_call and delete_dead_assignment.
24888 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24891 * substring-locations.c (format_warning_va): Strengthen case 1 so
24892 that both endpoints of the substring must be within the format
24893 range for just the substring to be printed.
24895 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24897 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24898 * config/i386/i386.c (ix86_target_string): Add missing options
24899 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24900 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24901 flags_other and ix86_target_other to flags2_other. Display unknown
24903 (ix86_valid_target_attribute_inner_p): Add missing options and
24904 reorder options by implied ISAs, as in ix86_target_string.
24906 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24908 * hash-table.h (hash_table::too_empty_p): New function.
24909 (hash_table::expand): Use it.
24910 (hash_table::traverse): Likewise.
24911 (hash_table::empty_slot): Use sizeof (value_type) instead of
24912 sizeof (PTR) to convert bytes to elements. Shrink the table
24913 if the current size is excessive for the current number of
24916 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24918 * ira-costs.c (record_reg_classes): Break from the inner loop
24919 early once alt_fail is known to be true. Update outer loop
24920 handling accordingly.
24922 2017-01-13 Jeff Law <law@redhat.com>
24924 * tree-ssa-dse.c (decrement_count): New function.
24925 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24926 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24927 when we know the partially dead statement is a mem* function.
24929 PR tree-optimization/61912
24930 PR tree-optimization/77485
24931 * tree-ssa-dse.c: Include expr.h.
24932 (maybe_trim_constructor_store): New function.
24933 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24935 PR tree-optimization/33562
24936 PR tree-optimization/61912
24937 PR tree-optimization/77485
24938 * doc/invoke.texi: Document new dse-max-object-size param.
24939 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24940 * tree-ssa-dse.c: Include params.h.
24941 (dse_store_status): New enum.
24942 (initialize_ao_ref_for_dse): New, partially extracted from
24944 (valid_ao_ref_for_dse, normalize_ref): New.
24945 (setup_live_bytes_from_ref, compute_trims): Likewise.
24946 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24947 (maybe_trim_partially_dead_store): Likewise.
24948 (maybe_trim_complex_store): Likewise.
24949 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24950 Track what bytes live from the original store. Return tri-state
24951 for dead, partially dead or live.
24952 (dse_dom_walker): Add constructor, destructor and new private members.
24953 (delete_dead_call, delete_dead_assignment): New extracted from
24955 (dse_optimize_stmt): Make a member of dse_dom_walker.
24956 Use initialize_ao_ref_for_dse.
24958 PR tree-optimization/33562
24959 PR tree-optimization/61912
24960 PR tree-optimization/77485
24961 * sbitmap.h (bitmap_count_bits): Prototype.
24962 (bitmap_clear_range, bitmap_set_range): Likewise.
24963 * sbitmap.c (bitmap_clear_range): New function.
24964 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24966 2017-01-13 Martin Liska <mliska@suse.cz>
24969 * function.c (set_cfun): Add new argument force.
24970 * function.h (set_cfun): Likewise.
24971 * ipa-inline-transform.c (inline_call): Use the function when
24972 strict alising from is dropped for function we inline to.
24974 2017-01-13 Richard Biener <rguenther@suse.de>
24976 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24977 for dumping GIMPLE INTEGER_CSTs.
24979 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24981 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24982 to 201112L since C++17.
24984 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24987 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24988 if -fsanitize=kernel-address is present.
24990 2017-01-13 Richard Biener <rguenther@suse.de>
24992 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24993 as _Literal ( type ) number in case usual suffixes do not
24994 preserve all information.
24996 2017-01-13 Richard Biener <rguenther@suse.de>
24998 PR tree-optimization/77283
24999 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25000 and ssa-iterators.h.
25001 (is_feasible_trace): Implement a cost model based on joiner
25004 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25007 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25008 char or short to __float128/_Float128 directly.
25010 2017-01-12 Martin Sebor <msebor@redhat.com>
25012 to -Wformat-overflow.
25013 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25014 (min_bytes_remaining): Same.
25015 (get_string_length): Same.
25016 (format_string): Same.
25017 (format_directive): Same.
25019 (pass_sprintf_length::handle_gimple_call): Same.
25021 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25023 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25024 info.nowrite calls with no lhs that can't throw. Return bool
25025 whether gsi_remove has been called or not.
25026 (pass_sprintf_length::handle_gimple_call): Return bool whether
25027 try_substitute_return_value called gsi_remove. Formatting fix.
25028 (pass_sprintf_length::execute): Don't use gsi_remove if
25029 handle_gimple_call returned true.
25032 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25033 be removed due to side-effects, don't remove following barrier nor
25034 turn the successor edge into fallthru edge.
25036 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25039 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25040 element-reversing loads and stores as not swappable.
25042 2017-01-12 Nathan Sidwell <nathan@acm.org>
25043 Nicolai Stange <nicstange@gmail.com>
25045 * combine.c (try_combine): Don't ignore result of overlap checking
25046 loop. Combine overlap & asm check into single loop.
