1 2017-09-13 Richard Biener <rguenther@suse.de>
3 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
6 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11 Disable pc relative literal load irrespective of
12 TARGET_FIX_ERR_A53_84341 for default.
14 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16 * config/sparc/sparc.c (output_return): Output the source location of
17 the insn in the delay slot, if any.
18 (output_sibcall): Likewise.
20 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
23 * common.opt (-static-pie): New alias.
24 (shared): Negate static-pie.
25 (-no-pie): Update help text.
27 (static-pie): New option.
28 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
30 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
31 (LINK_EH_SPEC): Likewise.
32 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
33 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
34 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
35 * gcc.c (LINK_COMMAND_SPEC): Likewise.
36 (init_gcc_specs): Likewise.
37 (init_spec): Likewise.
38 (display_help): Update help message for -pie.
39 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
42 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
44 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
45 (movdi_aarch64): Likewise.
46 (movti_aarch64): Likewise.
48 2017-09-12 Simon Wright <simon@pushface.org>
51 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
52 calculation of the minor version, always output as 0.
54 2017-09-12 Jakub Jelinek <jakub@redhat.com>
57 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
58 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
59 on it early, rather than manual conversion late. For
60 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
61 instead of performing manual conversion.
63 2017-09-12 Carl Love <cel@us.ibm.com>
65 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
66 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
68 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
69 VMULOSW): Add definitions.
70 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
71 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
72 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
73 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
74 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
76 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
78 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
80 (movti_aarch64): Likewise.
81 (movdf_aarch64): Likewise.
82 (movtf_aarch64): Likewise.
83 (load_pairdi): Likewise.
84 (store_pairdi): Likewise.
85 (load_pairdf): Likewise.
86 (store_pairdf): Likewise.
87 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
88 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
89 (ldr_got_small_<mode>): Likewise.
90 (ldr_got_small_28k_<mode>): Likewise.
91 (ldr_got_tiny): Likewise.
92 * config/aarch64/iterators.md (ldst_sz): New.
93 (ldpstp_sz): Likewise.
94 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
96 (thunderx_load): Split load_8 to load_16.
97 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
99 (thunderx2t99_storepair_basic): Split store_8 to store_16.
100 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
101 (xgene1_store_pair): Split store_8 to store_16.
102 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
103 (falkor_st_0_st_sd): Split store_8 to store_16.
105 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
107 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
108 and store1/2/3/4 to store_4/8/12/16.
109 * config/aarch64/aarch64.md: Update for rename.
110 * config/arm/arm.md: Likewise.: Likewise.
111 * config/arm/arm.c: Likewise.
112 * config/arm/thumb1.md: Likewise.
113 * config/arm/thumb2.md: Likewise.
114 * config/arm/vfp.md: Likewise.
115 * config/arm/arm-generic.md: Likewise.
116 * config/arm/arm1020e.md: Likewise.
117 * config/arm/arm1026ejs.md: Likewise.
118 * config/arm/arm1136jfs.md: Likewise.
119 * config/arm/arm926ejs.md: Likewise.
120 * config/arm/cortex-a15.md: Likewise.
121 * config/arm/cortex-a17.md: Likewise.
122 * config/arm/cortex-a5.md: Likewise.
123 * config/arm/cortex-a53.md: Likewise.
124 * config/arm/cortex-a57.md: Likewise.
125 * config/arm/cortex-a7.md: Likewise.
126 * config/arm/cortex-a8.md: Likewise.
127 * config/arm/cortex-a9.md: Likewise.
128 * config/arm/cortex-m4.md: Likewise.
129 * config/arm/cortex-m7.md: Likewise.
130 * config/arm/cortex-r4.md: Likewise.
131 * config/arm/exynos-m1.md: Likewise.
132 * config/arm/fa526.md: Likewise.
133 * config/arm/fa606te.md: Likewise.
134 * config/arm/fa626te.md: Likewise.
135 * config/arm/fa726te.md: Likewise.
136 * config/arm/fmp626.md: Likewise.
137 * config/arm/iwmmxt.md: Likewise.
138 * config/arm/ldmstm.md: Likewise.
139 * config/arm/marvell-pj4.md: Likewise.
140 * config/arm/xgene1.md: Likewise.
141 * config/aarch64/thunderx.md: Likewise.
142 * config/aarch64/thunderx2t99.md: Likewise.
143 * config/aarch64/falkor.md: Likewise.
145 2017-09-12 Martin Liska <mliska@suse.cz>
147 * attribs.c (private_lookup_attribute): New function.
148 * attribs.h (private_lookup_attribute): Declared here.
149 (lookup_attribute): Called from this place.
151 2017-09-12 Richard Biener <rguenther@suse.de>
153 PR tree-optimization/82157
154 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
155 stmts with side-effects.
157 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
158 Alan Hayward <alan.hayward@arm.com>
159 David Sherwood <david.sherwood@arm.com>
161 * target.def (hard_regno_nregs): New hook.
162 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
163 * targhooks.h (default_hard_regno_nregs): Declare.
164 * targhooks.c (default_hard_regno_nregs): New function.
165 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
166 (TARGET_HARD_REGNO_NREGS): ...this hook.
167 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
168 (CLASS_MAX_NREGS): Likewise.
169 * doc/tm.texi: Regenerate.
170 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
171 instead of HARD_REGNO_NREGS.
172 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
173 HARD_REGNO_NREGS in the comment.
174 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
175 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
176 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
177 Return an unsigned int.
178 (TARGET_HARD_REGNO_NREGS): Redefine.
179 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
180 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
181 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
182 (arc_hard_regno_nregs): New function.
183 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
184 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
185 (arm_hard_regno_nregs): New function.
186 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
187 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
188 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
189 (TARGET_HARD_REGNO_NREGS): Redefine.
190 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
191 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
192 (HARD_REGNO_NREGS): Delete.
193 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
194 (cr16_hard_regno_nregs): New function.
195 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
196 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
197 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
198 (cris_hard_regno_nregs): New function.
199 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
200 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
201 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
202 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
203 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
204 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
205 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
206 (frv_hard_regno_nregs): Make static. Take and return an
208 (frv_class_max_nregs): Remove outdated copy of documentation.
209 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
210 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
211 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
212 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
213 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
214 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
215 (TARGET_HARD_REGNO_NREGS): Redefine.
216 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
217 (CLASS_MAX_NREGS): Update comment.
218 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
219 (ia64_hard_regno_nregs): New function.
220 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
221 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
222 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
223 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
224 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
226 (m32c_hard_regno_nregs): Likewise. Make static.
227 (TARGET_HARD_REGNO_NREGS): Redefine.
228 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
229 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
230 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
231 (m68k_hard_regno_nregs): New function.
232 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
233 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
234 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
235 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
236 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
237 Take and return an unsigned int.
238 (TARGET_HARD_REGNO_NREGS): Redefine.
239 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
240 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
241 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
242 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
243 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
244 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
245 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
246 (msp430_hard_regno_nregs): Make static. Take and return an
248 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
249 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
250 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
251 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
252 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
253 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
254 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
255 (TARGET_HARD_REGNO_NREGS): Redefine.
256 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
257 (PA_HARD_REGNO_NREGS): ...this.
258 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
259 (PA_HARD_REGNO_NREGS): ...this.
260 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
261 (pa_hard_regno_nregs): New function.
262 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
263 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
264 (pdp11_hard_regno_nregs): New function.
265 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
266 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
267 (rs6000_hard_regno_nregs_hook): New function.
268 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
269 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
270 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
271 Take and return an unsigned int. Move earlier in file.
272 (TARGET_HARD_REGNO_NREGS): Redefine.
273 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
274 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
275 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
276 (rl78_hard_regno_nregs): Make static. Take and return an
278 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
279 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
280 (rs6000_hard_regno_nregs_hook): New function.
281 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
282 * config/rx/rx.c (rx_hard_regno_nregs): New function.
283 (TARGET_HARD_REGNO_NREGS): Redefine.
284 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
285 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
286 instead of HARD_REGNO_NREGS.
287 (s390_hard_regno_nregs): New function.
288 (s390_hard_regno_mode_ok): Add comment from s390.h.
289 (TARGET_HARD_REGNO_NREGS): Redefine.
290 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
291 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
292 (sh_hard_regno_nregs): New function.
293 (sh_pass_in_reg_p): Use it.
294 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
295 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
296 (sparc_hard_regno_nregs): New function.
297 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
298 * config/spu/spu.c (spu_hard_regno_nregs): New function.
299 (spu_function_arg_advance): Use it, supplying a valid register number.
300 (TARGET_HARD_REGNO_NREGS): Redefine.
301 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
302 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
303 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
304 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
305 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
306 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
307 (CLASS_MAX_NREGS): Remove copy of old documentation.
308 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
309 (visium_hard_regno_nregs): New function.
310 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
311 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
312 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
313 xtensa_hard_regno_nregs): New function.
314 * system.h (HARD_REGNO_NREGS): Poison.
316 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
318 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
319 hard_regno_nregs instead of HARD_REGNO_NREGS.
320 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
321 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
322 (c6x_expand_epilogue): Likewise.
323 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
324 (frv_read_iacc_argument): Likewise.
325 * config/sh/sh.c: Include regs.h.
326 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
327 (regs_used): Likewise.
328 (output_stack_adjust): Likewise.
329 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
330 * expmed.c: Include regs.h.
331 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
332 * ree.c: Include regs.h.
333 (combine_reaching_defs): Use hard_regno_nregs instead of
335 (add_removable_extension): Likewise.
337 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
339 * regs.h (hard_regno_nregs): Turn into a function.
340 (end_hard_regno): Update accordingly.
341 * caller-save.c (setup_save_areas): Likewise.
342 (save_call_clobbered_regs): Likewise.
343 (replace_reg_with_saved_mem): Likewise.
344 (insert_restore): Likewise.
345 (insert_save): Likewise.
346 * combine.c (can_change_dest_mode): Likewise.
347 (move_deaths): Likewise.
348 (distribute_notes): Likewise.
349 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
350 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
351 (rs6000_split_multireg_move): Likewise.
352 (rs6000_register_move_cost): Likewise.
353 (rs6000_memory_move_cost): Likewise.
354 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
355 (rs6000_split_multireg_move): Likewise.
356 (rs6000_register_move_cost): Likewise.
357 (rs6000_memory_move_cost): Likewise.
358 * cselib.c (cselib_reset_table): Likewise.
359 (cselib_lookup_1): Likewise.
360 * emit-rtl.c (set_mode_and_regno): Likewise.
361 * function.c (aggregate_value_p): Likewise.
362 * ira-color.c (setup_profitable_hard_regs): Likewise.
363 (check_hard_reg_p): Likewise.
364 (calculate_saved_nregs): Likewise.
365 (assign_hard_reg): Likewise.
366 (improve_allocation): Likewise.
367 (calculate_spill_cost): Likewise.
368 * ira-emit.c (modify_move_list): Likewise.
369 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
370 (ira_hard_reg_in_set_p): Likewise.
371 * ira.c (setup_reg_mode_hard_regset): Likewise.
372 (clarify_prohibited_class_mode_regs): Likewise.
373 (check_allocation): Likewise.
374 * lra-assigns.c (find_hard_regno_for_1): Likewise.
375 (lra_setup_reg_renumber): Likewise.
376 (setup_try_hard_regno_pseudos): Likewise.
377 (spill_for): Likewise.
378 (assign_hard_regno): Likewise.
379 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
380 * lra-constraints.c (in_class_p): Likewise.
381 (lra_constraint_offset): Likewise.
382 (simplify_operand_subreg): Likewise.
383 (lra_constraints): Likewise.
384 (split_reg): Likewise.
385 (split_if_necessary): Likewise.
386 (invariant_p): Likewise.
387 (inherit_in_ebb): Likewise.
388 * lra-lives.c (process_bb_lives): Likewise.
389 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
390 (get_hard_regs): Likewise.
391 (do_remat): Likewise.
392 * lra-spills.c (assign_spill_hard_regs): Likewise.
393 * mode-switching.c (create_pre_exit): Likewise.
394 * postreload.c (reload_combine_recognize_pattern): Likewise.
395 * recog.c (peep2_find_free_register): Likewise.
396 * regcprop.c (kill_value_regno): Likewise.
397 (set_value_regno): Likewise.
398 (copy_value): Likewise.
399 (maybe_mode_change): Likewise.
400 (find_oldest_value_reg): Likewise.
401 (copyprop_hardreg_forward_1): Likewise.
402 * regrename.c (check_new_reg_p): Likewise.
403 (regrename_do_replace): Likewise.
404 * reload.c (push_reload): Likewise.
405 (combine_reloads): Likewise.
406 (find_dummy_reload): Likewise.
407 (operands_match_p): Likewise.
408 (find_reloads): Likewise.
409 (find_equiv_reg): Likewise.
410 (reload_adjust_reg_for_mode): Likewise.
411 * reload1.c (count_pseudo): Likewise.
412 (count_spilled_pseudo): Likewise.
413 (find_reg): Likewise.
414 (clear_reload_reg_in_use): Likewise.
415 (free_for_value_p): Likewise.
416 (allocate_reload_reg): Likewise.
417 (choose_reload_regs): Likewise.
418 (reload_adjust_reg_for_temp): Likewise.
419 (emit_reload_insns): Likewise.
420 (delete_output_reload): Likewise.
421 * rtlanal.c (subreg_get_info): Likewise.
422 * sched-deps.c (sched_analyze_reg): Likewise.
423 * sel-sched.c (init_regs_for_mode): Likewise.
424 (mark_unavailable_hard_regs): Likewise.
425 (choose_best_reg_1): Likewise.
426 (verify_target_availability): Likewise.
427 * valtrack.c (dead_debug_insert_temp): Likewise.
428 * var-tracking.c (track_loc_p): Likewise.
429 (emit_note_insn_var_location): Likewise.
430 * varasm.c (make_decl_rtl): Likewise.
431 * reginfo.c (choose_hard_reg_mode): Likewise.
432 (init_reg_modes_target): Refer directly to
433 this_target_regs->x_hard_regno_nregs.
435 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
437 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
438 instead of hard_regno_nregs.
440 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
442 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
443 end_hard_regno instead of hard_regno_nregs.
444 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
445 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
446 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
447 * ira-color.c (improve_allocation): Likewise.
448 * lra-assigns.c (find_hard_regno_for_1): Likewise.
449 * lra-lives.c (mark_regno_live): Likewise.
450 (mark_regno_dead): Likewise.
451 * lra-remat.c (operand_to_remat): Likewise.
452 * lra.c (collect_non_operand_hard_regs): Likewise.
453 * postreload.c (reload_combine_note_store): Likewise.
454 (move2add_valid_value_p): Likewise.
455 * reload.c (regno_clobbered_p): Likewise.
457 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
459 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
461 * config/v850/v850.c (v850_reorg): Likewise.
462 * reload.c (refers_to_regno_for_reload_p): Likewise.
463 (find_equiv_reg): Likewise.
464 * reload1.c (reload_reg_reaches_end_p): Likewise.
466 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
468 * caller-save.c (add_used_regs): Use REG_NREGS instead of
470 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
471 * config/arm/arm.c (output_move_neon): Likewise.
472 (arm_attr_length_move_neon): Likewise.
473 (neon_split_vcombine): Likewise.
474 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
475 (c6x_mark_reg_written): Likewise.
476 (c6x_dwarf_register_span): Likewise.
477 * config/i386/i386.c (ix86_save_reg): Likewise.
478 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
479 (rws_access_reg): Likewise.
480 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
481 * mode-switching.c (create_pre_exit): Likewise.
482 * ree.c (combine_reaching_defs): Likewise.
483 (add_removable_extension): Likewise.
484 * regcprop.c (find_oldest_value_reg): Likewise.
485 (copyprop_hardreg_forward_1): Likewise.
486 * reload.c (reload_inner_reg_of_subreg): Likewise.
487 (push_reload): Likewise.
488 (combine_reloads): Likewise.
489 (find_dummy_reload): Likewise.
490 (reload_adjust_reg_for_mode): Likewise.
491 * reload1.c (find_reload_regs): Likewise.
492 (forget_old_reloads_1): Likewise.
493 (reload_reg_free_for_value_p): Likewise.
494 (reload_adjust_reg_for_temp): Likewise.
495 (emit_reload_insns): Likewise.
496 (delete_output_reload): Likewise.
497 * sel-sched.c (choose_best_reg_1): Likewise.
498 (choose_best_pseudo_reg): Likewise.
500 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
501 Alan Hayward <alan.hayward@arm.com>
502 David Sherwood <david.sherwood@arm.com>
504 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
505 * target.def (slow_unaligned_access): New hook.
506 * targhooks.h (default_slow_unaligned_access): Declare.
507 * targhooks.c (default_slow_unaligned_access): New function.
508 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
509 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
510 * doc/tm.texi: Regenerate.
511 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
512 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
513 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
515 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
516 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
518 (rs6000_slow_unaligned_access): New function.
519 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
520 (expand_block_compare): Likewise.
521 (expand_strn_compare): Likewise.
522 (rs6000_rtx_costs): Likewise.
523 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
524 (riscv_slow_unaligned_access): Likewise.
525 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
526 (riscv_slow_unaligned_access_p): ...this and make static.
527 (riscv_option_override): Update accordingly.
528 (riscv_slow_unaligned_access): New function.
529 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
530 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
531 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
532 (rs6000_slow_unaligned_access): New function.
533 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
534 (rs6000_rtx_costs): Likewise.
535 * config/rs6000/rs6000-string.c (expand_block_compare)
536 (expand_strn_compare): Use targetm.slow_unaligned_access instead
537 of SLOW_UNALIGNED_ACCESS.
538 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
539 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
540 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
541 of SLOW_UNALIGNED_ACCESS.
542 * expmed.c (simple_mem_bitfield_p): Likewise.
543 * expr.c (alignment_for_piecewise_move): Likewise.
544 (emit_group_load_1): Likewise.
545 (emit_group_store): Likewise.
546 (copy_blkmode_from_reg): Likewise.
547 (emit_push_insn): Likewise.
548 (expand_assignment): Likewise.
549 (store_field): Likewise.
550 (expand_expr_real_1): Likewise.
551 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
552 * lra-constraints.c (simplify_operand_subreg): Likewise.
553 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
554 * gimple-ssa-store-merging.c: Likewise in block comment at start
556 * tree-ssa-strlen.c: Include target.h.
557 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
558 of SLOW_UNALIGNED_ACCESS.
559 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
561 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
563 PR rtl-optimization/82185
564 * expmed.c (emit_store_flag_int): Only test tem if it has been
567 2017-09-12 Richard Biener <rguenther@suse.de>
570 * match.pd ((FTYPE) N CMP CST): Fix typo.
572 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
574 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
576 (mips_near_type_p): Add 'short_call' attribute as a synonym
578 * doc/extend.texi (short_call): Document new function attribute.
580 2017-09-12 Jakub Jelinek <jakub@redhat.com>
583 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
584 assertion check that in the condition.
585 (get_atomic_generic_size): Likewise. Before testing if parameter
586 has pointer type, if it has array type, call for C++
587 default_conversion to perform array-to-pointer conversion.
589 2017-09-12 Richard Biener <rguenther@suse.de>
591 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
592 for operations we cannot scalarize.
594 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
596 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
597 vectors heap vectors. Clean up comments.
598 Make visited_bbs a reference.
599 (profitable_jump_thread_path): Make GC
600 vectors heap vectors. Clean up comments.
602 (convert_and_register_jump_thread_path): Make GC vectors heap
604 (check_subpath_and_update_thread_path): Same. Clean up comments.
605 Make visited_bbs a reference.
606 (handle_phi): Abstract common code to to
607 register_jump_thread_path_if_profitable.
608 Rename VAR_BB to DEF_BB.
610 Make GC vectors heap vectors.
611 Make visited_bbs a reference.
612 (handle_assignment): Same.
613 (register_jump_thread_path_if_profitable): New.
614 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
616 Make GC vectors heap vectors. Clean up comments.
617 Make visited_bbs a reference.
618 (find_jump_threads_backwards): Make visited_bbs live in the stack.
619 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
622 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
625 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
626 words of E_DImode object are reachable by xtensa_uimm8x4 access.
628 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
630 Revert r251800 and r251799.
632 2017-09-11 Martin Jambor <mjambor@suse.cz>
635 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
636 arguments in advance.
637 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
638 use it to find predecessor edges.
639 (naive_outof_ssa): Collect vector of predecessors.
641 2017-09-08 Jason Merrill <jason@redhat.com>
643 PR c++/70029 - ICE with ref-qualifier and -flto
644 * langhooks.h (struct lang_hooks_for_types): Add
645 copy_lang_qualifiers.
646 * attribs.c (build_type_attribute_qual_variant): Use it.
647 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
649 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
650 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
652 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
655 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
656 (*mulsi3_sp64): New instruction.
657 (mulsi3): New expander.
659 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
661 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
663 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
665 * sancov.c: Include memmodel.h.
667 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
670 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
673 2017-09-07 Carl Love <cel@us.ibm.com>
675 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
676 the sldi instruction.
678 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
680 * sancov.c: Include tm_p.h.
682 2017-09-07 Jakub Jelinek <jakub@redhat.com>
685 * output.h (switch_to_other_text_partition): New declaration.
686 * varasm.c (switch_to_other_text_partition): New function.
687 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
688 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
689 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
690 to the other text partition before emitting LCL label and switch back
691 after emitting the word after it.
693 2017-09-07 Richard Biener <rguenther@suse.de>
695 * passes.def (pass_split_crit_edges): Remove instance before PRE.
696 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
697 critical edges here, after loop init.
698 (pass_data_pre): Remove PROP_no_crit_edges flags.
699 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
700 for valueization of call args to avoid leaking VN_TOP.
701 (visit_use): Assert we do not visit default defs.
702 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
703 Use error_mark_node to more easily detect leaking VN_TOP.
704 All default-defs are varying, not VN_TOP. Mark them visited.
705 (run_scc_vn): Make code match comment.
707 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
709 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
710 OPTION_MASK_FLOAT128_KEYWORD.
711 (POWERPC_MASKS): Likewise.
712 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
713 support for the -mfloat128-type option, and make -mfloat128
714 default on PowerPC Linux systems. Define or undefine
715 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
716 Define __float128 to be __ieee128 if IEEE 128-bit support is
717 enabled, or undefine it.
718 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
719 Delete defining __FLOAT128_TYPE__.
720 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
721 -mfloat128-type option and make -mfloat128 default on PowerPC
723 (TARGET_FLOAT128_TYPE): Likewise.
724 (-mfloat128-type): Likewise.
725 * config/rs6000/rs6000.c (rs6000_option_override_internal):
726 Delete the -mfloat128-type option and make -mfloat128 default on
727 PowerPC Linux systems. Always use __ieee128 to be the keyword for
728 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
729 128-bit floating point is enabled. Change tests from using
730 -mfloat128-type to -mfloat128.
731 (rs6000_mangle_type): Use the correct mangling for the __float128
732 type even if normal long double is restricted to 64-bits.
733 (floatn_mode): Enable the _Float128 type by default on VSX Linux
735 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
736 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
737 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
739 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
740 documentation for -mfloat128.
742 2017-09-06 Olivier Hainque <hainque@adacore.com>
744 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
746 2017-09-06 Wish Wu <wishwu007@gmail.com>
747 Jakub Jelinek <jakub@redhat.com>
749 * asan.c (initialize_sanitizer_builtins): Add
750 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
751 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
752 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
753 BT_FN_VOID_UINT64_PTR variables.
754 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
755 (BT_FN_VOID_UINT16_UINT16): Likewise.
756 (BT_FN_VOID_UINT32_UINT32): Likewise.
757 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
758 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
759 (BT_FN_VOID_UINT64_PTR): Likewise.
760 * common.opt (flag_sanitize_coverage): New variable.
761 (fsanitize-coverage=trace-pc): Remove.
762 (fsanitize-coverage=): Add.
763 * flag-types.h (enum sanitize_coverage_code): New enum.
764 * fold-const.c (fold_range_test): Disable non-short-circuit
765 optimization if flag_sanitize_coverage.
766 (fold_truth_andor): Likewise.
767 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
768 * opts.c (COVERAGE_SANITIZER_OPT): Define.
769 (coverage_sanitizer_opts): New array.
770 (get_closest_sanitizer_option): Add OPTS argument, handle also
771 OPT_fsanitize_coverage_.
772 (parse_sanitizer_options): Adjusted to also handle
773 OPT_fsanitize_coverage_.
774 (common_handle_option): Add OPT_fsanitize_coverage_.
775 * sancov.c (instrument_comparison, instrument_switch): New function.
776 (sancov_pass): Add trace-cmp support.
777 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
778 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
779 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
780 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
781 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
782 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
783 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
784 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
785 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
786 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
788 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
790 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
791 error. Only quit immediately if parsing is complete.
792 (BEGIN): Initialize fatal_err and parse_done.
793 (begin fpu, end fpu): Check number of arguments.
794 (begin arch, end arch): Likewise.
795 (begin cpu, end cpu): Likewise.
796 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
797 (optalias): Likewise.
799 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
801 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
802 * config/arm/arm-isa.h: Delete. Move definitions to ...
803 * arm-cpus.in: ... here. Use new feature and fgroup values.
804 * config/arm/arm.c (arm_option_override): Use lower case for feature
806 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
807 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
808 * config/arm/parsecpu.awk (END): Add new command 'isa'.
810 (print_isa_bits_for): New function.
811 (gen_isa): New function.
812 (gen_comm_data): Use print_isa_bits_for.
813 (define feature): New keyword.
814 (define fgroup): New keyword.
815 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
816 (arm-isa.h): Add rule to generate file.
817 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
818 case for feature bit names.
820 2017-09-06 Richard Biener <rguenther@suse.de>
822 * tree-ssa-pre.c (NECESSARY): Remove.
823 (create_expression_by_pieces): Do not touch pass-local flags.
824 (insert_into_preds_of_block): Likewise.
825 (do_pre_regular_insertion): Likewise.
826 (eliminate_insert): Likewise.
827 (eliminate_dom_walker::before_dom_children): Likewise.
828 (fini_eliminate): Do not look at inserted_exprs.
829 (mark_operand_necessary): Remove.
830 (remove_dead_inserted_code): Replace with simple work-list
831 algorithm based on inserted_exprs and SSA uses.
832 (pass_pre::execute): Re-order fini_eliminate and
833 remove_dead_inserted_code.
835 2017-09-06 Olivier Hainque <hainque@adacore.com>
837 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
838 for VxWorks 7. Adjust surrounding comments.
840 2017-09-06 Richard Biener <rguenther@suse.de>
842 * gimple-ssa-strength-reduction.c
843 (find_candidates_dom_walker::before_dom_children): Also allow
846 2017-09-06 Richard Biener <rguenther@suse.de>
848 PR tree-optimization/82108
849 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
850 for gap in the non-permutation SLP case.
852 2017-09-06 Martin Jambor <mjambor@suse.cz>
854 PR tree-optimization/82078
855 * tree-sra.c (sort_and_splice_var_accesses): Move call to
856 add_access_to_work_queue...
857 (build_accesses_from_assign): ...here.
858 (propagate_all_subaccesses): Make sure racc is the group
859 representative, if there is one.
861 2017-09-06 Jakub Jelinek <jakub@redhat.com>
864 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
867 2017-09-06 Richard Biener <rguenther@suse.de>
869 * gimple-ssa-strength-reduction.c
870 (find_candidates_dom_walker::before_doom_children): Use a
871 type and not a mode check.
873 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
876 * config/arm/predicates.md (arm_general_adddi_operand): Create new
878 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
880 2017-09-05 Jeff Law <law@redhat.com>
882 PR tree-optimization/64910
883 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
884 cases where we have 3 or more operands.
886 2017-09-05 Jakub Jelinek <jakub@redhat.com>
889 * omp-low.c (lower_omp_for): Recompute tree invariant if
890 gimple_omp_for_initial/final is ADDR_EXPR.
893 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
894 into gimple val before gimplification fo the COND_EXPR.
896 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
898 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
899 REGION_COPY argument.
900 (thread_through_all_blocks): Remove unused argument to
901 duplicate_thread_path.
903 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
904 Alan Hayward <alan.hayward@arm.com>
905 David Sherwood <david.sherwood@arm.com>
907 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
908 Take a scalar_mode rather than a machine_mode.
909 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
910 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
911 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
912 (aarch64_gen_adjusted_ldpstp): Likewise.
913 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
915 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
916 Alan Hayward <alan.hayward@arm.com>
917 David Sherwood <david.sherwood@arm.com>
919 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
920 Take a scalar_int_mode instead of a machine_mode.
921 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
922 (aarch64_output_scalar_simd_mov_immediate): Likewise.
923 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
924 (aarch64_simd_attr_length_rglist): Delete.
925 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
926 a scalar_int_mode instead of a machine_mode.
927 (aarch64_add_offset): Likewise.
928 (aarch64_internal_mov_immediate): Likewise
929 (aarch64_add_constant_internal): Likewise.
930 (aarch64_add_constant): Likewise.
931 (aarch64_movw_imm): Likewise.
932 (aarch64_rtx_arith_op_extract_p): Likewise.
933 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
934 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
935 Remove assert that the mode isn't a vector.
936 (aarch64_output_scalar_simd_mov_immediate): Likewise.
937 (aarch64_expand_mov_immediate): Update calls after above changes.
938 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
939 (aarch64_and_bitmask_imm): Check for scalar integer modes.
940 (aarch64_move_imm): Likewise.
941 (aarch64_can_const_movi_rtx_p): Likewise.
942 (aarch64_strip_extend): Likewise.
943 (aarch64_extr_rtx_p): Likewise.
944 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
945 a CONST_INT when the mode parameter is VOIDmode.
946 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
948 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
950 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
951 * stor-layout.c (bitwise_mode_for_mode): Likewise.
952 (bitwise_type_for_mode): Update accordingly.
954 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
956 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
957 * stor-layout.c (mode_for_size_tree): Likewise.
958 (mode_for_array): Update accordingly.
959 (layout_decl): Likewise.
960 (compute_record_mode): Likewise. Only set the mode once.
962 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
964 * target.def (get_mask_mode): Change return type to opt_mode.
966 * doc/tm.texi: Regenerate.
967 * targhooks.h (default_get_mask_mode): Return an opt_mode.
968 * targhooks.c (default_get_mask_mode): Likewise.
969 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
970 * optabs-query.c (can_vec_mask_load_store_p): Update use of
971 targetm.get_mask_mode.
972 * tree.c (build_truth_vector_type): Likewise.
974 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
976 * machmode.h (mode_for_vector): Return an opt_mode.
977 * stor-layout.c (mode_for_vector): Likewise.
978 (mode_for_int_vector): Update accordingly.
979 (layout_type): Likewise.
980 * config/i386/i386.c (emit_memmov): Likewise.
981 (ix86_expand_set_or_movmem): Likewise.
982 (ix86_expand_vector_init): Likewise.
983 (ix86_get_mask_mode): Likewise.
984 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
986 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
987 * expmed.c (extract_bit_field_1): Likewise.
988 * expr.c (expand_expr_real_2): Likewise.
989 * optabs-query.c (can_vec_perm_p): Likewise.
990 (can_vec_mask_load_store_p): Likewise.
991 * optabs.c (expand_vec_perm): Likewise.
992 * targhooks.c (default_get_mask_mode): Likewise.
993 * tree-vect-stmts.c (vectorizable_store): Likewise.
994 (vectorizable_load): Likewise.
995 (get_vectype_for_scalar_type_and_size): Likewise.
997 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
999 * machmode.h (mode_for_int_vector): New function.
1000 * stor-layout.c (mode_for_int_vector): Likewise.
1001 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1002 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1003 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1004 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1005 (s390_expand_vcond): Likewise.
1007 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1009 * machmode.h (opt_machine_mode): New type.
1010 (opt_mode<T>): Allow construction from anything that can be
1012 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1013 (mode_for_size): Return an opt_machine_mode.
1014 * stor-layout.c (mode_for_size): Likewise.
1015 (mode_for_size_tree): Update call accordingly.
1016 (bitwise_mode_for_mode): Likewise.
1017 (make_fract_type): Likewise.
1018 (make_accum_type): Likewise.
1019 * caller-save.c (replace_reg_with_saved_mem): Update call
1021 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1022 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1023 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1024 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1025 * expmed.c (extract_bit_field_1): Likewise.
1026 * reload.c (get_secondary_mem): Likewise.
1027 * varasm.c (assemble_integer): Likewise.
1028 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1031 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1033 * machmode.h (decimal_float_mode_for_size): New function.
1034 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1035 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1036 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1037 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1038 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1039 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1041 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1043 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1044 (get_builtin_sync_mode): Likewise.
1045 (expand_ifn_atomic_compare_exchange): Likewise.
1046 (expand_builtin_atomic_clear): Likewise.
1047 (expand_builtin_atomic_test_and_set): Likewise.
1048 (fold_builtin_atomic_always_lock_free): Likewise.
1049 * calls.c (compute_argument_addresses): Likewise.
1050 (emit_library_call_value_1): Likewise.
1051 (store_one_arg): Likewise.
1052 * combine.c (combine_instructions): Likewise.
1053 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1054 * config/arm/arm.c (arm_function_value): Likewise.
1055 (aapcs_allocate_return_reg): Likewise.
1056 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1057 * config/i386/i386.c (construct_container): Likewise.
1058 (ix86_gimplify_va_arg): Likewise.
1059 (ix86_expand_sse_cmp): Likewise.
1060 (emit_memmov): Likewise.
1061 (emit_memset): Likewise.
1062 (expand_small_movmem_or_setmem): Likewise.
1063 (ix86_expand_pextr): Likewise.
1064 (ix86_expand_pinsr): Likewise.
1065 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1066 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1068 * config/mips/mips.c (mips_function_value_1) Likewise.
1069 (mips_block_move_straight): Likewise.
1070 (mips_expand_ins_as_unaligned_store): Likewise.
1071 * config/powerpcspe/powerpcspe.c
1072 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1073 (rs6000_darwin64_record_arg_flush): Likewise.
1074 * config/rs6000/rs6000.c
1075 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1076 (rs6000_darwin64_record_arg_flush): Likewise.
1077 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1078 (sparc_function_value_1): Likewise.
1079 * config/spu/spu.c (adjust_operand): Likewise.
1080 (spu_emit_branch_or_set): Likewise.
1081 (arith_immediate_p): Likewise.
1082 * emit-rtl.c (gen_lowpart_common): Likewise.
1083 * expr.c (expand_expr_real_1): Likewise.
1084 * function.c (assign_parm_setup_block): Likewise.
1085 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1086 * reload1.c (alter_reg): Likewise.
1087 * stor-layout.c (mode_for_vector): Likewise.
1088 (layout_type): Likewise.
1090 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1092 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1093 (spu_convert_move): Likewise.
1094 * lower-subreg.c (resolve_simple_move): Likewise.
1096 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1099 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1100 define_insn to a define_expand.
1101 (altivec_vsum2sws_direct): New define_insn.
1102 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1104 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1106 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1109 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1114 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1115 realign stack if stack alignment needed is less than incoming
1118 2017-09-05 Marek Polacek <polacek@redhat.com>
1121 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1124 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1126 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1128 2017-09-05 Richard Biener <rguenther@suse.de>
1130 PR tree-optimization/82084
1131 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1133 2017-09-05 Richard Biener <rguenther@suse.de>
1135 PR tree-optimization/82102
1136 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1138 2017-09-05 Martin Liska <mliska@suse.cz>
1140 PR tree-optimization/82032
1141 * tree-cfg.c (generate_range_test): New function.
1142 * tree-cfg.h (generate_range_test): Declared here.
1143 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1144 (cleanup_control_expr_graph): Use it.
1145 * tree-switch-conversion.c (try_switch_expansion): Remove
1147 (emit_case_nodes): Use generate_range_test.
1149 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1152 * config/i386/i386.md (*<btsc><mode>_mask): Add
1153 TARGET_USE_BT to insn constraint.
1154 (*btr<mode>_mask): Ditto.
1156 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1158 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1159 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1161 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1164 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1165 TARGET_NEON and TARGET_IWMMXT.
1166 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1167 TARGET_NEON and TARGET_IWMMXT.
1168 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1170 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1172 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1173 (ix86_rewrite_tls_address): Ditto.
1174 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1175 (ix86_rewrite_tls_address_1): Ditto.
1176 (ix86_rewrite_tls_address): Ditto.
1177 * config/i386/predicates.md (tls_address_pattern): New predicate.
1178 * config/i386/i386.md (TLS address splitter): New splitter.
1180 2017-09-04 Richard Biener <rguenther@suse.de>
1182 PR tree-optimization/82084
1183 * fold-const.h (can_native_encode_string_p): Declare.
1184 * fold-const.c (can_native_encode_string_p): Factor out from ...
1185 (native_encode_string): ... here.
1186 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1187 vectorizing stores from constants we later cannot handle.
1189 2017-09-04 Marek Polacek <polacek@redhat.com>
1192 * doc/invoke.texi: Update -Wtautological-compare documentation.
1194 2017-09-04 Jeff Law <law@redhat.com>
1196 PR tree-optimization/64910
1197 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1198 swap the first and last operand if the last is a constant.
1200 2017-09-04 Marek Polacek <polacek@redhat.com>
1203 * convert.c (do_narrow): When sanitizing signed integer overflows,
1204 bail out for signed types.
1205 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1207 2017-09-04 Richard Biener <rguenther@suse.de>
1209 PR tree-optimization/82060
1210 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1211 Move devirtualization after stmt folding and before EH/AB/noreturn
1212 cleanup to get the stmt refs canonicalized. Use a bool instead
1213 of gimple_modified_p since that doesn't work for NOPs. Schedule
1214 NOPs generated by folding for removal.
1216 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1217 Alan Hayward <alan.hayward@arm.com>
1218 David Sherwood <david.sherwood@arm.com>
1220 * coretypes.h (pad_direction): New enum.
1221 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1222 (FUNCTION_ARG_PADDING): Likewise.
1223 * target.def (function_arg_padding): New hook.
1224 * targhooks.h (default_function_arg_padding): Declare.
1225 * targhooks.c (default_function_arg_padding): New function.
1226 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1227 (TARGET_FUNCTION_ARG_PADDING): ...this.
1228 * doc/tm.texi: Regenerate.
1229 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1230 instead of direction.
1231 (compute_argument_addresses): Likewise.
1232 (load_register_parameters): Likewise.
1233 (emit_library_call_value_1): Likewise.
1234 (store_one_arg): Use targetm.calls.function_arg_padding instead
1235 of FUNCTION_ARG_PADDING.
1236 (must_pass_in_stack_var_size_or_pad): Likewise.
1237 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1238 (emit_group_store): Likewise.