25048 2017-01-12 Richard Biener <rguenther@suse.de>
25050 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25051 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25053 2017-01-12 Richard Biener <rguenther@suse.de>
25055 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25056 and TS_TARGET_OPTION directly derive from TS_BASE.
25057 * tree-core.h (tree_optimization_option): Derive from tree_base.
25058 (tree_target_option): Likewise.
25060 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25062 * config/i386/i386.c (memory_address_length): Increase len
25063 only when rip_relative_addr_p returns false.
25065 2017-01-11 Julia Koval <julia.koval@intel.com>
25067 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25068 (OPTION_MASK_ISA_SGX_SET): New.
25069 (ix86_handle_option): Handle OPT_msgx.
25070 * config.gcc: Added sgxintrin.h.
25071 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25072 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25073 * config/i386/i386.c (ix86_target_string): Add -msgx.
25075 (ix86_option_override_internal): Handle new options.
25076 (ix86_valid_target_attribute_inner_p): Add sgx.
25077 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25078 * config/i386/i386.opt: Add msgx.
25079 * config/i386/sgxintrin.h: New file.
25080 * config/i386/x86intrin.h: Add sgxintrin.h.
25082 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25085 * fold-const.c (maybe_nonzero_address): Return 1 for function
25087 (tree_single_nonzero_warnv_p): Don't handle function local objects
25091 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25094 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25097 * opts.c: Include "spellcheck.h"
25098 (struct string_fragment): New struct.
25099 (struct edit_distance_traits<const string_fragment &>): New
25101 (get_closest_sanitizer_option): New function.
25102 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25104 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25106 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25108 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25109 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25110 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25111 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25112 for initial die_offset if dwarf_split_debug_info.
25113 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25114 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25116 (output_skeleton_debug_sections): Formatting fix. Use
25117 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25118 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25120 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25122 * config/arm/cortex-a53.md: Add bypasses for
25123 cortex_a53_r2f_cvt.
25124 (cortex_a53_r2f): Only use for transfers.
25125 (cortex_a53_f2r): Likewise.
25126 (cortex_a53_r2f_cvt): Add reservation for conversions.
25127 (cortex_a53_f2r_cvt): Likewise.
25129 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25131 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25132 to all inlined functions, change static to extern.
25134 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25137 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25139 (arm_assemble_integer): Likewise.
25141 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25143 * config.gcc: Use new awk script to check CPU, FPU and architecture
25144 parameters for --with-... options.
25145 * config/arm/parsecpu.awk: New file
25146 * config/arm/arm-cpus.in: New file.
25147 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25149 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25151 * config/arm/t-arm: Update dependency rules.
25152 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25153 of processing .def files.
25154 * config/arm/genopt.sh: Deleted.
25155 * config/arm/gentune.sh: Deleted.
25156 * config/arm/arm-cores.def: Deleted.
25157 * config/arm/arm-arches.def: Deleted.
25158 * config/arm/arm-fpus.def: Deleted.
25159 * config/arm/arm-tune.md: Regenerated.
25160 * config/arm/arm-tables.opt: Regenerated.
25161 * config/arm/arm-cpu.h: New generated file.
25162 * config/arm/arm-cpu-data.h: New generated file.
25163 * config/arm/arm-cpu-cdata.h: New generated file.
25165 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25168 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25170 (input_varpool_node): Unpack dynamically_initialized bit.
25172 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25174 PR rtl-optimization/79032
25175 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25176 the alignment of the adjusted memory reference against that of MODE,
25177 instead of the alignment of the original memory reference.
25179 2017-01-11 Martin Jambor <mjambor@suse.cz>
25181 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25183 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25184 decorated functions.
25186 2017-01-11 Richard Biener <rguenther@suse.de>
25188 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25189 set range/nonnull info for PHI results. Do not set it on
25190 stmts marked for removal.
25192 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25194 * expr.c (store_field): In the bitfield case, fetch the return value
25195 from the registers before applying a single big-endian adjustment.
25196 Always do a final load for a BLKmode value not larger than a word.
25198 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25201 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25202 that we correctly handle column numbers greater than
25203 LINE_MAP_MAX_COLUMN_NUMBER.
25205 2017-01-10 Martin Sebor <msebor@redhat.com>
25207 PR middle-end/78245
25208 * gimple-ssa-sprintf.c (get_destination_size): Call
25209 {init,fini}object_sizes.
25210 * tree-object-size.c (addr_object_size): Adjust.
25211 (pass_through_call): Adjust.
25212 (pass_object_sizes::execute): Adjust.
25213 * tree-object-size.h (fini_object_sizes): Declare.
25215 2017-01-10 Martin Sebor <msebor@redhat.com>
25217 PR tree-optimization/78775
25218 * builtins.c (get_size_range): Move...
25219 * calls.c: ...to here.
25220 (alloc_max_size): Accept zero argument.
25221 (operand_signed_p): Remove.
25222 (maybe_warn_alloc_args_overflow): Call get_size_range.
25223 * calls.h (get_size_range): Declare.
25225 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25227 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25228 from TI's devices.csv file as of September 2016.
25229 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25231 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25233 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25234 * doc/invoke.texi: Likewise.
25235 * doc/md.texi: Likewise.
25236 * doc/objc.texi: Likewise.
25238 2017-01-10 Joshua Conner <joshconner@google.com>
25240 * config/arm/fuchsia-elf.h: New file.
25241 * config/fuchsia.h: New file.
25242 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25243 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25245 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25247 2016-01-10 Richard Biener <rguenther@suse.de>
25249 PR tree-optimization/79034
25250 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25251 Propagate out degenerate PHIs in the joiner.
25253 2017-01-10 Martin Liska <mliska@suse.cz>
25255 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25256 (sort_congruence_classes_by_decl_uid): Likewise.
25257 (sort_congruence_class_groups_by_decl_uid): Likewise.
25258 (sem_item_optimizer::merge_classes): Sort class, groups in these
25259 classes and members in the groups by DECL_UID of declarations.
25260 This would make merge operations stable.
25262 2017-01-10 Martin Liska <mliska@suse.cz>
25264 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25265 usage of m_classes_vec.
25266 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25267 (sem_item_optimizer::get_group_by_hash): Likewise.
25268 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25269 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25270 (sem_item_optimizer::verify_classes): Likewise.
25271 (sem_item_optimizer::process_cong_reduction): Likewise.
25272 (sem_item_optimizer::dump_cong_classes): Likewise.
25273 (sem_item_optimizer::merge_classes): Likewise.
25274 * ipa-icf.h (congruence_class_hash): Rename from
25275 congruence_class_group_hash. Remove declaration of m_classes_vec.
25277 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25279 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25280 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25281 * config.gcc: Add avx512vpopcntdqintrin.h.
25282 * config/i386/avx512vpopcntdqintrin.h: New.
25283 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25284 * config/i386/i386-builtin-types.def: Add new types.
25285 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25286 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25287 __builtin_ia32_vpopcountq_v8di_mask): New.
25288 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25289 __AVX512VPOPCNTDQ__.
25290 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25291 (PTA_AVX512VPOPCNTDQ): Define.
25292 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25293 TARGET_AVX512VPOPCNTDQ_P): Define.
25294 * config/i386/i386.opt: Add mavx512vpopcntdq.
25295 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25296 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25298 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25300 PR middle-end/77484
25301 * predict.def (PRED_CALL): Set to 67.
25303 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25305 * expr.c (store_field): In the bitfield case, if the value comes from
25306 a function call and is of an aggregate type returned in registers, do
25307 not modify the field mode; extract the value in all cases if the mode
25308 is BLKmode and the size is not larger than a word.
25310 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25313 * config/i386/cpuid.h: Fix undefined behavior.
25315 2017-01-04 Jeff Law <law@redhat.com>
25317 PR tree-optimization/79007
25318 PR tree-optimization/67955
25319 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25320 conservative for pt.null when flag_non_call_exceptions is on.
25322 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25324 PR translation/79019
25325 PR translation/79020
25326 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25327 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25328 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25330 * config/avr/avr.opt (maccumulate-args): Likewise.
25331 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25332 * common.opt (freport-bug): Likewise.
25333 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25334 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25335 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25336 translatable string.
25337 * config/i386/i386.c (function_value_32): Likewise.
25338 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25339 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25341 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25342 * common/config/msp430/msp430-common.c (msp430_handle_option):
25344 * symtab.c (symtab_node::verify_base): Likewise.
25345 * opts.c (set_debug_level): Likewise.
25346 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25347 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25348 missing whitespace to translatable strings.
25349 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25350 * config/sh/superh.h: Likewise.
25351 * config/i386/xopintrin.h: Likewise.
25352 * config/i386/znver1.md: Likewise.
25353 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25354 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25355 * double-int.h (struct double_int): Likewise.
25356 * double-int.c (div_and_round_double): Likewise.
25357 * wide-int.cc: Likewise.
25358 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25359 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25360 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25361 (crossjumps_occurred): ... this.
25362 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25365 PR tree-optimization/78899
25366 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25367 returning bool return struct loop *, NULL for failure and the new
25369 (versionable_outer_loop_p): Don't version outer loop if it has
25370 dont_vectorized bit set.
25371 (tree_if_conversion): When versioning outer loop, ensure
25372 tree_if_conversion is performed also on the inner loop of the
25373 non-vectorizable outer loop copy.
25374 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25375 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25376 prevent vectorization of it.
25377 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25378 the outer loop vectorization of the non-scalar version is attempted
25379 before vectorization of the inner loop in scalar version. If
25380 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25381 vectorization of its inner loop.