1239 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1240 instead of FUNCTION_ARG_PADDING.
1241 (emit_push_insn): Likewise, and propagate enum change throughout
1243 * function.h (direction): Delete.
1244 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1246 * function.c (assign_parm_find_stack_rtl): Likewise.
1247 (assign_parm_setup_block_p): Likewise.
1248 (assign_parm_setup_block): Likewise.
1249 (gimplify_parameters): Likewise.
1250 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1251 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1253 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1254 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1255 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1256 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1257 (aarch64_function_arg_padding): ...this new function.
1258 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1259 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1260 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1261 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1262 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1263 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1264 (arm_pad_arg_upward): Replace with...
1265 (arm_function_arg_padding): ...this new function.
1266 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1268 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1269 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1270 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1271 (ia64_hpux_function_arg_padding): Replace with...
1272 (ia64_function_arg_padding): ...this new function. Use pad_direction
1273 instead of direction. Check for TARGET_HPUX.
1274 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1275 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1276 (iq2000_function_arg_padding): New function.
1277 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1278 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1279 (mips_function_arg_padding): ...this new function.
1280 (mips_pad_reg_upward): Update accordingly.
1281 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1282 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1283 targetm.calls.function_arg_padding.
1284 (FUNCTION_ARG_PADDING): Delete.
1285 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1286 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1287 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1288 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1289 (nios2_block_reg_padding): Return pad_direction instead of direction.
1290 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1291 instead of direction.
1292 (nios2_function_arg_padding): Likewise. Make static.
1293 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1294 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1295 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1296 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1297 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1298 (pa_function_arg_padding): Make static. Return pad_direction instead
1300 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1301 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1302 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1303 instead of direction. Use targetm.calls.function_arg_padding.
1304 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1305 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1306 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1307 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1308 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1310 (function_arg_padding): Rename to...
1311 (rs6000_function_arg_padding): ...this. Make static. Return
1312 pad_direction instead of direction.
1313 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1314 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1315 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1316 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1317 instead of direction. Use targetm.calls.function_arg_padding.
1318 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1319 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1320 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1321 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1322 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1323 (function_arg_padding): Rename to...
1324 (rs6000_function_arg_padding): ...this. Make static. Return
1325 pad_direction instead of direction.
1326 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1327 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1328 * config/s390/s390.c (s390_function_arg_padding): New function.
1329 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1330 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1331 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1332 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1333 (function_arg_padding): Rename to...
1334 (sparc_function_arg_padding): ...this. Make static. Return
1335 pad_direction instead of direction.
1336 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1337 * config/spu/spu.c (spu_function_arg_padding): New function.
1338 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1339 * system.h (FUNCTION_ARG_PADDING): Poison.
1341 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1342 Alan Hayward <alan.hayward@arm.com>
1343 David Sherwood <david.sherwood@arm.com>
1345 * target.def (modes_tieable_p): New hook.
1346 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1347 (TARGET_MODES_TIEABLE_P): ...this.
1348 * doc/tm.texi.in: Regenerate.
1349 * hooks.h (hook_bool_mode_mode_true): Declare.
1350 * hooks.c (hook_bool_mode_mode_true): New function.
1351 * combine.c (subst): Use targetm.modes_tieable_p instead of
1353 * dse.c (find_shift_sequence): Likewise.
1354 * expmed.c (extract_low_bits): Likewise.
1355 * lower-subreg.c: Include target.h.
1356 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1358 * rtlanal.c (rtx_cost): Likewise.
1359 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1360 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1361 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1362 (TARGET_MODES_TIEABLE_P): Redefine.
1363 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1364 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1365 (TARGET_MODES_TIEABLE_P): Redefine.
1366 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1367 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1368 (arc_modes_tieable_p): New function.
1369 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1370 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1371 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1372 (arm_modes_tieable_p): Make static.
1373 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1374 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1375 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1376 (TARGET_MODES_TIEABLE_P): Redefine.
1377 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1378 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1379 (TARGET_MODES_TIEABLE_P): Redefine.
1380 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1381 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1382 (cr16_modes_tieable_p): New function.
1383 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1384 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1385 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1386 (TRULY_NOOP_TRUNCATION): Update comment.
1387 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1388 (TRULY_NOOP_TRUNCATION): Update comment.
1389 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1390 (frv_modes_tieable_p): New function.
1391 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1392 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1393 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1394 (TARGET_MODES_TIEABLE_P): Redefine.
1395 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1396 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1397 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1398 (TARGET_MODES_TIEABLE_P): Redefine.
1399 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1400 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1401 (ia64_modes_tieable_p): New function.
1402 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1403 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1404 (iq2000_modes_tieable_p): New function.
1405 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1406 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1407 (lm32_modes_tieable_p): New function.
1408 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1409 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1410 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1411 (TARGET_MODES_TIEABLE_P): Redefine.
1412 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1413 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1414 (m32r_modes_tieable_p): New function.
1415 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1416 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1417 (m68k_modes_tieable_p): New function.
1418 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1419 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1420 (mcore_modes_tieable_p): New function.
1421 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1422 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1424 (TARGET_MODES_TIEABLE_P): Redefine.
1425 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1426 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1427 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1428 (TARGET_MODES_TIEABLE_P): Redefine.
1429 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1430 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1431 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1432 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1433 (mn10300_modes_tieable_p): ...this and make static.
1434 (TARGET_MODES_TIEABLE_P): Redefine.
1435 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1436 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1437 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1438 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1439 (msp430_modes_tieable_p): Make static.
1440 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1441 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1442 (TARGET_MODES_TIEABLE_P): Redefine.
1443 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1444 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1445 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1446 (TARGET_MODES_TIEABLE_P): Redefine.
1447 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1448 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1449 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1450 (TARGET_MODES_TIEABLE_P): Redefine.
1451 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1452 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1453 (pdp11_modes_tieable_p): New function.
1454 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1455 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1456 (rs6000_modes_tieable_p): New function.
1457 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1458 * config/powerpcspe/powerpcspe.md: Update comment.
1459 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1460 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1461 (TARGET_MODES_TIEABLE_P): Redefine.
1462 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1463 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1464 (rl78_modes_tieable_p): New function.
1465 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1466 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1467 (rs6000_modes_tieable_p): New function.
1468 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1469 * config/rs6000/rs6000.md: Update comment.
1470 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1471 * config/rx/rx.c (rx_modes_tieable_p): New function.
1472 (TARGET_MODES_TIEABLE_P): Redefine.
1473 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1474 * config/s390/s390.c (s390_modes_tieable_p): New function.
1475 (TARGET_MODES_TIEABLE_P): Redefine.
1476 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1477 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1478 (sh_modes_tieable_p): New function.
1479 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1480 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1481 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1482 (sparc_modes_tieable_p): Make static.
1483 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1484 * config/spu/spu.c (spu_modes_tieable_p): New function.
1485 (TARGET_MODES_TIEABLE_P): Redefine.
1486 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1487 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1488 (TARGET_MODES_TIEABLE_P): Redefine.
1489 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1490 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1491 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1492 * config/v850/v850.c (v850_modes_tieable_p): New function.
1493 (TARGET_MODES_TIEABLE_P): Redefine.
1494 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1495 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1496 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1497 (visium_modes_tieable_p): New function.
1498 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1499 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1500 (xtensa_modes_tieable_p): New function.
1501 * system.h (MODES_TIEABLE_P): Poison.
1503 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1504 Alan Hayward <alan.hayward@arm.com>
1505 David Sherwood <david.sherwood@arm.com>
1507 * target.def (hard_regno_mode_ok): New hook.
1508 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1509 (TARGET_HARD_REGNO_MODE_OK): ...this.
1510 * doc/tm.texi.in: Regenerate.
1511 * hooks.h (hook_bool_uint_mode_true): Declare.
1512 * hooks.c (hook_bool_uint_mode_true): New function.
1513 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1515 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1516 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1517 instead of HARD_REGNO_MODE_OK.
1518 * caller-save.c: Include target.h.
1519 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1521 * combine.c (can_combine_p): Likewise.
1522 (combinable_i3pat): Likewise.
1523 (can_change_dest_mode): Likewise.
1524 * expr.c (init_expr_target): Likewise.
1525 (convert_move): Likewise.
1526 (convert_modes): Likewise.
1527 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1528 (setup_prohibited_mode_move_regs): Likewise.
1529 * ira.h (target_ira): Likewise.
1530 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1531 * lra-constraints.c (process_alt_operands): Likewise.
1532 (split_reg): Likewise.
1533 * recog.c (peep2_find_free_register): Likewise.
1534 * ree.c (combine_reaching_defs): Likewise.
1535 * regcprop.c (maybe_mode_change): Likewise.
1536 * reginfo.c (init_reg_sets_1): Likewise.
1537 (choose_hard_reg_mode): Likewise.
1538 (simplifiable_subregs): Likewise.
1539 * regrename.c (check_new_reg_p): Likewise.
1540 * reload.c (find_valid_class): Likewise.
1541 (find_valid_class_1): Likewise.
1542 (reload_inner_reg_of_subreg): Likewise.
1543 (push_reload): Likewise.
1544 (combine_reloads): Likewise.
1545 (find_dummy_reload): Likewise.
1546 (find_reloads): Likewise.
1547 * reload1.c (find_reg): Likewise.
1548 (set_reload_reg): Likewise.
1549 (allocate_reload_reg): Likewise.
1550 (choose_reload_regs): Likewise.
1551 (reload_adjust_reg_for_temp): Likewise.
1552 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1553 (simplify_subreg_regno): Likewise.
1554 * sel-sched.c (init_regs_for_mode): Likewise.
1555 * varasm.c (make_decl_rtl): Likewise.
1556 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1557 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1559 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1560 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1561 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1562 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1563 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1565 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1566 (arc_mode_class): Delete.
1567 (HARD_REGNO_MODE_OK): Delete.
1568 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1569 (arc_hard_regno_mode_ok): Rename old array to...
1570 (arc_hard_regno_mode_ok_modes): ...this.
1571 (arc_conditional_register_usage): Update accordingly.
1572 (arc_mode_class): Make static.
1573 (arc_hard_regno_mode_ok): New function.
1574 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1575 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1576 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1577 (arm_hard_regno_mode_ok): Make static.
1578 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1580 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1581 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1582 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1584 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1585 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1586 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1587 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1588 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1589 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1590 * config/bfin/predicates.md (valid_reg_operand): Use
1591 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1592 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1593 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1594 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1595 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1596 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1597 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1598 (cr16_hard_regno_mode_ok): Make static and return a bool.
1599 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1600 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1601 (cris_hard_regno_mode_ok): New function.
1602 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1603 (epiphany_mode_class): Delete.
1604 (HARD_REGNO_MODE_OK): Delete.
1605 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1606 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1607 (hard_regno_mode_ok): Rename to...
1608 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1609 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1610 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1612 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1613 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1614 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1615 (frv_hard_regno_mode_ok): Make static and return a bool.
1616 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1618 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1619 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1620 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1621 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1624 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1625 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1626 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1628 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1629 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1630 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1631 (ia64_hard_regno_mode_ok): New function.
1632 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1633 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1634 (iq2000_hard_regno_mode_ok): New function.
1635 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1636 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1637 (lm32_hard_regno_mode_ok): New function.
1638 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1639 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1640 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1641 instead of HARD_REGNO_MODE_OK.
1642 (m32c_hard_regno_ok): Rename to...
1643 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1644 (m32c_cannot_change_mode_class): Update accordingly.
1645 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1646 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1647 (m32r_mode_class): Delete.
1648 (HARD_REGNO_MODE_OK): Delete.
1649 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1650 (m32r_hard_regno_mode_ok): Rename to...
1651 (m32r_hard_regno_modes): ...this.
1652 (m32r_mode_class): Make static.
1653 (m32r_hard_regno_mode_ok): New function.
1654 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1655 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1656 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1657 (m68k_hard_regno_mode_ok): Make static.
1658 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1659 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1660 (mcore_hard_regno_mode_ok): New function.
1661 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1662 (HARD_REGNO_MODE_OK): Delete.
1663 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1665 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1666 (microblaze_hard_regno_mode_ok): New function.
1667 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1668 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1669 (mips_hard_regno_mode_ok): Delete.
1670 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1671 (mips_hard_regno_mode_ok_p): ...this and make static.
1672 (mips_hard_regno_mode_ok_p): Rename to...
1673 (mips_hard_regno_mode_ok_uncached): ...this.
1674 (mips_hard_regno_mode_ok): New function.
1675 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1676 of HARD_REGNO_MODE_OK.
1677 (mips_option_override): Update after above name changes.
1678 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1679 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1680 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1681 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1682 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1683 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1684 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1685 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1686 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1687 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1688 (msp430_hard_regno_mode_ok): Make static and return a bool.
1689 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1690 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1691 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1693 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1694 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1695 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1696 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1697 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1698 (PA_HARD_REGNO_MODE_OK): ...this
1699 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1700 (PA_HARD_REGNO_MODE_OK): ...this.
1701 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1702 (pa_hard_regno_mode_ok): New function.
1703 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1704 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1705 (pdp11_hard_regno_mode_ok): New function.
1706 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1707 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1709 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1711 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1712 (rs6000_hard_regno_mode_ok): Rename to...
1713 (rs6000_hard_regno_mode_ok_uncached): ...this.
1714 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1715 (rs6000_hard_regno_mode_ok): New function.
1716 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1717 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1718 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1719 (riscv_hard_regno_mode_ok): ...this and make static.
1720 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1721 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1722 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1723 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1724 (rl78_hard_regno_mode_ok): Make static and return bool.
1725 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1726 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1728 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1729 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1730 (rs6000_hard_regno_mode_ok): Rename to...
1731 (rs6000_hard_regno_mode_ok_uncached): ...this.
1732 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1733 (rs6000_hard_regno_mode_ok): New function.
1734 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1735 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1736 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1737 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1738 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1739 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1741 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1742 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1743 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1744 (sh_hard_regno_mode_ok): Make static.
1745 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1746 instead of HARD_REGNO_MODE_OK.
1747 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1748 (sparc_mode_class): Delete.
1749 (HARD_REGNO_MODE_OK): Delete.
1750 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1751 (hard_regno_mode_classes): Make static.
1752 (sparc_mode_class): Likewise.
1753 (sparc_hard_regno_mode_ok): New function.
1754 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1755 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1756 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1758 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1759 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1760 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1761 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1762 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1763 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1764 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1765 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1766 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1767 (visium_hard_regno_mode_ok): New function.
1768 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1769 instead of HARD_REGNO_MODE_OK.
1770 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1771 (HARD_REGNO_MODE_OK): Delete.
1772 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1773 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1774 (xtensa_option_override): Update accordingly.
1775 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1776 (xtensa_hard_regno_mode_ok): New function.
1777 * system.h (HARD_REGNO_MODE_OK): Poison.
1779 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1780 Alan Hayward <alan.hayward@arm.com>
1781 David Sherwood <david.sherwood@arm.com>
1783 * target.def (hard_regno_call_part_clobbered): New hook.
1784 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1785 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1786 * doc/tm.texi: Regenerate.
1787 * hooks.h (hook_bool_uint_mode_false): Declare.
1788 * hooks.c (hook_bool_uint_mode_false): New function.
1789 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1790 * cselib.c (cselib_process_insn): Use
1791 targetm.hard_regno_call_part_clobbered instead of
1792 HARD_REGNO_CALL_PART_CLOBBERED.
1793 * ira-conflicts.c (ira_build_conflicts): Likewise.
1794 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1795 * lra-constraints.c (need_for_call_save_p): Likewise.
1796 * lra-lives.c: Include target.h.
1797 (check_pseudos_live_through_calls): Use
1798 targetm.hard_regno_call_part_clobbered instead of
1799 HARD_REGNO_CALL_PART_CLOBBERED.
1800 * regcprop.c: Include target.h.
1801 (copyprop_hardreg_forward_1): Use
1802 targetm.hard_regno_call_part_clobbered instead of
1803 HARD_REGNO_CALL_PART_CLOBBERED.
1804 * reginfo.c (choose_hard_reg_mode): Likewise.
1805 * regrename.c (check_new_reg_p): Likewise.
1806 * reload.c (find_equiv_reg): Likewise.
1807 * reload1.c (emit_reload_insns): Likewise.
1808 * sched-deps.c (deps_analyze_insn): Likewise.
1809 * sel-sched.c (init_regs_for_mode): Likewise.
1810 (mark_unavailable_hard_regs): Likewise.
1811 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1812 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1813 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1815 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1816 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1817 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1819 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1821 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1822 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1823 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1825 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1826 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1827 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1829 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1830 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1832 * config/powerpcspe/powerpcspe.c
1833 (rs6000_hard_regno_call_part_clobbered): New function.
1834 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1835 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1836 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1838 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1839 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1840 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1842 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1843 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1844 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1846 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1847 Alan Hayward <alan.hayward@arm.com>
1848 David Sherwood <david.sherwood@arm.com>
1850 * rtl.h (subreg_memory_offset): Declare.
1851 * emit-rtl.c (subreg_memory_offset): New function.
1852 * expmed.c (store_bit_field_1): Use it.
1853 * expr.c (undefined_operand_subword_p): Likewise.
1854 * simplify-rtx.c (simplify_subreg): Likewise.
1856 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1858 PR rtl-optimization/57448
1861 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1862 using atomic_load pattern.
1863 (expand_atomic_store): Likewise.
1865 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1868 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1869 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1870 handling. Use replace_call_with_value with NULL instead of
1871 gsi_replace, unlink_stmt_vdef and release_defs.
1873 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1876 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1878 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1881 * rtl.h (emit_library_call_value_1): Declare.
1882 (emit_library_call): Replace declaration with a series of overloads.
1883 Remove the parameter count argument.
1884 (emit_library_call_value): Likewise.
1885 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1886 with an "rtx_mode_t *".
1887 (emit_library_call_value): Delete.
1888 (emit_library_call): Likewise.
1889 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1890 (asan_emit_allocas_unpoison): Likewise.
1891 * builtins.c (expand_builtin_powi): Likewise.
1892 (expand_asan_emit_allocas_unpoison): Likewise.
1893 * cfgexpand.c (expand_main_function): Likewise.
1894 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1895 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1896 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1897 * config/arm/arm.c (arm_trampoline_init): Likewise.
1898 (arm_call_tls_get_addr): Likewise.
1899 (arm_expand_divmod_libfunc): Likewise.
1900 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1901 (smulsi3_highpart): Likewise.
1902 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1903 (c6x_expand_compare): Likewise.
1904 (c6x_expand_movmem): Likewise.
1905 * config/frv/frv.c (frv_trampoline_init): Likewise.
1906 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1907 (ix86_expand_divmod_libfunc): Likewise.
1908 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1909 (ia64_expand_compare): Likewise.
1910 (ia64_profile_hook): Likewise.
1911 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1912 (nonlocal_goto): Likewise.
1913 (restore_stack_nonlocal): Likewise.
1914 * config/m32r/m32r.c (block_move_call): Likewise.
1915 (m32r_trampoline_init): Likewise.
1916 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1917 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1918 (m68k_call_m68k_read_tp): Likewise.
1919 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1920 (microblaze_expand_divide): Likewise.
1921 * config/mips/mips.h (mips_args): Likewise.
1922 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1923 (MIPS_ICACHE_SYNC): Likewise.
1924 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1925 (nios2_trampoline_init): Likewise.
1926 * config/pa/pa.c (hppa_tls_call): Likewise.
1927 (pa_trampoline_init): Likewise.
1928 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1929 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1930 (expand_strn_compare): Likewise.
1931 (rs6000_generate_compare): Likewise.
1932 (rs6000_expand_float128_convert): Likewise.
1933 (output_profile_hook): Likewise.
1934 (rs6000_trampoline_init): Likewise.
1935 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1936 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1937 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1938 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1939 (rs6000_generate_compare): Likewise.
1940 (rs6000_expand_float128_convert): Likewise.
1941 (output_profile_hook): Likewise.
1942 (rs6000_trampoline_init): Likewise.
1943 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1944 * config/sh/sh.c (sh_trampoline_init): Likewise.
1945 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1946 (sparc_emit_float_lib_cmp): Likewise.
1947 (sparc32_initialize_trampoline): Likewise.
1948 (sparc64_initialize_trampoline): Likewise.
1949 (sparc_profile_hook): Likewise.
1950 * config/spu/spu.c (ea_load_store): Likewise.
1951 * config/spu/spu.md (floatunssidf2): Likewise.
1952 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1953 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1954 * config/visium/visium.c (expand_block_move_4): Likewise.
1955 (expand_block_move_2): Likewise.
1956 (expand_block_move_1): Likewise.
1957 (expand_block_set_4): Likewise.
1958 (expand_block_set_2): Likewise.
1959 (expand_block_set_1): Likewise.
1960 (visium_trampoline_init): Likewise.
1961 (visium_profile_hook): Likewise.
1962 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1963 (xtensa_setup_frame_addresses): Likewise.
1964 (xtensa_trampoline_init): Likewise.
1965 * except.c (sjlj_emit_function_enter): Likewise.
1966 (sjlj_emit_function_exit): Likewise.
1967 * explow.c (allocate_dynamic_stack_space): Likewise.
1968 (probe_stack_range): Likewise.
1969 * expr.c (convert_mode_scalar): Likewise.
1970 * optabs.c (expand_binop): Likewise.
1971 (expand_twoval_binop_libfunc): Likewise.
1972 (expand_unop): Likewise.
1973 (prepare_cmp_insn): Likewise.
1974 (prepare_float_lib_cmp): Likewise.
1975 (expand_float): Likewise.
1976 (expand_fix): Likewise.
1977 (expand_fixed_convert): Likewise.
1978 (maybe_emit_sync_lock_test_and_set): Likewise.
1979 (expand_atomic_compare_and_swap): Likewise.
1980 (expand_mem_thread_fence): Likewise.
1981 (expand_atomic_fetch_op): Likewise.
1983 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1985 * doc/generic.texi (OpenACC): Adjust URL.
1986 * doc/invoke.texi (C Dialect Options): Ditto.
1988 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1990 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1991 predicate for operand 1. Add (m,<S>) constraint.
1992 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1993 Prevent memory operand 1 with register operand 2.
1995 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1997 PR rtl-optimization/82024
1998 * combine.c (try_combine): If the combination result is a PARALLEL,
1999 and we only need to retain the SET in there that would be placed
2000 at I2, check that we can place that at I3 instead, before doing so.
2002 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2005 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2007 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2008 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2011 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2012 Jeff Law <law@redhat.com>
2014 * varasm.c (bss_initializer_p): Do not put constants into .bss
2015 (categorize_decl_for_section): Handle bss_initializer_p returning
2016 false when DECL_INITIAL is NULL.
2018 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2021 * config/s390/s390.c (s390_can_inline_p): New function.
2023 2017-09-01 Jeff Law <law@redhat.com>
2025 PR tree-optimization/82052
2026 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2027 Always initialize the returned slot after a hash table miss
2028 when INSERT is true.
2030 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2032 * config/s390/s390.md (mem_signal_fence): Remove.
2033 * doc/md.texi (mem_signal_fence): Remove.
2034 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2036 * target-insns.def (mem_signal_fence): Remove.
2038 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2041 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2044 * asan.c (create_odr_indicator): Strip name encoding from assembler
2045 name before appending it after __odr_asan_.
2047 2017-09-01 Martin Liska <mliska@suse.cz>
2049 PR tree-optimization/82059
2050 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2051 frequency only when an edge is redirected.
2053 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2055 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2056 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2057 (arc_conditional_register_usage): Remove ARC600 lp_count
2059 (arc_file_start): Emit Tag_ARC_CPU_variation.
2060 (arc_can_use_doloop_p): New conditions to use ZOLs.
2061 (hwloop_fail): New function.
2062 (hwloop_optimize): Likewise.
2063 (hwloop_pattern_reg): Likewise.
2064 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2065 (arc_reorg_loops): New function, calls reorg_loops.
2066 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2067 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2069 (arc_loop_hazard): Remove function, functionality moved into
2071 (arc_hazard): Remove arc_loop_hazard call.
2072 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2073 into hwloop_optimize.
2074 (arc_label_align): Remove ZOL handling.
2075 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2076 * config/arc/arc.md (doloop_begin): Remove pattern.
2077 (doloop_begin_i): Likewise.
2078 (doloop_end_i): Likewise.
2079 (doloop_fallback): Likewise.
2080 (doloop_fallback_m): Likewise.
2081 (doloop_end): Reimplement expand.
2082 (arc_lp): New pattern for LP instruction.
2083 (loop_end): New pattern.
2084 (loop_fail): Likewise.
2085 (decrement_and_branch_until_zero): Likewise.
2086 * config/arc/arc.opt (mlpc-width): New option.
2087 * doc/invoke.texi (mlpc-width): Document option.
2089 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2091 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2092 (arc_ccfsm_advance): Fix checking for delay slots.
2093 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2095 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2097 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2099 (movhi_insn): Update store instruction constraint which are saving
2100 6-bit short immediates.
2101 (movsi_insn): Consider also short scaled load operations.
2102 (zero_extendhisi2_i): Use Usd constraint instead of T.
2103 (extendhisi2_i): Add q constraint.
2104 (arc_clzsi2): Add type and length attributes.
2105 (arc_ctzsi2): Likewise.
2106 * config/arc/constraints.md (Usc): Update constraint, the
2107 assembler can parse two relocations for a single instruction.
2109 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2111 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2112 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2114 2017-08-31 Olivier Hainque <hainque@adacore.com>
2116 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2117 match as powerpc-wrs-vxworks*.
2119 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2121 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2122 register constraint for by-element operand.
2123 (aarch64_mls_elt_merge<mode>): Likewise.
2125 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2127 * config/arc/arc.c (arc_can_follow_jump): Check for short
2130 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2132 * config.gcc: Use g.opt for arc.
2133 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2134 functionality moved to ...
2135 (legitimate_scaled_address_p): New function, ...here.
2136 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2137 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2138 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2140 (arc_override_options): Handle G option.
2141 (arc_output_pic_addr_const): Correct function definition.
2142 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2143 (arc_decl_anon_ns_mem_p): Delete.
2144 (arc_in_small_data_p): Overhaul this function to take into
2145 consideration the value given via G option.
2146 (arc_rewrite_small_data_1): Renamed and corrected old
2147 arc_rewrite_small_data function.
2148 (arc_rewrite_small_data): New function.
2149 (small_data_pattern): Don't use pic_offset_table_rtx.
2150 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2151 * config/arc/simdext.md (movmisalignv2hi): Use
2152 prepare_move_operands function.
2154 (movmisalign*): Likewise.
2155 * doc/invoke.texi (ARC options): Document -G option.
2157 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2159 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2161 * config/arc/arc.c (arc_print_operand): Output scalled address for
2162 sdata whenever is possible.
2163 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2164 load/stores are available.
2165 (compact_sda_memory_operand): Check for the alignment required by
2166 code density instructions.
2167 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2169 * config/arc/constraints.md (Usd): Update constraint.
2170 (Us0): New constraint.
2171 (Usc): Update constraint.
2173 2017-08-31 Richard Biener <rguenther@suse.de>
2176 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2179 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2181 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2182 Resize type_signature.
2184 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2185 Alan Hayward <alan.hayward@arm.com>
2186 David Sherwood <david.sherwood@arm.com>
2188 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2189 subregs whose inner modes can be stored in GPRs.
2190 (aarch64_classify_index): Likewise.
2192 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2193 Alan Hayward <alan.hayward@arm.com>
2194 David Sherwood <david.sherwood@arm.com>
2196 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2197 (V_INT_EQUIV): ...this.
2198 (v_cmp_result): Rename to...
2199 (v_int_equiv): ...this.
2200 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2201 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2202 (copysign<mode>3): Likewise.
2203 (aarch64_simd_bsl<mode>_internal): Likewise.
2204 (aarch64_simd_bsl<mode>): Likewise.
2205 (vec_cmp<mode><mode>): Likewise.
2206 (vcond<mode><mode>): Likewise.
2207 (vcond<v_cmp_mixed><mode>): Likewise.
2208 (vcondu<mode><v_cmp_mixed>): Likewise.
2209 (aarch64_cm<optab><mode>): Likewise.
2210 (aarch64_cmtst<mode>): Likewise.
2211 (aarch64_fac<optab><mode>): Likewise.
2212 (vec_perm_const<mode>): Likewise.
2213 (vcond_mask_<mode><v_cmp_result>): Rename to...
2214 (vcond_mask_<mode><v_int_equiv>): ...this.
2215 (vec_cmp<mode><v_cmp_result>): Rename to...
2216 (vec_cmp<mode><v_int_equiv>): ...this.
2218 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2219 Alan Hayward <alan.hayward@arm.com>
2220 David Sherwood <david.sherwood@arm.com>
2222 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2224 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2225 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2226 (UNSPEC_LD4_DREG): New unspecs.
2227 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2228 (aarch64_ld2<mode>_dreg_be): Replace with...
2229 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2231 (aarch64_ld3<mode>_dreg_le)
2232 (aarch64_ld3<mode>_dreg_be): Replace with...
2233 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2235 (aarch64_ld4<mode>_dreg_le)
2236 (aarch64_ld4<mode>_dreg_be): Replace with...
2237 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2240 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2242 PR tree-optimization/81987
2243 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2244 insert an initializer in a location not dominated by the stride
2247 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2249 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2250 on the entire header of the finally block in the fallthru case.
2252 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2254 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2256 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2258 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2259 rs6000_emit_move_from_cr and call renamed function.
2260 (rs6000_emit_prologue): Call renamed functions.
2261 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2262 movesi_from_cr, remove volatile CRs.
2264 2017-08-30 Jon Beniston <jon@beniston.com>
2265 Richard Biener <rguenther@suse.de>
2267 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2268 of VECTOR_MODE_P check.
2269 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2270 element vector types.
2272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2274 * df.h (df_read_modify_subreg_p): Remove in favor of...
2275 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2276 const_rtx instead of an rtx.
2277 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2278 * df-problems.c (df_word_lr_mark_ref): Likewise.
2279 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2280 (mark_pseudo_reg_dead): Likewise.
2281 (mark_ref_dead): Likewise.
2282 * reginfo.c (init_subregs_of_mode): Likewise.
2283 * sched-deps.c (sched_analyze_1): Likewise.
2284 * df-scan.c (df_def_record_1): Likewise.
2285 (df_uses_record): Likewise.
2286 (df_read_modify_subreg_p): Remove in favor of...
2287 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2288 const_rtx instead of an rtx.
2290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2291 Alan Hayward <alan.hayward@arm.com>
2292 David Sherwood <david.sherwood@arm.com>
2294 * rtl.h (partial_subreg_p): New function.
2295 * caller-save.c (save_call_clobbered_regs): Use it.
2296 * calls.c (expand_call): Likewise.
2297 * combine.c (combinable_i3pat): Likewise.
2298 (simplify_set): Likewise.
2299 (make_extraction): Likewise.
2300 (make_compound_operation_int): Likewise.
2301 (gen_lowpart_or_truncate): Likewise.
2302 (force_to_mode): Likewise.
2303 (make_field_assignment): Likewise.
2304 (reg_truncated_to_mode): Likewise.
2305 (record_truncated_value): Likewise.
2306 (move_deaths): Likewise.
2307 * cse.c (record_jump_cond): Likewise.
2308 (cse_insn): Likewise.
2309 * cselib.c (cselib_lookup_1): Likewise.
2310 * expmed.c (extract_bit_field_using_extv): Likewise.
2311 * function.c (assign_parm_setup_reg): Likewise.
2312 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2313 * ira-build.c (create_insn_allocnos): Likewise.
2314 * lra-coalesce.c (merge_pseudos): Likewise.
2315 * lra-constraints.c (match_reload): Likewise.
2316 (simplify_operand_subreg): Likewise.
2317 (curr_insn_transform): Likewise.
2318 * lra-lives.c (process_bb_lives): Likewise.
2319 * lra.c (new_insn_reg): Likewise.
2320 (lra_substitute_pseudo): Likewise.
2321 * regcprop.c (mode_change_ok): Likewise.
2322 (maybe_mode_change): Likewise.
2323 (copyprop_hardreg_forward_1): Likewise.
2324 * reload.c (push_reload): Likewise.
2325 (find_reloads): Likewise.
2326 (find_reloads_subreg_address): Likewise.
2327 * reload1.c (alter_reg): Likewise.
2328 (eliminate_regs_1): Likewise.
2329 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2331 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2333 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2334 back to if statements, including unpack.
2336 2017-08-30 Martin Liska <mliska@suse.cz>
2339 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2341 (func_checker::compare_asm_inputs_outputs): ... this function.
2342 (func_checker::compare_gimple_asm): Use the function to compare
2343 also ASM constrains.
2344 * ipa-icf-gimple.h: Rename the function.
2346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2347 Alan Hayward <alan.hayward@arm.com>
2348 David Sherwood <david.sherwood@arm.com>
2350 * coretypes.h (complex_mode): New type.
2351 * gdbhooks.py (build_pretty_printer): Handle it.
2352 * machmode.h (complex_mode): New class.
2353 (complex_mode::includes_p): New function.
2354 (is_complex_int_mode): Likewise.
2355 (is_complex_float_mode): Likewise.
2356 * genmodes.c (get_mode_class): Handle complex mode classes.
2357 * function.c (expand_function_end): Use is_complex_int_mode.
2359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2360 Alan Hayward <alan.hayward@arm.com>
2361 David Sherwood <david.sherwood@arm.com>
2363 * coretypes.h (scalar_mode_pod): New typedef.
2364 * gdbhooks.py (build_pretty_printer): Handle it.
2365 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2366 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2367 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2368 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2372 Alan Hayward <alan.hayward@arm.com>
2373 David Sherwood <david.sherwood@arm.com>
2375 * machmode.h (mode_for_vector): Take a scalar_mode instead
2377 * stor-layout.c (mode_for_vector): Likewise.
2378 * explow.c (promote_mode): Use as_a <scalar_mode>.
2379 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2382 Alan Hayward <alan.hayward@arm.com>
2383 David Sherwood <david.sherwood@arm.com>
2385 * target.def (preferred_simd_mode): Take a scalar_mode
2386 instead of a machine_mode.
2387 * targhooks.h (default_preferred_simd_mode): Likewise.
2388 * targhooks.c (default_preferred_simd_mode): Likewise.
2389 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2390 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2391 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2392 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2393 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2394 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2395 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2396 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2398 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2399 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2400 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2401 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2402 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2403 * doc/tm.texi: Regenerate.
2404 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2408 Alan Hayward <alan.hayward@arm.com>
2409 David Sherwood <david.sherwood@arm.com>
2411 * target.def (scalar_mode_supported_p): Take a scalar_mode
2412 instead of a machine_mode.
2413 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2414 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2415 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2416 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2417 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2418 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2419 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2420 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2421 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2422 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2423 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2424 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2425 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2426 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2428 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2429 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2430 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2431 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2432 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2434 * doc/tm.texi: Regenerate.
2436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2437 Alan Hayward <alan.hayward@arm.com>
2438 David Sherwood <david.sherwood@arm.com>
2440 * coretypes.h (opt_scalar_mode): New typedef.
2441 * gdbhooks.py (build_pretty_printers): Handle it.
2442 * machmode.h (mode_iterator::get_2xwider): Add overload for
2444 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2446 * expr.c (convert_mode_scalar): Likewise.
2447 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2448 * optabs.c (expand_float): Likewise.
2449 (expand_fix): Likewise.
2450 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2453 Alan Hayward <alan.hayward@arm.com>
2454 David Sherwood <david.sherwood@arm.com>
2456 * optabs.c (expand_float): Explicitly check for scalars before
2457 using a branching expansion.
2458 (expand_fix): Likewise.
2460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2461 Alan Hayward <alan.hayward@arm.com>
2462 David Sherwood <david.sherwood@arm.com>
2464 * expr.c (convert_mode): Split scalar handling out into...
2465 (convert_mode_scalar): ...this new function. Treat the modes
2468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2469 Alan Hayward <alan.hayward@arm.com>
2470 David Sherwood <david.sherwood@arm.com>
2472 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2474 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2477 Alan Hayward <alan.hayward@arm.com>
2478 David Sherwood <david.sherwood@arm.com>
2480 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2481 rather than a machine_mode.
2482 (fixed_from_string): Likewise.
2483 (fixed_convert): Likewise.
2484 (fixed_convert_from_int): Likewise.
2485 (fixed_convert_from_real): Likewise.
2486 (real_convert_from_fixed): Likewise.
2487 * fixed-value.c (fixed_from_double_int): Likewise.
2488 (fixed_from_string): Likewise.
2489 (fixed_convert): Likewise.
2490 (fixed_convert_from_int): Likewise.
2491 (fixed_convert_from_real): Likewise.
2492 (real_convert_from_fixed): Likewise.
2493 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2496 Alan Hayward <alan.hayward@arm.com>
2497 David Sherwood <david.sherwood@arm.com>
2499 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2500 of separate mode class checks. Do not allow vector modes here.
2501 (immed_wide_int_const): Use as_a <scalar_mode>.
2502 * explow.c (trunc_int_for_mode): Likewise.
2503 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2504 (wi::shwi): Likewise.
2505 (wi::min_value): Likewise.
2506 (wi::max_value): Likewise.
2507 * dwarf2out.c (loc_descriptor): Likewise.
2508 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2512 Alan Hayward <alan.hayward@arm.com>
2513 David Sherwood <david.sherwood@arm.com>
2515 * tree.h (SCALAR_TYPE_MODE): New macro.
2516 * expr.c (expand_expr_addr_expr_1): Use it.
2517 (expand_expr_real_2): Likewise.
2518 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2519 (fold_convert_const_fixed_from_int): Likewise.
2520 (fold_convert_const_fixed_from_real): Likewise.
2521 (native_encode_fixed): Likewise
2522 (native_encode_complex): Likewise
2523 (native_encode_vector): Likewise.
2524 (native_interpret_fixed): Likewise.
2525 (native_interpret_real): Likewise.
2526 (native_interpret_complex): Likewise.
2527 (native_interpret_vector): Likewise.
2528 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2529 (simd_clone_adjust_argument_types): Likewise.
2530 (simd_clone_init_simd_arrays): Likewise.
2531 (simd_clone_adjust): Likewise.
2532 * stor-layout.c (layout_type): Likewise.
2533 * tree.c (build_minus_one_cst): Likewise.
2534 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2535 * tree-inline.c (estimate_move_cost): Likewise.
2536 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2537 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2538 (vectorizable_reduction): Likewise.
2539 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2540 (vect_recog_mixed_size_cond_pattern): Likewise.
2541 (check_bool_pattern): Likewise.
2542 (adjust_bool_pattern): Likewise.
2543 (search_type_for_mask_1): Likewise.
2544 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2545 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2546 (vectorizable_load): Likewise.
2547 (vectorizable_store): Likewise.
2548 * ubsan.c (ubsan_encode_value): Likewise.
2549 * varasm.c (output_constant): Likewise.
2551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2552 Alan Hayward <alan.hayward@arm.com>
2553 David Sherwood <david.sherwood@arm.com>
2555 * coretypes.h (scalar_mode): New class.
2556 * machmode.h (scalar_mode): Likewise.
2557 (scalar_mode::includes_p): New function.
2558 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2559 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2560 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2561 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2562 * expmed.c (store_bit_field_1): Likewise.
2563 (extract_bit_field_1): Likewise.
2564 * expr.c (write_complex_part): Likewise.
2565 (read_complex_part): Likewise.
2566 (emit_move_complex_push): Likewise.
2567 (expand_expr_real_2): Likewise.
2568 * function.c (assign_parm_setup_reg): Likewise.
2569 (assign_parms_unsplit_complex): Likewise.
2570 * optabs.c (expand_binop): Likewise.
2571 * rtlanal.c (subreg_get_info): Likewise.
2572 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2573 * varasm.c (output_constant_pool_2): Likewise.
2575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2576 Alan Hayward <alan.hayward@arm.com>
2577 David Sherwood <david.sherwood@arm.com>
2579 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2581 (expmed_mult_highpart_optab): Likewise.
2582 (expmed_mult_highpart): Likewise.
2584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2585 Alan Hayward <alan.hayward@arm.com>
2586 David Sherwood <david.sherwood@arm.com>
2588 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2589 instead of a machine_mode.
2590 (builtin_memset_read_str): Likewise.
2591 * builtins.c (c_readstr): Likewise.
2592 (builtin_memcpy_read_str): Likewise.
2593 (builtin_strncpy_read_str): Likewise.
2594 (builtin_memset_read_str): Likewise.
2595 (builtin_memset_gen_str): Likewise.
2596 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2597 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2598 instead of a machine_mode.
2599 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2601 (make_extraction): Likewise.
2602 (try_widen_shift_mode): Take and return scalar_int_modes instead
2604 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2605 a scalar_int_mode instead of a machine_mode.
2606 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2607 (avr_addr_space_pointer_mode): Likewise.
2608 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2609 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2610 (msp430_unwind_word_mode): Likewise.
2611 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2612 (spu_addr_space_pointer_mode): Likewise.
2613 (spu_addr_space_address_mode): Likewise.
2614 (spu_libgcc_cmp_return_mode): Likewise.
2615 (spu_libgcc_shift_count_mode): Likewise.
2616 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2617 (rl78_addr_space_pointer_mode): Likewise.
2618 (fl78_unwind_word_mode): Likewise.
2619 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2621 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2622 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2623 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2624 (mips_valid_pointer_mode): Likewise.
2625 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2626 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2627 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2629 (ft32_addr_space_address_mode): Likewise.
2630 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2631 scalar_int_mode instead of a machine_mode.
2632 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2634 (m32c_addr_space_address_mode): Likewise.
2635 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2636 (rs6000_eh_return_filter_mode): Likewise.
2637 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2638 (rs6000_eh_return_filter_mode): Likewise.
2639 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2640 (s390_libgcc_shift_count_mode): Likewise.
2641 (s390_unwind_word_mode): Likewise.
2642 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2644 * target.def (mode_rep_extended): Likewise.
2645 (valid_pointer_mode): Likewise.
2646 (addr_space.valid_pointer_mode): Likewise.
2647 (eh_return_filter_mode): Return a scalar_int_mode rather than
2649 (libgcc_cmp_return_mode): Likewise.
2650 (libgcc_shift_count_mode): Likewise.
2651 (unwind_word_mode): Likewise.
2652 (addr_space.pointer_mode): Likewise.
2653 (addr_space.address_mode): Likewise.
2654 * doc/tm.texi: Regenerate.
2655 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2657 (do_jump): Use scalar_int_mode for local variables.
2658 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2659 rather than a machine_mode.
2660 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2661 (scompare_loc_descriptor_wide): Likewise.
2662 (scompare_loc_descriptor_narrow): Likewise.
2663 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2665 * except.c (sjlj_emit_dispatch_table): Likewise.
2666 (expand_builtin_eh_copy_values): Likewise.
2667 * explow.c (convert_memory_address_addr_space_1): Likewise.
2668 Take a scalar_int_mode rather than a machine_mode.
2669 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2670 than a machine_mode.
2671 (memory_address_addr_space): Use scalar_int_mode for local variables.
2672 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2673 rather than a machine_mode.
2674 * expmed.c (mask_rtx): Likewise.
2675 (init_expmed_one_conv): Likewise.
2676 (expand_mult_highpart_adjust): Likewise.
2677 (extract_high_half): Likewise.
2678 (expmed_mult_highpart_optab): Likewise.
2679 (expmed_mult_highpart): Likewise.
2680 (expand_smod_pow2): Likewise.
2681 (expand_sdiv_pow2): Likewise.
2682 (emit_store_flag_int): Likewise.
2683 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2685 (extract_low_bits): Likewise.
2686 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2688 * expr.c (pieces_addr::adjust): Likewise.
2689 (can_store_by_pieces): Likewise.
2690 (store_by_pieces): Likewise.
2691 (clear_by_pieces_1): Likewise.
2692 (expand_expr_addr_expr_1): Likewise.
2693 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2694 (expand_expr_real_1): Likewise.
2695 (try_casesi): Likewise.
2696 * final.c (shorten_branches): Likewise.
2697 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2698 type of "mode" to machine_mode.
2699 * internal-fn.c (expand_arith_overflow_result_store): Take a
2700 scalar_int_mode rather than a machine_mode.
2701 (expand_mul_overflow): Use scalar_int_mode for local variables.
2702 * loop-doloop.c (doloop_modify): Likewise.
2703 (doloop_optimize): Likewise.
2704 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2705 than a machine_mode.
2706 (expand_doubleword_shift_condmove): Likewise.
2707 (expand_doubleword_shift): Likewise.
2708 (expand_doubleword_clz): Likewise.
2709 (expand_doubleword_popcount): Likewise.
2710 (expand_doubleword_parity): Likewise.
2711 (expand_absneg_bit): Use scalar_int_mode for local variables.
2712 (prepare_float_lib_cmp): Likewise.
2713 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2714 rather than a machine_mode.
2715 (convert_memory_address_addr_space): Likewise.
2716 (get_mode_bounds): Likewise.
2717 (get_address_mode): Return a scalar_int_mode rather than a
2719 * rtlanal.c (get_address_mode): Likewise.
2720 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2721 than a machine_mode.
2722 * targhooks.c (default_mode_rep_extended): Likewise.
2723 (default_valid_pointer_mode): Likewise.
2724 (default_addr_space_valid_pointer_mode): Likewise.
2725 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2726 than a machine_mode.
2727 (default_libgcc_cmp_return_mode): Likewise.
2728 (default_libgcc_shift_count_mode): Likewise.
2729 (default_unwind_word_mode): Likewise.
2730 (default_addr_space_pointer_mode): Likewise.
2731 (default_addr_space_address_mode): Likewise.
2732 * targhooks.h (default_eh_return_filter_mode): Likewise.
2733 (default_libgcc_cmp_return_mode): Likewise.
2734 (default_libgcc_shift_count_mode): Likewise.
2735 (default_unwind_word_mode): Likewise.
2736 (default_addr_space_pointer_mode): Likewise.
2737 (default_addr_space_address_mode): Likewise.
2738 (default_mode_rep_extended): Take a scalar_int_mode rather than
2740 (default_valid_pointer_mode): Likewise.
2741 (default_addr_space_valid_pointer_mode): Likewise.
2742 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2744 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2745 rather than a machine_mode.
2746 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2747 for local variables.
2748 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2749 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2750 than a machine_mode.
2752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2753 Alan Hayward <alan.hayward@arm.com>
2754 David Sherwood <david.sherwood@arm.com>
2756 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2757 the mode argument to scalar_int_mode.
2758 (do_jump_by_parts_zero_rtx): Likewise.
2759 (do_jump_by_parts_equality_rtx): Likewise.
2760 (do_jump_by_parts_greater): Take a mode argument.
2761 (do_jump_by_parts_equality): Likewise.
2762 (do_jump_1): Update calls accordingly.
2764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2765 Alan Hayward <alan.hayward@arm.com>
2766 David Sherwood <david.sherwood@arm.com>
2768 * is-a.h (safe_dyn_cast): New function.
2769 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2770 (jump_table_for_label): Likewise.
2771 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2772 instead of an rtx_insn *.
2773 (shorten_branches): Use dyn_cast instead of LABEL_P and
2774 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2775 rtx_jump_table_data::get_data_mode.
2776 (final_scan_insn): Likewise.
2778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2779 Alan Hayward <alan.hayward@arm.com>
2780 David Sherwood <david.sherwood@arm.com>
2782 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2783 trying to combine a full-register integer set with a subreg
2786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2787 Alan Hayward <alan.hayward@arm.com>
2788 David Sherwood <david.sherwood@arm.com>
2790 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2791 that is always either address_mode or pointer_mode.
2793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2794 Alan Hayward <alan.hayward@arm.com>
2795 David Sherwood <david.sherwood@arm.com>
2797 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2798 when the two are known to be equal.
2800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2801 Alan Hayward <alan.hayward@arm.com>
2802 David Sherwood <david.sherwood@arm.com>
2804 * simplify-rtx.c (simplify_const_unary_operation): Use
2805 is_a <scalar_int_mode> instead of checking for a nonzero
2806 precision. Forcibly convert op_mode to a scalar_int_mode
2807 in that case. More clearly differentiate the operand and
2808 result modes and use the former when deciding what the value
2809 of a count-bits operation should be. Use is_int_mode instead
2810 of checking for a MODE_INT. Remove redundant check for whether
2811 this mode has a zero precision.
2813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2814 Alan Hayward <alan.hayward@arm.com>
2815 David Sherwood <david.sherwood@arm.com>
2817 * optabs.c (widen_leading): Change the type of the mode argument
2818 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2819 (widen_bswap): Likewise.
2820 (expand_parity): Likewise.
2821 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2822 (expand_ffs): Likewise.
2823 (epand_unop): Check for scalar integer modes before calling the
2826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2827 Alan Hayward <alan.hayward@arm.com>
2828 David Sherwood <david.sherwood@arm.com>
2830 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2832 (expand_expr_real_1): Update call accordingly.
2834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2835 Alan Hayward <alan.hayward@arm.com>
2836 David Sherwood <david.sherwood@arm.com>
2838 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2839 value_mode arguments. Use scalar_int_mode internally.
2840 (store_bit_field_1): Rename the new integer mode from imode
2841 to op0_mode and use it instead of GET_MODE (op0). Update calls
2842 to store_split_bit_field, store_bit_field_using_insv and
2843 store_fixed_bit_field.
2844 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2845 Use scalar_int_mode internally. Use a bit count rather than a mode
2846 when calculating the largest bit size for get_best_mode.
2847 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2848 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2849 Remove assertion that OP0 has a scalar integer mode.
2850 (store_split_bit_field): Add op0_mode and value_mode arguments.
2851 Update calls to extract_fixed_bit_field.
2852 (extract_bit_field_using_extv): Add an op0_mode argument.
2853 Use scalar_int_mode internally.
2854 (extract_bit_field_1): Rename the new integer mode from imode to
2855 op0_mode and use it instead of GET_MODE (op0). Update calls to
2856 extract_split_bit_field, extract_bit_field_using_extv and
2857 extract_fixed_bit_field.
2858 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2859 to extract_split_bit_field and extract_fixed_bit_field_1.
2860 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2861 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2863 (extract_split_bit_field): Add an op0_mode argument. Update call
2864 to extract_fixed_bit_field.
2866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2867 Alan Hayward <alan.hayward@arm.com>
2868 David Sherwood <david.sherwood@arm.com>
2870 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2871 * explow.c (hard_function_value): Likewise.
2872 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2873 convert_to_mode call outside the loop.
2874 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2875 for the mode iterator. Require the mode specified by max_pieces
2877 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2879 (copy_blkmode_to_reg): Likewise.
2880 (set_storage_via_setmem): Likewise.
2881 * optabs.c (prepare_cmp_insn): Likewise.
2882 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2883 * stor-layout.c (finish_bitfield_representative): 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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2890 * expr.c (convert_move): Use them.
2891 (convert_modes): Likewise.
2892 (store_expr_with_bounds): Likewise.
2894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2895 Alan Hayward <alan.hayward@arm.com>
2896 David Sherwood <david.sherwood@arm.com>
2898 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2899 parameter for the mode of "x". Remove the "known_x", "known_mode"
2900 and "known_ret" arguments. Change the type of the mode argument
2902 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2903 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2904 (reg_num_sign_bit_copies_for_combine): Likewise.
2905 * rtlanal.c (nonzero_bits1): Likewise.
2906 (num_sign_bit_copies1): Likewise.
2907 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2908 (reg_num_sign_bit_copies_general): Likewise.
2909 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2910 (reg_nonzero_bits_general): Likewise.
2912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2913 Alan Hayward <alan.hayward@arm.com>
2914 David Sherwood <david.sherwood@arm.com>
2916 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2917 than in subroutines. Return 1 for non-integer modes.
2918 (cached_num_sign_bit_copies): Change the type of the mode parameter
2920 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2921 classes. Handle CONST_INT_P first and then check whether X also
2922 has a scalar integer mode. Check the same thing for inner registers
2923 of a SUBREG and for values that are being extended or truncated.
2925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2926 Alan Hayward <alan.hayward@arm.com>
2927 David Sherwood <david.sherwood@arm.com>
2929 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2930 in subroutines. Return the mode mask for non-integer modes.
2931 (cached_nonzero_bits): Change the type of the mode parameter
2933 (nonzero_bits1): Likewise. Remove early exit for other mode
2934 classes. Handle CONST_INT_P first and then check whether X
2935 also has a scalar integer mode.
2937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2938 Alan Hayward <alan.hayward@arm.com>
2939 David Sherwood <david.sherwood@arm.com>
2941 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2942 Return a scalar_int_mode and assert that the size is greater than
2944 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2945 (op_by_pieces_d::op_by_pieces_d): Likewise.
2946 (op_by_pieces_d::run): Likewise.
2947 (can_store_by_pieces): Likewise.
2949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2950 Alan Hayward <alan.hayward@arm.com>
2951 David Sherwood <david.sherwood@arm.com>
2953 * combine.c (extract_left_shift): Add a mode argument and update
2955 (make_compound_operation_int): Change the type of the mode parameter
2956 to scalar_int_mode and update the call to extract_left_shift.
2958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2959 Alan Hayward <alan.hayward@arm.com>
2960 David Sherwood <david.sherwood@arm.com>
2962 * combine.c (simplify_and_const_int): Change the type of the mode
2963 parameter to scalar_int_mode.
2964 (simplify_and_const_int_1): Likewise. Update recursive call.
2966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2967 Alan Hayward <alan.hayward@arm.com>
2968 David Sherwood <david.sherwood@arm.com>
2970 * combine.c (simplify_compare_const): Check that the mode is a
2971 scalar_int_mode (rather than VOIDmode) before testing its
2973 (simplify_comparison): Move COMPARISON_P handling out of the
2974 loop and restrict the latter part of the loop to scalar_int_modes.
2975 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2976 and when considering SUBREG_REGs. Use is_int_mode instead of
2977 checking GET_MODE_CLASS against MODE_INT.
2979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2980 Alan Hayward <alan.hayward@arm.com>
2981 David Sherwood <david.sherwood@arm.com>
2983 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2984 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2985 shift_unit_mode and for modes involved in scalar shifts.
2987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2988 Alan Hayward <alan.hayward@arm.com>
2989 David Sherwood <david.sherwood@arm.com>
2991 * combine.c (force_int_to_mode): New function, split out from...
2992 (force_to_mode): ...here. Keep xmode up-to-date and use it
2993 instead of GET_MODE (x).
2995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2996 Alan Hayward <alan.hayward@arm.com>
2997 David Sherwood <david.sherwood@arm.com>
2999 * optabs-query.h (extraction_insn::struct_mode): Change type to
3000 opt_scalar_int_mode and update comment.
3001 (extraction_insn::field_mode): Change type to scalar_int_mode.
3002 (extraction_insn::pos_mode): Likewise.
3003 * combine.c (make_extraction): Update accordingly.
3004 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3005 (get_optab_extraction_insn): Likewise.
3006 * recog.c (simplify_while_replacing): Likewise.
3007 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3008 parameter to opt_scalar_int_mode.
3010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3011 Alan Hayward <alan.hayward@arm.com>
3012 David Sherwood <david.sherwood@arm.com>
3014 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3015 to a scalar_int_mode instead of a machine_mode.
3016 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3017 (get_best_mode): Return a boolean and use a pointer argument to store
3018 the selected mode. Replace the limit mode parameter with a bit limit.
3019 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3020 for the values returned by bit_field_mode_iterator::next_mode.
3021 (store_bit_field): Update call to get_best_mode.
3022 (store_fixed_bit_field): Likewise.
3023 (extract_fixed_bit_field): Likewise.
3024 * expr.c (optimize_bitfield_assignment_op): Likewise.
3025 * fold-const.c (optimize_bit_field_compare): Likewise.
3026 (fold_truth_andor_1): Likewise.
3027 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3028 Update for new type of m_mode.
3029 (get_best_mode): As above.
3031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3032 Alan Hayward <alan.hayward@arm.com>
3033 David Sherwood <david.sherwood@arm.com>
3035 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3036 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3037 (store_bit_field): Check is_a <scalar_int_mode> before calling
3038 strict_volatile_bitfield_p.
3039 (extract_bit_field): Likewise.
3041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3042 Alan Hayward <alan.hayward@arm.com>
3043 David Sherwood <david.sherwood@arm.com>
3045 * target.def (cstore_mode): Return a scalar_int_mode.
3046 * doc/tm.texi: Regenerate.
3047 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3048 * targhooks.h (default_cstore_mode): Likewise.
3049 * targhooks.c (default_cstore_mode): Likewise, using a forced
3051 * expmed.c (emit_cstore): Expect the target of the cstore to be
3054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3055 Alan Hayward <alan.hayward@arm.com>
3056 David Sherwood <david.sherwood@arm.com>
3058 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3060 (niter_desc): Likewise mode.
3061 (iv_analyze): Add a mode parameter.
3063 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3064 and change its type to scalar_int_mode.
3065 * loop-iv.c: Update commentary at head of file.
3066 (iv_constant): Pass the mode paraeter before the rtx it describes
3067 and change its type to scalar_int_mode. Remove VOIDmode handling.
3068 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3069 (iv_extend): Likewise.
3070 (shorten_into_mode): Likewise.
3071 (iv_add): Use scalar_int_mode.
3072 (iv_mult): Likewise.
3073 (iv_shift): Likewise.
3074 (canonicalize_iv_subregs): Likewise.
3075 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3076 it describes and change its mode to scalar_int_mode. Also change
3077 the type of the returned inner_mode to scalar_int_mode.
3078 (get_biv_step): Likewise, turning outer_mode from a pointer
3079 into a direct parameter. Update call to get_biv_step_1.
3080 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3081 iv_constant and get_biv_step.
3082 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3083 and change its type to scalar_int_mode. Don't initialise iv->mode
3084 to VOIDmode and remove later checks for its still being VOIDmode.
3085 Update calls to iv_analyze_op and iv_analyze_expr. Check
3086 is_a <scalar_int_mode> when changing the mode under consideration.
3087 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3088 Update call to iv_analyze_expr.
3089 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3090 inner register is not also a scalar_int_mode. Update call to
3092 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3093 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3094 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3095 separate mode class checks. Update calls to iv_analyze. Remove
3096 fix-up of VOIDmodes after iv_analyze_biv.
3097 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3098 don't have a scalar_int_mode. Update call to biv_p.
3100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3101 Alan Hayward <alan.hayward@arm.com>
3102 David Sherwood <david.sherwood@arm.com>
3104 * cfgexpand.c (convert_debug_memory_address): Use
3105 as_a <scalar_int_mode>.
3106 * combine.c (expand_compound_operation): Likewise.
3107 (make_extraction): Likewise.
3108 (change_zero_ext): Likewise.
3109 (simplify_comparison): Likewise.
3110 * cse.c (cse_insn): Likewise.
3111 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3112 (mem_loc_descriptor): Likewise.
3113 (loc_descriptor): Likewise.
3114 * expmed.c (init_expmed_one_mode): Likewise.
3115 (synth_mult): Likewise.
3116 (emit_store_flag_1): Likewise.
3117 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3118 of a comparison with size.
3119 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3120 (reduce_to_bit_field_precision): Likewise.
3121 * function.c (expand_function_end): Likewise.
3122 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3123 * loop-doloop.c (doloop_modify): Likewise.
3124 * optabs.c (expand_binop): Likewise.
3125 (expand_unop): Likewise.
3126 (expand_copysign_absneg): Likewise.
3127 (prepare_cmp_insn): Likewise.
3128 (maybe_legitimize_operand): Likewise.
3129 * recog.c (const_scalar_int_operand): Likewise.
3130 * rtlanal.c (get_address_mode): Likewise.
3131 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3132 (simplify_cond_clz_ctz): Likewise.
3133 * tree-nested.c (get_nl_goto_field): Likewise.
3134 * tree.c (build_vector_type_for_mode): Likewise.
3135 * var-tracking.c (use_narrower_mode): Likewise.
3137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3138 Alan Hayward <alan.hayward@arm.com>
3139 David Sherwood <david.sherwood@arm.com>
3141 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3142 * builtins.c (expand_builtin_signbit): Use it.
3143 * cfgexpand.c (expand_debug_expr): Likewise.
3144 * dojump.c (do_jump): Likewise.
3145 (do_compare_and_jump): Likewise.
3146 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3147 * expmed.c (make_tree): Likewise.
3148 * expr.c (expand_expr_real_2): Likewise.
3149 (expand_expr_real_1): Likewise.
3150 (try_casesi): Likewise.
3151 * fold-const-call.c (fold_const_call_ss): Likewise.
3152 * fold-const.c (unextend): Likewise.
3153 (extract_muldiv_1): Likewise.
3154 (fold_single_bit_test): Likewise.
3155 (native_encode_int): Likewise.
3156 (native_encode_string): Likewise.
3157 (native_interpret_int): Likewise.
3158 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3159 * internal-fn.c (expand_addsub_overflow): Likewise.
3160 (expand_neg_overflow): Likewise.
3161 (expand_mul_overflow): Likewise.
3162 (expand_arith_overflow): Likewise.
3163 * match.pd: Likewise.
3164 * stor-layout.c (layout_type): Likewise.
3165 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3166 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3167 * tree-ssanames.c (get_range_info): Likewise.
3168 * tree-switch-conversion.c (array_value_type) Likewise.
3169 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3170 (vect_recog_divmod_pattern): Likewise.
3171 (vect_recog_mixed_size_cond_pattern): Likewise.
3172 * tree-vrp.c (extract_range_basic): Likewise.
3173 (simplify_float_conversion_using_ranges): Likewise.
3174 * tree.c (int_fits_type_p): Likewise.
3175 * ubsan.c (instrument_bool_enum_load): Likewise.
3176 * varasm.c (mergeable_string_section): Likewise.
3177 (narrowing_initializer_constant_valid_p): Likewise.
3178 (output_constant): Likewise.
3180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3181 Alan Hayward <alan.hayward@arm.com>
3182 David Sherwood <david.sherwood@arm.com>
3184 * machmode.h (NARROWEST_INT_MODE): New macro.
3185 * expr.c (alignment_for_piecewise_move): Use it instead of
3186 GET_CLASS_NARROWEST_MODE (MODE_INT).
3187 (push_block): Likewise.
3188 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3190 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3193 Alan Hayward <alan.hayward@arm.com>
3194 David Sherwood <david.sherwood@arm.com>
3196 * postreload.c (move2add_valid_value_p): Change the type of the
3197 mode parameter to scalar_int_mode.
3198 (move2add_use_add2_insn): Add a mode parameter and use it instead
3200 (move2add_use_add3_insn): Likewise.
3201 (reload_cse_move2add): Update accordingly.
3203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3204 Alan Hayward <alan.hayward@arm.com>
3205 David Sherwood <david.sherwood@arm.com>
3207 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3209 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3210 * optabs.c (expand_unop): Likewise.
3212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3213 Alan Hayward <alan.hayward@arm.com>
3214 David Sherwood <david.sherwood@arm.com>
3216 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3217 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3218 (popcount_loc_descriptor): Likewise.
3219 (bswap_loc_descriptor): Likewise.
3220 (rotate_loc_descriptor): Likewise.
3221 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3222 calling the functions above.
3224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3225 Alan Hayward <alan.hayward@arm.com>
3226 David Sherwood <david.sherwood@arm.com>
3228 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3230 (try_combine): Likewise.
3231 (simplify_if_then_else): Likewise.
3232 * cse.c (cse_insn): Likewise.
3233 * dwarf2out.c (mem_loc_descriptor): Likewise.
3234 * emit-rtl.c (gen_lowpart_common): Likewise.
3235 * simplify-rtx.c (simplify_truncation): Likewise.
3236 (simplify_binary_operation_1): Likewise.
3237 (simplify_const_relational_operation): Likewise.
3238 (simplify_ternary_operation): Likewise.
3239 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3242 Alan Hayward <alan.hayward@arm.com>
3243 David Sherwood <david.sherwood@arm.com>
3245 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3246 * reload.c (push_reload): Likewise.
3247 (find_reloads): Likewise.
3249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3250 Alan Hayward <alan.hayward@arm.com>
3251 David Sherwood <david.sherwood@arm.com>
3253 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3254 (make_compound_operation_int): Likewise.
3255 (change_zero_ext): Likewise.
3256 * expr.c (convert_move): Likewise.
3257 (convert_modes): Likewise.
3258 * fwprop.c (forward_propagate_subreg): Likewise.
3259 * loop-iv.c (get_biv_step_1): Likewise.
3260 * optabs.c (widen_operand): Likewise.
3261 * postreload.c (move2add_valid_value_p): Likewise.
3262 * recog.c (simplify_while_replacing): Likewise.
3263 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3264 (simplify_binary_operation_1): Likewise. Remove redundant
3265 mode equality check.
3267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3268 Alan Hayward <alan.hayward@arm.com>
3269 David Sherwood <david.sherwood@arm.com>
3271 * combine.c (combine_simplify_rtx): Add checks for
3272 is_a <scalar_int_mode>.
3273 (simplify_if_then_else): Likewise.
3274 (make_field_assignment): Likewise.
3275 (simplify_comparison): Likewise.
3276 * ifcvt.c (noce_try_bitop): Likewise.
3277 * loop-invariant.c (canonicalize_address_mult): Likewise.
3278 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3281 Alan Hayward <alan.hayward@arm.com>
3282 David Sherwood <david.sherwood@arm.com>
3284 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3285 is_a <scalar_int_mode> instead of != BLKmode.
3287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3288 Alan Hayward <alan.hayward@arm.com>
3289 David Sherwood <david.sherwood@arm.com>
3291 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3292 instead of != VOIDmode.
3293 * combine.c (if_then_else_cond): Likewise.
3294 (change_zero_ext): Likewise.
3295 * dwarf2out.c (mem_loc_descriptor): Likewise.
3296 (loc_descriptor): Likewise.
3297 * rtlanal.c (canonicalize_condition): Likewise.
3298 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3301 Alan Hayward <alan.hayward@arm.com>
3302 David Sherwood <david.sherwood@arm.com>
3304 * simplify-rtx.c (simplify_binary_operation_1): Use
3305 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3308 Alan Hayward <alan.hayward@arm.com>
3309 David Sherwood <david.sherwood@arm.com>
3311 * wide-int.h (int_traits<unsigned char>) New class.
3312 (int_traits<unsigned short>) Likewise.
3313 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3314 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3316 * combine.c (set_nonzero_bits_and_sign_copies): Use
3317 is_a <scalar_int_mode>.
3318 (find_split_point): Likewise.
3319 (combine_simplify_rtx): Likewise.
3320 (simplify_logical): Likewise.
3321 (expand_compound_operation): Likewise.
3322 (expand_field_assignment): Likewise.
3323 (make_compound_operation): Likewise.
3324 (extended_count): Likewise.
3325 (change_zero_ext): Likewise.
3326 (simplify_comparison): Likewise.
3327 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3328 (ucompare_loc_descriptor): Likewise.
3329 (minmax_loc_descriptor): Likewise.
3330 (mem_loc_descriptor): Likewise.
3331 (loc_descriptor): Likewise.
3332 * expmed.c (init_expmed_one_mode): Likewise.
3333 * lra-constraints.c (lra_constraint_offset): Likewise.
3334 * optabs.c (prepare_libcall_arg): Likewise.
3335 * postreload.c (move2add_note_store): Likewise.
3336 * reload.c (operands_match_p): Likewise.
3337 * rtl.h (load_extend_op): Likewise.
3338 * rtlhooks.c (gen_lowpart_general): Likewise.
3339 * simplify-rtx.c (simplify_truncation): Likewise.
3340 (simplify_unary_operation_1): Likewise.
3341 (simplify_binary_operation_1): Likewise.
3342 (simplify_const_binary_operation): Likewise.
3343 (simplify_const_relational_operation): Likewise.
3344 (simplify_subreg): Likewise.
3345 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3346 * var-tracking.c (adjust_mems): Likewise.
3347 (prepare_call_arguments): Likewise.
3349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3350 Alan Hayward <alan.hayward@arm.com>
3351 David Sherwood <david.sherwood@arm.com>
3353 * machmode.h (is_int_mode): New fuction.
3354 * combine.c (find_split_point): Use it.
3355 (combine_simplify_rtx): Likewise.
3356 (simplify_if_then_else): Likewise.
3357 (simplify_set): Likewise.
3358 (simplify_shift_const_1): Likewise.
3359 (simplify_comparison): Likewise.
3360 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3361 * cse.c (notreg_cost): Likewise.
3362 (cse_insn): Likewise.
3363 * cselib.c (cselib_lookup_1): Likewise.
3364 * dojump.c (do_jump_1): Likewise.
3365 (do_compare_rtx_and_jump): Likewise.
3366 * dse.c (get_call_args): Likewise.
3367 * dwarf2out.c (rtl_for_decl_init): Likewise.
3368 (native_encode_initializer): Likewise.
3369 * expmed.c (emit_store_flag_1): Likewise.
3370 (emit_store_flag): Likewise.
3371 * expr.c (convert_modes): Likewise.
3372 (store_field): Likewise.
3373 (expand_expr_real_1): Likewise.
3374 * fold-const.c (fold_read_from_constant_string): Likewise.
3375 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3376 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3377 * optabs.c (expand_binop): Likewise.
3378 (expand_unop): Likewise.
3379 (expand_abs_nojump): Likewise.
3380 (expand_one_cmpl_abs_nojump): Likewise.
3381 * simplify-rtx.c (mode_signbit_p): Likewise.
3382 (val_signbit_p): Likewise.
3383 (val_signbit_known_set_p): Likewise.
3384 (val_signbit_known_clear_p): Likewise.
3385 (simplify_relational_operation_1): Likewise.
3386 * tree.c (vector_type_mode): Likewise.
3388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3389 Alan Hayward <alan.hayward@arm.com>
3390 David Sherwood <david.sherwood@arm.com>
3392 * machmode.h (smallest_mode_for_size): Fix formatting.
3393 (smallest_int_mode_for_size): New function.
3394 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3395 instead of smallest_mode_for_size.
3396 * combine.c (make_extraction): Likewise.
3397 * config/arc/arc.c (arc_expand_movmem): Likewise.
3398 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3399 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3400 * config/s390/s390.c (s390_expand_insv): Likewise.
3401 * config/sparc/sparc.c (assign_int_registers): Likewise.
3402 * config/spu/spu.c (spu_function_value): Likewise.
3403 (spu_function_arg): Likewise.
3404 * coverage.c (get_gcov_type): Likewise.
3405 (get_gcov_unsigned_t): Likewise.
3406 * dse.c (find_shift_sequence): Likewise.
3407 * expmed.c (store_bit_field_1): Likewise.
3408 * expr.c (convert_move): Likewise.
3409 (store_field): Likewise.
3410 * internal-fn.c (expand_arith_overflow): Likewise.
3411 * optabs-query.c (get_best_extraction_insn): Likewise.
3412 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3413 * stor-layout.c (layout_type): Likewise.
3414 (initialize_sizetypes): Likewise.
3415 * targhooks.c (default_get_mask_mode): Likewise.
3416 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
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 * machmode.h (opt_mode::else_blk): New function.
3423 (int_mode_for_mode): Declare.
3424 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3425 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3427 * cfgexpand.c (expand_debug_expr): Likewise.
3428 * combine.c (gen_lowpart_or_truncate): Likewise.
3429 (gen_lowpart_for_combine): Likewise.
3430 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3431 * config/avr/avr.c (avr_to_int_mode): Likewise.
3432 (avr_out_plus_1): Likewise.
3433 (avr_out_plus): Likewise.
3434 (avr_out_round): Likewise.
3435 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3436 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3437 (s390_expand_vcond): Likewise.
3438 * config/spu/spu.c (spu_split_immediate): Likewise.
3439 (spu_expand_mov): Likewise.
3440 * dse.c (get_stored_val): Likewise.
3441 * expmed.c (store_bit_field_1): Likewise.
3442 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3444 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3445 (extract_low_bits): Likewise.
3446 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3447 handling rather than repeating the check.
3448 (emit_group_store): Likewise.
3449 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3450 * optabs.c (expand_absneg_bit): Likewise.
3451 (expand_copysign_absneg): Likewise.
3452 (expand_copysign_bit): Likewise.
3453 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3454 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3455 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3456 * var-tracking.c (prepare_call_arguments): Likewise.
3457 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3458 int_mode_for_mode instead of mode_for_size.
3459 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3462 Alan Hayward <alan.hayward@arm.com>
3463 David Sherwood <david.sherwood@arm.com>
3465 * machmode.h (int_mode_for_size): New function.
3466 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3467 instead of mode_for_size.
3468 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3470 * combine.c (expand_field_assignment): Use int_mode_for_size
3471 instead of mode_for_size.
3472 (make_extraction): Likewise.
3473 (simplify_shift_const_1): Likewise.
3474 (simplify_comparison): Likewise.
3475 * dojump.c (do_jump): Likewise.
3476 * dwarf2out.c (mem_loc_descriptor): Likewise.
3477 * emit-rtl.c (init_derived_machine_modes): Likewise.
3478 * expmed.c (flip_storage_order): Likewise.
3479 (convert_extracted_bit_field): Likewise.
3480 * expr.c (copy_blkmode_from_reg): Likewise.
3481 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3482 * internal-fn.c (expand_mul_overflow): Likewise.
3483 * lower-subreg.c (simple_move): Likewise.
3484 * optabs-libfuncs.c (init_optabs): Likewise.
3485 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3486 * tree.c (vector_type_mode): Likewise.
3487 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3488 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3489 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3490 * tree-vect-stmts.c (vectorizable_load): Likewise.
3491 (vectorizable_store): Likewise.
3493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3494 Alan Hayward <alan.hayward@arm.com>
3495 David Sherwood <david.sherwood@arm.com>
3497 * coretypes.h (pod_mode): New type.
3498 (scalar_int_mode_pod): New typedef.
3499 * machmode.h (pod_mode): New class.
3500 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3501 * genmodes.c (emit_mode_int_n): Update accordingly.
3502 * lower-subreg.h (target_lower_subreg): Change type to
3503 scalar_int_mode_pod.
3504 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3505 scalar_int_mode_pod.
3507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3508 Alan Hayward <alan.hayward@arm.com>
3509 David Sherwood <david.sherwood@arm.com>
3511 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3512 machine_mode to scalar_int_mode.
3513 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3514 (rs6000_option_override_internal): Remove cast to int.
3515 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3516 machine_mode to scalar_int_mode.
3517 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3518 (rs6000_option_override_internal): Remove cast to int.
3519 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3520 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3522 * config/s390/s390.c (s390_expand_builtin): Likewise.
3523 * coretypes.h (scalar_int_mode): New type.
3524 (opt_scalar_int_mode): New typedef.
3525 * machmode.h (scalar_int_mode): New class.
3526 (scalar_int_mode::includes_p): New function.
3527 (byte_mode): Change type to scalar_int_mode.
3528 (word_mode): Likewise.
3529 (ptr_mode): Likewise.
3530 * emit-rtl.c (byte_mode): Likewise.
3531 (word_mode): Likewise.
3532 (ptr_mode): Likewise.
3533 (init_derived_machine_modes): Update accordingly.
3534 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3535 and MODE_PARTIAL_INT.
3536 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3537 opt_scalar_int_mode.
3539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3540 Alan Hayward <alan.hayward@arm.com>
3541 David Sherwood <david.sherwood@arm.com>
3543 * target.def (libgcc_floating_mode_supported_p): Take a
3545 * doc/tm.texi: Regenerate.
3546 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3548 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3549 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3553 Alan Hayward <alan.hayward@arm.com>
3554 David Sherwood <david.sherwood@arm.com>
3556 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3557 * doc/tm.texi: Regenerate.
3558 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3559 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3560 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3561 * targhooks.h (default_floatn_mode): Likewise.
3562 * targhooks.c (default_floatn_mode): Likewise.
3563 * tree.c (build_common_tree_nodes): Update accordingly.
3565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3566 Alan Hayward <alan.hayward@arm.com>
3567 David Sherwood <david.sherwood@arm.com>
3569 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3570 (mode_iterator::iterate_p): Likewise.
3571 (mode_iterator::get_wider): Likewise.
3572 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3575 Alan Hayward <alan.hayward@arm.com>
3576 David Sherwood <david.sherwood@arm.com>
3578 * coretypes.h (opt_scalar_float_mode): New typedef.
3579 * machmode.h (float_mode_for_size): New function.
3580 * emit-rtl.c (double_mode): Delete.
3581 (init_emit_once): Use float_mode_for_size.
3582 * stor-layout.c (layout_type): Likewise.
3583 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3586 Alan Hayward <alan.hayward@arm.com>
3587 David Sherwood <david.sherwood@arm.com>
3589 * output.h (assemble_real): Take a scalar_float_mode.
3590 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3591 * config/arm/arm.md (consttable_4): Likewise.
3592 (consttable_8): Likewise.
3593 (consttable_16): Likewise.
3594 * config/mips/mips.md (consttable_float): Likewise.
3595 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3596 * varasm.c (assemble_real): Take a scalar_float_mode.
3597 (output_constant_pool_2): Update accordingly.
3598 (output_constant): Likewise.