25382 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25383 has 2 inner loops, rename also on edges from bb whose single pred
25384 is outer_loop->header. Fix typo in function comment.
25386 2017-01-09 Martin Sebor <msebor@redhat.com>
25389 * asan.c (asan_emit_stack_protection): Increase local buffer size
25390 to avoid snprintf truncation warning.
25392 2017-01-09 Andrew Pinski <apinski@cavium.com>
25394 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25395 to reference thunderx2t99 for the tuning structure
25396 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25398 (thunderx2t99_extra_costs): This.
25399 * config/aarch64/aarch64-tune.md: Regenerate.
25400 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25401 (vulcan_addrcost_table): This.
25402 (vulcan_regmove_cost): Rename to ...
25403 (thunderx2t99_regmove_cost): This.
25404 (vulcan_vector_cost): Rename to ...
25405 (thunderx2t99_vector_cost): this.
25406 (vulcan_branch_cost): Rename to ...
25407 (thunderx2t99_branch_cost): This.
25408 (vulcan_tunings): Rename to ...
25409 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25410 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25412 2017-01-09 Martin Jambor <mjambor@suse.cz>
25416 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25417 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25418 (propagate_vr_accross_jump_function): Use the above function for all
25419 value range computations for pass-through jump functions and type
25420 converasion from explicit value range values.
25421 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25422 parameters from TYPE_ARG_TYPES.
25423 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25424 (ipa_write_node_info): Stream type of the actual argument.
25425 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25427 2017-01-09 Martin Liska <mliska@suse.cz>
25430 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25431 (lookup_compiler): Do not show error message with have_E.
25433 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25435 PR tree-optimization/78938
25436 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25437 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25438 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25439 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25442 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25444 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25447 2017-01-09 Richard Biener <rguenther@suse.de>
25449 PR tree-optimization/78997
25450 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25451 name condition properly.
25453 2017-01-09 Richard Biener <rguenther@suse.de>
25456 * dwarf2out.c (is_cxx): New overload with context.
25457 (is_naming_typedef_decl): Use it.
25459 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25461 * invoke.texi (Option Summary): Correct spacing in option lists
25462 and add line breaks to fix over-long lines.
25464 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25466 PR middle-end/17660
25468 * extend.texi (Common Variable Attributes): Add xref to GCC
25469 Internals manual to explain mode attribute keywords.
25471 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25474 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25475 and Preprocessor Options.
25476 (Options for Linking): Document -pthread here....
25477 (RS/6000 and PowerPC Options): ...not here.
25478 (Solaris 2 Options): ...or here.
25479 * doc/cppopts.texi: Document -pthread.
25481 2017-01-08 Martin Sebor <msebor@redhat.com>
25483 PR middle-end/77708
25484 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25485 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25486 New member functions.
25487 (format_directive): Used them.
25489 (pass_sprintf_length::handle_gimple_call): Same.
25490 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25491 to avoid truncation for any argument.
25492 (extract_affine_mul): Same.
25493 * tree.c (get_file_function_name): Same.
25495 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25497 PR middle-end/77484
25498 * predict.def (PRED_INDIR_CALL): Set to 86.
25500 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25502 PR preprocessor/54124
25503 * doc/cppopts.texi: Reformat -d subtable to list the full name
25504 of the options. Add cross-reference to the docs for the general
25505 compiler -d options.
25506 * doc/invoke.texi (Developer Options): Add cross-reference to the
25507 preprocessor-specific -d option documentation.
25509 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25511 PR preprocessor/13498
25512 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25513 redudant material, and reflect new command-line options.
25514 (System Headers): Likewise.
25516 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25518 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25519 -isystem, and -idirafter. Copy-edit.
25520 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25521 default for -ftrack-macro-expansion. Delete obsolete and
25522 badly-formatted implementation details about -fdebug-cpp output.
25523 * doc/cppwarnopts.texi: Copy-edit.
25525 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25528 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25529 that the transition from a max line width >= 1<<10 to narrower
25530 lines works correctly.
25532 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25534 * doc/options.texi (PerFunction): New.
25535 * opt-functions.awk (switch_flags): Map both Optimization and
25536 PerFunction to CL_OPTIMIZATION.
25537 * opth-gen.awk: Test for PerFunction flag along with
25539 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25540 it only when the latter is present. Skip those that don't in
25541 the hash function generator.
25542 * common.opt (fvar-tracking): Mark as PerFunction instead of
25544 (fvar-tracking-assignments): Likewise.
25545 (fvar-tracking-assignments-toggle): Likewise.
25546 (fvar-tracking-uninit): Likewise.
25548 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25550 PR translation/79018
25551 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25554 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25557 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25558 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25559 TARGET_LONG_JUMP_TABLE_OFFSETS.
25560 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25561 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25562 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25563 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25564 * config/m68k/m68k.md (tablejump expander): Likewise.