3600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3601 Alan Hayward <alan.hayward@arm.com>
3602 David Sherwood <david.sherwood@arm.com>
3604 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3605 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3606 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3607 (native_encode_real): Likewise.
3608 (native_interpret_real): Likewise.
3609 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3610 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3613 Alan Hayward <alan.hayward@arm.com>
3614 David Sherwood <david.sherwood@arm.com>
3616 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3617 <scalar_float_mode>. Simplify.
3618 (gen_extend_conv_libfunc): Likewise.
3620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3621 Alan Hayward <alan.hayward@arm.com>
3622 David Sherwood <david.sherwood@arm.com>
3624 * coretypes.h (scalar_float_mode): New type.
3625 * machmode.h (mode_traits::from_int): Use machine_mode if
3626 USE_ENUM_MODES is defined.
3627 (is_a): New function.
3629 (dyn_cast): Likewise.
3630 (scalar_float_mode): New class.
3631 (scalar_float_mode::includes_p): New function.
3632 (is_float_mode): Likewise.
3633 * gdbhooks.py (MachineModePrinter): New class.
3634 (build_pretty_printer): Use it for scalar_float_mode.
3635 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3636 (format_helper::format_helper): Turn into a template.
3637 * genmodes.c (get_mode_class): New function.
3638 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3639 or machine_mode if none.
3640 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3641 as_a <scalar_float_mode>.
3642 * dwarf2out.c (mem_loc_descriptor): Likewise.
3643 (insert_float): Likewise.
3644 (add_const_value_attribute): Likewise.
3645 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3646 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3647 (expand_unop): Update accordingly.
3648 (expand_abs_nojump): Likewise.
3649 (expand_copysign_absneg): Take a scalar_float_mode.
3650 (expand_copysign_bit): Likewise.
3651 (expand_copysign): Update accordingly.
3653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3654 Alan Hayward <alan.hayward@arm.com>
3655 David Sherwood <david.sherwood@arm.com>
3657 * coretypes.h (opt_mode): New class.
3658 * machmode.h (opt_mode): Likewise.
3659 (opt_mode::else_void): New function.
3660 (opt_mode::require): Likewise.
3661 (opt_mode::exists): Likewise.
3662 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3663 (GET_MODE_2XWIDER_MODE): Likewise.
3664 (mode_iterator::get_wider): Update accordingly.
3665 (mode_iterator::get_2xwider): Likewise.
3666 (mode_iterator::get_known_wider): Likewise, turning into a template.
3667 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3668 forcing a wider mode to exist.
3669 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3670 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3671 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3672 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3673 * lower-subreg.c (init_lower_subreg): Likewise.
3674 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3675 on the final iteration.
3676 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3677 a wider mode exists before asking for a move pattern.
3678 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3679 forcing a wider mode to exist.
3680 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3681 returning false if no such mode exists.
3682 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3683 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3684 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3685 Avoid checking for a MODE_INT if we already know the mode is not a
3687 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3688 forcing a wider mode to exist.
3689 (expmed_mult_highpart_optab): Likewise.
3690 (expmed_mult_highpart): Likewise.
3691 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3693 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3694 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3695 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3696 * internal-fn.c (expand_mul_overflow): Update use of
3697 GET_MODE_2XWIDER_MODE.
3698 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3699 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3700 GET_MODE_WIDER_MODE.
3701 (convert_plusminus_to_widen): Likewise.
3702 * tree-switch-conversion.c (array_value_type): Likewise.
3703 * var-tracking.c (emit_note_insn_var_location): Likewise.
3704 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3705 Return false inside rather than outside the loop if no wider mode
3707 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3708 and GET_MODE_2XWIDER_MODE
3709 (can_compare_p): Use else_void.
3710 * gdbhooks.py (OptMachineModePrinter): New class.
3711 (build_pretty_printer): Use it for opt_mode.
3713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3714 Alan Hayward <alan.hayward@arm.com>
3715 David Sherwood <david.sherwood@arm.com>
3717 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3718 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3721 Alan Hayward <alan.hayward@arm.com>
3722 David Sherwood <david.sherwood@arm.com>
3724 * machmode.h (mode_traits): New structure.
3725 (get_narrowest_mode): New function.
3726 (mode_iterator::start): Likewise.
3727 (mode_iterator::iterate_p): Likewise.
3728 (mode_iterator::get_wider): Likewise.
3729 (mode_iterator::get_known_wider): Likewise.
3730 (mode_iterator::get_2xwider): Likewise.
3731 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3732 (FOR_EACH_MODE): Likewise.
3733 (FOR_EACH_MODE_FROM): Likewise.
3734 (FOR_EACH_MODE_UNTIL): Likewise.
3735 (FOR_EACH_WIDER_MODE): Likewise.
3736 (FOR_EACH_2XWIDER_MODE): Likewise.
3737 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3738 * combine.c (simplify_comparison): Likewise
3739 * config/i386/i386.c (type_natural_mode): Likewise.
3740 * cse.c (cse_insn): Likewise.
3741 * dse.c (find_shift_sequence): Likewise.
3742 * emit-rtl.c (init_derived_machine_modes): Likewise.
3743 (init_emit_once): Likewise.
3744 * explow.c (hard_function_value): Likewise.
3745 * expmed.c (extract_fixed_bit_field_1): Likewise.
3746 (extract_bit_field_1): Likewise.
3747 (expand_divmod): Likewise.
3748 (emit_store_flag_1): Likewise.
3749 * expr.c (init_expr_target): Likewise.
3750 (convert_move): Likewise.
3751 (alignment_for_piecewise_move): Likewise.
3752 (widest_int_mode_for_size): Likewise.
3753 (emit_block_move_via_movmem): Likewise.
3754 (copy_blkmode_to_reg): Likewise.
3755 (set_storage_via_setmem): Likewise.
3756 (compress_float_constant): Likewise.
3757 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3758 * optabs-query.c (get_best_extraction_insn): Likewise.
3759 * optabs.c (expand_binop): Likewise.
3760 (expand_twoval_unop): Likewise.
3761 (expand_twoval_binop): Likewise.
3762 (widen_leading): Likewise.
3763 (widen_bswap): Likewise.
3764 (expand_parity): Likewise.
3765 (expand_unop): Likewise.
3766 (prepare_cmp_insn): Likewise.
3767 (prepare_float_lib_cmp): Likewise.
3768 (expand_float): Likewise.
3769 (expand_fix): Likewise.
3770 (expand_sfix_optab): Likewise.
3771 * postreload.c (move2add_use_add2_insn): Likewise.
3772 * reg-stack.c (reg_to_stack): Likewise.
3773 * reginfo.c (choose_hard_reg_mode): Likewise.
3774 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3775 * stor-layout.c (mode_for_size): Likewise.
3776 (smallest_mode_for_size): Likewise.
3777 (mode_for_vector): Likewise.
3778 (finish_bitfield_representative): Likewise.
3779 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3780 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3781 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3782 * var-tracking.c (prepare_call_arguments): Likewise.
3784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3785 Alan Hayward <alan.hayward@arm.com>
3786 David Sherwood <david.sherwood@arm.com>
3788 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3789 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3790 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3791 * machmode.h (mode_size): Move earlier in file.
3792 (mode_precision): Likewise.
3793 (mode_inner): Likewise.
3794 (mode_nunits): Likewise.
3795 (mode_unit_size): Likewise.
3796 (unit_unit_precision): Likewise.
3797 (mode_wider): Likewise.
3798 (mode_2xwider): Likewise.
3799 (machine_mode): New class.
3800 (mode_to_bytes): New function.
3801 (mode_to_bits): Likewise.
3802 (mode_to_precision): Likewise.
3803 (mode_to_inner): Likewise.
3804 (mode_to_unit_size): Likewise.
3805 (mode_to_unit_precision): Likewise.
3806 (mode_to_nunits): Likewise.
3807 (GET_MODE_SIZE): Use mode_to_bytes.
3808 (GET_MODE_BITSIZE): Use mode_to_bits.
3809 (GET_MODE_PRECISION): Use mode_to_precision.
3810 (GET_MODE_INNER): Use mode_to_inner.
3811 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3812 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3813 (GET_MODE_NUNITS): Use mode_to_nunits.
3814 * system.h (ALWAYS_INLINE): New macro.
3815 * config/powerpcspe/powerpcspe-c.c
3816 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3817 int for arg1_mode and arg2_mode.
3819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3820 Alan Hayward <alan.hayward@arm.com>
3821 David Sherwood <david.sherwood@arm.com>
3823 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3824 Prefix mode names with E_ in case statements.
3825 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3826 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3827 (aarch64_split_simd_move): Likewise.
3828 (aarch64_gen_storewb_pair): Likewise.
3829 (aarch64_gen_loadwb_pair): Likewise.
3830 (aarch64_gen_store_pair): Likewise.
3831 (aarch64_gen_load_pair): Likewise.
3832 (aarch64_get_condition_code_1): Likewise.
3833 (aarch64_constant_pool_reload_icode): Likewise.
3834 (get_rsqrte_type): Likewise.
3835 (get_rsqrts_type): Likewise.
3836 (get_recpe_type): Likewise.
3837 (get_recps_type): Likewise.
3838 (aarch64_gimplify_va_arg_expr): Likewise.
3839 (aarch64_simd_container_mode): Likewise.
3840 (aarch64_emit_load_exclusive): Likewise.
3841 (aarch64_emit_store_exclusive): Likewise.
3842 (aarch64_expand_compare_and_swap): Likewise.
3843 (aarch64_gen_atomic_cas): Likewise.
3844 (aarch64_emit_bic): Likewise.
3845 (aarch64_emit_atomic_swap): Likewise.
3846 (aarch64_emit_atomic_load_op): Likewise.
3847 (aarch64_evpc_trn): Likewise.
3848 (aarch64_evpc_uzp): Likewise.
3849 (aarch64_evpc_zip): Likewise.
3850 (aarch64_evpc_ext): Likewise.
3851 (aarch64_evpc_rev): Likewise.
3852 (aarch64_evpc_dup): Likewise.
3853 (aarch64_gen_ccmp_first): Likewise.
3854 (aarch64_gen_ccmp_next): Likewise.
3855 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3856 (alpha_emit_xfloating_libcall): Likewise.
3857 (emit_insxl): Likewise.
3858 (alpha_arg_type): Likewise.
3859 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3860 (arc_preferred_simd_mode): Likewise.
3861 (arc_secondary_reload): Likewise.
3862 (get_arc_condition_code): Likewise.
3863 (arc_print_operand): Likewise.
3864 (arc_legitimate_constant_p): Likewise.
3865 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3866 * config/arc/arc.md (casesi_load): Likewise.
3867 (casesi_compact_jump): Likewise.
3868 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3869 (cc_use_register): Likewise.
3870 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3871 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3872 (arm_init_iwmmxt_builtins): Likewise.
3873 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3874 (neon_expand_vector_init): Likewise.
3875 (arm_attr_length_move_neon): Likewise.
3876 (maybe_get_arm_condition_code): Likewise.
3877 (arm_emit_vector_const): Likewise.
3878 (arm_preferred_simd_mode): Likewise.
3879 (arm_output_iwmmxt_tinsr): Likewise.
3880 (thumb1_output_casesi): Likewise.
3881 (thumb2_output_casesi): Likewise.
3882 (arm_emit_load_exclusive): Likewise.
3883 (arm_emit_store_exclusive): Likewise.
3884 (arm_expand_compare_and_swap): Likewise.
3885 (arm_evpc_neon_vuzp): Likewise.
3886 (arm_evpc_neon_vzip): Likewise.
3887 (arm_evpc_neon_vrev): Likewise.
3888 (arm_evpc_neon_vtrn): Likewise.
3889 (arm_evpc_neon_vext): Likewise.
3890 (arm_validize_comparison): Likewise.
3891 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3892 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3893 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3894 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3895 (c6x_preferred_simd_mode): Likewise.
3896 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3897 (epiphany_rtx_costs): Likewise.
3898 * config/epiphany/predicates.md (proper_comparison_operator):
3900 * config/frv/frv.c (condexec_memory_operand): Likewise.
3901 (frv_emit_move): Likewise.
3902 (output_move_single): Likewise.
3903 (output_condmove_single): Likewise.
3904 (frv_hard_regno_mode_ok): Likewise.
3905 (frv_matching_accg_mode): Likewise.
3906 * config/h8300/h8300.c (split_adds_subs): Likewise.
3907 (h8300_rtx_costs): Likewise.
3908 (h8300_print_operand): Likewise.
3909 (compute_mov_length): Likewise.
3910 (output_logical_op): Likewise.
3911 (compute_logical_op_length): Likewise.
3912 (compute_logical_op_cc): Likewise.
3913 (h8300_shift_needs_scratch_p): Likewise.
3914 (output_a_shift): Likewise.
3915 (compute_a_shift_length): Likewise.
3916 (compute_a_shift_cc): Likewise.
3917 (expand_a_rotate): Likewise.
3918 (output_a_rotate): Likewise.
3919 * config/i386/i386.c (classify_argument): Likewise.
3920 (function_arg_advance_32): Likewise.
3921 (function_arg_32): Likewise.
3922 (function_arg_64): Likewise.
3923 (function_value_64): Likewise.
3924 (ix86_gimplify_va_arg): Likewise.
3925 (ix86_legitimate_constant_p): Likewise.
3926 (put_condition_code): Likewise.
3927 (split_double_mode): Likewise.
3928 (ix86_avx256_split_vector_move_misalign): Likewise.
3929 (ix86_expand_vector_logical_operator): Likewise.
3930 (ix86_split_idivmod): Likewise.
3931 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3932 (ix86_build_const_vector): Likewise.
3933 (ix86_build_signbit_mask): Likewise.
3934 (ix86_match_ccmode): Likewise.
3935 (ix86_cc_modes_compatible): Likewise.
3936 (ix86_expand_branch): Likewise.
3937 (ix86_expand_sse_cmp): Likewise.
3938 (ix86_expand_sse_movcc): Likewise.
3939 (ix86_expand_int_sse_cmp): Likewise.
3940 (ix86_expand_vec_perm_vpermi2): Likewise.
3941 (ix86_expand_vec_perm): Likewise.
3942 (ix86_expand_sse_unpack): Likewise.
3943 (ix86_expand_int_addcc): Likewise.
3944 (ix86_split_to_parts): Likewise.
3945 (ix86_vectorize_builtin_gather): Likewise.
3946 (ix86_vectorize_builtin_scatter): Likewise.
3947 (avx_vpermilp_parallel): Likewise.
3948 (inline_memory_move_cost): Likewise.
3949 (ix86_tieable_integer_mode_p): Likewise.
3950 (x86_maybe_negate_const_int): Likewise.
3951 (ix86_expand_vector_init_duplicate): Likewise.
3952 (ix86_expand_vector_init_one_nonzero): Likewise.
3953 (ix86_expand_vector_init_one_var): Likewise.
3954 (ix86_expand_vector_init_concat): Likewise.
3955 (ix86_expand_vector_init_interleave): Likewise.
3956 (ix86_expand_vector_init_general): Likewise.
3957 (ix86_expand_vector_set): Likewise.
3958 (ix86_expand_vector_extract): Likewise.
3959 (emit_reduc_half): Likewise.
3960 (ix86_emit_i387_round): Likewise.
3961 (ix86_mangle_type): Likewise.
3962 (ix86_expand_round_sse4): Likewise.
3963 (expand_vec_perm_blend): Likewise.
3964 (canonicalize_vector_int_perm): Likewise.
3965 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3966 (expand_vec_perm_1): Likewise.
3967 (expand_vec_perm_interleave3): Likewise.
3968 (expand_vec_perm_even_odd_pack): Likewise.
3969 (expand_vec_perm_even_odd_1): Likewise.
3970 (expand_vec_perm_broadcast_1): Likewise.
3971 (ix86_vectorize_vec_perm_const_ok): Likewise.
3972 (ix86_expand_vecop_qihi): Likewise.
3973 (ix86_expand_mul_widen_hilo): Likewise.
3974 (ix86_expand_sse2_abs): Likewise.
3975 (ix86_expand_pextr): Likewise.
3976 (ix86_expand_pinsr): Likewise.
3977 (ix86_preferred_simd_mode): Likewise.
3978 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3979 * config/i386/sse.md (*andnot<mode>3): Likewise.
3980 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3981 (*<code><mode>3): Likewise.
3982 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3983 (ia64_expand_atomic_op): Likewise.
3984 (ia64_arg_type): Likewise.
3985 (ia64_mode_to_int): Likewise.
3986 (ia64_scalar_mode_supported_p): Likewise.
3987 (ia64_vector_mode_supported_p): Likewise.
3988 (expand_vec_perm_broadcast): Likewise.
3989 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3990 (iq2000_function_arg_advance): Likewise.
3991 (iq2000_function_arg): Likewise.
3992 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3993 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3994 (m68k_libcall_value): Likewise.
3995 (m68k_function_value): Likewise.
3996 (sched_attr_op_type): Likewise.
3997 * config/mcore/mcore.c (mcore_output_move): Likewise.
3998 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4000 (microblaze_function_arg): Likewise.
4001 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4002 (mips_print_operand): Likewise.
4003 (mips_mode_ok_for_mov_fmt_p): Likewise.
4004 (mips_vector_mode_supported_p): Likewise.
4005 (mips_preferred_simd_mode): Likewise.
4006 (mips_expand_vpc_loongson_even_odd): Likewise.
4007 (mips_expand_vec_unpack): Likewise.
4008 (mips_expand_vi_broadcast): Likewise.
4009 (mips_expand_vector_init): Likewise.
4010 (mips_expand_vec_reduc): Likewise.
4011 (mips_expand_msa_cmp): Likewise.
4012 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4013 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4014 (cc_flags_for_mode): Likewise.
4015 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4016 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4017 (nds32_output_casesi_pc_relative): Likewise.
4018 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4019 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4020 (nvptx_gen_unpack): Likewise.
4021 (nvptx_gen_pack): Likewise.
4022 (nvptx_gen_shuffle): Likewise.
4023 (nvptx_gen_wcast): Likewise.
4024 (nvptx_preferred_simd_mode): Likewise.
4025 * config/pa/pa.c (pa_secondary_reload): Likewise.
4026 * config/pa/predicates.md (base14_operand): Likewise.
4027 * config/powerpcspe/powerpcspe-c.c
4028 (altivec_resolve_overloaded_builtin): Likewise.
4029 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4031 (rs6000_preferred_simd_mode): Likewise.
4032 (output_vec_const_move): Likewise.
4033 (rs6000_expand_vector_extract): Likewise.
4034 (rs6000_split_vec_extract_var): Likewise.
4035 (reg_offset_addressing_ok_p): Likewise.
4036 (rs6000_legitimate_offset_address_p): Likewise.
4037 (rs6000_legitimize_address): Likewise.
4038 (rs6000_emit_set_const): Likewise.
4039 (rs6000_const_vec): Likewise.
4040 (rs6000_emit_move): Likewise.
4041 (spe_build_register_parallel): Likewise.
4042 (rs6000_darwin64_record_arg_recurse): Likewise.
4043 (swap_selector_for_mode): Likewise.
4044 (spe_init_builtins): Likewise.
4045 (paired_init_builtins): Likewise.
4046 (altivec_init_builtins): Likewise.
4047 (do_load_for_compare): Likewise.
4048 (rs6000_generate_compare): Likewise.
4049 (rs6000_expand_float128_convert): Likewise.
4050 (emit_load_locked): Likewise.
4051 (emit_store_conditional): Likewise.
4052 (rs6000_output_function_epilogue): Likewise.
4053 (rs6000_handle_altivec_attribute): Likewise.
4054 (rs6000_function_value): Likewise.
4055 (emit_fusion_gpr_load): Likewise.
4056 (emit_fusion_p9_load): Likewise.
4057 (emit_fusion_p9_store): Likewise.
4058 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4059 (fusion_gpr_mem_load): Likewise.
4060 (fusion_addis_mem_combo_load): Likewise.
4061 (fusion_addis_mem_combo_store): Likewise.
4062 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4063 (fusion_gpr_mem_load): Likewise.
4064 (fusion_addis_mem_combo_load): Likewise.
4065 (fusion_addis_mem_combo_store): Likewise.
4066 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4068 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4069 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4070 (rs6000_preferred_simd_mode): Likewise.
4071 (output_vec_const_move): Likewise.
4072 (rs6000_expand_vector_extract): Likewise.
4073 (rs6000_split_vec_extract_var): Likewise.
4074 (reg_offset_addressing_ok_p): Likewise.
4075 (rs6000_legitimate_offset_address_p): Likewise.
4076 (rs6000_legitimize_address): Likewise.
4077 (rs6000_emit_set_const): Likewise.
4078 (rs6000_const_vec): Likewise.
4079 (rs6000_emit_move): Likewise.
4080 (rs6000_darwin64_record_arg_recurse): Likewise.
4081 (swap_selector_for_mode): Likewise.
4082 (paired_init_builtins): Likewise.
4083 (altivec_init_builtins): Likewise.
4084 (rs6000_expand_float128_convert): Likewise.
4085 (emit_load_locked): Likewise.
4086 (emit_store_conditional): Likewise.
4087 (rs6000_output_function_epilogue): Likewise.
4088 (rs6000_handle_altivec_attribute): Likewise.
4089 (rs6000_function_value): Likewise.
4090 (emit_fusion_gpr_load): Likewise.
4091 (emit_fusion_p9_load): Likewise.
4092 (emit_fusion_p9_store): Likewise.
4093 * config/rx/rx.c (rx_gen_move_template): Likewise.
4094 (flags_from_mode): Likewise.
4095 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4096 (s390_slb_comparison): Likewise.
4097 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4098 (s390_vector_mode_supported_p): Likewise.
4099 (s390_cc_modes_compatible): Likewise.
4100 (s390_match_ccmode_set): Likewise.
4101 (s390_canonicalize_comparison): Likewise.
4102 (s390_emit_compare_and_swap): Likewise.
4103 (s390_branch_condition_mask): Likewise.
4104 (s390_rtx_costs): Likewise.
4105 (s390_secondary_reload): Likewise.
4106 (__SECONDARY_RELOAD_CASE): Likewise.
4107 (s390_expand_cs): Likewise.
4108 (s390_preferred_simd_mode): Likewise.
4109 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4110 * config/sh/sh.c (sh_print_operand): Likewise.
4111 (dump_table): Likewise.
4112 (sh_secondary_reload): Likewise.
4113 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4114 * config/sh/sh.md (casesi_worker_1): Likewise.
4115 (casesi_worker_2): Likewise.
4116 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4117 (fcc_comparison_operator): Likewise.
4118 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4119 (emit_soft_tfmode_cvt): Likewise.
4120 (sparc_preferred_simd_mode): Likewise.
4121 (output_cbranch): Likewise.
4122 (sparc_print_operand): Likewise.
4123 (sparc_expand_vec_perm_bmask): Likewise.
4124 (vector_init_bshuffle): Likewise.
4125 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4126 (spu_vector_mode_supported_p): Likewise.
4127 (spu_expand_insv): Likewise.
4128 (spu_emit_branch_or_set): Likewise.
4129 (spu_handle_vector_attribute): Likewise.
4130 (spu_builtin_splats): Likewise.
4131 (spu_builtin_extract): Likewise.
4132 (spu_builtin_promote): Likewise.
4133 (spu_expand_sign_extend): Likewise.
4134 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4135 (tilegx_simd_int): Likewise.
4136 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4137 (tilepro_simd_int): Likewise.
4138 * config/v850/v850.c (const_double_split): Likewise.
4139 (v850_print_operand): Likewise.
4140 (ep_memory_offset): Likewise.
4141 * config/vax/vax.c (vax_rtx_costs): Likewise.
4142 (vax_output_int_move): Likewise.
4143 (vax_output_int_add): Likewise.
4144 (vax_output_int_subtract): Likewise.
4145 * config/visium/predicates.md (visium_branch_operator): Likewise.
4146 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4147 (visium_print_operand_address): Likewise.
4148 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4149 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4150 (xtensa_expand_conditional_branch): Likewise.
4151 (xtensa_copy_incoming_a7): Likewise.
4152 (xtensa_output_literal): Likewise.
4153 * dfp.c (decimal_real_maxval): Likewise.
4154 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4157 Alan Hayward <alan.hayward@arm.com>
4158 David Sherwood <david.sherwood@arm.com>
4160 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4161 (mode_nunits_inline): Likewise.
4162 (mode_inner_inline): Likewise.
4163 (mode_unit_size_inline): Likewise.
4164 (mode_unit_precision_inline): Likewise.
4165 (emit_insn_modes_h): Likewise. Also emit a #define of the
4167 (emit_mode_wider): Add an E_ prefix to mode names.
4168 (emit_mode_complex): Likewise.
4169 (emit_mode_inner): Likewise.
4170 (emit_mode_adjustments): Likewise.
4171 (emit_mode_int_n): Likewise.
4172 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4173 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4174 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4175 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4176 (CRC32_BUILTIN, ENTRY): Likewise.
4177 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4178 (aarch64_pop_regs): Likewise.
4179 (aarch64_process_components): Likewise.
4180 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4181 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4182 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4183 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4184 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4185 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4187 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4188 (pcmpistr): Likewise.
4189 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4190 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4191 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4193 * config/rl78/rl78.c (mduc_regs): Likewise.
4194 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4195 (htm_expand_builtin): Likewise.
4196 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4197 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4198 * config/xtensa/xtensa.c (print_operand): Likewise.
4199 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4200 (NUM_MODE_VECTOR_INT): Likewise.
4201 * genoutput.c (null_operand): Likewise.
4202 (output_operand_data): Likewise.
4203 * genrecog.c (print_parameter_value): Likewise.
4204 * lra.c (debug_operand_data): Likewise.
4206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4208 * dwarf2out.c (scompare_loc_descriptor_wide)
4209 (scompare_loc_descriptor_narrow): New functions, split out from...
4210 (scompare_loc_descriptor): ...here.
4211 * expmed.c (emit_store_flag_int): New function, split out from...
4212 (emit_store_flag): ...here.
4214 2017-08-30 Richard Biener <rguenther@suse.de>
4216 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4217 (dwarf2out_early_finish): Move setting of AT_pubnames from
4218 early debug output to early finish.
4220 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4222 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4223 and -mdata-region to the assembler.
4225 2017-08-30 Richard Biener <rguenther@suse.de>
4227 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4229 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4230 (dwarf2out_early_global_decl): Only generate a DIE for the
4231 abstract origin if it doesn't already exist or is a declaration DIE.
4232 (resolve_addr): Do not add the linkage name twice when
4233 generating a stub DIE for the DW_TAG_GNU_call_site target.
4235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4237 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4238 Use machine_mode rather than int for arg1_mode.
4240 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4243 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4244 that the second argument of the built-in functions to unpack
4245 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4246 switch statement instead a lot of if statements.
4247 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4248 Allow 64-bit values to be in Altivec registers as well as
4249 traditional floating point registers.
4250 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4252 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4254 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4255 MAX_REGS_PER_ADDRESS == 1.
4257 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4259 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4260 * config/i386/i386.c (ix86_option_override_internal): Simplify
4261 setting of opts->x_flag_entry.
4263 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4264 Jakub Jelinek <jakub@redhat.com>
4265 Richard Biener <rguenther@suse.de>
4267 PR tree-optimization/81503
4268 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4269 folded constant fits in the target type; reorder tests for clarity.
4271 2017-08-29 Martin Liska <mliska@suse.cz>
4273 * passes.def: Include pass_lower_switch.
4274 * stmt.c (dump_case_nodes): Remove and move to
4275 tree-switch-conversion.
4276 (case_values_threshold): Likewise.
4277 (expand_switch_as_decision_tree_p): Likewise.
4278 (emit_case_decision_tree): Likewise.
4279 (expand_case): Likewise.
4280 (balance_case_nodes): Likewise.
4281 (node_has_low_bound): Likewise.
4282 (node_has_high_bound): Likewise.
4283 (node_is_bounded): Likewise.
4284 (emit_case_nodes): Likewise.
4285 (struct simple_case_node): New struct.
4286 (add_case_node): Remove.
4287 (emit_case_dispatch_table): Use vector instead of case_list.
4288 (reset_out_edges_aux): Remove.
4289 (compute_cases_per_edge): Likewise.
4290 (expand_case): Build list of simple_case_node.
4291 (expand_sjlj_dispatch_table): Use it.
4292 * tree-switch-conversion.c (struct case_node): Moved from
4293 stmt.c and adjusted.
4294 (emit_case_nodes): Likewise.
4295 (node_has_low_bound): Likewise.
4296 (node_has_high_bound): Likewise.
4297 (node_is_bounded): Likewise.
4298 (case_values_threshold): Likewise.
4299 (reset_out_edges_aux): Likewise.
4300 (compute_cases_per_edge): Likewise.
4301 (add_case_node): Likewise.
4302 (dump_case_nodes): Likewise.
4303 (balance_case_nodes): Likewise.
4304 (expand_switch_as_decision_tree_p): Likewise.
4305 (emit_jump): Likewise.
4306 (emit_case_decision_tree): Likewise.
4307 (try_switch_expansion): Likewise.
4308 (do_jump_if_equal): Likewise.
4309 (emit_cmp_and_jump_insns): Likewise.
4310 (fix_phi_operands_for_edge): New function.
4311 (record_phi_operand_mapping): Likewise.
4312 (class pass_lower_switch): New pass.
4313 (pass_lower_switch::execute): New function.
4314 (make_pass_lower_switch): Likewise.
4315 (conditional_probability):
4316 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4317 * tree-pass.h: Add make_pass_lower_switch.
4319 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4322 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4325 2017-08-29 Richard Biener <rguenther@suse.de>
4327 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4328 we do not add a DW_AT_inline attribute twice.
4329 (gen_subprogram_die): Remove code setting DW_AT_inline on
4330 DECL_ABSTRACT_P nodes.
4332 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4334 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4335 calls to internal functions.
4336 (gimplify_modify_expr): Likewise.
4337 * tree-call-cdce.c (use_internal_fn): Likewise.
4338 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4339 (convert_to_divmod): Set the nothrow flag.
4340 * tree-if-conv.c (predicate_mem_writes): Likewise.
4341 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4342 (vectorizable_call): Likewise.
4343 (vectorizable_store): Likewise.
4344 (vectorizable_load): Likewise.
4345 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4346 (vect_recog_mask_conversion_pattern): Likewise.
4348 2017-08-29 Martin Liska <mliska@suse.cz>
4351 * gcc.c (driver_handle_option): Add new argument.
4352 * opts-common.c (handle_option): Pass
4353 target_option_override_hook.
4354 * opts-global.c (lang_handle_option): Add new option.
4355 (set_default_handlers): Add new argument.
4356 (decode_options): Likewise.
4357 * opts.c (target_handle_option): Likewise.
4358 (common_handle_option): Call target_option_override_hook.
4359 * opts.h (struct cl_option_handler_func): Add hook for
4360 target option override.
4361 (struct cl_option_handlers): Likewise.
4362 (set_default_handlers): Add new argument.
4363 (decode_options): Likewise.
4364 (common_handle_option): Likewise.
4365 (target_handle_option): Likewise.
4366 * toplev.c (toplev::main): Pass targetm.target_option.override
4369 2017-08-29 Richard Biener <rguenther@suse.de>
4370 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4372 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4373 life to the active subtree.
4375 2017-08-28 Jeff Law <law@redhat.com>
4377 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4378 derive_equivalences.
4379 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4381 (edge_info::derive_equivalences): New private member function
4383 * tree-ssa-dom.c (class edge_info): Changed from a struct
4384 to a class. Add ctor/dtor, methods and data members.
4385 (edge_info::edge_info): Renamed from allocate_edge_info.
4386 Initialize additional members.
4387 (edge_info::~edge_info): New.
4388 (free_dom_edge_info): Delete the edge info.
4389 (record_edge_info): Use new class & associated member functions.
4390 Tighten forms for testing for edge equivalences.
4391 (record_temporary_equivalences): Iterate over the simple
4392 equivalences rather than assuming there's only one per edge.
4393 (cprop_into_successor_phis): Iterate over the simple
4394 equivalences rather than assuming there's only one per edge.
4395 (optimize_stmt): Use operand_equal_p rather than pointer
4396 equality for mini-DSE code.
4398 2017-08-28 Nathan Sidwell <nathan@acm.org>
4400 * gcc.c (execute): Fold SIGPIPE handling into switch
4401 statement. Adjust internal error message.
4403 2017-08-28 Richard Biener <rguenther@suse.de>
4406 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4407 Do nothing for removed DIEs.
4409 2017-08-28 Richard Biener <rguenther@suse.de>
4411 PR tree-optimization/81977
4412 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4415 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4418 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4419 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4420 using targetm.gen_mem_thread_fence.
4422 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4425 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4426 predicate to register_operand. Reorder operands.
4427 (*btr<mode>): Ditto.
4428 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4429 (*btr<mode>_mask): Ditto.
4431 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4433 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4434 * config/rs6000/xmmintrin.h: New file.
4435 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4437 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4440 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4441 parameter and_insn and return it.
4442 (recombine_lvx_pattern): Insert a copy to ensure availability of
4443 the base register of the copied masking operation at the point of
4444 the instruction replacement.
4445 (recombine_stvx_pattern): Likewise.
4447 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4449 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4450 undocumented switches.
4451 (-mpower9-dform-vector): Likewise.
4452 (-mpower9-dform): Likewise.
4453 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4454 comments to delete references to -mpower9-dform* switches.
4455 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4456 Delete reference to -mpower9-dform* switches, test for
4457 -mpower9-vector instead.
4458 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4459 (OTHER_P9_VECTOR_MASKS): Likewise.
4460 (POWERPC_MASKS): Likewise.
4461 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4462 tests against -mpower9-dform* to -mpower9-vector. Delete code
4463 that checked for -mpower9-dform* consistancy with other options.
4464 Add test for -mpower9-misc to enable other power9 switches.
4465 (rs6000_init_hard_regno_mode_ok): Likewise.
4466 (rs6000_option_override_internal): Likewise.
4467 (rs6000_emit_prologue): Likewise.
4468 (rs6000_emit_epilogue): Likewise.
4469 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4470 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4471 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4473 (emit_fusion_p9_store): Likewise.
4474 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4475 resetting these macros if the assembler does not support ISA 3.0
4477 (TARGET_P9_DFORM_VECTOR): Likewise.
4478 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4479 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4481 2017-08-25 Alan Modra <amodra@gmail.com>
4483 PR rtl-optimization/81747
4484 * cse.c (cse_extended_basic_block): Don't attempt to record
4485 equivalences for degenerate conditional branches that branch
4486 to their fall-through.
4488 2017-08-24 Martin Sebor <msebor@redhat.com>
4491 * gimple-fold.c (size_must_be_zero_p): New function.
4492 (gimple_fold_builtin_memory_op): Call it.
4494 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4496 * config/rs6000/mm_malloc.h: New file.
4498 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4500 PR tree-optimization/81913
4501 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4502 analysis when either IVs in condition can wrap.
4504 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4506 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4507 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4509 2017-08-24 Richard Biener <rguenther@suse.de>
4512 * targhooks.c (default_target_can_inline_p): Properly
4513 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4514 is present and always compare.
4515 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4516 infer -mfpmath=sse from TARGET_SSE_P.
4517 (ix86_can_inline_p): Properly use target_option_default_node when
4518 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4520 2017-08-24 Richard Biener <rguenther@suse.de>
4523 * dwarf2out.c (output_die): Handle flag_generate_offload like
4525 (output_comp_unit): Likewise.
4526 (gen_array_type_die): Likewise.
4527 (dwarf2out_early_finish): Likewise.
4528 (note_variable_value_in_expr): Likewise.
4529 (dwarf2out_finish): Likewise. Adjust assert.
4530 * cgraphunit.c (symbol_table::compile): Move setting of
4531 flag_generate_offload earlier ...
4532 (symbol_table::finalize_compilation_unit): ... here, before
4533 early debug finalization.
4535 2017-08-24 Richard Biener <rguenther@suse.de>
4537 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4538 and ipa-fnsummary.h.
4539 (ix86_can_inline_p): When ix86_fpmath flags do not match
4540 check whether the callee uses FP math at all.
4542 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4545 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4546 instead of TYPE_PRECISION.
4548 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4549 Alan Hayward <alan.hayward@arm.com>
4550 David Sherwood <david.sherwood@arm.com>
4552 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4554 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4556 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4558 (ix86_frame::stack_realign_allocate): New field.
4559 (struct machine_frame_state): Modify comments.
4560 (machine_frame_state::sp_realigned_fp_end): New field.
4561 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4563 (sp_valid_at): Add assertion to assure no attempt to access invalid
4564 offset of a realigned stack.
4565 (fp_valid_at): Likewise.
4566 (choose_baseaddr): Modify comments.
4567 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4568 ix86_expand_prologue.
4569 (ix86_expand_prologue): Modify stack realignment and allocation.
4570 (ix86_expand_epilogue): Modify comments.
4571 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4572 avx2_runtime, avx512f, and avx512f_runtime.
4574 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4576 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4577 (mstackrealign): Do not init to -1.
4578 * config/i386/i386.c (ix86_option_override_internal):
4579 Check opts_set, not opts when setting default value of
4580 opts->x_ix86_force_align_arg_pointer.
4582 2017-08-23 Richard Biener <rguenther@suse.de>
4584 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4585 lang_hooks.decl_printable_name.
4586 * print-rtl-function.c (print_rtx_function): Likewise.
4587 * tree-pretty-print.c (dump_function_header): Likewise.
4589 2017-08-23 Richard Biener <rguenther@suse.de>
4592 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4593 -g0 at compile-time.
4595 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4598 * doc/sourcebuild.texi (Other hardware attributes):
4601 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4604 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4605 Fix single-use check.
4607 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4609 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4611 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4613 * doc/install.texi: Modify to add more details on running selected
4616 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4618 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4619 is combined with -mabi=ms.
4620 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4623 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4625 PR tree-optimization/81488
4626 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4627 and cached_basis fields.
4628 (MAX_SPREAD): New constant.
4629 (alloc_cand_and_find_basis): Initialize new fields.
4630 (clear_visited): New function.
4631 (create_phi_basis_1): Rename from create_phi_basis, set visited
4632 and cached_basis fields.
4633 (create_phi_basis): New wrapper function.
4634 (phi_add_costs_1): Rename from phi_add_costs, add spread
4635 parameter, set visited field, short-circuit when limits reached.
4636 (phi_add_costs): New wrapper function.
4637 (record_phi_increments_1): Rename from record_phi_increments, set
4639 (record_phi_increments): New wrapper function.
4640 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4641 (phi_incr_cost): New wrapper function.
4642 (all_phi_incrs_profitable_1): Rename from
4643 all_phi_incrs_profitable, set visited field.
4644 (all_phi_incrs_profitable): New wrapper function.