25565 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25566 TARGET_LONG_JUMP_TABLE_OFFSETS.
25567 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25568 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25570 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25571 David Holsgrove <david.holsgrove@xilinx.com>
25573 * common/config/microblaze/microblaze-common.c
25574 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25575 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25577 * config/microblaze/microblaze.c (microblaze_must_save_register)
25578 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25580 (microblaze_eh_return): New function.
25581 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25582 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25583 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25584 * config/microblaze/microblaze.md (eh_return): New pattern.
25586 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25588 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25589 GCC_DIAGNOSTIC_STRINGIFY): Define.
25591 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25593 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25595 * config/arm/arm.md (<mcrr>): New.
25597 * config/arm/arm.c (arm_arch5te): New.
25598 (arm_option_override): Set arm_arch5te.
25599 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25601 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25602 (arm_mcrr_qualifiers): ... this. New.
25603 (MRRC_QUALIFIERS): Define to...
25604 (arm_mrrc_qualifiers): ... this. New.
25605 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25607 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25608 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25609 (MRRCI, mrrc, MRRC): New.
25610 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25611 VUNSPEC_MRRC2): New.
25613 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25615 * config/arm/arm.md (<mcr>): New.
25617 * config/arm/arm.c (arm_coproc_builtin_available): Add
25618 support for mcr, mrc, mcr2 and mrc2.
25619 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25620 (arm_mcr_qualifiers): ... this. New.
25621 (MRC_QUALIFIERS): Define to ...
25622 (arm_mrc_qualifiers): ... this. New.
25623 (MCR_QUALIFIERS): Define to ...
25624 (arm_mcr_qualifiers): ... this. New.
25625 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25627 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25628 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25629 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25630 VUNSPEC_MRC2): New.
25632 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25634 * config/arm/arm.md (*ldc): New.
25638 * config/arm/arm.c (arm_coproc_builtin_available): Add
25639 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25640 (arm_coproc_ldc_stc_legitimate_address): New.
25641 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25642 'qualifier_const_pointer'.
25643 (LDC_QUALIFIERS): Define to...
25644 (arm_ldc_qualifiers): ... this. New.
25645 (STC_QUALIFIERS): Define to...
25646 (arm_stc_qualifiers): ... this. New.
25647 * config/arm/arm-protos.h
25648 (arm_coproc_ldc_stc_legitimate_address): New.
25649 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25650 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25651 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25652 stc2, stcl, stc2l): New.
25653 * config/arm/constraints.md (Uz): New.
25654 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25655 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25656 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25657 VUNSPEC_STC2L): New.
25659 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25661 * config/arm/arm.md (<cdp>): New.
25662 * config/arm/arm.c (neon_const_bounds): Rename this ...
25663 (arm_const_bounds): ... this.
25664 (arm_coproc_builtin_available): New.
25665 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25666 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25667 (CDP_QUALIFIERS): Define to...
25668 (arm_cdp_qualifiers): ... this. New.
25670 (arm_expand_builtin_args): Add case for 6 arguments.
25671 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25672 (arm_const_bounds): ... this.
25673 (arm_coproc_builtin_available): New.
25674 * config/arm/arm_acle.h (__arm_cdp): New.
25676 * config/arm/arm_acle_builtins.def (cdp): New.
25678 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25679 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25680 'arm_const_bounds'.
25681 * config/arm/types.md (coproc): New.
25682 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25683 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25684 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25685 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25687 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25689 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25690 (UBINOP_QUALIFIERS): New.
25692 (acle_builtin_data): New. Change comment.
25693 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25694 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25695 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25696 arm_acle_builtins.def.
25697 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25698 (arm_init_acle_builtins): New.
25699 (CRC32_BUILTIN): Remove.
25700 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25701 crc32cb, crc32ch and crc32cw.
25702 (arm_init_crc32_builtins): Remove.
25703 (arm_init_builtins): Use arm_init_acle_builtins rather
25704 than arm_init_crc32_builtins.
25705 (arm_expand_acle_builtin): New.
25706 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25707 * config/arm/arm_acle_builtins.def: New.
25709 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25711 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25712 (arm_builtin_datum): ... this.
25713 (arm_init_neon_builtin): Rename to ...
25714 (arm_init_builtin): ... this. Add a new parameters PREFIX
25715 and USE_SIG_IN_NAME.
25716 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25717 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25718 'arm_builtin_datum'.
25719 (arm_init_vfp_builtins): Likewise.
25720 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25721 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25722 (arm_expand_neon_args): Rename to ...
25723 (arm_expand_builtin_args): ... this. Rename builtin_arg
25724 enum values and differentiate between ARG_BUILTIN_MEMORY
25725 and ARG_BUILTIN_NEON_MEMORY.
25726 (arm_expand_neon_builtin_1): Rename to ...
25727 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25728 values, arm_expand_builtin_args and add bool parameter NEON.
25729 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25730 (arm_expand_vfp_builtin): Likewise.