4646 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4647 Alan Hayward <alan.hayward@arm.com>
4648 David Sherwood <david.sherwood@arm.com>
4650 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4651 that takes the outer and inner modes.
4652 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4653 comparison as the canonical test for a paradoxical subreg.
4654 * combine.c (simplify_set): Use paradoxical_subreg_p.
4655 (make_extraction): Likewise.
4656 (force_to_mode): Likewise.
4657 (rtx_equal_for_field_assignment_p): Likewise.
4658 (gen_lowpart_for_combine): Likewise.
4659 (simplify_comparison): Likewise.
4660 * cse.c (equiv_constant): Likewise.
4661 * expmed.c (store_bit_field_1): Likewise.
4662 * final.c (alter_subreg): Likewise.
4663 * fwprop.c (propagate_rtx): Likewise.
4664 (forward_propagate_subreg): Likewise.
4665 * ira-conflicts.c (ira_build_conflicts): Likewise.
4666 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4667 * lra-constraints.c (curr_insn_transform): Likewise.
4668 (split_reg): Likewise.
4669 * lra-eliminations.c (move_plus_up): Likewise.
4670 (lra_eliminate_regs_1): Likewise.
4671 * recog.c (general_operand): Likewise.
4672 * ree.c (combine_reaching_defs): Likewise.
4673 * reload.c (push_reload): Likewise.
4674 (find_reloads): Likewise.
4675 * reload1.c (elimination_effects): Likewise.
4676 (compute_reload_subreg_offset): Likewise.
4677 (choose_reload_regs): Likewise.
4678 * rtlanal.c (subreg_lsb_1): Likewise.
4679 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4680 (simplify_subreg): Likewise.
4681 * var-tracking.c (track_loc_p): Likewise.
4682 * emit-rtl.c (byte_lowpart_offset): Likewise.
4683 (paradoxical_subreg_p): Delete out-of-line definition.
4685 2017-08-22 Jeff Law <law@redhat.com>
4687 PR tree-optimization/81741
4688 PR tree-optimization/71947
4689 * tree-ssa-dom.c: Include tree-inline.h.
4690 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4691 equivalences if one is more expensive to compute than the other.
4692 * tree-ssa-scopedtables.h (class const_or_copies): Make
4693 record_const_or_copy_raw method private.
4694 (class avail_exprs_stack): New method simplify_binary_operation.
4695 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4696 avail_exprs_stack::simplify_binary_operation as needed.
4697 (avail_exprs_stack::simplify_binary_operation): New function.
4699 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4701 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4702 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4703 (DOT_SYMBOLS): Likewise.
4704 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4705 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4706 (RS6000_ABI_NAME): Likewise.
4707 (TARGET_CMODEL): Likewise.
4708 (TOC_SECTION_ASM_OP): Likewise.
4709 (SET_CMODEL): New macro.
4710 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4712 2017-08-22 Richard Biener <rguenther@suse.de>
4714 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4715 to allow for free-lang-data replacements similar to verify_type_variant.
4717 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4718 Alan Hayward <alan.hayward@arm.com>
4719 David Sherwood <david.sherwood@arm.com>
4721 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4722 VOIDmode for the LABEL_REF.
4724 2017-08-22 Richard Biener <rguenther@suse.de>
4726 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4728 2017-08-22 Richard Biener <rguenther@suse.de>
4730 * common.opt (feliminate-dwarf2-dups): Ignore.
4731 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4732 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4733 same_die_p_wrap, compute_section_prefix,
4734 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4735 (comdat_symbol_id, comdat_symbol_number): Likewise.
4736 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4738 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4739 (output_die): Mark unreachable path unreachable.
4740 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4741 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4742 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4743 (dwarf2out_early_finish): Likewise.
4745 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4747 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4749 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4752 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4753 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4754 (avr_attribute_table) <io, io_low, address>: Initialize
4755 .decl_required with true.
4757 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4759 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4760 undocumented debugging options.
4761 (-mvsx-scalar-double): Likewise.
4762 (-mallow-df-permute): Likewise.
4763 (-mvectorize-builtins): Likewise.
4764 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4765 (rs6000_builtin_vectorized_function): Likewise.
4766 (rs6000_builtin_md_vectorized_function): Likewise.
4767 (rs6000_opt_vars): Likewise.
4769 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4772 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4773 (*btrq_imm): Rename from *btrq.
4774 (*btcq_imm): Rename from *btcq.
4775 (btsc): New code attribute.
4776 (*<btsc><mode>): New insn pattern.
4777 (*btr<mode>): Ditto.
4778 (*<btsc><mode>_mask): New insn_and_split pattern.
4779 (*btr<mode>_mask): Ditto.
4781 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4782 Alan Hayward <alan.hayward@arm.com>
4783 David Sherwood <david.sherwood@arm.com>
4785 * function.c (pad_below): Simplify padding calculation.
4787 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4788 Alan Hayward <alan.hayward@arm.com>
4789 David Sherwood <david.sherwood@arm.com>
4791 * target.def (function_prologue): Remove frame size argument.
4792 (function_epilogue): Likewise.
4793 * doc/tm.texi: Regenerate.
4794 * final.c (final_start_function): Update call to function_prologue.
4795 (final_end_function): Update call to function_epilogue.
4796 (default_function_pro_epilogue): Remove frame size argument.
4797 * output.h (default_function_pro_epilogue): Likewise.
4798 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4799 (arm_output_function_prologue): Likewise.
4800 * config/frv/frv.c (frv_function_prologue): Likewise.
4801 (frv_function_epilogue): Likewise.
4802 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4803 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4804 (ia64_output_function_epilogue): Likewise.
4805 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4806 (m32r_output_function_epilogue): Likewise.
4807 * config/microblaze/microblaze.c (microblaze_function_prologue)
4808 (microblaze_function_epilogue): Likewise.
4809 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4810 (mips_output_function_epilogue): Likewise.
4811 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4812 (mmix_target_asm_function_epilogue): Likewise.
4813 * config/msp430/msp430.c (msp430_start_function): Likewise.
4814 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4815 (nds32_asm_function_epilogue): Likewise.
4816 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4817 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4818 (pa_output_function_epilogue): Likewise.
4819 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4820 (rs6000_output_function_epilogue): Likewise.
4821 * config/rl78/rl78.c (rl78_start_function): Likewise.
4822 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4823 (rs6000_output_function_epilogue): Likewise.
4824 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4825 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4826 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4827 (sparc_asm_function_epilogue): Likewise.
4829 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4831 * tree.h (type_has_mode_precision_p): New function.
4832 * convert.c (convert_to_integer_1): Use it.
4833 * expr.c (expand_expr_real_2): Likewise.
4834 (expand_expr_real_1): Likewise.
4835 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4836 * match.pd: Likewise.
4837 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4838 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4839 * tree-tailcall.c (process_assignment): Likewise.
4840 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4841 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4842 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4843 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4844 (vectorizable_assignment): Likewise.
4845 (vectorizable_shift): Likewise.
4846 (vectorizable_operation): Likewise.
4847 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4849 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4851 * match.pd: Add pow (C, x) simplification.
4853 2017-08-21 Richard Biener <rguenther@suse.de>
4855 PR tree-optimization/81900
4856 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4857 for blocks with abnormal predecessors.
4858 (compute_antic): Do not set visited flag prematurely.
4860 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4863 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4865 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4867 * stor-layout.h (vector_type_mode): Move to...
4868 * tree.h (vector_type_mode): ...here.
4869 * stor-layout.c (vector_type_mode): Move to...
4870 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4872 2017-08-21 Richard Biener <rguenther@suse.de>
4874 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4875 register_external_die hooks.
4876 (debug_false_tree_charstarstar_uhwistar): Declare.
4877 (debug_nothing_tree_charstar_uhwi): Likewise.
4878 * debug.c (do_nothing_debug_hooks): Adjust.
4879 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4880 (debug_nothing_tree_charstar_uhwi): Likewise.
4881 * dbxout.c (dbx_debug_hooks): Adjust.
4882 (xcoff_debug_hooks): Likewise.
4883 * sdbout.c (sdb_debug_hooks): Likewise.
4884 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4885 * dwarf2out.c (macinfo_label_base): New global.
4886 (dwarf2out_register_external_die): New function for the
4887 register_external_die hook.
4888 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4889 (dwarf2_debug_hooks): Use them.
4890 (dwarf2_lineno_debug_hooks): Adjust.
4891 (struct die_struct): Add with_offset flag.
4892 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4893 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4894 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4895 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4896 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4897 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4898 defining section names for the early LTO debug variants.
4899 (reset_indirect_string): New helper.
4900 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4901 (print_dw_val): Add support for offsetted symbol references.
4902 (get_ultimate_context): Split out from is_cxx.
4903 (is_cxx): Use get_ultimate_context.
4904 (is_fortran): Add decl overload.
4905 (compute_comp_unit_symbol): Split out worker from
4906 compute_section_prefix.
4907 (compute_section_prefix): Call compute_comp_unit_symbol and
4908 set comdat_type_p here.
4909 (output_die): Skip DIE symbol output for the LTO added one.
4910 Handle DIE symbol references with offset.
4911 (output_comp_unit): Guard section name mangling properly.
4912 For LTO debug sections emit a symbol at the section beginning
4913 which we use to refer to its DIEs.
4914 (add_abstract_origin_attribute): For DIEs registered via
4915 dwarf2out_register_external_die directly refer to the early
4916 DIE rather than indirectly through the shadow one we created.
4917 Remove obsolete call to dwarf2out_abstract_function for
4918 non-function/block origins.
4919 (gen_array_type_die): When generating early LTO debug do
4920 not emit DW_AT_string_length.
4921 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4922 late when in LTO. As suggested place a gcc_unreachable for
4923 the DECL_ABSTRACT_P case.
4924 (gen_subprogram_die): Avoid another specification DIE
4925 for early built declarations/definitions for the late LTO case.
4926 (gen_variable_die): Add type references for late duplicated VLA dies
4928 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4929 we have the abstract instance already.
4930 (process_scope_var): Adjust decl DIE contexts in LTO which
4931 first puts them in limbo.
4932 (gen_decl_die): Do not generate type DIEs late apart from
4933 types for VLAs or for decls we do not yet have a DIE. Do not
4934 call dwarf2out_abstract_function late.
4935 (dwarf2out_early_global_decl): Make sure to create DIEs
4936 for abstract instances of a decl first.
4937 (dwarf2out_late_global_decl): Adjust comment.
4938 (output_macinfo_op): With multiple macro sections use
4939 macinfo_label_base to distinguish labels.
4940 (output_macinfo): Likewise. Update macinfo_label_base.
4941 Pass in the line info label.
4942 (note_variable_value_in_expr): When generating LTO resolve
4943 all variable values here by generating DIEs as needed.
4944 (init_sections_and_labels): Add early LTO debug flag parameter
4945 and generate different sections and names if set. Add generation
4946 counter for the labels so we can have multiple of them.
4947 (reset_dies): Helper to allow DIEs to be output multiple times.
4948 (dwarf2out_finish): When outputting DIEs to the fat part of an
4949 LTO object first reset DIEs.
4950 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4951 (modified_type_die): Check for decl_ultimate_origin being self
4953 (gen_type_die_with_usage): Likewise.
4954 (gen_typedef_die): Allow decl_ultimate_origin being self.
4955 (set_decl_abstract_flags): Remove.
4956 (set_block_abstract_flags): Likewise.
4957 (dwarf2out_abstract_function): Treat the early generated DIEs
4958 as the abstract copy and only add DW_AT_inline and
4959 DW_AT_artificial here and call set_decl_origin_self.
4960 If the DIE has an abstract origin don't do anything.
4961 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4962 if we have none yet (Go fails to build one, PR78628).
4963 (variably_modified_type_p): Prevent endless recursion for Ada
4964 cyclic pointer types.
4965 * lto-streamer-in.c: Include debug.h.
4966 (dref_queue): New global.
4967 (lto_read_tree_1): Stream in DIE references.
4968 (lto_input_tree): Register DIE references.
4969 (input_function): Stream DECL_DEBUG_ARGS.
4970 * lto-streamer-out.c: Include debug.h.
4971 (lto_write_tree_1): Output DIE references.
4972 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4973 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4974 (output_function): Stream DECL_DEBUG_ARGS.
4975 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4976 Stream DECL_ABSTRACT_ORIGIN.
4977 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4978 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4979 DECL_CONTEXT for file-scope decls.
4980 * lto-streamer.h (struct dref_entry): Declare.
4981 (dref_queue): Likewise.
4982 * cfgexpand.c (pass_expand::execute): Do not call the
4983 outlining_inline_function hook here.
4984 * lto-wrapper.c (debug_obj): New global.
4985 (tool_cleanup): Unlink it if required.
4986 (debug_objcopy): New function.
4987 (run_gcc): Handle early debug sections in the IL files by
4988 extracting them to separate files, partially linkin them and
4989 feeding the result back as result to the linker.
4990 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4991 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4992 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4993 sections into a separate segment.
4994 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4996 (darwin_asm_dwarf_section): Likewise.
4997 (darwin_asm_output_dwarf_offset): Likewise.
4998 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5000 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5001 Alan Hayward <alan.hayward@arm.com>
5002 David Sherwood <david.sherwood@arm.com>
5004 * read-md.h (md_reader::record_potential_iterator_use): Replace
5005 pointer argument with an rtx and an index.
5006 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5007 (apply_mode_iterator): Likewise.
5008 (apply_code_iterator): Likewise.
5009 (apply_int_iterator): Likewise.
5010 (apply_subst_iterator): Likewise.
5011 (record_iterator_use): Likewise.
5012 (record_attribute_use): Likewise.
5013 (md_reader::record_potential_iterator_use): Likewise. Update calls
5014 to record_iterator_use and apply_iterator.
5015 (iterator_use): Replace ptr with x and index.
5016 (attribute_use): Likewise.
5017 (apply_attribute_uses): Update calls to apply_iterator.
5018 (apply_iterators): Likewise. Update initialization of iterator_use.
5019 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5020 and record_potential_iterator_use.
5021 (rtx_reader::read_rtx_operand): Likewise.
5023 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5024 Alan Hayward <alan.hayward@arm.com>
5025 David Sherwood <david.sherwood@arm.com>
5027 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5030 2017-08-21 Richard Biener <rguenther@suse.de>
5033 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5034 at struct end conservatively when comparing common bases.
5036 2017-08-21 Richard Biener <rguenther@suse.de>
5038 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5039 (mem_ref_in_stmt): Remove.
5040 (determine_max_movement): Use ref index to get at the reference.
5041 (invariantness_dom_walker::before_dom_children): Deal with
5042 lim data already initialized.
5043 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5045 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5047 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5048 (TARGET_ISA_ROUND): Ditto.
5049 (TARGET_ROUND): Ditto.
5050 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5051 * config/i386/i386.md: Ditto.
5052 * config/i386/sse.md: Ditto.
5053 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5054 with OPTION_MASK_ISA_SSE4_1.
5056 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5059 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5060 __builtin_ia32_lzcnt_u16.
5062 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5065 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5066 (rs6000_set_current_function): Rewrite function to use it.
5068 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5071 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5072 and TYPE_WARN_IF_NOT_ALIGN.
5073 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5074 (handle_warn_if_not_align): New.
5075 (place_union_field): Call handle_warn_if_not_align.
5076 (place_field): Call handle_warn_if_not_align.
5077 Copy TYPE_WARN_IF_NOT_ALIGN.
5078 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5079 (layout_type): Likewise.
5080 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5082 (tree_decl_common): Add warn_if_not_align.
5083 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5084 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5085 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5086 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5087 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5088 * doc/extend.texi: Document warn_if_not_aligned attribute.
5089 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5091 2017-08-17 Martin Liska <mliska@suse.cz>
5094 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5095 (get_data_dependence): Use it as pointer type.
5096 (distribute_loop): Likewise.
5098 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5100 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5101 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5102 (p8_vmrgow_<mode>_direct): New define_insn.
5103 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5104 handle endianness for vmrgew and vmrgow permute patterns.
5106 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5108 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5109 * config/rs6000/rs6000-cpus.def: Remove comment.
5110 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5111 (POWERPC_MASKS): Likewise.
5112 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5113 use of TARGET_VSX_TIMODE.
5114 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5115 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5116 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5117 (rs6000_option_override_internal): Remove dead code.
5118 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5119 (rs6000_legitimize_reload_address): Likewise.
5120 (rs6000_legitimate_address_p): Likewise.
5121 (rs6000_opt_masks): Delete "vsx-timode".
5122 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5123 from function comment.
5124 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5125 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5126 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5128 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5129 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5130 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5133 (define_peephole2): Likewise.
5135 2017-08-17 Martin Sebor <msebor@redhat.com>
5138 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5139 past the end of an array.
5140 (test_pp_format): Add test cases.
5142 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5144 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5145 missing ECF_NOTHROW flags.
5147 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5150 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5151 operands residing in integer registers.
5152 (*vsx_le_perm_load_<mode>): Likewise.
5153 (*vsx_le_perm_store_<mode>): Likewise.
5154 (define_peephole2): Add peepholes to optimize the above.
5156 2017-08-17 Marek Polacek <polacek@redhat.com>
5159 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5160 to mimic what shorten_compare did. Change the return type to bool.
5161 (fold_cond_expr_with_comparison): Update call to
5162 operand_equal_for_comparison_p.
5163 (fold_ternary_loc): Likewise.
5165 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5167 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5169 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5171 2017-08-17 Richard Biener <rguenther@suse.de>
5173 * tree-ssa-structalias.c (solve_graph): When propagating
5174 to successors update the graphs succ edges and avoid duplicate work.
5176 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5179 * config/i386/i386.c (ix86_option_override_internal): Save target
5180 specific options after ix86_stack_protector_guard_reg was changed.
5182 2017-08-17 Richard Biener <rguenther@suse.de>
5184 PR tree-optimization/81827
5185 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5186 (new_var_info): Initialize it conservatively.
5187 (get_call_vi): Mark register vars.
5188 (new_scalar_tmp_constraint_exp): Likewise.
5189 (handle_rhs_call): Likewise.
5190 (handle_const_call): Likewise.
5191 (create_function_info_for): Likewise.
5192 (solve_constraints): Sort varinfos to separate register from
5193 non-register vars to pack points-to solution bitmaps during
5196 2017-08-17 Marek Polacek <polacek@redhat.com>
5198 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5200 2017-08-17 Richard Biener <rguenther@suse.de>
5202 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5203 to true when overflow is undefined and we saturated the result.
5205 2017-08-17 Alan Modra <amodra@gmail.com>
5208 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5209 Don't use store multiple if only one reg needs saving.
5210 (interesting_frame_related_regno): New function.
5211 (rs6000_frame_related): Don't emit frame info for regs that
5213 (rs6000_emit_epilogue): Likewise.
5215 2017-08-16 Nathan Sidwell <nathan@acm.org>
5217 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5218 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5219 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5220 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5221 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5222 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5223 (verify_type): Adjust for TYPE_BINFO move.
5224 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5225 process TYPE_BINFO directly.
5226 (hash_tree): Likewise.
5227 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5229 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5232 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5234 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5236 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5239 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5240 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5241 Add ix86_binary_operator_ok to insn constraint.
5242 (*iordi_1_bts): Ditto.
5243 (*xordi_1_btc): Ditto.
5244 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5245 Update corresponding peephole2 pattern.
5249 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5251 PR tree-optimization/81832
5252 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5253 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5255 2017-08-16 Marek Polacek <polacek@redhat.com>
5258 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5259 regarding size_zero_node.
5261 2017-08-16 Martin Liska <mliska@suse.cz>
5264 * config.gcc: Respect previously set extra_objs in case
5267 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5269 PR tree-optimization/81835
5270 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5271 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5272 not depend on the phi.
5274 2017-08-16 Alan Modra <amodra@gmail.com>
5276 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5279 2017-08-16 Alan Modra <amodra@gmail.com>
5281 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5282 (save_reg_p): ..into this. Update all callers.
5283 (first_reg_to_save): Simplify.
5285 2017-08-16 Alan Modra <amodra@gmail.com>
5287 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5290 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5294 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5295 constant count if that count is less than 32.
5297 2017-08-15 Nathan Sidwell <nathan@acm.org>
5299 * gcc.c (execute): Emit friendlier message if inferior is killed
5300 by an external cause.
5302 2017-08-15 Richard Biener <rguenther@suse.de>
5304 PR tree-optimization/81790
5305 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5306 CONSTRUCTORs from simplifying and VN.
5308 2017-08-14 Martin Sebor <msebor@redhat.com>
5310 * builtin-attrs.def: Add comments.
5312 2017-08-14 Martin Sebor <msebor@redhat.com>
5315 * doc/extend.texi (attribute nonstring): Document new attribute.
5317 2017-08-14 Martin Sebor <msebor@redhat.com>
5320 * tree-diagnostic.c (default_tree_printer): Handle %G.
5321 * gimple-pretty-print.h (percent_G_format): Declare new function.
5322 * gimple-pretty-print.c (percent_G_format): Define.
5323 * tree-pretty-print.c (percent_K_format): Add argument.
5325 2017-08-14 Martin Sebor <msebor@redhat.com>
5327 PR translation/79998
5328 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5329 Remove a stray space.
5331 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5334 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5335 (*iordi_1_bts): Ditto.
5336 (*xordi_1_btc): Ditto.
5338 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5341 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5342 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5344 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5345 quoted strings, and make more translator-friendly.
5346 (darwin_rs6000_override_options): Likewise.
5347 (rs6000_option_override_internal): Likewise.
5348 (rs6000_return_in_memory): Fix overlong line.
5349 (init_cmulative_args): Use quoted strings, and make more
5350 translator-friendly.
5351 (rs6000_pass_by_reference): Fix overlong line.
5352 (def_builtin): Use quoted strings.
5353 (altivec_expand_predicate_builtin): Use quoted strings, and make
5354 more translator-friendly.
5355 (htm_expand_builtin): Use quoted strings.
5356 (cpu_expand_builtin): Use quoted strings, and make more
5357 translator-friendly.
5358 (altivec_expand_builtin): Likewise.
5359 (paired_expand_predicate_builtin): Likewise.
5360 (rs6000_invalid_builtin): Likewise.
5361 (builtin_function_type): Use quoted strings.
5362 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5363 more translator-friendly.
5364 (rs6000_trampoline_init): Likewise.
5365 (rs6000_handle_altivec_attribute): Likewise.
5366 (rs6000_inner_target_options): Use quoted strings.
5367 (rs6000_disable_incompatible_switches): Likewise.
5368 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5369 strings, and make more translator-friendly.
5370 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5372 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5374 PR tree-optimization/81799
5375 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5376 cond_expr to simple gimple operand.
5378 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5381 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5383 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5387 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5388 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5389 TARGET_GASISR_PROLOGUES.
5390 * config/avr/avr.c (avr_option_override): Same.
5391 (avr_pass_pre_proep::execute): Same.
5393 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5396 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5397 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5399 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5401 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5402 define_insn_and_split. Split to a memory load from 0 in
5403 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5404 using PTR mode iterator.
5405 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5406 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5407 (*add_tp_<mode>): Redefine as define_insn_and_split.
5408 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5409 address space. Merge with *add_tp_x32 using PTR mode iterator.
5410 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5411 Split to an add with a memory load from 0 in
5412 DEFAULT_TLS_SEG_REG address space.
5414 2017-08-12 Andrew Pinski <apinski@cavium.com>
5416 * config/aarch64/aarch64-option-extensions.def (rdma):
5417 Fix feature string to what Linux prints out in /proc/cpuinfo.
5419 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5422 * dwarf2out.c (modified_type_die): For C typedef types that have
5423 an ultimate origin, process the ultimate origin instead of the
5425 (gen_typedef_die): Assert that input DECLs have no ultimate
5427 (gen_type_die_with_usage): For typedef variants that have an
5428 ultimate origin, just call gen_decl_die on the original DECL.
5429 (process_scope_var): Avoid creating DIEs for local typedefs and
5430 concrete static variables.
5432 2017-08-12 Alan Modra <amodra@gmail.com>
5436 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5437 match gnu-user.h startfile.
5438 (ENDFILE_LINUX_SPEC): Similarly.
5440 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5443 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5445 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5447 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5448 * config/aarch64/aarch64.md (mov<mode>): Change.
5449 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5450 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5451 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5453 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5455 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5456 for storage order barriers.
5458 2017-08-11 Martin Liska <mliska@suse.cz>
5461 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5462 variables of void type.
5464 2017-08-11 Martin Liska <mliska@suse.cz>
5466 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5467 TARGET_SUPPORTS_ALIASES.
5468 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5469 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5470 (optimize_weakref): Likewise.
5471 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5472 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5473 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5475 2017-08-11 Martin Liska <mliska@suse.cz>
5478 * config/i386/i386.c (make_resolver_func): Do complete
5479 refactoring of the function.
5481 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5484 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5485 * config/i386/i386.c (ix86_stack_protect_guard): Use
5486 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5487 * doc/invoke.texi (x86 Options): Document
5488 -mstack-protector-guard-symbol= option.
5490 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5492 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5493 * config/i386/i386.c (ix86_split_stack_guard): New function.
5494 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5495 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5496 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5497 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5498 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5499 (split_stack_space_check): Call ix86_split_stack_guard.
5501 2017-08-10 Martin Sebor <msebor@redhat.com>
5503 * print-tree.c (print_node): Print location using the established
5505 Replace spaces with colons.
5506 (debug_raw, debug): Ditto.
5508 2017-08-10 Martin Sebor <msebor@redhat.com>
5511 * pretty-print.c (pp_format): Correct the handling of %s precision.
5513 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5516 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5518 (ix86_finalize_stack_frame_flags): This. Also clear
5519 frame_pointer_needed if -fno-omit-frame-pointer is used without
5521 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5522 with ix86_finalize_stack_frame_flags.
5523 (ix86_expand_epilogue): Likewise.
5524 (ix86_expand_split_stack_prologue): Likewise.
5525 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5527 2017-08-10 Martin Liska <mliska@suse.cz>
5530 * c-attribs.c (handle_target_attribute):
5531 Report warning for an empty string argument of target attribute.
5533 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5536 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5538 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5539 or DECL_NONLOCAL labels.
5540 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5541 or DECL_NONLOCAL labels here.
5543 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5545 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5546 to indicate when early gimple folding has been disabled.
5547 (rs6000_gimple_fold_builtin): Add debug content.
5548 (rs6000_invalid_builtin): Fix whitespace.
5549 (rs6000_expand_builtin): Fix whitespace.
5550 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5552 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5555 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5556 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5558 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5560 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5561 (qdf24xx): Likewise.
5562 * config/aarch64/aarch64.md: Include falkor.md.
5563 * config/aarch64/falkor.md: New.
5565 2017-08-09 Marek Polacek <polacek@redhat.com>
5568 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5569 * diagnostic.c (emit_diagnostic): Add a comment.
5570 (emit_diagnostic_valist): New function.
5572 2017-08-09 Marek Polacek <polacek@redhat.com>
5575 * input.c (make_location): New overload.
5576 * input.h (make_location): Declare.
5578 2017-08-08 Alan Modra <amodra@gmail.com>
5579 H.J. Lu <hongjiu.lu@intel.com>
5582 * gcc.c (NO_PIE_SPEC): Delete.
5583 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5585 (LINK_PIE_SPEC): ..to here.
5586 (LINK_COMMAND_SPEC): Support -no-pie.
5587 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5588 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5590 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5591 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5592 (ENDFILE_CRTEND_SPEC): Similarly.
5594 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5597 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5598 (mstack-protector-guard-offset=): Ditto.
5599 * config/i386/i386.c (ix86_option_override): Handle
5600 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5602 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5603 ix86_stack_protect_guard_offset variables.
5604 (TARGET_STACK_PROTECT_GUARD): Always define.
5605 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5606 and -mstack-protector-guard-offset= options.
5608 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5610 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5611 boundary case for the last candidate.
5613 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5615 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5616 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5618 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5621 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5622 * config/aarch64/aarch64-builtins.c
5623 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5624 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5625 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5627 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5628 Andrew Pinski <pinskia@gmail.com>
5631 * internal-fn.def (XORSIGN): New.
5632 * optabs.def (xorsign_optab): New.
5633 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5634 (convert_expand_mult_copysign): New.
5635 (pass_optimize_widening_mul::execute): Call
5636 convert_expand_mult_copysign.
5638 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5640 PR tree-optimization/81354
5641 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5642 Insert on edges rather than explicitly creating landing pads.
5643 (analyze_candidates_and_replace): Commit edge inserts.
5645 2017-08-08 Richard Biener <rguenther@suse.de>
5648 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5649 (expand_simple_operations): Also look through ADDR_EXPRs with
5650 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5652 2017-08-08 Richard Biener <rguenther@suse.de>
5654 PR tree-optimization/81723
5655 * tree-vect-slp.c (struct bst_traits): New hash traits.
5656 (bst_fail): New global.
5657 (vect_build_slp_tree_2): New worker, split out from ...
5658 (vect_build_slp_tree): ... this now wrapping it with using
5659 bst_fail set to cache SLP tree build fails. Properly handle
5661 (vect_analyze_slp_instance): Allocate and free bst_fail.
5663 2017-08-08 Martin Liska <mliska@suse.cz>
5666 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5667 LABEL_DECLs that can be from a different function.
5669 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5671 PR tree-optimization/81744
5672 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5673 loop's number of iterations.
5675 2017-08-08 Martin Liska <mliska@suse.cz>
5677 * asan.c: Include header files.
5678 * attribs.c (build_decl_attribute_variant): New function moved
5680 (build_type_attribute_qual_variant): Likewise.
5681 (cmp_attrib_identifiers): Likewise.
5682 (simple_cst_list_equal): Likewise.
5683 (omp_declare_simd_clauses_equal): Likewise.
5684 (attribute_value_equal): Likewise.
5685 (comp_type_attributes): Likewise.
5686 (build_type_attribute_variant): Likewise.
5687 (lookup_ident_attribute): Likewise.
5688 (remove_attribute): Likewise.
5689 (merge_attributes): Likewise.
5690 (merge_type_attributes): Likewise.
5691 (merge_decl_attributes): Likewise.
5692 (merge_dllimport_decl_attributes): Likewise.
5693 (handle_dll_attribute): Likewise.
5694 (attribute_list_equal): Likewise.
5695 (attribute_list_contained): Likewise.
5696 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5697 (lookup_attribute_by_prefix): Likewise.
5698 * bb-reorder.c: Include header files.
5699 * builtins.c: Likewise.
5700 * calls.c: Likewise.
5701 * cfgexpand.c: Likewise.
5702 * cgraph.c: Likewise.
5703 * cgraphunit.c: Likewise.
5704 * convert.c: Likewise.
5705 * dwarf2out.c: Likewise.
5706 * final.c: Likewise.
5707 * fold-const.c: Likewise.
5708 * function.c: Likewise.
5709 * gimple-expr.c: Likewise.
5710 * gimple-fold.c: Likewise.
5711 * gimple-pretty-print.c: Likewise.
5712 * gimple.c: Likewise.
5713 * gimplify.c: Likewise.
5714 * hsa-common.c: Likewise.
5715 * hsa-gen.c: Likewise.
5716 * internal-fn.c: Likewise.
5717 * ipa-chkp.c: Likewise.
5718 * ipa-cp.c: Likewise.
5719 * ipa-devirt.c: Likewise.
5720 * ipa-fnsummary.c: Likewise.
5721 * ipa-inline.c: Likewise.
5722 * ipa-visibility.c: Likewise.
5724 * lto-cgraph.c: Likewise.
5725 * omp-expand.c: Likewise.
5726 * omp-general.c: Likewise.
5727 * omp-low.c: Likewise.
5728 * omp-offload.c: Likewise.
5729 * omp-simd-clone.c: Likewise.
5730 * opts-global.c: Likewise.
5731 * passes.c: Likewise.
5732 * predict.c: Likewise.
5733 * sancov.c: Likewise.
5734 * sanopt.c: Likewise.
5735 * symtab.c: Likewise.
5736 * toplev.c: Likewise.
5737 * trans-mem.c: Likewise.
5738 * tree-chkp.c: Likewise.
5739 * tree-eh.c: Likewise.
5740 * tree-into-ssa.c: Likewise.
5741 * tree-object-size.c: Likewise.
5742 * tree-parloops.c: Likewise.
5743 * tree-profile.c: Likewise.
5744 * tree-ssa-ccp.c: Likewise.
5745 * tree-ssa-live.c: Likewise.
5746 * tree-ssa-loop.c: Likewise.
5747 * tree-ssa-sccvn.c: Likewise.
5748 * tree-ssa-structalias.c: Likewise.
5749 * tree-ssa.c: Likewise.
5750 * tree-streamer-in.c: Likewise.
5751 * tree-vectorizer.c: Likewise.
5752 * tree-vrp.c: Likewise.
5754 * ubsan.c: Likewise.
5755 * varasm.c: Likewise.
5756 * varpool.c: Likewise.
5757 * tree.c: Remove functions moved to attribs.[ch].
5759 * config/aarch64/aarch64.c: Add attrs.h header file.
5760 * config/alpha/alpha.c: Likewise.
5761 * config/arc/arc.c: Likewise.
5762 * config/arm/arm.c: Likewise.
5763 * config/avr/avr.c: Likewise.
5764 * config/bfin/bfin.c: Likewise.
5765 * config/c6x/c6x.c: Likewise.
5766 * config/cr16/cr16.c: Likewise.
5767 * config/cris/cris.c: Likewise.
5768 * config/darwin.c: Likewise.
5769 * config/epiphany/epiphany.c: Likewise.
5770 * config/fr30/fr30.c: Likewise.
5771 * config/frv/frv.c: Likewise.
5772 * config/ft32/ft32.c: Likewise.
5773 * config/h8300/h8300.c: Likewise.
5774 * config/i386/winnt.c: Likewise.
5775 * config/ia64/ia64.c: Likewise.
5776 * config/iq2000/iq2000.c: Likewise.
5777 * config/lm32/lm32.c: Likewise.
5778 * config/m32c/m32c.c: Likewise.
5779 * config/m32r/m32r.c: Likewise.
5780 * config/m68k/m68k.c: Likewise.
5781 * config/mcore/mcore.c: Likewise.
5782 * config/microblaze/microblaze.c: Likewise.
5783 * config/mips/mips.c: Likewise.
5784 * config/mmix/mmix.c: Likewise.
5785 * config/mn10300/mn10300.c: Likewise.
5786 * config/moxie/moxie.c: Likewise.
5787 * config/msp430/msp430.c: Likewise.
5788 * config/nds32/nds32-isr.c: Likewise.
5789 * config/nds32/nds32.c: Likewise.
5790 * config/nios2/nios2.c: Likewise.
5791 * config/nvptx/nvptx.c: Likewise.
5792 * config/pa/pa.c: Likewise.
5793 * config/pdp11/pdp11.c: Likewise.
5794 * config/powerpcspe/powerpcspe.c: Likewise.
5795 * config/riscv/riscv.c: Likewise.
5796 * config/rl78/rl78.c: Likewise.
5797 * config/rx/rx.c: Likewise.
5798 * config/s390/s390.c: Likewise.
5799 * config/sh/sh.c: Likewise.
5800 * config/sol2.c: Likewise.
5801 * config/sparc/sparc.c: Likewise.
5802 * config/spu/spu.c: Likewise.
5803 * config/stormy16/stormy16.c: Likewise.
5804 * config/tilegx/tilegx.c: Likewise.
5805 * config/tilepro/tilepro.c: Likewise.
5806 * config/v850/v850.c: Likewise.
5807 * config/vax/vax.c: Likewise.
5808 * config/visium/visium.c: Likewise.
5809 * config/xtensa/xtensa.c: Likewise.
5811 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5814 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5815 constraints since the -mupper-regs-* switches have been
5817 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5818 into a vector from a double word element that was extracted from
5819 another vector, and eliminate extra XXPERMDI instructions.
5820 (vsx_concat_<mode>_2): Likewise.
5821 (vsx_concat_<mode>_3): Likewise.
5822 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5823 concat to allow optimizing inserts from previous extracts.
5825 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5827 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5828 memory reference to a SSP offset in TLS address space.
5829 (ix86_print_operand) <case '@'>: Remove.
5830 (ix86_print_operand_punct_valid_p): Remove '@' code.
5831 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5833 (stack_tls_protect_set_<mode>): Remove.
5834 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5835 (stack_tls_protect_test_<mode>): Remove.
5836 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5838 2017-08-07 Olivier Hainque <hainque@adacore.com>
5841 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5843 2017-08-07 Douglas Rupp <rupp@adacore.com>
5845 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5846 variable was referenced as multidir in command.
5848 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5851 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5854 2017-08-07 Martin Liska <mliska@suse.cz>
5856 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5857 * config/rl78/rl78.c: Add include of attribs.h.
5858 * config/sh/sh.c: Likewise.
5859 * config/v850/v850.c: Likewise.
5861 2017-08-07 Tom de Vries <tom@codesourcery.com>
5864 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5866 2017-08-07 Martin Liska <mliska@suse.cz>
5868 * config/mips/mips.c: Include attribs.h.
5870 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5873 * doc/invoke.texi: Document change in behvaior for -Ofast for
5876 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5878 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5880 (aarch64_pop_regs): Likewise.
5881 (aarch64_gen_load_pair): Likewise.
5882 (aarch64_save_callee_saves): Likewise.
5883 (aarch64_restore_callee_saves): Likewise.
5885 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5887 * config/i386/i386.c: Revert the last change.
5889 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5892 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5894 (ix86_finalize_stack_frame_flags): This. Also clear
5895 frame_pointer_needed if -fno-omit-frame-pointer is used without
5897 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5898 with ix86_finalize_stack_frame_flags.
5899 (ix86_expand_epilogue): Likewise.
5900 (ix86_expand_split_stack_prologue): Likewise.
5902 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5905 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5906 to P_AES for Westmere.
5908 2017-08-07 Jonathan Yong <10walls@gmail.com>
5910 * config/i386/mingw.opt (fset-stack-executable): Removed.
5911 * config/i386/cygming.opt (fset-stack-executable): Moved
5913 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5915 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5917 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5919 2017-08-07 Marek Polacek <polacek@redhat.com>
5922 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5924 2017-08-07 Martin Liska <mliska@suse.cz>
5926 * attribs.h (canonicalize_attr_name): New function.
5927 (cmp_attribs): Move from c-format.c and adjusted.
5928 (is_attribute_p): Moved from tree.h.
5929 * tree-inline.c: Add new includes.
5930 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5931 (private_is_attribute_p): Remove.
5932 (private_lookup_attribute): Likewise.
5933 (private_lookup_attribute_by_prefix): Simplify.
5934 (remove_attribute): Use is_attribute_p.
5935 * tree.h: Remove removed declarations.
5937 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5940 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5941 instead of computing it in the function. Formatting fix.
5942 (expand_case): Don't rely on default_edge being the first edge,
5943 clear it if removing it, pass default_edge to
5944 emit_case_dispatch_table.
5945 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5948 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5950 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5951 insn in the function, emit NOP after the insn.
5953 2017-08-06 Tom de Vries <tom@codesourcery.com>
5955 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5958 2017-08-06 Tom de Vries <tom@codesourcery.com>
5960 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5963 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5965 PR tree-optimization/57371
5966 * match.pd: New pattern.
5968 2017-08-04 Marek Polacek <polacek@redhat.com>
5971 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5972 perform the computation in offset_int.
5974 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5976 PR tree-optimization/81136
5977 * tree-vectorizer.h: Include tree-hash-traits.h.
5978 (vec_base_alignments): New typedef.
5979 (vec_info): Add a base_alignments field.
5980 (vect_record_base_alignments): Declare.
5981 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5983 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5984 (create_data_ref): Add an is_conditional_in_stmt argument.
5985 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5986 the is_conditional_in_stmt field.
5987 (data_ref_loc): Add an is_conditional_in_stmt field.
5988 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5989 (find_data_references_in_stmt): Update call to create_data_ref.
5990 (graphite_find_data_references_in_stmt): Likewise.
5991 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5992 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5993 (vect_record_base_alignment): New function.
5994 (vect_record_base_alignments): Likewise.
5995 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5996 for nested statements even if we fail to compute a misalignment.
5997 Use pooled base alignments for unconditional references.
5998 (vect_find_same_alignment_drs): Compare base addresses instead
6000 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6001 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6003 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6005 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6006 Add an explicit name for the enum. Use auto_vec for slp_instances
6008 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6010 (_bb_vec_info): Add a constructor and destructor.
6011 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6012 (destroy_loop_vec_info): Delete.
6013 (vect_destroy_datarefs): Likewise.
6014 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6015 (vec_info::vec_info): New function.
6016 (vec_info::~vec_info): Likewise.
6017 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6018 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6019 destroy_loop_vec_info.
6020 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6021 (_loop_vec_info::_loop_vec_info): ...this.
6022 (destroy_loop_vec_info): Replace with...
6023 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6024 the stmt_vec_infos. Leave handling of vec_info information to its
6025 destructor. Remove explicit vector releases.
6026 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6027 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6028 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6029 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6030 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6031 (destroy_bb_vec_info): Replace with...
6032 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6033 information to its destructor.
6034 (vect_slp_analyze_bb_1): Use new and delete instead of
6035 new_bb_vec_info and destroy_bb_vec_info.
6036 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6039 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6041 * tree-data-ref.h (subscript): Add access_fn field.
6042 (data_dependence_relation): Add could_be_independent_p.
6043 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6044 (same_access_functions): Move to tree-data-ref.c.
6045 * tree-data-ref.c (ref_contains_union_access_p): New function.
6046 (access_fn_component_p): Likewise.
6047 (access_fn_components_comparable_p): Likewise.
6048 (dr_analyze_indices): Add a reference to access_fn_component_p.
6049 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6051 (constant_access_functions): Likewise.
6052 (add_other_self_distances): Likewise.
6053 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6054 (initialize_data_dependence_relation): Use XCNEW and remove
6055 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6056 of access functions that have the same type. Allow the
6057 subsequence to end with different bases in some circumstances.
6058 Record the chosen access functions in SUB_ACCESS_FN.
6059 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6060 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6061 (subscript_dependence_tester_1): Likewise dra and drb.
6062 (build_classic_dist_vector): Update calls accordingly.
6063 (subscript_dependence_tester): Likewise.
6064 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6065 DDR_COULD_BE_INDEPENDENT_P.
6066 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6067 comp_alias_ddrs instead of may_alias_ddrs.
6068 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6070 (vect_analyze_data_ref_dependence): Use it if
6071 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6072 distance vectors if that fails.
6073 (dependence_distance_ge_vf): New function.
6074 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6075 LOOP_VINFO_MAY_ALIAS_DDRS.
6077 2017-08-04 Richard Biener <rguenther@suse.de>
6080 * fold-const.c (fold_binary_loc): Properly restrict
6081 minus_var0 && minus_var1 case when associating undefined overflow
6084 2017-08-04 Tom de Vries <tom@codesourcery.com>
6086 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6088 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6090 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6091 Don't start diagnostic messages with a capital letter.
6092 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6094 (rs6000_invalid_builtin): Likewise.
6095 (rs6000_trampoline_init): Likewise.
6097 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6100 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6101 after setting changeable df flags.
6103 2017-08-03 Richard Biener <rguenther@suse.de>
6105 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6106 up if the use is in USE - X.
6108 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6110 * toplev.c (dumpfile.h): New include.
6111 (internal_error_reentered): New static function. Use it...
6112 (internal_error_function): ...here to handle reentered internal_error.
6114 2017-08-03 Richard Biener <rguenther@suse.de>
6117 * fold-const.c (split_tree): Add minus_var and minus_con
6118 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6119 here but always use minus_*.
6120 (associate_trees): Assert we never associate with MINUS_EXPR
6121 and NULL first operand. Do not recurse for PLUS_EXPR operands
6122 when associating as MINUS_EXPR either.
6123 (fold_binary_loc): Track minus_var and minus_con.
6125 2017-08-03 Tom de Vries <tom@codesourcery.com>
6128 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6129 ACCEL_COMPILER, apply finish_options on
6130 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6132 2017-08-03 Tom de Vries <tom@codesourcery.com>
6135 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6136 function_entry_patch_area_size > 0.
6138 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6141 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6142 instead of 10??LU, perform unit multiplication in wide_int,
6143 don't change alloc_object_size_limit if the limit is larger
6146 PR tree-optimization/81655
6147 PR tree-optimization/81588
6148 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6149 the case when ranges[i].low and high are 1 for unsigned type with
6153 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6154 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6156 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6158 * tree-vrp.h: Add include guard.
6160 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6163 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6164 (ud2): New insn pattern.
6165 * config/i386/i386.c (ix86_expand_epilogue):
6166 For naked functions, generate ud2 instead of trap insn.
6168 2017-08-02 Marek Polacek <polacek@redhat.com>
6171 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6173 2017-08-02 Tom de Vries <tom@codesourcery.com>
6174 Cesar Philippidis <cesar@codesourcery.com>
6176 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6177 Add missing edge probabilities.
6179 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6181 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6184 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6187 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6188 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6189 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6190 prologue_seq sequences - if any.
6192 2017-08-02 Richard Biener <rguenther@suse.de>
6194 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6195 via vectors if supported, integer extracts via punning if supported
6196 or otherwise vector extracts.
6198 2017-08-02 Richard Biener <rguenther@suse.de>
6200 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6202 (bitmap_insert_into_set): ... this.
6204 2017-08-02 Richard Biener <rguenther@suse.de>
6206 PR tree-optimization/81633
6208 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6210 PR tree-optimization/71752
6211 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6213 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6215 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6216 (machine_function::call_ms2sysv_pad_out): Remove field.
6217 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6218 (ix86_compute_frame_layout): Likewise.
6220 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6223 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6224 attribute with interrupt attribute.
6226 2017-08-01 Andrew Pinski <apinski@cavium.com>
6228 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6229 BIT_INSERT_EXPR's operand 1
6230 to see if the types precision matches.
6232 2017-08-01 Martin Liska <mliska@suse.cz>
6235 * builtins.c (expand_builtin_memcpy_args): Remove.
6236 (expand_builtin_memcpy): Call newly added function
6237 expand_builtin_memory_copy_args.
6238 (expand_builtin_memcpy_with_bounds): Likewise.
6239 (expand_builtin_mempcpy): Remove last argument.
6240 (expand_builtin_mempcpy_with_bounds): Likewise.
6241 (expand_builtin_memory_copy_args): New function created from
6242 expand_builtin_mempcpy_args with small modifications.
6243 (expand_builtin_mempcpy_args): Remove.
6244 (expand_builtin_stpcpy): Remove unused argument.
6245 (expand_builtin): Likewise.
6246 (expand_builtin_with_bounds): Likewise.
6248 2017-08-01 Martin Liska <mliska@suse.cz>
6251 Make mempcpy more optimal (PR middle-end/70140).
6253 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6256 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6257 __builtin_vec_cmpne verify both arguments are compatible vectors
6258 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6259 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6260 move computation of aligned to after checking the argument types.
6264 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6265 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6268 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6269 Doug Rupp <rupp@adacore.com>
6270 Olivier Hainque <hainque@adacore.com>
6272 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6273 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6275 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6276 for TARGET_OS_CPP_BUILTIN.
6277 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6278 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6280 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6281 passing required abi options to the assembler for EABI configurations.
6282 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6283 of .text.hot and .text.unlikely sections for kernel modules when
6284 using ARM style exceptions.
6285 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6286 options. Add EXTRA_CC1_SPEC.
6287 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6289 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6291 (ARM_TARGET2_DWARF_FORMAT): Define.
6292 * config/arm/t-vxworks: Adjust multilib control to removal of the
6293 Diab command line options.
6295 2017-08-01 Martin Liska <mliska@suse.cz>
6297 PR gcov-profile/81561
6298 * gcov.c (unblock): Make unblocking safe as we need to preserve
6299 index correspondence of blocks and block_lists.
6301 2017-08-01 Richard Biener <rguenther@suse.de>
6303 PR tree-optimization/81181
6304 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6305 (compute_antic): ... end of iteration here.
6307 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6309 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6310 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6311 (ftree-slp-vectorize): Likewise.
6312 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6313 can no longer be set independent of flag_tree_loop_vectorize.
6314 * omp-general.c (emp_max_vf): Likewise.
6315 * opts.c (enable_fdo_optimizations): Remove references to
6316 flag_tree_vectorize, these are now implicit.
6317 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6318 and leave it for the options machinery.
6320 2017-08-01 Martin Liska <mliska@suse.cz>
6323 * builtins.c (expand_builtin_memcpy_args): Remove.
6324 (expand_builtin_memcpy): Call newly added function
6325 expand_builtin_memory_copy_args.
6326 (expand_builtin_memcpy_with_bounds): Likewise.
6327 (expand_builtin_mempcpy): Remove last argument.
6328 (expand_builtin_mempcpy_with_bounds): Likewise.
6329 (expand_builtin_memory_copy_args): New function created from
6330 expand_builtin_mempcpy_args with small modifications.
6331 (expand_builtin_mempcpy_args): Remove.
6332 (expand_builtin_stpcpy): Remove unused argument.
6333 (expand_builtin): Likewise.
6334 (expand_builtin_with_bounds): Likewise.
6336 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6339 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6340 print "ds:" only for immediates in generic address space.
6342 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6345 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6346 (ix86_function_ok_for_sibcall): Return false for naked functions.
6348 2017-08-01 Richard Biener <rguenther@suse.de>
6350 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6351 (compute_antic): Seed worklist with exit block predecessors.
6352 * cfganal.c (dfs_find_deadend): For a cycle return the source
6353 of the edge closing it.
6355 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6357 * config/aarch64/aarch64.c
6358 (aarch64_can_const_movi_rtx_p): Move 0 check.
6360 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6362 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6363 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6365 * match.pd: Ditto in address comparison pattern.
6367 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6369 PR tree-optimization/81627
6370 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6371 closed ssa form for store-store chain.
6373 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6375 PR tree-optimization/81620
6376 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6377 for store-store chain.
6379 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6381 PR tree-optimization/81588
6382 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6383 ranges[i].in_p, invert comparison code ccode. For >/>=,
6384 swap rhs1 and rhs2 and comparison code unconditionally,
6385 for </<= don't do that. Don't swap rhs1/rhs2 again if
6386 ranges[i].in_p, instead invert comparison code ccode if
6387 opcode or oe->rank is BIT_IOR_EXPR.
6390 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6391 a direct optab to conversion optab.
6392 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6393 with GET_MODE_INNER as last argument instead of optab_handler.
6394 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6395 vector extraction if possible and optab is available.
6396 * expr.c (store_constructor): Use convert_optab_handler instead
6397 of optab_handler. Use vector initialization from smaller
6398 vectors if possible and optab is available.
6399 * tree-vect-stmts.c (vectorizable_load): Likewise.
6400 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6402 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6403 of vec_init from half-sized vectors with the same element mode.
6404 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6405 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6406 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6407 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6408 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6409 after mode in gen_vec_extract* calls.
6410 (vec_extract<mode>): Renamed to ...
6411 (vec_extract<mode><ssescalarmodelower>): ... this.
6412 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6413 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6414 element mode after mode in gen_vec_init* calls.
6415 (VEC_INIT_HALF_MODE): New mode iterator.
6416 (vec_init<mode>): Renamed to ...
6417 (vec_init<mode><ssescalarmodelower>): ... this.
6418 (vec_init<mode><ssehalfvecmodelower>): New expander.
6419 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6420 (vec_extractv2sfsf): ... this.
6421 (vec_initv2sf): Renamed to ...
6422 (vec_initv2sfsf): ... this.
6423 (vec_extractv2si): Renamed to ...
6424 (vec_extractv2sisi): ... this.
6425 (vec_initv2si): Renamed to ...
6426 (vec_initv2sisi): ... this.
6427 (vec_extractv4hi): Renamed to ...
6428 (vec_extractv4hihi): ... this.
6429 (vec_initv4hi): Renamed to ...
6430 (vec_initv4hihi): ... this.
6431 (vec_extractv8qi): Renamed to ...
6432 (vec_extractv8qiqi): ... this.
6433 (vec_initv8qi): Renamed to ...
6434 (vec_initv8qiqi): ... this.
6435 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6436 (vec_init<mode>): Renamed to ...
6437 (vec_init<mode><VEC_base_l>): ... this.
6438 (vec_extract<mode>): Renamed to ...
6439 (vec_extract<mode><VEC_base_l>): ... this.
6440 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6441 (vec_initv2sfsf): ... this.
6442 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6443 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6444 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6445 element mode after mode in gen_vec_init* calls.
6446 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6447 (vec_init<mode><Vel>): ... this.
6448 (vec_extract<mode>): Renamed to ...
6449 (vec_extract<mode><Vel>): ... this.
6450 * config/aarch64/iterators.md (Vel): New mode attribute.
6451 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6452 Add element mode after mode in gen_vec_extract* calls.
6453 * config/s390/vector.md (non_vec_l): New mode attribute.
6454 (vec_extract<mode>): Renamed to ...
6455 (vec_extract<mode><non_vec_l>): ... this.
6456 (vec_init<mode>): Renamed to ...
6457 (vec_init<mode><non_vec_l>): ... this.
6458 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6459 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6461 * config/arm/iterators.md (V_elem_l): New mode attribute.
6462 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6463 (vec_extract<mode><V_elem_l>): ... this.
6464 (vec_extractv2di): Renamed to ...
6465 (vec_extractv2didi): ... this.
6466 (vec_init<mode>): Renamed to ...
6467 (vec_init<mode><V_elem_l>): ... this.
6468 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6469 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6470 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6471 Add element mode after gen_vec_extract* calls.
6472 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6473 (vec_init<mode><unitmode>): ... this.
6474 (vec_extract<mode>): Renamed to ...
6475 (vec_extract<mode><unitmode>): ... this.
6476 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6477 (vec_init<mode><unitmode>): ... this.
6478 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6479 (vec_initv2sfsf): ... this.
6480 (vec_extractv2sf): Renamed to ...
6481 (vec_extractv2sfsf): ... this.
6482 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6483 Add element mode after gen_vec_extract* calls.
6484 * config/mips/mips.md (unitmode): New mode iterator.
6485 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6486 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6487 * config/spu/spu.md (inner_l): New mode attribute.
6488 (vec_init<mode>): Renamed to ...
6489 (vec_init<mode><inner_l>): ... this.
6490 (vec_extract<mode>): Renamed to ...
6491 (vec_extract<mode><inner_l>): ... this.
6492 * config/sparc/sparc.md (veltmode): New mode iterator.
6493 (vec_init<VMALL:mode>): Renamed to ...
6494 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6495 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6496 (vec_initv2sisi): ... this.
6497 (vec_initv2sf): Renamed to ...
6498 (vec_initv2sfsf): ... this.
6499 (vec_extractv2sf): Renamed to ...
6500 (vec_extractv2sfsf): ... this.
6501 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6502 (vec_init<mode>): Renamed to ...
6503 (vec_init<mode><VEC_base_l>): ... this.
6504 (vec_extract<mode>): Renamed to ...
6505 (vec_extract<mode><VEC_base_l>): ... this.
6506 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6507 (vec_initv2sfsf): ... this.
6508 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6509 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6510 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6511 gen_vec_init* calls.
6513 2017-08-01 Richard Biener <rguenther@suse.de>
6515 PR tree-optimization/81297
6516 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6517 TREE_OVERFLOW from INTEGER_CSTs.
6519 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6521 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6523 2017-07-31 Carl Love <cel@us.ibm.com>
6525 * config/rs6000/rs6000-c: Add support for built-in functions
6526 vector signed char vec_xl_be (signed long long, signed char *);
6527 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6528 vector signed int vec_xl_be (signed long long, signed int *);
6529 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6530 vector signed long long vec_xl_be (signed long long, signed long long *);
6531 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6532 vector signed short vec_xl_be (signed long long, signed short *);
6533 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6534 vector double vec_xl_be (signed long long, double *);
6535 vector float vec_xl_be (signed long long, float *);
6536 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6537 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6538 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6540 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6541 (altivec_expand_builtin): Add switch statement to call
6542 altivec_expand_xl_be for each builtin.
6543 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6544 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6545 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6546 __builtin_vsx_le_be_v16qi.
6547 * doc/extend.texi: Update the built-in documentation file for the
6548 new built-in functions.
6550 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6553 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6555 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6557 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6559 * config.gcc: Add z14.
6560 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6561 CPU model numbers for z13s and z14.
6562 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6564 * config/s390/s390-opts.h (enum processor_type): Rename
6565 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6566 * config/s390/s390.c (processor_table): Add field for CPU name to
6567 be passed to Binutils.
6568 (s390_asm_output_machine_for_arch): Use the new field in
6569 processor_table for Binutils.
6570 (s390_expand_builtin): Replace arch12 with z14.
6571 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6572 (s390_get_sched_attrmask): Likewise.
6573 (s390_get_unit_mask): Likewise.
6574 * config/s390/s390.opt: Add z14 to processor_type enum.
6576 2017-07-31 Martin Jambor <mjambor@suse.cz>
6579 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6580 regardless of optimization level.
6582 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6583 Martin Liska <mliska@suse.cz>
6585 * predict.def: Remove old comment and adjust probability.
6586 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6589 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6592 * config/i386/i386.c (ix86_function_naked): New function.
6593 (ix86_can_use_return_insn_p): Return false for naked functions.
6594 (ix86_expand_prologue): Skip prologue for naked functions.
6595 (ix86_expand_epilogue): Skip epilogue for naked functions
6596 and emit trap instruction.
6597 (ix86_warn_func_return): New function.
6598 (ix86_attribute_table): Add "naked" attribute specification.
6599 (TARGET_WARN_FUNC_RETURN): Define.
6600 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6602 2017-07-31 Martin Liska <mliska@suse.cz>
6604 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6605 (dump_gimple_bb_header): Always dump BB info.
6606 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6608 2017-07-31 Martin Liska <mliska@suse.cz>
6611 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6612 also with current_function_decl non-null equality.
6614 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6617 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6618 change type to the element type, instead add eltype variable and
6619 use it where we are interested in the element type.
6621 PR tree-optimization/81603
6622 * ipa-polymorphic-call.c
6623 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6624 offset arithmetic in offset_int, bail out if the resulting bit offset
6625 doesn't fit into shwi.
6627 2017-07-31 Martin Liska <mliska@suse.cz>
6629 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6630 (gimplify_save_expr): Fix comment.
6632 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6635 * config/i386/i386.c (ix86_function_arg): Update arguments for
6637 (ix86_compute_frame_layout): Set the initial stack offset to
6638 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6639 INCOMING_FRAME_SP_OFFSET.
6640 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6641 stack before exception handler returns.
6642 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6643 the 'ERROR_CODE' for exception handler.
6645 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6647 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6648 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6649 (ASM_OUTPUT_REG_POP): Ditto.
6650 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6651 instead of asm_fprintf to output pure string.
6653 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6655 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6656 to imported_module_or_decl hook.
6657 (debug_nothing_tree_tree_tree_bool): Remove.
6658 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6659 * debug.c (do_nothing_debug_hooks): Use
6660 debug_nothing_tree_tree_tree_bool_bool instead of
6661 debug_nothing_tree_tree_tree_bool.
6662 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6663 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6664 * sdbout.c (sdb_debug_hooks): Likewise.
6665 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6666 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6668 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6669 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6670 attribute, don't add anything.
6672 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6674 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6675 (fold_build2_stat_loc): Likewise.
6676 (fold_build3_stat_loc): Likewise.
6677 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6678 (fold_build1_loc): Remove macro.
6679 (fold_build2_loc): Likewise.
6680 (fold_build3_loc): Likewise.
6682 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6684 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6685 (gimple_build_debug_bind_source_stat): Likewise.
6686 * gimple.h (gimple_build_debug_bind): Remove macro.
6687 (gimple_build_debug_bind_source): Likewise.
6689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6691 * bitmap.c (bitmap_alloc): Adjust.
6692 (bitmap_gc_alloc): Likewise.
6693 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6697 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6698 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6699 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6700 (bitmap_gc_alloc_stat): Likewise.
6701 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6705 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6706 * rtl.h (shallow_copy_rtx): Remove macro.
6708 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6710 * emit-rtl.c (gen_raw_REG): Adjust.
6711 * gengenrtl.c (gendef): Likewise.
6712 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6713 * rtl.h (rtx_alloc): Remove macro.
6715 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6717 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6718 (build_tree_list_stat): Likewise.
6719 * tree.h (build_tree_list): Remove macro.
6720 (build_tree_list_vec): Likewise.
6722 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6724 * tree.c (make_vector_stat): Remove _stat from name.
6725 (build_vector_stat): Likewise.
6726 * tree.h (make_vector_stat): Remove macro.
6727 (build_vector_stat): Likewise.
6729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6731 * tree.h (build_var_debug_value): Remove prototype.
6733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6735 * tree.c (tree_cons_stat): Remove _stat from name.
6736 * tree.h (tree_cons): Remove macro.
6738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6740 * tree.c (build_vl_exp_stat): Remove _stat from name.
6741 * tree.h (build_vl_exp): Remove macro.
6743 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6745 * tree.c (build_decl_stat): Remove _stat from name.
6746 * tree.h (build_decl): Remove macro.
6748 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6750 * gimple.c (gimple_build_with_ops_stat): Adjust.
6751 (gimple_alloc_stat): Remove _stat from name.
6752 * gimple.h (gimple_alloc): Remove macro.
6754 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6756 * tree.c (make_tree_vec_stat): Remove _stat from name.
6757 (grow_tree_vec_stat): Likewise.
6758 * tree.h (make_tree_vec_stat): Adjust prototype.
6759 (grow_tree_vec_stat): Likewise.
6760 (make_tree_vec): Remove macro.
6761 (grow_tree_vec): Likewise.
6763 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6765 * fold-const.c (fold_build1_stat_loc): Adjust.
6766 (fold_build2_stat_loc): Likewise.
6767 (fold_build3_stat_loc): Likewise.
6768 * tree.c (build0_stat): Remove _stat from name.
6769 (build1_stat): Likewise.
6770 (build2_stat): Likewise.
6771 (build3_stat): Likewise.
6772 (build4_stat): Likewise.
6773 (build5_stat): Likewise.
6774 * tree.h (build1_loc): Remove macro, and rename _stat function
6776 (build2_loc): Likewise.
6777 (build3_loc): Likewise.
6778 (build4_loc): Likewise.
6779 (build5_loc): Likewise.
6781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6783 * tree.c (make_int_cst_stat): Remove _stat from name.
6784 * tree.h (make_int_cst_stat): Adjust prototype.
6785 (make_int_cst): Remove macro.
6787 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6789 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6790 * tree.h (make_tree_binfo_stat): Adjust prototype.
6791 (make_tree_binfo): Remove.
6793 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6795 * tree.c (copy_node_stat): Rename to copy_node.
6796 (build_distinct_type_copy): Adjust.
6797 * tree.h (copy_node_stat): Adjust prototype.
6798 (copy_node): Remove macro.
6800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6802 * tree.c (make_node_stat): rename to make_node.
6803 (build_tree_list_stat): Adjust.
6804 (build0_stat): Likewise.
6805 (build2_stat): Likewise.
6806 (build3_stat): Likewise.
6807 (build4_stat): Likewise.
6808 (build5_stat): Likewise.
6809 (build_decl_stat): Likewise.
6810 * tree.h (make_node_stat): Adjust prototype.
6811 (make_node): remove macro.
6813 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6815 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6816 (PPC_FEATURE2_SCV): Likewise.
6817 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6819 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6821 * config/aarch64/aarch64.c
6822 (aarch64_internal_mov_immediate): Add new special pattern.
6823 * config/aarch64/aarch64.md (*movdi_aarch64):
6824 Add reg/32bit const mov case.
6826 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6827 Richard Sandiford <richard.sandiford@linaro.org>
6829 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6830 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6831 Add integer and movi cases.
6832 (movi-split-hf-df-sf split, fp16): New.
6833 (enabled): Added TARGET_FP_F16INST.
6834 * config/aarch64/iterators.md (GPF_HF): New.
6835 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6837 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6839 * config/aarch64/aarch64.c
6840 (aarch64_simd_container_mode): Add prototype.
6841 (aarch64_expand_mov_immediate): Add HI support.
6842 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6843 (aarch64_can_const_movi_rtx_p): New.
6844 (aarch64_preferred_reload_class):
6845 Remove restrictions of using FP registers for certain SIMD operations.
6846 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6847 (aarch64_valid_floating_const): Add integer move validation.
6848 (aarch64_simd_imm_scalar_p): Remove.
6849 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6850 (aarch64_legitimate_constant_p): Expand list of supported cases.
6851 * config/aarch64/aarch64-protos.h
6852 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6853 (aarch64_reinterpret_float_as_int): New.
6854 (aarch64_simd_imm_scalar_p): Remove.
6855 * config/aarch64/constraints.md (Uvi): New.
6856 (Dd): Split into Ds and new Dd.
6857 * config/aarch64/aarch64.md (*movsi_aarch64):
6859 (*movdi_aarch64): Add SIMD mov case.
6861 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6863 * tree-predcom.c: (struct chain): Handle store-store chain in which
6864 stores for elimination only store loop invariant values.
6865 (execute_pred_commoning_chain): Ditto.
6866 (prepare_initializers_chain_store_elim): Ditto.
6867 (prepare_finalizers): Ditto.
6868 (is_inv_store_elimination_chain): New function.
6869 (initialize_root_vars_store_elim_1): New function.
6871 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6873 * tree-predcom.c: Revise general description of the pass.
6874 (enum chain_type): New enum type for store elimination.
6875 (struct chain): New field supporting store elimination.
6876 (struct component): Ditto.
6877 (dump_chain): Dump store-stores chain.
6878 (release_chain): Release resources.
6879 (split_data_refs_to_components): Compute and create component
6880 contains only stores for elimination.
6881 (get_chain_last_ref_at): New function.
6882 (make_invariant_chain): Initialization.
6883 (make_rooted_chain): Specify chain type in parameter and record it.
6884 (add_looparound_copies): Skip for store-stores chain.
6885 (determine_roots_comp): Compute type of chain and pass it to
6887 (initialize_root_vars_store_elim_2): New function.
6888 (finalize_eliminated_stores): New function.
6889 (remove_stmt): Handle store for elimination.
6890 (execute_pred_commoning_chain): Execute predictive commoning on
6892 (determine_unroll_factor): Skip unroll for store-stores chain.
6893 (prepare_initializers_chain_store_elim): New function.
6894 (prepare_initializers_chain): Hanlde store-store chain.
6895 (prepare_finalizers_chain, prepare_finalizers): New function.
6896 (tree_predictive_commoning_loop): Return integer value indicating
6897 if loop is unrolled or lcssa form is corrupted.
6898 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6900 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6902 * tree-predcom.c (initialize_root): Delete.
6903 (execute_pred_commoning_chain): Initialize root vars and replace
6904 reference of non-combined chain directly, rather than call above
6907 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6909 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6910 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6912 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6914 * tree-predcom.c (struct chain): New field init_seq.
6915 (release_chain): Release init_seq.
6916 (prepare_initializers_chain): Record intialization stmts in above
6918 (insert_init_seqs): New function.
6919 (tree_predictive_commoning_loop): Call insert_init_seqs.
6921 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6923 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6925 2017-07-28 Richard Biener <rguenther@suse.de>
6927 * match.pd: Remove superfluous :c.
6928 * genmatch.c (simplify::id): Add member.
6929 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6931 (current_id): New global.
6932 (dt_node::parent): Move from ...
6933 (dt_operand::parent): ... here. Add for_id member.
6934 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6935 (decision_tree::find_node): Relax order requirement when
6936 merging DT_TRUE nodes to ones inbetween the current simplify
6937 and the one we try to merge with. Add diagnostic whenever
6938 we need to enforce pattern order by not merging.
6939 (decision_tree::insert): Set current_id.
6940 (decision_tree::print_node): Dump parent node and for_id.
6941 (parser::last_id): Add member.
6942 (parser::push_simplify): Assign unique id.
6943 (parser::parser): Initialize last_id.
6945 2017-07-28 Martin Liska <mliska@suse.cz>
6948 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6949 gimple_build_debug_bind.
6951 2017-07-28 Richard Biener <rguenther@suse.de>
6953 PR tree-optimization/81502
6954 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6956 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6958 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6959 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6960 for BIT_FIELD_REF args.
6961 * fold-const.c (make_bit_field_ref): Likewise.
6962 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6964 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6967 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6968 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6969 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6970 Or it into SANITIZER_UNDEFINED.
6971 * ubsan.c: Include gimple-fold.h and varasm.h.
6972 (ubsan_expand_ptr_ifn): New function.
6973 (instrument_pointer_overflow): New function.
6974 (maybe_instrument_pointer_overflow): New function.
6975 (instrument_object_size): Formatting fix.
6976 (pass_ubsan::execute): Call instrument_pointer_overflow
6977 and maybe_instrument_pointer_overflow.
6978 * internal-fn.c (expand_UBSAN_PTR): New function.
6979 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6980 * sanitizer.def (__ubsan_handle_pointer_overflow,
6981 __ubsan_handle_pointer_overflow_abort): New builtins.
6982 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6983 * internal-fn.def (UBSAN_PTR): New internal function.
6984 * opts.c (sanitizer_opts): Add pointer-overflow.
6985 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6986 * fold-const.c (build_range_check): Compute pointer range check in
6987 integral type if pointer arithmetics would be needed. Formatting
6990 2017-07-28 Martin Liska <mliska@suse.cz>
6993 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6994 parameters that are of a variable-length.
6996 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6998 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7000 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7001 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7002 (CRT_CALL_STATIC_FUNCTION): Likewise.
7003 (ASM_DEFAULT_SPEC): New define.
7004 (ASM_SPEC32): Likewise.
7005 (ASM_SPEC64): Likewise.
7006 (ASM_SPEC_COMMON): Likewise.
7007 (ASM_SPEC): Likewise.
7008 (INVALID_64BIT): Likewise.
7009 (LINK_OS_DEFAULT_SPEC): Likewise.
7010 (LINK_OS_SPEC32): Likewise.
7011 (LINK_OS_SPEC64): Likewise.
7012 (POWERPC_LINUX): Likewise.
7013 (PTRDIFF_TYPE): Likewise.
7014 (RESTORE_FP_PREFIX): Likewise.
7015 (RESTORE_FP_SUFFIX): Likewise.
7016 (SAVE_FP_PREFIX): Likewise.
7017 (SAVE_FP_SUFFIX): Likewise.
7018 (SIZE_TYPE): Likewise.
7019 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7020 (TARGET_64BIT): Likewise.
7021 (TARGET_64BIT): Likewise.
7022 (TARGET_AIX): Likewise.
7023 (WCHAR_TYPE_SIZE): Likewise.
7024 (WCHAR_TYPE): Undefine.
7025 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7026 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7027 (CPP_OS_RTEMS_SPEC): Delete.
7028 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7029 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7031 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7033 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7035 PR tree-optimization/81578
7036 * tree-parloops.c (build_new_reduction): Bail out if
7037 reduction_code isn't one of the standard OpenMP reductions.
7038 Move the details printing after that decision.
7040 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7042 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7043 related to reload_in_progress.
7044 (splat_input_operand): Likewise.
7045 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7047 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7049 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7050 (TARGET_INSTANTIATE_DECLS): Likewise.
7051 (legitimate_indexed_address_p): Delete reload_in_progress code.
7052 (rs6000_debug_legitimate_address_p): Likewise.
7053 (rs6000_eliminate_indexed_memrefs): Likewise.
7054 (rs6000_emit_le_vsx_store): Likewise.
7055 (rs6000_emit_move_si_sf_subreg): Likewise.
7056 (rs6000_emit_move): Likewise.
7057 (register_to_reg_type): Likewise.
7058 (rs6000_pre_atomic_barrier): Likewise.
7059 (rs6000_machopic_legitimize_pic_address): Likewise.
7060 (rs6000_allocate_stack_temp): Likewise.
7061 (rs6000_address_for_fpconvert): Likewise.
7062 (rs6000_address_for_altivec): Likewise.
7063 (rs6000_secondary_memory_needed_rtx): Delete function.
7064 (rs6000_check_sdmode): Likewise.
7065 (rs6000_alloc_sdmode_stack_slot): Likewise.
7066 (rs6000_instantiate_decls): Likewise.
7067 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7068 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7069 Delete reload_in_progress.
7070 (*vec_reload_and_plus_<mptrsize>): Likewise.
7071 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7072 (vsx_div_v2di): Likewise.
7073 (vsx_udiv_v2di): Likewise.
7075 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7077 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7078 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7079 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7080 (rs6000_debug_reg_global): Delete print of LRA status.
7081 (rs6000_option_override_internal): Delete dead LRA related code.
7082 (rs6000_lra_p): Delete function.
7083 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7085 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7087 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7088 * config/riscv/rtems.h: New file.
7090 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7091 Sudakshina Das <sudi.das@arm.com>
7093 * config/aarch64/aarch64.md
7094 (define_split for and<mode>3nr_compare): Move
7095 non aarch64_logical_operand to a register.
7096 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7097 register immediate operand to a register.
7098 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7100 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7103 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7105 2017-07-27 Richard Biener <rguenther@suse.de>
7107 PR tree-optimization/81573
7108 PR tree-optimization/81494
7109 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7110 multi defuse cycle case.
7112 2017-07-27 Richard Biener <rguenther@suse.de>
7114 PR tree-optimization/81571
7115 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7118 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7120 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7121 earlier and only if MASK_FPU is set. Adjust formatting.
7123 2017-07-27 Martin Liska <mliska@suse.cz>
7125 * opt-functions.awk: Add validation of value of Init.
7126 * optc-gen.awk: Pass new argument.
7128 2017-07-27 Martin Liska <mliska@suse.cz>
7130 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7131 Fix wrong condition.
7133 2017-07-27 Martin Liska <mliska@suse.cz>
7135 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7136 BBs and edges seen by autoFDO.
7138 2017-07-27 Richard Biener <rguenther@suse.de>
7140 PR tree-optimization/81502
7141 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7142 with incompatible but same sized type.
7143 (execute_update_addresses_taken): Likewise.
7145 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7147 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7148 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7150 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7153 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7154 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7155 Change s_operand to memory_operand.
7157 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7159 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7160 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7161 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7162 Emit instructions rather than returning an expression. Handle TFmode
7163 and KFmode by casting to TImode.
7164 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7165 (rs6000_emit_le_vsx_store): Likewise.
7166 * config/rs6000/vsx.md (VSX_TI): New iterator.
7167 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7168 (*vsx_le_undo_permute_<mode>): Likewise.
7169 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7170 emit the split sequence.
7171 (*vsx_le_perm_store_<mode>): Likewise.
7173 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7175 PR tree-optimization/81555
7176 PR tree-optimization/81556
7177 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7178 if true, force CHANGED for the recursive invocation.
7179 (reassociate_bb): Remember original length of ops array, pass
7180 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7182 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7183 attributes for noipa attribute. For naked attribute use
7184 lookup_attribute first before lookup_attribute_spec.
7185 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7187 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7188 for functions with noipa attribute.
7189 (cgraph_externally_visible_p): Return true for functions with noipa
7191 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7192 for functions with noipa attribute.
7193 * doc/extend.texi: Document noipa function attribute.
7194 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7195 also for functions with noipa attribute.
7196 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7198 2017-07-26 Andrew Pinski <apinski@cavium.com>
7200 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7201 vec_unalign_load_cost and vec_unalign_store_cost.
7203 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7205 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7206 -mvsx-small-integer option.
7207 (ISA_3_0_MASKS_IEEE): Likewise.
7208 (OTHER_VSX_VECTOR_MASKS): Likewise.
7209 (POWERPC_MASKS): Likewise.
7210 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7211 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7212 code, only testing for DImode being allowed in non-VSX floating
7214 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7215 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7217 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7218 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7219 TARGET_P8_VECTOR test.
7220 (rs6000_secondary_reload_simple_move): Likewise.
7221 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7222 since TARGET_P9_VECTOR was already tested.
7223 (rs6000_opt_masks): Remove -mvsx-small-integer.
7224 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7225 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7227 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7228 test for TARGET_VEXTRACTUB was used, and that uses
7230 (p9 extract splitter): Likewise.
7231 (vsx_extract_<mode>_di_p9): Likewise.
7232 (vsx_extract_<mode>_store_p9): Likewise.
7233 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7234 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7235 the elimination of TARGET_VSX_SMALL_INTEGER.
7236 (vsx_extract_<mode>_p8): Likewise.
7237 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7238 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7239 (vsx_set_<mode>_p9): Likewise.
7240 (vsx_set_v4sf_p9): Likewise.
7241 (vsx_set_v4sf_p9_zero): Likewise.
7242 (vsx_insert_extract_v4sf_p9): Likewise.
7243 (vsx_insert_extract_v4sf_p9_2): Likewise.
7244 * config/rs6000/rs6000.md (sign extend splitter): Change
7245 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7246 (floatsi<mode>2_lfiwax_mem): Likewise.
7247 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7248 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7249 since a test for TARGET_P9_VECTOR was used.
7250 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7251 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7252 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7253 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7254 TARGET_P8_VECTOR test.
7255 (fix_trunc<mode>si2_stfiwx): Likewise.