25731 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25733 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25735 PR middle-end/77484
25736 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25737 * predict.c (tree_estimate_probability_bb): Reverse direction of
25738 polymorphic call predictor.
25740 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25742 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25743 (determine_pass_name_match): ...this new function and...
25744 (should_skip_pass_p): ...this new function.
25746 2017-01-06 Nathan Sidwell <nathan@acm.org>
25748 * ipa-visibility.c (function_and_variable_visibility): Reformat
25749 comments and long lines. Remove extrneous if.
25750 * symtab.c (symtab_node::make_decl_local): Fix code format.
25751 (symtab_node::set_section_for_node): Fix comment typo.
25753 2017-01-06 Martin Liska <mliska@suse.cz>
25756 * lra-constraints.c: Rename invariant to lra_invariant.
25757 * predict.c (set_even_probabilities): Initialize e to NULL.
25759 2017-01-05 Martin Sebor <msebor@redhat.com>
25761 PR tree-optimization/78910
25762 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25763 (format_integer): Correct off-by-one error in the handling
25764 of precision with negative numbers in signed conversions..
25766 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25768 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25770 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25772 PR tree-optimization/71016
25773 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25774 factor_out_conditional_conversion. Formatting fix.
25775 (factor_out_conditional_conversion): Add cond_stmt argument.
25776 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25777 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25780 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25782 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25783 read-rtl-function.o, and selftest-rtl.o.
25784 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25785 (selftest::aarch64_test_loading_full_dump): New function.
25786 (selftest::aarch64_run_selftests): New function.
25787 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25788 selftest::aarch64_run_selftests.
25789 * config/i386/i386.c
25790 (selftest::ix86_test_loading_dump_fragment_1): New function.
25791 (selftest::ix86_test_loading_call_insn): New function.
25792 (selftest::ix86_test_loading_full_dump): New function.
25793 (selftest::ix86_test_loading_unspec): New function.
25794 (selftest::ix86_run_selftests): Call the new functions.
25795 * emit-rtl.c (maybe_set_max_label_num): New function.
25796 * emit-rtl.h (maybe_set_max_label_num): New decl.
25797 * function.c (instantiate_decls): Guard call to
25798 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25799 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25801 * gensupport.c (gen_reader::gen_reader): Pass "false"
25802 for new "compact" param of rtx_reader.
25803 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25804 rather than an empty string for NULL strings.
25805 * read-md.c: Potentially include config.h rather than bconfig.h.
25806 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25807 (have_error): New global, copied from errors.c.
25808 (md_reader::read_name): Rename to...
25809 (md_reader::read_name_1): ...this, adding "out_loc" param,
25810 and converting "missing name or number" to returning false, rather
25812 (md_reader::read_name): Reimplement in terms of read_name_1.
25813 (md_reader::read_name_or_nil): New function.
25814 (md_reader::read_string): Handle "(nil)" by returning NULL.
25815 (md_reader::md_reader): Add new param "compact".
25816 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25817 (md_reader::read_file): New method.
25818 * read-md.h (md_reader::md_reader): Add new param "compact".
25819 (md_reader::read_file): New method.
25820 (md_reader::is_compact): New accessor.
25821 (md_reader::read_name): Convert return type from void to file_location.
25822 (md_reader::read_name_or_nil): New decl.
25823 (md_reader::read_name_1): New decl.
25824 (md_reader::m_compact): New field.
25825 (noop_reader::noop_reader): Pass "false" for new "compact" param
25827 (rtx_reader::rtx_reader): Add new "compact" param.
25828 (rtx_reader::read_rtx_operand): Make virtual and convert return
25829 type from void to rtx.
25830 (rtx_reader::read_until): New decl.
25831 (rtx_reader::handle_any_trailing_information): New virtual function.
25832 (rtx_reader::postprocess): New virtual function.
25833 (rtx_reader::finalize_string): New virtual function.
25834 (rtx_reader::m_in_call_function_usage): New field.
25835 (rtx_reader::m_reuse_rtx_by_id): New field.
25836 * read-rtl-function.c: New file.
25837 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25838 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25839 (selftest::verify_three_block_rtl_cfg): New decl.
25840 * read-rtl-function.h: New file.
25841 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25842 For host, include function.h, memmodel.h, and emit-rtl.h.
25843 (one_time_initialization): New function.
25844 (struct compact_insn_name): New struct.
25845 (compact_insn_names): New array.
25846 (find_code): Handle insn codes in compact dumps.
25847 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25848 (bind_subst_iter_and_attr): Likewise.
25849 (add_condition_to_string): Likewise.
25850 (add_condition_to_rtx): Likewise.
25851 (apply_attribute_uses): Likewise.
25852 (add_current_iterators): Likewise.
25853 (apply_iterators): Likewise.
25854 (initialize_iterators): Guard usage of apply_subst_iterator with
25855 #ifdef GENERATOR_FILE.