7256 (fix_trunc<mode>si2_internal): Likewise.
7257 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7258 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7260 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7261 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7262 TARGET_P8_VECTOR test.
7263 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7264 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7265 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7267 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7268 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7269 since a test for TARGET_P9_VECTOR was used.
7270 (splitter for loading small constants): Likewise.
7272 2017-07-26 Andrew Pinski <apinski@cavium.com>
7274 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7277 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7280 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7281 (fp_valid_at): Likewise.
7283 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7285 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7286 (qdf24xx_addrcost_table): Likewise.
7287 (cortexa57_tunings): Update to use generic_branch_cost.
7288 (cortexa72_tunings): Likewise.
7289 (cortexa73_tunings): Likewise.
7290 (qdf24xx_tunings): Likewise.
7292 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7294 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7295 (thunderx2t99_branch_cost): Likewise.
7296 (cortexa35_tunings): Update to use generic_branch_cost.
7297 (cortexa53_tunings): Likewise.
7298 (cortexa57_tunings): Likewise.
7299 (cortexa72_tunings): Likewise.
7300 (cortexa73_tunings): Likewise.
7301 (thunderx2t99_tunings): Likewise.
7303 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7305 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7306 (sparc_option_override): Honour MASK_FSMULD.
7307 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7308 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7309 * config/sparc/sparc.opt (mfsmuld): New option.
7310 * doc/invoke.texi (mfsmuld): Document option.
7312 2017-07-26 Marek Polacek <polacek@redhat.com>
7315 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7317 2017-07-26 Richard Biener <rguenther@suse.de>
7319 * gimple-match-head.c (do_valueize): Return OP if valueize
7321 (get_def): New helper to get at the def stmt of a SSA name
7323 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7324 do_valueize to get at the def stmt.
7325 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7327 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7330 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7332 2017-07-26 Martin Liska <mliska@suse.cz>
7335 * function.c (expand_function_start): Make expansion of
7336 nonlocal_goto_save_area after parm_birth_insn.
7338 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7340 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7341 from all CPU target flags enable members.
7343 2017-07-26 Richard Biener <rguenther@suse.de>
7345 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7346 (decision_tree::gen): Make 'type' const.
7347 (write_predicate): Likewise.
7349 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7351 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7352 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7353 (rs6000_option_override_internal): Likewise.
7354 (rs6000_expand_vector_set): Likewise.
7355 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7356 (TARGET_UPPER_REGS_SF): Likewise.
7357 (TARGET_UPPER_REGS_DI): Likewise.
7358 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7359 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7360 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7361 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7362 (Splitters for DI constants in Altivec registers): Likewise.
7363 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7364 (vsx_set_v4sf_p9): Likewise.
7365 (vsx_set_v4sf_p9_zero): Likewise.
7366 (vsx_insert_extract_v4sf_p9): Likewise.
7367 (vsx_insert_extract_v4sf_p9_2): Likewise.
7369 2017-07-25 Carl Love <cel@us.ibm.com>
7371 * doc/extend.texi: Update the built-in documentation file for the
7372 existing built-in functions
7373 vector signed char vec_cnttz (vector signed char);
7374 vector unsigned char vec_cnttz (vector unsigned char);
7375 vector signed short vec_cnttz (vector signed short);
7376 vector unsigned short vec_cnttz (vector unsigned short);
7377 vector signed int vec_cnttz (vector signed int);
7378 vector unsigned int vec_cnttz (vector unsigned int);
7379 vector signed long long vec_cnttz (vector signed long long);
7380 vector unsigned long long vec_cnttz (vector unsigned long long);
7382 2017-07-25 Andrew Pinski <apinski@cavium.com>
7384 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7385 accesses where the use is for the first operand of a BIT_INSERT.
7387 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7390 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7391 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7393 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7395 * config/i386/gstabs.h: Delete.
7396 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7398 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7400 * config/i386/i386.c (ix86_decompose_address): Do not check for
7401 register RTX when looking at index_reg or base_reg.
7402 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7404 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7406 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7407 to update EH info here.
7409 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7411 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7413 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7415 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7417 2017-07-25 Torsten Duwe <duwe@suse.de>
7419 * common.opt: Introduce -fpatchable-function-entry
7420 command line option, and its variables function_entry_patch_area_size
7421 and function_entry_patch_area_start.
7422 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7423 including a two-value parser.
7424 * target.def (print_patchable_function_entry): New target hook.
7425 * targhooks.h (default_print_patchable_function_entry): New function.
7426 * targhooks.c (default_print_patchable_function_entry): Likewise.
7427 * toplev.c (process_options): Switch off IPA-RA if
7428 patchable function entries are being generated.
7429 * varasm.c (assemble_start_function): Look at the
7430 patchable-function-entry command line switch and current
7431 function attributes and maybe generate NOP instructions by
7432 calling the print_patchable_function_entry hook.
7433 * doc/extend.texi: Document patchable_function_entry attribute.
7434 * doc/invoke.texi: Document -fpatchable_function_entry
7435 command line option.
7436 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7438 * doc/tm.texi: Re-generate.
7440 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7443 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7444 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7446 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7448 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7450 2017-07-25 Richard Biener <rguenther@suse.de>
7452 PR tree-optimization/81455
7453 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7454 not walk in cycles when looking for guards.
7456 2017-07-25 Richard Biener <rguenther@suse.de>
7458 PR tree-optimization/81529
7459 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7460 when optimizing backedge uses.
7462 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7464 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7466 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7467 to dl_section_ref. On AIX, append an expression to subtract
7468 the size of the section length to dl_section_ref.
7470 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7472 * configure.ac: If any of the config.* scripts fail, exit 1.
7473 * configure: Regenerate.
7475 2017-07-25 Richard Biener <rguenther@suse.de>
7478 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7479 of immediate uses, be more verbose on errors.
7481 2017-07-25 Richard Biener <rguenther@suse.de>
7483 PR tree-optimization/81510
7484 * tree-vect-loop.c (vect_is_simple_reduction): When the
7485 reduction stmt is not inside the loop bail out.
7487 2017-07-25 Richard Biener <rguenther@suse.de>
7489 PR tree-optimization/81303
7490 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7491 profitability check against LOOP_VINFO_NITERSM1.
7493 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7495 * domwalk.c (cmp_bb_postorder): Simplify.
7496 (sort_bbs_postorder): New function. Use it...
7497 (dom_walker::walk): ...here to optimize common cases.
7499 2017-07-25 Martin Liska <mliska@suse.cz>
7502 * ipa-visibility.c (function_and_variable_visibility): Make the
7503 redirection just on target that supports aliasing.
7504 Fix GNU coding style.
7506 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7509 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7511 * config/arm/rtems.h: Likewise.
7512 * config/bfin/rtems.h: Likewise.
7513 * config/i386/rtemself.h: Likewise.
7514 * config/lm32/rtems.h: Likewise.
7515 * config/m32c/rtems.h: Likewise.
7516 * config/m68k/rtemself.h: Likewise.
7517 * config/microblaze/rtems.h: Likewise.
7518 * config/mips/rtems.h: Likewise.
7519 * config/moxie/rtems.h: Likewise.
7520 * config/nios2/rtems.h: Likewise.
7521 * config/powerpcspe/rtems.h: Likewise.
7522 * config/rs6000/rtems.h: Likewise.
7523 * config/rtems.h: Likewise.
7524 * config/sh/rtems.h: Likewise.
7525 * config/sh/rtemself.h: Likewise.
7526 * config/sparc/rtemself.h: Likewise.
7528 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7531 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7532 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7533 * tree-ssa-structalias.c (alias_get_name): Same.
7535 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7538 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7539 instructions if no du chain is found.
7541 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7543 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7545 2017-07-25 Richard Biener <rguenther@suse.de>
7548 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7551 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7553 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7555 (ISA_2_7_MASKS_SERVER): Likewise.
7556 (ISA_3_0_MASKS_IEEE): Likewise.
7557 (OTHER_P8_VECTOR_MASKS): Likewise.
7558 (OTHER_VSX_VECTOR_MASKS): Likewise.
7559 (POWERPC_MASKS): Likewise.
7560 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7561 duplicate list of options.
7562 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7563 explicit -mupper-regs options.
7564 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7565 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7566 alias for -mupper-regs-df.
7567 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7568 (rs6000_init_hard_regno_mode_ok): Likewise.
7569 (rs6000_option_override_internal): Likewise.
7570 (rs6000_opt_masks): Likewise.
7571 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7572 options in terms of whether -mvsx or -mpower8-vector was used.
7573 (TARGET_UPPER_REGS_DI): Likewise.
7574 (TARGET_UPPER_REGS_SF): Likewise.
7575 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7576 -mupper-regs-* options.
7578 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7580 * passes.c (emergency_dump_function): Print some empty lines and a
7581 header before the RTL dump.
7583 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7585 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7587 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7590 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7591 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7593 2017-07-24 Carl Love <cel@us.ibm.com>
7595 * config/rs6000/rs6000-c.c: Add support for built-in functions
7596 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7597 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7598 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7599 vec_extract_fp_from_shortl): Add defines for the two builtins.
7600 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7601 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7603 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7604 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7605 * doc/extend.texi: Update the built-in documentation file for the
7606 new built-in function.
7608 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7611 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7613 * doc/generic.texi: Likewise.
7614 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7615 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7617 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7619 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7620 (aarch64_mls_elt_merge<mode>): Likewise.
7622 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7624 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7625 having __cxa_atexit.
7627 2017-07-23 Michael Collison <michael.collison@arm.com>
7629 * config/arm/arm.c (arm_option_override): Deprecate
7630 use of -mstructure-size-boundary.
7631 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7632 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7634 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7637 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7638 Reduce cost estimate for direct moves.
7640 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7643 * config/i386/i386.c (ix86_option_override_internal): Disable
7644 BMI, BMI2 and TBM instructions for -m16.
7646 2017-07-21 Carl Love <cel@us.ibm.com>
7648 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7649 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7650 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7652 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7653 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7654 VMULOSW): New enum "unspec" values.
7655 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7656 altivec_vmulosw): New patterns.
7657 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7658 VMULOSW): Add definitions.
7660 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7662 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7663 (qdf24xx): Likewise.
7664 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7665 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7666 (AARCH64_FL_V8_1): Renumber.
7667 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7668 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7669 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7670 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7671 rdma to feature modifiers list.
7673 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7676 * ipa-visibility (function_and_variable_visibility): Convert
7677 recursive PLT call to direct call if appropriate.
7679 2017-07-21 Andrew Pinski <apinski@cavium.com>
7681 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7682 operand 1 to see if the types precision matches.
7683 * fold-const.c (operand_equal_p): Likewise.
7685 2017-07-21 Richard Biener <rguenther@suse.de>
7687 PR tree-optimization/81303
7688 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7689 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7690 (vect_peeling_hash_get_lowest_cost): Adjust.
7691 (vect_enhance_data_refs_alignment): Likewise. Use
7692 vect_get_peeling_costs_all_drs to compute the penalty for no
7693 peeling to match up costs.
7695 2017-07-21 Richard Biener <rguenther@suse.de>
7697 PR tree-optimization/81500
7698 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7699 we didn't identify a reduction path.
7701 2017-07-21 Tom de Vries <tom@codesourcery.com>
7702 Cesar Philippidis <cesar@codesourcery.com>
7704 PR gcov-profile/81442
7705 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7708 2017-07-21 Tom de Vries <tom@codesourcery.com>
7711 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7713 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7714 nvptx_override_options_after_change.
7716 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7718 * dwarf2out.c (output_file_names): Avoid double testing for
7721 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7723 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7725 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7727 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7729 (try_crossjump_to_edge): Do not punt on partitioned functions.
7731 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7733 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7734 Put all BBs reachable only via paths crossing cold region to cold
7736 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7738 2016-07-21 Richard Biener <rguenther@suse.de>
7740 PR tree-optimization/81303
7741 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7742 into account prologue and epilogue iterations when raising
7743 min_profitable_iters to sth at least covering one vector iteration.
7745 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7747 * config/arm/arm.c (arm_test_cpu_arch_dat):
7750 2017-07-20 Nathan Sidwell <nathan@acm.org>
7752 Remove TYPE_METHODS.
7753 * tree.h (TYPE_METHODS): Delete.
7754 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7755 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7756 (dbxout_type_methods): Scan TYPE_FIELDS.
7757 (dbxout_type): Don't check TYPE_METHODS here.
7758 * function.c (use_register_for_decl): Always ignore register for
7759 class types when not optimizing.
7760 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7761 * tree.c (free_lang_data_in_type): Stitch out member functions and
7762 templates from TYPE_FIELDS.
7763 (build_distinct_type_copy, verify_type_variant,
7764 verify_type): Member fns are on TYPE_FIELDS.
7765 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7766 * tree-pretty-print.c (dump_generic_node): Likewise.
7768 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7771 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7772 V2TImode and V4TImode.
7773 (ix86_expand_vector_extract): Likewise.
7774 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7775 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7776 (ssescalarmode): Handle V4TImode and V2TImode.
7777 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7778 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7779 (VEXTRACTI128_MODE): New mode iterator.
7780 (splitter for *vec_extractv?ti first element): New.
7781 (VEC_INIT_MODE): New mode iterator.
7782 (vec_init<mode>): Consolidate 3 expanders into one using
7783 VEC_INIT_MODE mode iterator.
7785 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7787 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7788 non_spilled_static_chain_regno_p.
7790 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7792 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7794 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7796 * bb-reorder.c (connect_traces): Allow copying of blocks within
7799 2017-07-20 Richard Biener <rguenther@suse.de>
7801 * gimple.h (gimple_phi_result): Add gphi * overload.
7802 (gimple_phi_result_ptr): Likewise.
7803 (gimple_phi_arg): Likewise. Adjust index assert to only
7804 allow actual argument accesses rather than all slots available
7806 (gimple_phi_arg_def): Add gphi * overload.
7807 * tree-phinodes.c (make_phi_node): Initialize only actual
7809 (resize_phi_node): Clear memory not covered by old node,
7810 do not initialize excess argument slots.
7811 (reserve_phi_args_for_new_edge): Initialize new argument slot
7814 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7816 PR tree-optimization/81388
7818 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7820 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7821 by removing computation of may_be_zero.
7823 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7824 Tom de Vries <tom@codesourcery.com>
7827 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7828 when gimple level profile disagrees with what RTL expander did.
7830 2017-07-20 Richard Biener <rguenther@suse.de>
7832 PR tree-optimization/61171
7833 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7834 (vect_analyze_stmt): Add slp instance parameter.
7835 (vectorizable_reduction): Likewise.
7836 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7837 (vect_is_simple_reduction): Deal with chains not detected
7838 as SLP reduction chain, specifically not properly associated
7839 chains containing a mix of plus/minus.
7840 (get_reduction_op): Remove.
7841 (get_initial_defs_for_reduction): Simplify, pass in whether
7842 this is a reduction chain, pass in the SLP node for the PHIs.
7843 (vect_create_epilog_for_reduction): Get the SLP instance as
7845 (vectorizable_reduction): Get the SLP instance as arg.
7846 During analysis remember the SLP node with the PHIs in the
7847 instance. Simplify getting at the vectorized reduction PHIs.
7848 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7849 through SLP instance.
7850 (vect_slp_analyze_operations): Likewise.
7851 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7852 (vect_transform_stmt): Likewise.
7854 2017-07-20 Tom de Vries <tom@codesourcery.com>
7856 PR tree-optimization/81489
7857 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7858 read of phi arg location to before loop that modifies phi.
7860 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7862 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7865 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7868 * except.c (execute): Fix ordering issue.
7870 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7872 PR rtl-optimization/81423
7873 * combine.c (make_compound_operation_int): Don't try to optimize
7874 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7876 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7878 PR rtl-optimization/81423
7879 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7880 with a constant that is -1 in the truncated to mode.
7882 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7884 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7885 (determine_unlikely_bbs): ... here.
7886 * predict.h (propagate_unlikely_bbs_forward): Declare.
7887 * cfgexpand.c (pass_expand::execute): Use it.
7888 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7890 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7891 propagate_unlikely_bbs_forward.
7893 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7896 * except.c (maybe_add_nop_after_section_switch): New function.
7899 2017-07-19 Tom de Vries <tom@codesourcery.com>
7901 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7903 2017-07-19 Tom de Vries <tom@codesourcery.com>
7905 * gimple.h (gimple_phi_arg): Make assert more strict.
7907 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7909 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7910 * config/rs6000/mmintrin.h: New file.
7911 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7913 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7915 PR tree-optimization/81346
7916 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7918 2017-07-19 Tom de Vries <tom@codesourcery.com>
7920 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7922 2017-07-19 Tom de Vries <tom@codesourcery.com>
7924 * config/nvptx/nvptx-modes.def: Add V2DImode.
7925 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7926 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7927 (nvptx_output_mov_insn): Handle lack of mov.b128.
7928 (nvptx_print_operand): Handle 'H' and 'L' codes.
7929 (nvptx_vector_mode_supported): Allow V2DImode.
7930 (nvptx_preferred_simd_mode): New function.
7931 (nvptx_data_alignment): New function.
7932 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7933 nvptx_preferred_simd_mode.
7934 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7936 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7938 2017-07-19 Tom de Vries <tom@codesourcery.com>
7940 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7941 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7942 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7943 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7944 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7945 (mov<VECIM>_insn): New define_insn.
7946 (define_expand "mov<VECIM>): New define_expand.
7948 2017-07-19 Tom de Vries <tom@codesourcery.com>
7950 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7952 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7954 PR tree-optimization/81346
7955 * fold-const.h (fold_div_compare, range_check_type): Declare.
7956 * fold-const.c (range_check_type): New function.
7957 (build_range_check): Use range_check_type.
7958 (fold_div_compare): No longer static, rewritten into
7959 a match.pd helper function.
7960 (fold_comparison): Don't call fold_div_compare here.
7961 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7964 2017-07-19 Nathan Sidwell <nathan@acm.org>
7966 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7967 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7968 * tree.c (find_decls_types_r, verify_type): Use
7969 TYPE_{MIN,MAX}_VALUE_RAW.
7970 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7971 (hash_tree): Likewise.
7972 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7974 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7977 2017-07-18 Tom de Vries <tom@codesourcery.com>
7980 * omp-expand.c (expand_omp_for_static_chunk): Handle
7981 equal-argument loop exit phi.
7983 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7986 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7987 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7988 operand 2 predicate.
7989 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7990 operand 2 predicate.
7991 (ror,rol -> rorx splitters): Use const_int_operand as
7992 operand 2 predicate.
7994 2017-06-18 Richard Biener <rguenther@suse.de>
7996 PR tree-optimization/81410
7997 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7998 the gap in the ! slp_perm SLP case after each group.
8000 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8003 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8006 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8009 * predict.c (set_even_probabilities): Cleanup; do not affect
8010 probabilities that are already known.
8011 (combine_predictions_for_bb): Call even when count is set.
8013 2017-07-18 Nathan Sidwell <nathan@acm.org>
8015 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8018 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8021 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8022 optimization for loop niter analysis.
8024 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8027 * config/avr/avr.c (avr_optimize_casesi): Don't use
8028 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8030 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8032 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8033 body_cost_vec from _vect_peel_extended_info.
8034 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8035 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8038 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8040 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8042 2017-07-18 Richard Biener <rguenther@suse.de>
8044 PR tree-optimization/80620
8045 PR tree-optimization/81403
8046 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8047 info when re-using a VN table entry.
8049 2017-07-18 Richard Biener <rguenther@suse.de>
8051 PR tree-optimization/81418
8052 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8053 vectype_in. Verify that with lane-reducing reduction operations
8054 we have a single def-use cycle.
8056 2017-07-17 Carl Love <cel@us.ibm.com>
8058 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8060 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8061 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8062 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8063 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8064 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8065 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8066 VMULOSW): New enum "unspec" values.
8067 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8068 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8069 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8070 altivec_vmulosw): New patterns.
8071 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8072 VMULOSW): Add definitions.
8073 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8075 * config/alpha/alpha.c: Include predict.h.
8077 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8079 * tree-vrp.c (compare_assert_loc): Fix comparison function
8080 to return predictable results.
8082 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8084 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8087 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8088 * doc/invoke.texi (mexpand-adddi): Update text.
8090 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8092 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8093 that also clobbers the CC register. The old expand code is moved
8095 (*arc_clzsi2): ... here.
8096 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8097 the CC register. The old expand code is moved to ...
8098 (arc_ctzsi2): ... here.
8100 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8102 * config/arc/arc.opt (mindexed-loads): Use initial value
8103 TARGET_INDEXED_LOADS_DEFAULT.
8104 (mauto-modify-reg): Use initial value
8105 TARGET_AUTO_MODIFY_REG_DEFAULT.
8106 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8107 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8108 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8109 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8111 2017-07-17 Martin Liska <mliska@suse.cz>
8114 * opts.c (finish_options): Do not allow -fgnu-tm
8115 w/ -fsanitize={kernel-,}address. Say sorry.
8117 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8120 * tree-loop-distribution.c (classify_partition): Only assert on
8121 numer of iterations.
8122 (merge_dep_scc_partitions): Delete prameter. Update function call.
8123 (distribute_loop): Remove code handling loop with unknown niters.
8124 (pass_loop_distribution::execute): Skip loop with unknown niters.
8126 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8129 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8130 function sort_partitions_by_post_order.
8132 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8134 PR tree-optimization/81374
8135 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8136 the max index of basic blocks, rather than number of basic blocks.
8138 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8140 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8142 (arc_legitimate_pic_operand_p): Likewise.
8143 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8145 (arc_needs_pcl_p): Likewise.
8146 (arc_legitimate_pc_offset_p): Likewise.
8147 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8148 function is also used in constrains.md.
8149 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8150 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8151 PLUS. Only return true/false in known cases, otherwise assert.
8152 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8153 is already called in arc_legitimate_constant_p.
8154 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8156 (LEGITIMATE_PIC_OPERAND_P): Use
8157 arc_raw_symbolic_reference_mentioned_p function.
8158 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8163 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8164 Andrew Burgess <andrew.burgess@embecosm.com>
8166 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8167 (arc_return_address_register): New function.
8168 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8169 (arc_handle_fndecl_attribute): Add naked attribute.
8170 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8171 (TARGET_WARN_FUNC_RETURN): Likewise.
8172 (arc_allocate_stack_slots_for_args): New function.
8173 (arc_warn_func_return): Likewise.
8174 (machine_function): Change type fn_type.
8175 (arc_compute_function_type): Consider new naked function type,
8176 change function return type.
8177 (arc_must_save_register): Adapt to handle new
8178 arc_compute_function_type's return type.
8179 (arc_expand_prologue): Likewise.
8180 (arc_expand_epilogue): Likewise.
8181 (arc_return_address_regs): Delete.
8182 (arc_return_address_register): New function.
8183 (arc_epilogue_uses): Use above function.
8184 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8185 (arc_function_type): Change encoding, add naked type.
8186 (ARC_INTERRUPT_P): Change to handle the new encoding.
8187 (ARC_FAST_INTERRUPT_P): Likewise.
8188 (ARC_NORMAL_P): Define.
8189 (ARC_NAKED_P): Likewise.
8190 (arc_compute_function_type): Delete prototype.
8191 * config/arc/arc.md (in_ret_delay_slot): Use
8192 arc_return_address_register function.
8193 (simple_return): Likewise.
8194 (p_return_i): Likewise.
8196 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8198 PR tree-optimization/81428
8199 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8200 can't be built for those types.
8202 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8204 Remove stuff dead since r239246.
8206 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8207 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8208 (avr_inform_devices): Remove dead stuff.
8210 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8212 * config/arm/arm_neon.h: Fix softp typo.
8214 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8216 PR tree-optimization/81365
8217 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8218 aggregate moves onto bb predecessor edges, make sure there are no
8219 loads that could alias the lhs in between the start of bb and the
8222 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8225 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8226 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8227 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8229 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8231 PR tree-optimization/81396
8232 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8233 (init_symbolic_number): Initialize it to 1.
8234 (perform_symbolic_merge): Add n_ops from both operands into the new
8236 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8237 without base_addr as useless if they need more than one operation.
8238 (bswap_replace): Handle !bswap case for NULL base_addr.
8240 2017-07-17 Tom de Vries <tom@codesourcery.com>
8243 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8246 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8248 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8249 conditional builtin define __FIX_LEON3FT_B2BST.
8251 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8253 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8254 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8257 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8259 PR rtl-optimization/81424
8260 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8261 to remove potential trapping from operands if -fnon-call-exceptions.
8263 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8265 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8266 profile_proability for scalling.
8267 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8269 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8271 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8273 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8275 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8276 fixpoint arithmetics.
8278 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8280 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8281 fixpoint arithmetics.
8283 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8285 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8286 fixpoint arithmetics.
8288 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8290 * profile-count.h (profile_probability::from_reg_br_prob_note,
8291 profile_probability::to_reg_br_prob_note): New functions.
8292 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8293 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8294 * predict.c (probability_reliable_p): Update.
8295 (edge_probability_reliable_p): Update.
8296 (br_prob_note_reliable_p): Update.
8297 (invert_br_probabilities): Update.
8298 (add_reg_br_prob_note): New function.
8299 (combine_predictions_for_insn): Update.
8300 * asan.c (asan_clear_shadow): Update.
8301 * cfgbuild.c (compute_outgoing_frequencies): Update.
8302 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8303 (update_br_prob_note): Update.
8304 (rtl_verify_edges): Update.
8305 (purge_dead_edges): Update.
8306 (fixup_reorder_chain): Update.
8307 * emit-rtl.c (try_split): Update.
8308 * ifcvt.c (cond_exec_process_insns): Update.
8309 (cond_exec_process_if_block): Update.
8310 (dead_or_predicable): Update.
8311 * internal-fn.c (expand_addsub_overflow): Update.
8312 (expand_neg_overflow): Update.
8313 (expand_mul_overflow): Update.
8314 * loop-doloop.c (doloop_modify): Update.
8315 * loop-unroll.c (compare_and_jump_seq): Update.
8316 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8317 * predict.h: Update.
8318 * reorg.c (mostly_true_jump): Update.
8320 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8321 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8322 * config/arc/arc.c: (emit_unlikely_jump): Update.
8323 * config/arm/arm.c: (emit_unlikely_jump): Update.
8324 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8325 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8326 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8327 (ix86_print_operand): Update.
8328 (ix86_split_fp_branch): Update.
8329 (predict_jump): Update.
8330 * config/ia64/ia64.c (ia64_print_operand): Update.
8331 * config/mmix/mmix.c (mmix_print_operand): Update.
8332 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8333 (rs6000_expand_split_stack_prologue): Update.
8334 * config/rs6000/rs6000.c: Update.
8335 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8336 (s390_expand_vec_movstr): Update.
8337 (s390_expand_cs_tdsi): Update.
8338 (s390_expand_split_stack_prologue): Update.
8339 * config/sh/sh.c (sh_print_operand): Update.
8340 (expand_cbranchsi4): Update.
8341 (expand_cbranchdi4): Update.
8342 * config/sparc/sparc.c (output_v9branch): Update.
8343 * config/spu/spu.c (get_branch_target): Update.
8344 (ea_load_store_inline): Update.
8345 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8346 * config/tilepro/tilepro.c: Update.
8348 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8350 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8351 (gimplify_save_expr): Likewise.
8353 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8355 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8357 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8359 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8361 * timevar.def (TV_IPA_FNSUMMARY): Define.
8363 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8365 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8366 to back store errata sensitive sequence from being generated.
8367 (sqrtdf2_fix): Likewise.
8369 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8371 * tree-ssa-threadupdate.c (compute_path_counts,
8372 update_joiner_offpath_counts): Use profile_probability.
8374 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8377 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8379 * config/arm/arm-c.c (arm_cpu_builtins): Define
8380 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8382 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8384 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8385 array entries to represent __ieee128 versions of the
8386 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8387 scalar_extract_sig, and scalar_insert_exp built-in functions.
8388 (altivec_resolve_overloaded_builtin): Add special case handling
8389 for the __builtin_scalar_insert_exp function, as represented by
8390 the P9V_BUILTIN_VEC_VSIEDP constant.
8391 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8392 exponent support for __ieee128 argument.
8393 (VSESQP): Add scalar extract signature support for __ieee128
8395 (VSTDCNQP): Add scalar test negative support for __ieee128
8397 (VSIEQP): Add scalar insert exponent support for __int128 argument
8398 with __ieee128 result.
8399 (VSIEQPF): Add scalar insert exponent support for __ieee128
8400 argument with __ieee128 result.
8401 (VSTDCQP): Add scalar test data class support for __ieee128
8403 (VSTDCNQP): Add overload support for scalar test negative with
8405 (VSTDCQP): Add overload support for scalar test data class
8407 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8409 (UNSPEC_VSX_SIEXPQP): New constant.
8410 (xsxexpqp): New insn for VSX scalar extract exponent quad
8412 (xsxsigqp): New insn for VSX scalar extract significand quad
8414 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8415 precision with floating point argument.
8416 (xststdcqp): New expand for VSX scalar test data class quad
8418 (xststdcnegqp): New expand for VSX scalar test negative quad
8420 (xststdcqp): New insn to match expansions for VSX scalar test data
8421 class quad precision and VSX scalar test negative quad precision.
8422 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8423 special case operand checking to enforce that second operand of
8424 VSX scalar test data class with quad precision argument is a 7-bit
8426 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8427 prototypes and descriptions of __ieee128 versions of
8428 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8429 scalar_test_data_class, and scalar_test_neg built-in functions.
8431 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8433 PR tree-optimization/81162
8434 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8435 replace a negate with an add.
8437 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8439 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8441 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8443 * config/arm/arm-c.c (arm_cpu_builtins): Define
8444 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8446 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8448 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8449 (armv8-r): Set ARM Cortex-R52 as default CPU.
8450 * config/arm/arm-tables.opt: Regenerate.
8451 * config/arm/arm-tune.md: Regenerate.
8452 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8454 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8455 extension for -mcpu=cortex-r52.
8457 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8459 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8460 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8461 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8462 (fp-armv8): Define it as FP_ARMv8 only.
8463 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8464 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8466 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8467 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8468 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8469 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8470 than TARGET_FPU_ARMV8.
8471 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8472 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8473 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8475 * config/arm/neon.md (neon_vrint): Likewise.
8476 (neon_vcvt): Likewise.
8477 (neon_<fmaxmin_op><mode>): Likewise.
8478 (<fmaxmin><mode>3): Likewise.
8479 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8480 * config/arm/predicates.md (arm_cond_move_operator): Check against
8481 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8483 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8485 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8488 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8490 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8491 loop in comment with memset.
8493 2017-07-14 Martin Liska <mliska@suse.cz>
8495 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8496 * dwarf2out.c (is_java): Remove the function.
8497 (output_pubname): Remove usage of the function.
8498 (lower_bound_default): Remove usage of DW_LANG_Java.
8499 (gen_compile_unit_die): Likewise.
8500 * gcc.c: Remove compiler defaults for .java and .zip files.
8501 * gimple-expr.c (remove_suffix): Change as there's no longer
8502 extension than 4-letter one.
8503 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8504 (gimplify_save_expr): Likewise.
8505 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8506 as it's possible even for other languages than Java.
8507 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8508 * lto-opts.c (lto_write_options): Remove reference to Java.
8509 * opts.c (strip_off_ending): Update file extension handling.
8510 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8511 * tree-eh.c (lower_resx): Likewise.
8512 * tree.c (free_lang_data_in_type): Remove dead code.
8513 (find_decls_types_r): Likewise.
8514 (build_common_builtin_nodes): Remove Java from a comment.
8515 (verify_type): Remove dead code.
8516 * varasm.c (assemble_external): Remove Java from a comment.
8518 2017-07-14 Martin Liska <mliska@suse.cz>
8520 * opts.c (finish_options): Add quotes.
8521 (common_handle_option): Likewise.
8523 2017-07-14 Martin Liska <mliska@suse.cz>
8525 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8526 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8528 * dwarf2out.c (lower_bound_default): Do not handle
8530 (gen_compile_unit_die): Likewise.
8531 * gcc.c: Remove default extension binding for GNU Pascal.
8532 * stmt.c: Remove Pascal language from a comment.
8533 * xcoffout.c: Likewise.
8535 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8538 * diagnostic-show-locus.c (fixit_cmp): New function.
8539 (layout::layout): Sort m_fixit_hints.
8540 (column_range::column_range): Assert that the values are valid.
8541 (struct char_span): New struct.
8542 (correction::overwrite): New method.
8543 (struct source_line): New struct.
8544 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8545 calls in terms of classes source_line and char_span, and
8546 correction::overwrite.
8547 (selftest::test_overlapped_fixit_printing_2): New function.
8548 (selftest::diagnostic_show_locus_c_tests): Call it.
8550 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8552 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8553 early if there is no lhs.
8555 2017-07-13 Martin Liska <mliska@suse.cz>
8557 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8558 (gen_reference_type_die): Likewise.
8559 * stor-layout.c: Remove Pascal-related comment.
8561 2017-07-13 Martin Liska <mliska@suse.cz>
8563 * opts.c (finish_options): Add quotes to error messages.
8564 (parse_sanitizer_options): Likewise.
8566 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8568 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8570 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8572 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8574 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8576 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8578 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8580 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8583 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8584 provides the hardware capability bits, define the macro
8585 __BUILTIN_CPU_SUPPORTS__.
8586 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8587 if GLIBC does not provide the hardware capability bits. Add a
8588 gcc_unreachable call if the built-in cpu function is neither
8589 __builtin_cpu_is nor __builtin_cpu_supports.
8590 (rs6000_get_function_versions_dispatcher): Change the warning
8591 that an old GLIBC is used which does not export the capability
8592 bits to be an error.
8593 * doc/extend.texi (target_clones attribute): Document the
8594 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8595 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8596 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8597 the macros defined by GCC if the newer GLIBC is available.
8599 2017-07-12 Jeff Law <law@redhat.com>
8601 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8602 remaining includes slightly.
8603 * config/riscv/riscv-builtins.c: Include profile-count.h.
8605 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8608 * config/avr/avr.c (avr_set_current_function): In diagnostic
8609 messages: Quote keywords and (parts of) identifiers.
8610 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8613 2017-07-12 Carl Love <cel@us.ibm.com>
8615 * config/rs6000/rs6000-c.c: Add support for built-in functions
8616 vector bool char vec_revb (vector bool char);
8617 vector bool short vec_revb (vector short char);
8618 vector bool int vec_revb (vector bool int);
8619 vector bool long long vec_revb (vector bool long long);
8620 * doc/extend.texi: Update the built-in documentation file for the
8621 new built-in functions.
8623 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8625 * config/s390/s390.md: Remove movcc splitter.
8627 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8629 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8630 load/store on condition.
8632 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8635 * config/avr/avr.c (avr_encode_section_info)
8636 [progmem && !TREE_READONLY]: Error if progmem object needs
8639 2017-07-11 Michael Collison <michael.collison@arm.com>
8641 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8644 2017-07-11 Carl Love <cel@us.ibm.com>
8646 * config/rs6000/rs6000-c.c: Add support for builtins
8647 vector unsigned int vec_parity_lsbb (vector signed int);
8648 vector unsigned int vec_parity_lsbb (vector unsigned int);
8649 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8650 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8651 vector unsigned long long vec_parity_lsbb (vector signed long long);
8652 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8653 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8654 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8655 * doc/extend.texi: Update the built-in documentation file for the
8656 new built-in functions.
8658 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8660 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8661 (layout::m_primary_loc): New field.
8662 (layout::layout): Initialize new field. Move location filtering
8663 logic from here to...
8664 (layout::maybe_add_location_range): ...this new method. Add
8665 support for filtering to just the lines already specified by other
8667 (layout::will_show_line_p): New method.
8668 (gcc_rich_location::add_location_if_nearby): New method.
8669 (selftest::test_add_location_if_nearby): New test function.
8670 (selftest::diagnostic_show_locus_c_tests): Call it.
8671 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8674 2017-07-11 Tom de Vries <tom@codesourcery.com>
8676 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8677 (bb_first_real_insn): New function.
8678 (nvptx_single): Add extra initialization of broadcasted condition
8681 2017-07-11 Nathan Sidwell <nathan@acm.org>
8683 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8685 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8687 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8688 Binutils doc as TEXI will mess them up.
8689 * doc/invoke.texi (AVR Options): Same here.
8691 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8693 * config/sparc/sparc.opt (mfix-ut700): New option.
8694 (mfix-gr712rc): Likewise.
8695 (sparc_fix_b2bst): New variable.
8696 * doc/invoke.texi (SPARC options): Document them.
8697 (ARM options): Fix warnings.
8698 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8699 instructions to prevent sequences that can trigger the store-store
8700 errata for certain LEON3FT processors.
8701 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8702 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8703 * config/sparc/sparc.md (fix_b2bst): New attribute.
8704 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8706 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8709 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8711 (*rsqrtsf2_sse): Ditto.
8713 (div<mode>3): Macroize insn from divdf3 and divsf3
8714 using MODEF mode iterator.
8716 2017-07-10 Martin Sebor <msebor@redhat.com>
8718 PR tree-optimization/80397
8719 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8720 instead of testing for equality to INTEGER_TYPE.
8722 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8724 * config.gcc: Remove uclibc from arc target spec.
8726 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8728 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8730 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8733 * lto-wrapper.c (remove_option): New function.
8734 (merge_and_complain): Merge PIC/PIE options more realistically.
8736 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8738 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8742 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8743 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8744 * config.in: Regenerate.
8745 * configure: Regenerate.
8746 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8747 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8748 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8749 (TARGET_GASISR_PROLOGUES): ...target mask.
8750 * common/config/avr/avr-common.c
8751 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8752 Set -mgas-isr-prologues.
8753 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8754 INSERT_PASS_BEFORE for it.
8755 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8756 * config/avr/avr.c (avr_option_override)
8757 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8758 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8759 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8760 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8761 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8763 (make_avr_pass_pre_proep): New function.
8764 (emit_push_sfr) <treg>: Add argument to function and use it
8766 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8767 and set machine->gasisr.yes.
8768 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8769 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8770 __gcc_isr.n_pushed to .L__stack_usage.
8771 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8772 (avr_asm_final_postscan_insn): ...this new static function.
8773 * config/avr/avr.h (machine_function)
8774 <is_no_gccisr, use_L__stack_usage>: New fields.
8775 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8776 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8777 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8778 (gasisr, *gasisr): New expander and insn.
8779 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8780 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8781 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8783 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8785 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8788 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8790 Move jump-tables out of .text again.
8793 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8794 (ASM_OUTPUT_ADDR_VEC): New function.
8795 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8796 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8797 INSN_ADDRESSes as asm comment.