25856 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25857 (md_reader::read_mapping): Likewise.
25858 (add_define_attr_for_define_subst): Likewise.
25859 (add_define_subst_attr): Likewise.
25860 (read_subst_mapping): Likewise.
25861 (check_code_iterator): Likewise.
25862 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25864 (one_time_initialization): New function.
25865 (rtx_reader::read_until): New method.
25866 (read_flags): New function.
25867 (parse_reg_note_name): New function.
25868 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25869 Handle reuse_rtx ids.
25870 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25871 Add parsing support for RTL dumps, mirroring the special-cases in
25872 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25873 values, and calling handle_any_trailing_information.
25874 (rtx_reader::read_rtx_operand): Convert return type from void
25875 to rtx, returning return_rtx. Handle case 'e'. Call
25876 finalize_string on XSTR and XTMPL fields.
25877 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25878 "(nil)" values were omitted. Call the postprocess vfunc on the
25880 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25881 class ctor. Initialize m_in_call_function_usage. Call
25882 one_time_initialization.
25883 * rtl-tests.c (selftest::test_uncond_jump): Call
25884 set_new_first_and_last_insn.
25885 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25886 * selftest-rtl.c: New file.
25887 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25888 (selftest::get_insn_by_uid): New decl.
25889 * selftest-run-tests.c (selftest::run_tests): Call
25890 read_rtl_function_c_tests.
25891 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25892 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25895 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25897 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25898 operands in a special way. Assert that pos+len <= mode precision.
25900 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25902 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25903 3 argument Alias with unlimited for the negative form.
25904 (fno-vect-cost-model): Removed.
25906 2017-01-05 Martin Liska <mliska@suse.cz>
25908 * hsa-gen.c (gen_hsa_divmod): New function.
25909 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25911 2017-01-05 Martin Liska <mliska@suse.cz>
25914 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25917 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25919 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25920 small constant length operands.
25922 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25924 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25925 between loop iterations.
25927 2017-01-05 Martin Liska <mliska@suse.cz>
25930 * gimplify.c (gimplify_decl_expr): Compare to
25931 asan_poisoned_variables instread of checking flags.
25932 (gimplify_target_expr): Likewise.
25933 (gimplify_expr): Likewise.
25934 (gimplify_function_tree): Conditionally initialize
25935 asan_poisoned_variables.
25937 2017-01-04 Jeff Law <law@redhat.com>
25939 PR tree-optimizatin/78812
25940 * rtl.h (contains_mem_rtx_p): Prototype.
25941 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25942 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25943 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25944 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25945 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25947 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25949 * input.c (assert_char_at_range): Default-initialize actual_range.
25951 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25953 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25954 to match the caller.
25956 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25958 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25959 insns after final jump in test to emit dummy move.
25961 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25963 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25964 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25966 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25968 * multiple_target.c (create_dispatcher_calls): Init e_next.
25969 * tree-ssa-loop-split.c (split_loop): Init border.
25970 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25973 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25978 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25979 (altivec_register_operand): Do not return true if the operand
25980 contains a SUBREG mixing SImode and SFmode.
25981 (vsx_register_operand): Likewise.
25982 (vsx_reg_sfsubreg_ok): New predicate.
25983 (vfloat_operand): Do not return true if the operand contains a
25984 SUBREG mixing SImode and SFmode.
25985 (vint_operand): Likewise.
25986 (vlogical_operand): Likewise.
25987 (gpc_reg_operand): Likewise.
25988 (int_reg_operand): Likewise.
25989 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25990 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25991 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25993 (rs6000_emit_move_si_sf_subreg): New helper function.
25994 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25995 fixup SUBREGs involving SImode and SFmode.
25996 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25997 numbers for the new peephole2 optimization.
25998 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25999 the GLIBC math library that do AND/IOR/XOR operations on single
26000 precision floating point.
26001 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26002 target macros to say whether we need to avoid SUBREGs mixing
26004 (TARGET_ALLOW_SF_SUBREG): Likewise.
26005 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26006 (UNSPEC_SI_FROM_SF): Likewise.
26007 (iorxor): Change spacing.
26008 (and_ior_xor): New iterator for AND, IOR, and XOR.
26009 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26010 (movdi_from_sf_zero_ext): Likewise.
26011 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26012 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26013 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26014 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26015 (fms<mode>4): Likewise.
26016 (fnma<mode>4): Likewise.
26017 (fnms<mode>4): Likewise.
26018 (nfma<mode>4): Likewise.
26019 (nfms<mode>4): Likewise.
26021 2017-01-04 Marek Polacek <polacek@redhat.com>
26024 * doc/invoke.texi: Document -Wpointer-compare.
26026 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26028 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26031 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26032 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26035 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26038 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26039 documentation of the powerpc_popcntb_ok attribute.
26040 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26041 code to issue warning messages if a requested CPU configuration is
26042 not supported by the binary (assembler and loader) toolchain.
26043 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26044 made to define a built-in function that has been disabled.
26045 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26046 made to define a built-in function that has been disabled.
26047 (altivec_init_builtins): Add comment explaining why definition
26048 of the DST built-in functions is not preceded by an assertion
26049 check. Add assertions to prevent ICE if attempts are made to
26050 define an altivec predicate or an abs* built-in function that has
26052 (htm_init_builtins): Add comment explaining why definition of the
26053 htm built-in functions is not preceded by an assertion check.
26055 2017-01-04 Jeff Law <law@redhat.com>
26057 PR tree-optimizatin/67955
26058 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26059 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26060 the points-to solution does not include pt_null. Use DECL_PT_UID
26063 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26065 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26066 Use gen_int_mode instead of gen_lopwart for const_int operands.
26068 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26070 PR tree-optimization/71563
26071 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26072 out of range value - has low bits known to be zero.
26074 2017-01-04 Alan Modra <amodra@gmail.com>
26076 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26077 * configure: Regenerate.
26078 * config.in: Regenerate.
26080 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26083 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26084 a substring of the message, but strcmp with the whole message. Ifdef
26085 ENABLE_NLS, translate the message first using dgettext.
26087 2017-01-03 Jeff Law <law@redhat.com>
26089 PR tree-optimizatin/78856
26090 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26091 (mark_threaded_blocks): Remove code to truncate thread paths that
26092 cross multiple loop headers. Instead invalidate the cached loop
26093 iteration information and handle case of a thread path walking
26094 into an irreducible region.
26096 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26099 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26100 assertions. Add support for doing the signbit if the IEEE 128-bit
26101 floating point value is in a GPR.
26102 * config/rs6000/rs6000.md (Fsignbit): Delete.
26103 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26104 Update the length attribute if the value is in a GPR.
26105 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26106 the sign or zero extension instruction, since the value is always 0/1.
26107 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26110 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26111 extracting SImode to a GPR register so that we can generate a
26112 store, limit the vector to be in a traditional Altivec register
26113 for the vextuwrx instruction.
26115 2017-01-03 Ian Lance Taylor <iant@google.com>
26117 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26119 2017-01-03 Martin Sebor <msebor@redhat.com>
26121 PR tree-optimization/78696
26122 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26123 precision. Use MPFR for %f for greater fidelity. Correct handling
26125 (pass_sprintf_length::compute_format_length): Set width and precision
26126 specified by asrerisk to void_node for vararg functions.
26127 (try_substitute_return_value): Adjust dump output.
26129 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26131 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26133 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26135 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26136 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26137 -mlra/-mno-lra was passed to the compiler.
26139 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26141 PR rtl-optimization/65618
26142 * emit-rtl.c (try_split): Move initialization of "before" and
26143 "after" to just before the call to emit_insn_after_setloc.
26145 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26147 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26149 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26151 * dwarf2out.c (gen_enumeration_type_die): When
26152 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26154 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26156 PR tree-optimization/78965
26157 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26158 Change first argument from const call_info & to call_info &. For %n
26159 set info.nowrite to false.
26161 PR middle-end/78901
26162 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26163 possibly throwing calls.
26165 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26166 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26167 and fns handling, rather than in a separate case SSA_NAME.
26169 2017-01-02 Jeff Law <law@redhat.com>
26171 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26172 fixes for first_period and second_period variables.
26174 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26177 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26178 (*insvqi_1): New insn pattern.
26179 (*insvqi_1_mem_rex64): Ditto.
26180 (*insvqi_2): Ditto.
26181 (*insvqi_3): Rename from *insvqi.
26183 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26185 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26187 * doc/cfg.texi (Edges): Remove reference to Java.
26188 (Maintaining the CFG): Ditto.
26190 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26192 PR middle-end/77674
26193 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26194 transparent aliases.
26196 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26198 PR middle-end/77484
26199 * predict.def (PRED_CALL): Update hitrate.
26200 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26201 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26202 into direct/indirect/polymorphic variants.
26204 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26206 Update copyright years.
26208 * gcc.c (process_command): Update copyright notice dates.
26209 * gcov-dump.c (print_version): Ditto.
26210 * gcov.c (print_version): Ditto.
26211 * gcov-tool.c (print_version): Ditto.
26212 * gengtype.c (create_file): Ditto.
26213 * doc/cpp.texi: Bump @copying's copyright year.
26214 * doc/cppinternals.texi: Ditto.
26215 * doc/gcc.texi: Ditto.
26216 * doc/gccint.texi: Ditto.
26217 * doc/gcov.texi: Ditto.
26218 * doc/install.texi: Ditto.
26219 * doc/invoke.texi: Ditto.
26221 Copyright (C) 2017 Free Software Foundation, Inc.
26223 Copying and distribution of this file, with or without modification,
26224 are permitted in any medium without royalty provided the copyright
26225 notice and this notice are preserved.