8798 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8799 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8800 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8801 * config/avr/avr.md (*tablejump): Adjust comment.
8802 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8803 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8805 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8806 (avr_output_addr_vec): New proto.
8807 (avr_log_t) <insn_addresses>: New field.
8809 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8812 * config/i386/i386.c (ix86_function_arg_advance): Set
8813 outgoing_args_on_stack to true if there are outgoing arguments
8815 (ix86_function_arg): Likewise.
8816 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8817 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8818 * config/i386/i386.h (machine_function): Add
8819 outgoing_args_on_stack.
8821 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8823 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8825 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8827 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8829 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8830 (REAL_H): Remove $(MACHMODE_H).
8831 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8833 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8834 $(MACHMODE_H) and double-int.h.
8835 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8837 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8840 2017-07-07 Andrew Pinski <apinski@cavium.com>
8842 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8843 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8845 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8847 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8848 Add warning if GCC was not configured to link against a GLIBC that
8849 exports the hardware capability bits.
8850 (make_resolver_func): Make resolver function private and not a
8851 COMDAT function. Create the name with clone_function_name instead
8852 of make_unique_name.
8855 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8856 correct operand in doing the split.
8858 2017-07-07 Carl Love <cel@us.ibm.com>
8860 * config/rs6000/rs6000-c: Add support for built-in function
8861 vector unsigned short vec_pack_to_short_fp32 (vector float,
8863 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8864 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8865 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8866 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8867 (convert_4f32_8i16): Add define_expand.
8868 * doc/extend.texi: Update the built-in documentation file for the
8869 new built-in function.
8871 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8873 * config/sparc/m8.md: New file.
8874 * config/sparc/sparc.md: Include m8.md.
8876 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8878 * config/sparc/sparc.opt: New option -mvis4b.
8879 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8880 (sparc_option_override): Handle VIS4B.
8881 (enum sparc_builtins): Define
8882 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8883 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8884 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8885 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8886 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8887 (check_constant_argument): New function.
8888 (sparc_vis_init_builtins): Define builtins
8889 __builtin_vis_dictunpack{8,16,32},
8890 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8891 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8892 __builtin_vis_fpcmpde{8,16,32}shl and
8893 __builtin_vis_fpcmpur{8,16,32}shl.
8894 (sparc_expand_builtin): Check that the constant operands to
8895 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8896 constant and in range.
8897 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8899 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8900 (SPARC_IMM5_P): Likewise.
8901 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8902 (enabled): Handle vis4b.
8903 (UNSPEC_DICTUNPACK): New unspec.
8904 (UNSPEC_FPCMPSHL): Likewise.
8905 (UNSPEC_FPUCMPSHL): Likewise.
8906 (UNSPEC_FPCMPDESHL): Likewise.
8907 (UNSPEC_FPCMPURSHL): Likewise.
8908 (cpu_feature): New CPU feature `vis4b'.
8909 (dictunpack{8,16,32}): New insns.
8910 (FPCSMODE): New mode iterator.
8911 (fpcscond): New code iterator.
8912 (fpcsucond): Likewise.
8913 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8914 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8915 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8916 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8917 * config/sparc/constraints.md: Define constraints `q' for unsigned
8918 2-bit integer constants and `t' for unsigned 5-bit integer
8920 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8922 (imm5_operand_dictunpack16): Likewise.
8923 (imm5_operand_dictunpack32): Likewise.
8924 (imm2_operand): Likewise.
8925 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8926 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8927 ditunpack* and fpcmp*shl builtins.
8929 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8931 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8932 * config.in: Add HAVE_AS_SPARC6 define.
8933 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8935 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8937 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8938 (CPP_CPU_SPEC): Handle m8.
8939 (ASM_CPU_SPEC): Likewise.
8940 * config/sparc/sparc-opts.h (enum processor_type): Add
8942 * config/sparc/sparc.c (m8_costs): New struct.
8943 (sparc_option_override): Handle TARGET_CPU_m8.
8944 (sparc32_initialize_trampoline): Likewise.
8945 (sparc64_initialize_trampoline): Likewise.
8946 (sparc_issue_rate): Likewise.
8947 (sparc_register_move_cost): Likewise.
8948 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8949 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8950 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8951 (CPP_CPU_SPEC): Handle M8.
8952 (ASM_CPU_SPEC): Likewise.
8953 (AS_M8_FLAG): Define.
8954 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8955 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8956 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8958 * configure: Regenerate.
8959 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8962 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8964 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8966 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8967 ("*movdi_insn_sp32"): Do not set v3pipe.
8968 ("*movsi_insn"): Likewise.
8969 ("*movdi_insn_sp64"): Likewise.
8970 ("*movsf_insn"): Likewise.
8971 ("*movdf_insn_sp32"): Likewise.
8972 ("*movdf_insn_sp64"): Likewise.
8973 ("*zero_extendsidi2_insn_sp64"): Likewise.
8974 ("*sign_extendsidi2_insn"): Likewise.
8975 ("*mov<VM32:mode>_insn"): Likewise.
8976 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8977 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8978 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8979 ("<vlop:code><VL:mode>3"): Likewise.
8980 ("*not_<vlop:code><VL:mode>3"): Likewise.
8981 ("*nand<VL:mode>_vis"): Likewise.
8982 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8983 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8984 ("one_cmpl<VL:mode>2"): Likewise.
8985 ("faligndata<VM64:mode>_vis"): Likewise.
8986 ("alignaddrsi_vis"): Likewise.
8987 ("alignaddrdi_vis"): Likweise.
8988 ("alignaddrlsi_vis"): Likewise.
8989 ("alignaddrldi_vis"): Likewise.
8990 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8991 ("bmaskdi_vis"): Likewise.
8992 ("bmasksi_vis"): Likewise.
8993 ("bshuffle<VM64:mode>_vis"): Likewise.
8994 ("cmask8<P:mode>_vis"): Likewise.
8995 ("cmask16<P:mode>_vis"): Likewise.
8996 ("cmask32<P:mode>_vis"): Likewise.
8997 ("pdistn<P:mode>_vis"): Likewise.
8998 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9000 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9002 * config/sparc/sparc.md ("subtype"): New insn attribute.
9003 ("*wrgsr_sp64"): Set insn subtype.
9004 ("*rdgsr_sp64"): Likewise.
9005 ("alignaddrsi_vis"): Likewise.
9006 ("alignaddrdi_vis"): Likewise.
9007 ("alignaddrlsi_vis"): Likewise.
9008 ("alignaddrldi_vis"): Likewise.
9009 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9010 ("fexpand_vis"): Likewise.
9011 ("fpmerge_vis"): Likewise.
9012 ("faligndata<VM64:mode>_vis"): Likewise.
9013 ("bshuffle<VM64:mode>_vis"): Likewise.
9014 ("cmask8<P:mode>_vis"): Likewise.
9015 ("cmask16<P:mode>_vis"): Likewise.
9016 ("cmask32<P:mode>_vis"): Likewise.
9017 ("fchksm16_vis"): Likewise.
9018 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9019 ("fmean16_vis"): Likewise.
9020 ("fp<plusminus_insn>64_vis"): Likewise.
9021 ("<plusminus_insn>v8qi3"): Likewise.
9022 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9023 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9024 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9025 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9026 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9027 ("*movqi_insn"): Likewise.
9028 ("*movhi_insn"): Likewise.
9029 ("*movsi_insn"): Likewise.
9030 ("movsi_pic_gotdata_op"): Likewise.
9031 ("*movdi_insn_sp32"): Likewise.
9032 ("*movdi_insn_sp64"): Likewise.
9033 ("movdi_pic_gotdata_op"): Likewise.
9034 ("*movsf_insn"): Likewise.
9035 ("*movdf_insn_sp32"): Likewise.
9036 ("*movdf_insn_sp64"): Likewise.
9037 ("*zero_extendhisi2_insn"): Likewise.
9038 ("*zero_extendqihi2_insn"): Likewise.
9039 ("*zero_extendqisi2_insn"): Likewise.
9040 ("*zero_extendqidi2_insn"): Likewise.
9041 ("*zero_extendhidi2_insn"): Likewise.
9042 ("*zero_extendsidi2_insn_sp64"): Likewise.
9043 ("ldfsr"): Likewise.
9044 ("prefetch_64"): Likewise.
9045 ("prefetch_32"): Likewise.
9046 ("tie_ld32"): Likewise.
9047 ("tie_ld64"): Likewise.
9048 ("*tldo_ldub_sp32"): Likewise.
9049 ("*tldo_ldub1_sp32"): Likewise.
9050 ("*tldo_ldub2_sp32"): Likewise.
9051 ("*tldo_ldub_sp64"): Likewise.
9052 ("*tldo_ldub1_sp64"): Likewise.
9053 ("*tldo_ldub2_sp64"): Likewise.
9054 ("*tldo_ldub3_sp64"): Likewise.
9055 ("*tldo_lduh_sp32"): Likewise.
9056 ("*tldo_lduh1_sp32"): Likewise.
9057 ("*tldo_lduh_sp64"): Likewise.
9058 ("*tldo_lduh1_sp64"): Likewise.
9059 ("*tldo_lduh2_sp64"): Likewise.
9060 ("*tldo_lduw_sp32"): Likewise.
9061 ("*tldo_lduw_sp64"): Likewise.
9062 ("*tldo_lduw1_sp64"): Likewise.
9063 ("*tldo_ldx_sp64"): Likewise.
9064 ("*mov<VM32:mode>_insn"): Likewise.
9065 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9066 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9068 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9070 * config/sparc/sparc.md ("type"): New insn type viscmp.
9071 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9073 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9074 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9075 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9076 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9078 ("n7_vis_logical_11cycle"): Likewise.
9079 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9080 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9081 * config/sparc/niagara.md ("niag_vis"): Likewise.
9082 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9083 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9085 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9087 * config/sparc/sparc.md: New instruction type `bmask'.
9088 (bmaskdi_vis): Use the `bmask' type.
9089 (bmasksi_vis): Likewise.
9090 * config/sparc/ultra3.md (us3_array): Likewise.
9091 * config/sparc/niagara7.md (n7_array): Likewise.
9092 * config/sparc/niagara4.md (n4_array): Likewise.
9093 * config/sparc/niagara2.md (niag2_vis): Likewise.
9094 (niag3_vis): Likewise.
9095 * config/sparc/niagara.md (niag_vis): Likewise.
9097 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9099 * ipa-comdats.c: Remove optimize check from gate.
9100 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9101 for functions not optimized.
9102 (ipa_fn_summary_read): Skip optimize check.
9103 (ipa_fn_summary_write): Likewise.
9104 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9106 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9108 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9109 (check_callers): Give up on caller which is not optimized.
9110 (inline_small_functions): Likewise.
9111 (ipa_inline): Do not give up when not optimizing.
9112 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9113 away unoptimizes cdtors.
9114 (whole_program_function_and_variable_visibility): Do
9115 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9116 * ipa.c (process_references): Do not check optimize.
9117 (symbol_table::remove_unreachable_nodes): Update optimize check.
9118 (set_writeonly_bit): Update optimize check.
9119 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9120 (pass_ipa_single_use::gate): Remove.
9122 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9124 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9125 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9126 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9127 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9128 permute_load, permute_store, adjust_extract, adjust_splat,
9129 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9130 replace_swap_with_copy, dump_swap_insn_table,
9131 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9132 recombine_lvx_pattern, recombine_stvx_pattern,
9133 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9134 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9135 to file rs6000-p8swap.c.
9136 * config/rs6000/rs6000-p8swap.c: New file.
9137 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9138 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9139 and rs6000*-*-* targets.
9141 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9143 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9145 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9147 * lto-wrapper.c (merge_and_complain): Do not merge
9148 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9149 fsigned_zeros, ftrapping_math, fwrapv.
9150 (append_compiler_options): Do not track these options.
9151 (append_linker_options): Likewie
9153 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9155 * cgraphunit.c (cgraph_node::finalize_function): When
9156 !flag_toplevel_reorde set no_reorder flag.
9157 (varpool_node::finalize_decl): Likewise.
9158 (symbol_table::compile): Drop no toplevel reorder path.
9160 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9162 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9163 edges; zero probability is not better than uninitialized.
9165 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9167 * asan.h (asan_sanitize_allocas_p): Declare.
9168 * asan.c (asan_sanitize_allocas_p): New function.
9169 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9170 (handle_builtin_alloca): Likewise.
9171 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9172 if !asan_sanitize_allocas_p.
9173 * params.def (asan-instrument-allocas): Add new option.
9174 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9175 * opts.c (common_handle_option): Disable allocas sanitization for
9178 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9180 * asan.c: Include gimple-fold.h.
9181 (get_last_alloca_addr): New function.
9182 (handle_builtin_stackrestore): Likewise.
9183 (handle_builtin_alloca): Likewise.
9184 (asan_emit_allocas_unpoison): Likewise.
9185 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9186 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9187 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9188 (instrument_builtin_call): Pass gimple iterator to
9189 get_mem_refs_of_builtin_call.
9190 (last_alloca_addr): New global.
9191 * asan.h (asan_emit_allocas_unpoison): Declare.
9192 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9193 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9194 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9195 if function calls alloca.
9196 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9197 * gimple-fold.h (replace_call_with_value): Declare.
9198 * internal-fn.c: Include asan.h.
9199 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9200 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9202 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9204 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9205 (C_SELFTEST_FLAGS): New.
9206 (CPP_SELFTEST_FLAGS): New.
9207 (SELFTEST_DEPS): New, from deps of s-selftest.
9208 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9209 (CPP_SELFTEST_DEPS): New.
9210 (selftest): Add dependency on s-selftest-c++.
9211 (s-selftest): Rename to...
9212 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9213 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9214 than SELFTEST_FLAGS.
9215 (selftest-gdb): Rename to...
9216 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9218 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9219 (selftest-valgrind): Rename to...
9220 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9222 (selftest-valgrind): Reintroduce as an alias for
9223 selftest-c-valgrind.
9224 (s-selftest-c++): New.
9225 (selftest-c++-gdb): New.
9226 (selftest-c++-valgrind): New.
9228 2017-07-06 Olivier Hainque <hainque@adacore.com>
9230 * gcc.c (process_command): When deciding if undefined variables
9231 should be ignored when processing specs, accept "gcc -v" as well.
9233 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9235 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9236 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9238 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9240 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9241 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9242 * config/arm/arm-tables.opt: Regenerate.
9243 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9245 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9247 2017-07-06 Carl Love <cel@us.ibm.com>
9249 * ChangeLog: Clean up from mid air collision
9251 2017-07-06 Carl Love <cel@us.ibm.com>
9253 * config/rs6000/rs6000-c.c: Add support for built-in functions
9254 vector signed int vec_subc (vector signed int, vector signed int);
9255 vector signed __int128 vec_subc (vector signed __int128,
9256 vector signed __int128);
9257 vector unsigned __int128 vec_subc (vector unsigned __int128,
9258 vector unsigned __int128);
9259 vector signed int vec_sube (vector signed int, vector signed int,
9261 vector unsigned int vec_sube (vector unsigned int,
9262 vector unsigned int,
9263 vector unsigned int);
9264 vector signed __int128 vec_sube (vector signed __int128,
9265 vector signed __int128,
9266 vector signed__int128);
9267 vector unsigned __int128 vec_sube (vector unsigned __int128,
9268 vector unsigned __int128,
9269 vector unsigned __int128);
9270 vector signed int vec_subec (vector signed int, vector signed int,
9272 vector unsigned int vec_subec (vector unsigned int,
9273 vector unsigned int,
9274 vector unsigned int);
9275 vector signed __int128 vec_subec (vector signed __int128,
9276 vector signed __int128,
9277 vector signed__int128);
9278 vector unsigned __int128 vec_subec (vector unsigned __int128,
9279 vector unsigned __int128,
9280 vector unsigned __int128);
9281 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9282 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9283 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9284 BU_ALTIVEC_OVERLOAD_X definitions.
9285 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9286 * doc/extend.texi: Update the built-in documentation file for the new
9289 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9292 * diagnostic-show-locus.c (layout::layout): Use start and finish
9293 spelling location for the start and finish of each range.
9294 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9295 Add unused aspect param.
9296 * input.c (expand_location_1): Add "aspect" param, and use it
9297 to access the correct part of the location.
9298 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9300 (expand_location_to_spelling_point): Likewise.
9301 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9302 param, and pass it to expand_location_1.
9304 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9306 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9307 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9308 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9309 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9310 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9311 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9312 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9313 _mm_maskz_getmant_ss): New intrinsics.
9314 (__builtin_ia32_getexpss128_mask): Changed to ...
9315 __builtin_ia32_getexpss128_round ... this.
9316 (__builtin_ia32_getexpsd128_mask): Changed to ...
9317 __builtin_ia32_getexpsd128_round ... this.
9318 * config/i386/i386-builtin-types.def
9319 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9320 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9321 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9322 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9323 __builtin_ia32_getmantss_mask_round): New builtins.
9324 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9325 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9326 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9327 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9328 * config/i386/sse.md
9329 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9330 avx512f_sgetexp<mode><mask_scalar_name>
9331 <round_saeonly_scalar_name> ... this.
9332 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9333 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9334 vgetexp<ssescalarmodesuffix>
9335 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9336 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9337 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9338 avx512f_vgetmant<mode><mask_scalar_name>
9339 <round_saeonly_scalar_name> ... this.
9340 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9341 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9342 vgetmant<ssescalarmodesuffix>
9343 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9344 %0<mask_scalar_operand4>, %1, %2
9345 <round_saeonly_scalar_mask_op4>, %3} ... this.
9346 * config/i386/subst.md (mask_scalar_operand4,
9347 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9348 round_saeonly_scalar_nimm_predicate): New subst attributes.
9350 2017-07-06 Julia Koval <julia.koval@intel.com>
9352 * config/i386/i386.c (ix86_erase_embedded_rounding):
9353 Remove code for old rounding pattern.
9355 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9357 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9359 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9361 * doc/sourcebuild.texi (Test Directives, Variants of
9362 dg-require-support): Add documentation for dg-require-stack-check.
9364 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9366 * config/i386/subst.md (mask_scalar, round_scalar,
9367 round_saeonly_scalar): New meta-templates.
9368 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9369 round_scalar_mask_operand3, round_scalar_mask_op3,
9370 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9371 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9372 round_saeonly_scalar_constraint,
9373 round_saeonly_scalar_prefix): New subst attribute.
9374 * config/i386/sse.md
9375 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9376 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9377 <round_scalar_name> ... this.
9378 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9379 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9380 <round_scalar_name> ... this.
9381 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9382 <sse>_vm<code><mode>3<mask_scalar_name>
9383 <round_saeonly_scalar_name> ... this.
9384 (v<plusminus_mnemonic><ssescalarmodesuffix>
9385 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9386 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9387 v<plusminus_mnemonic><ssescalarmodesuffix>
9388 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9389 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9390 (v<multdiv_mnemonic><ssescalarmodesuffix>
9391 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9392 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9393 v<multdiv_mnemonic><ssescalarmodesuffix>
9394 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9395 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9396 (v<maxmin_float><ssescalarmodesuffix>
9397 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9398 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9399 v<maxmin_float><ssescalarmodesuffix>
9400 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9401 %0<mask_scalar_operand3>, %1, %<iptr>2
9402 <round_saeonly_scalar_mask_op3>} ... this.
9404 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9406 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9407 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9409 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9410 Alan Hayward <alan.hayward@arm.com>
9411 David Sherwood <david.sherwood@arm.com>
9413 * combine.c (simplify_if_then_else): Remove "enum" before
9415 * compare-elim.c (can_eliminate_compare): Likewise.
9416 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9418 (aarch64_lookup_simd_builtin_type): Likewise.
9419 (aarch64_simd_builtin_type): Likewise.
9420 (aarch64_init_simd_builtin_types): Likewise.
9421 (aarch64_simd_expand_args): Likewise.
9422 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9424 (aarch64_reverse_mask): Likewise.
9425 (aarch64_simd_emit_reg_reg_move): Likewise.
9426 (aarch64_gen_adjusted_ldpstp): Likewise.
9427 (aarch64_ccmp_mode_to_code): Likewise.
9428 (aarch64_operands_ok_for_ldpstp): Likewise.
9429 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9430 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9432 (aarch64_min_divisions_for_recip_mul): Likewise.
9433 (aarch64_reassociation_width): Likewise.
9434 (aarch64_get_condition_code_1): Likewise.
9435 (aarch64_simd_emit_reg_reg_move): Likewise.
9436 (aarch64_simd_attr_length_rglist): Likewise.
9437 (aarch64_reverse_mask): Likewise.
9438 (aarch64_operands_ok_for_ldpstp): Likewise.
9439 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9440 (aarch64_gen_adjusted_ldpstp): Likewise.
9441 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9443 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9444 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9445 (arm_lookup_simd_builtin_type): Likewise.
9446 (arm_simd_builtin_type): Likewise.
9447 (arm_init_simd_builtin_types): Likewise.
9448 (arm_expand_builtin_args): Likewise.
9449 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9450 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9451 (ft32_setup_incoming_varargs): Likewise.
9452 (ft32_function_arg): Likewise.
9453 (ft32_function_arg_advance): Likewise.
9454 (ft32_pass_by_reference): Likewise.
9455 (ft32_arg_partial_bytes): Likewise.
9456 (ft32_valid_pointer_mode): Likewise.
9457 (ft32_addr_space_pointer_mode): Likewise.
9458 (ft32_addr_space_legitimate_address_p): Likewise.
9459 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9461 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9462 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9463 (iamcu_alignment): Likewise.
9464 (canonicalize_vector_int_perm): Likewise.
9465 (ix86_noce_conversion_profitable_p): Likewise.
9466 (ix86_mpx_bound_mode): Likewise.
9467 (ix86_operands_ok_for_move_multiple): Likewise.
9468 * config/microblaze/microblaze-protos.h
9469 (microblaze_expand_conditional_branch_reg): Likewise.
9470 * config/microblaze/microblaze.c
9471 (microblaze_expand_conditional_branch_reg): Likewise.
9472 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9474 (rs6000_reassociation_width): Likewise.
9475 (rs6000_invalid_binary_op): Likewise.
9476 (fusion_p9_p): Likewise.
9477 (emit_fusion_p9_load): Likewise.
9478 (emit_fusion_p9_store): Likewise.
9479 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9481 (riscv_hard_regno_mode_ok_p): Likewise.
9482 (riscv_address_insns): Likewise.
9483 (riscv_split_symbol): Likewise.
9484 (riscv_legitimize_move): Likewise.
9485 (riscv_function_value): Likewise.
9486 (riscv_hard_regno_nregs): Likewise.
9487 (riscv_expand_builtin): Likewise.
9488 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9489 (riscv_build_integer): Likewise.
9490 (riscv_split_integer): Likewise.
9491 (riscv_legitimate_constant_p): Likewise.
9492 (riscv_cannot_force_const_mem): Likewise.
9493 (riscv_regno_mode_ok_for_base_p): Likewise.
9494 (riscv_valid_base_register_p): Likewise.
9495 (riscv_valid_offset_p): Likewise.
9496 (riscv_valid_lo_sum_p): Likewise.
9497 (riscv_classify_address): Likewise.
9498 (riscv_legitimate_address_p): Likewise.
9499 (riscv_address_insns): Likewise.
9500 (riscv_load_store_insns): Likewise.
9501 (riscv_force_binary): Likewise.
9502 (riscv_split_symbol): Likewise.
9503 (riscv_force_address): Likewise.
9504 (riscv_legitimize_address): Likewise.
9505 (riscv_move_integer): Likewise.
9506 (riscv_legitimize_const_move): Likewise.
9507 (riscv_legitimize_move): Likewise.
9508 (riscv_address_cost): Likewise.
9509 (riscv_subword): Likewise.
9510 (riscv_output_move): Likewise.
9511 (riscv_canonicalize_int_order_test): Likewise.
9512 (riscv_emit_int_order_test): Likewise.
9513 (riscv_function_arg_boundary): Likewise.
9514 (riscv_pass_mode_in_fpr_p): Likewise.
9515 (riscv_pass_fpr_single): Likewise.
9516 (riscv_pass_fpr_pair): Likewise.
9517 (riscv_get_arg_info): Likewise.
9518 (riscv_function_arg): Likewise.
9519 (riscv_function_arg_advance): Likewise.
9520 (riscv_arg_partial_bytes): Likewise.
9521 (riscv_function_value): Likewise.
9522 (riscv_pass_by_reference): Likewise.
9523 (riscv_setup_incoming_varargs): Likewise.
9524 (riscv_print_operand): Likewise.
9525 (riscv_elf_select_rtx_section): Likewise.
9526 (riscv_save_restore_reg): Likewise.
9527 (riscv_for_each_saved_reg): Likewise.
9528 (riscv_register_move_cost): Likewise.
9529 (riscv_hard_regno_mode_ok_p): Likewise.
9530 (riscv_hard_regno_nregs): Likewise.
9531 (riscv_class_max_nregs): Likewise.
9532 (riscv_memory_move_cost): Likewise.
9533 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9534 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9535 (rl78_addr_space_address_mode): Likewise.
9536 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9538 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9539 (rs6000_reassociation_width): Likewise.
9540 (rs6000_invalid_binary_op): Likewise.
9541 (fusion_p9_p): Likewise.
9542 (emit_fusion_p9_load): Likewise.
9543 (emit_fusion_p9_store): Likewise.
9544 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9545 (ok_for_simple_move_operands): Likewise.
9546 (ok_for_simple_move_strict_operands): Likewise.
9547 (ok_for_simple_arith_logic_operands): Likewise.
9548 (visium_legitimize_reload_address): Likewise.
9549 (visium_select_cc_mode): Likewise.
9550 (output_cbranch): Likewise.
9551 (visium_split_double_move): Likewise.
9552 (visium_expand_copysign): Likewise.
9553 (visium_expand_int_cstore): Likewise.
9554 (visium_expand_fp_cstore): Likewise.
9555 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9556 (visium_function_arg): Likewise.
9557 (visium_function_arg_advance): Likewise.
9558 (visium_libcall_value): Likewise.
9559 (visium_setup_incoming_varargs): Likewise.
9560 (visium_legitimate_constant_p): Likewise.
9561 (visium_legitimate_address_p): Likewise.
9562 (visium_legitimize_address): Likewise.
9563 (visium_secondary_reload): Likewise.
9564 (visium_register_move_cost): Likewise.
9565 (visium_memory_move_cost): Likewise.
9566 (prepare_move_operands): Likewise.
9567 (ok_for_simple_move_operands): Likewise.
9568 (ok_for_simple_move_strict_operands): Likewise.
9569 (ok_for_simple_arith_logic_operands): Likewise.
9570 (visium_function_value_1): Likewise.
9571 (rtx_ok_for_offset_p): Likewise.
9572 (visium_legitimize_reload_address): Likewise.
9573 (visium_split_double_move): Likewise.
9574 (visium_expand_copysign): Likewise.
9575 (visium_expand_int_cstore): Likewise.
9576 (visium_expand_fp_cstore): Likewise.
9577 (visium_split_cstore): Likewise.
9578 (visium_select_cc_mode): Likewise.
9579 (visium_split_cbranch): Likewise.
9580 (output_cbranch): Likewise.
9581 (visium_print_operand_address): Likewise.
9582 * expmed.c (flip_storage_order): Likewise.
9583 * expmed.h (emit_cstore): Likewise.
9584 (flip_storage_order): Likewise.
9585 * genrecog.c (validate_pattern): Likewise.
9586 * hsa-gen.c (gen_hsa_addr): Likewise.
9587 * internal-fn.c (expand_arith_overflow): Likewise.
9588 * ira-color.c (allocno_copy_cost_saving): Likewise.
9589 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9590 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9591 (process_invariant_for_inheritance): Likewise.
9592 * lra-eliminations.c (move_plus_up): Likewise.
9593 * omp-low.c (lower_oacc_reductions): Likewise.
9594 * simplify-rtx.c (simplify_subreg): Likewise.
9595 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9596 (TARGET_CHKP_BOUND_MODE): Likewise..
9597 * targhooks.c (default_chkp_bound_mode): Likewise.
9598 (default_setup_incoming_vararg_bounds): Likewise.
9599 * targhooks.h (default_chkp_bound_mode): Likewise.
9600 (default_setup_incoming_vararg_bounds): Likewise.
9601 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9602 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9603 (have_whole_vector_shift): Likewise.
9604 * tree-vect-stmts.c (vectorizable_load): Likewise.
9605 * doc/tm.texi: Regenerate.
9607 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9609 Graceful degrade if Binutils PR21472 is not available.
9612 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9613 .rodata in flash test fails.
9614 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9615 * confgure: Regenerate.
9616 * config.in: Regenerate.
9617 * config/avr/avr.c (avr_asm_named_section)
9618 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9619 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9620 (avr_asm_init_sections): Same.
9622 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9624 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9625 (fma<VH:mode>4_intrinsic): Likewise.
9626 (*fmsub<VCVTF:mode>4): Likewise.
9627 (*fmsub<VH:mode>4_intrinsic): Likewise.
9629 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9632 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9633 Don't depend on "optimize > 0".
9634 (out_movhi_r_mr, out_movqi_mr_r): Same.
9635 (out_movhi_mr_r, out_movqi_r_mr): Same.
9636 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9637 io_address_operand on "optimize > 0".
9639 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9641 * tree-loop-distribution.c: Add general explanantion on the pass.
9642 (generate_loops_for_partition): Mark distributed loop.
9643 (pg_add_dependence_edges): New parameter. Handle alias data
9644 dependence specially and record it in the parameter if asked.
9645 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9646 (init_partition_graph_vertices, add_partition_graph_edge): New.
9647 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9648 (free_partition_graph_vdata, build_partition_graph): New.
9649 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9650 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9651 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9652 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9653 (version_for_distribution_p, finalize_partitions): New.
9654 (distribute_loop): Handle alias data dependence specially. Factor
9655 out loop fusion code as functions and call these functions.
9657 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9659 * tree-loop-distribution.c (classify_partition): New parameter and
9660 better handle reduction statement.
9661 (rdg_build_partitions): Revise comment.
9662 (distribute_loop): Compute statements in all partitions and pass it
9663 to classify_partition.
9665 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9667 * tree-loop-distribution.c (enum partition_type): New.
9668 (struct partition): New field type.
9669 (partition_merge_into): Add parameter. Update partition type.
9670 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9671 (build_rdg_partition_for_vertex): Compute partition type.
9672 (rdg_build_partitions): Dump partition type.
9673 (distribute_loop): Update calls to partition_merge_into.
9675 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9677 * tree-loop-distribution.c (struct ddr_hasher): New.
9678 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9680 (classify_partition): Call get_data_dependence.
9681 (pg_add_dependence_edges): Ditto.
9682 (distribute_loop): Release data dependence hash table.
9684 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9686 * tree-loop-distribution.c (ref_base_address): Delete.
9687 (similar_memory_accesses): Rename ...
9688 (share_memory_accesses): ... to this. Check if partitions access
9689 the same memory reference.
9690 (distribute_loop): Call share_memory_accesses.
9692 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9694 * tree-loop-distribution.c (struct partition): New field recording
9696 (partition_alloc, partition_free): Init and release data refs.
9697 (partition_merge_into): Merge data refs.
9698 (build_rdg_partition_for_vertex): Collect data refs for partition.
9699 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9701 (distribute_loop): Remve data refs from vertice data of partition
9704 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9706 * tree-loop-distribution.c (params.h): Include header file.
9707 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9708 (datarefs_vec): New global var.
9709 (create_rdg_vertices): Use datarefs_vec directly.
9710 (free_rdg): Don't free data references.
9711 (build_rdg): Update use. Don't free data references.
9712 (distribute_loop): Compute global variable for data references.
9713 Bail out if there are too many data references.
9715 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9717 * tree-loop-distribution.c (loop_nest): New global var.
9718 (build_rdg): Use loop directly, rather than loop nest.
9719 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9721 (distribute_loop): Compute global variable loop nest. Update use.
9723 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9725 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9726 (partition_merge_into): New parameter. Dump reason for fusion.
9727 (distribute_loop): Update use of partition_merge_into.
9729 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9731 * tree-loop-distribution.c (bb_top_order_index): New.
9732 (bb_top_order_index_size, bb_top_order_cmp): New.
9733 (stmts_from_loop): Use topological order.
9734 (pass_loop_distribution::execute): Compute and release topological
9735 order for basic blocks.
9737 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9739 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9742 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9744 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9745 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9746 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9747 * internal-fn.def (LOOP_DIST_ALIAS): New.
9748 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9749 (fold_loop_internal_call): ... this.
9750 (vect_loop_dist_alias_call): New function.
9751 (set_uid_loop_bbs): Call fold_loop_internal_call.
9752 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9755 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9758 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9759 Require dead FLAGS_REG at the beginning of a peephole.
9761 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9764 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9765 arguments in the call to __builtin_ia32_sbb_u32.
9766 (_subborrow_u64): Swap _X and _Y arguments in the call to
9767 __builtin_ia32_sbb_u64.
9769 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9772 * tree-vrp.c (compare_assert_loc): Turn into a function template
9773 with stable template parameter. Only test if a->e is NULL,
9774 !a->e == !b->e has been verified already. Use e == NULL or
9775 e != NULL instead of e or ! e tests. If stable is true, don't use
9776 iterative_hash_expr, on the other side allow a or b or both NULL
9777 and sort the NULLs last.
9778 (process_assert_insertions): Sort using compare_assert_loc<false>
9779 instead of compare_assert_loc, later sort using
9780 compare_assert_loc<true> before calling process_assert_insertions_for
9781 in a loop. Use break instead of continue once seen NULL pointer.
9783 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9785 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9786 Cortex-R7 and Cortex-R8 processors.
9788 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9790 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9791 uninitialized while src is not.
9793 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9795 * common/config/arm/arm-common.c: Adjust include path for
9797 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9798 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9799 (arm-cpu-data.h): Likewise.
9800 (arm-cpu-cdata.h): Likewise.
9801 * config/arm/arm-cpu.h: Delete.
9802 * config/arm/arm-cpu-cdata.h: Delete.
9803 * config/arm/arm-cpu-data.h: Delete.
9805 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9807 * config/arm/arm-cpus.in (cortex-a55): New.
9808 (cortex-a75): Likewise.
9809 (cortex-a75.cortex-a55): Likewise.
9810 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9812 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9813 * config/arm/arm-cpu-cdata.h: Regenerate.
9814 * config/arm/arm-cpu-data.h: Regenerate.
9815 * config/arm/arm-cpu.h: Regenerate.
9816 * config/arm/arm-tables.opt: Regenerate.
9817 * config/arm/arm-tune.md: Regenerate.
9819 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9821 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9823 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9825 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9828 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9830 PR tree-optimization/81292
9831 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9832 full_string_p, also call adjust_related_strinfos if the adjustment
9833 is simple, otherwise invalidate related strinfos.
9835 2017-07-04 Martin Liska <mliska@suse.cz>
9838 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9839 newly created variable as DECL_IGNORED_P.
9841 2017-07-04 Martin Liska <mliska@suse.cz>
9844 * ipa-inline.c (inline_small_functions):
9845 Use xstrdup_for_dump.
9847 2017-07-04 Tom de Vries <tom@codesourcery.com>
9849 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9851 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9854 * config/darwin.c (darwin_function_switched_text_sections):
9857 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9859 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9861 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9863 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9865 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9867 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9868 min_profitable_iters, and th as inclusive lower bounds.
9869 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9870 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9871 for min_profitable_iters and min_profitable_estimate.
9872 (vect_transform_loop): Treat th as an inclusive lower bound.
9873 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9875 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9878 * config/darwin.c (darwin_function_switched_text_sections):
9879 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9880 in two pieces, and suppress the use of buf.
9882 2017-07-03 Nathan Sidwell <nathan@acm.org>
9884 * hash-table.h (hash_table_mod1): Fix indentation.
9886 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9889 * predict.c (force_edge_cold): Be more careful about propagation
9891 * profile-count.h (profile_probability::guessed,
9892 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9894 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9896 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9898 * doc/invoke.texi (rcpc architecture extension): Document it.
9900 2017-07-03 Richard Biener <rguenther@suse.de>
9902 PR tree-optimization/60510
9903 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9904 the scalar reduction PHI and use it.
9905 (vectorizable_reduction): Properly guard the single_defuse_cycle
9906 path for non-SLP reduction chains where we cannot use it.
9907 Rework reduc_def/index and vector type deduction. Rework
9908 vector operand gathering during reduction op code-gen.
9909 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9910 chains dissolve the chain and leave it to non-SLP reduction
9913 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9915 * tree-data-ref.h (dr_alignment): Declare.
9916 * tree-data-ref.c (dr_alignment): New function.
9917 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9918 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9920 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9922 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9924 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9925 and base_misalignment fields.
9926 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9927 * tree-data-ref.c: Include builtins.h.
9928 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9929 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9930 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9931 * tree-vect-data-refs.c: Include tree-cfg.h.
9932 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9933 fields instead of calculating an alignment here.
9934 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9935 innermost_loop_behavior fields.
9937 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9939 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9941 (DR_STEP_ALIGNMENT): New macro.
9942 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9943 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9944 (create_data_ref): Print it.
9945 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9946 to tell whether the step preserves vector (mis)alignment.
9947 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9948 Move the check for an integer step and generalise to all INTEGER_CST.
9949 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9950 Print the outer step alignment.
9952 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9954 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9955 with offset_alignment.
9956 (DR_ALIGNED_TO): Delete.
9957 (DR_OFFSET_ALIGNMENT): New macro.
9958 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9959 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9960 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9961 (create_data_ref): Likewise.
9962 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9963 (vect_analyze_data_refs): Likewise.
9964 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9965 creating dummy innermost behavior.
9967 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9969 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9970 with a "innermost_loop_behavior *" and refeence tree.
9971 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9972 (create_data_ref): Update call accordingly.
9973 * tree-predcom.c (find_looparound_phi): Likewise.
9975 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9977 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9978 fields with dr_wrt_vec_loop.
9979 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9980 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9981 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9982 (vect_dr_behavior): New function.
9983 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9984 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9985 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9986 track whether the step preserves the misalignment.
9987 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9988 Use vect_dr_behavior.
9989 (vect_setup_realignment): Update call accordingly.
9990 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9991 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9992 call to vect_create_addr_base_for_vector_ref.
9993 (vect_create_cond_for_align_checks): Likewise.
9994 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9995 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9996 (vect_recog_mask_conversion_pattern): Likewise.
9997 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9998 (new_stmt_vec_info): Remove redundant zeroing.
10000 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10002 * common/config/arm/arm-common.c (arm_be8_option): New function.
10003 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10004 (ISA_ARMv6): Add isa_bit_be8.
10005 * config/arm/arm.h (arm_be8_option): Add prototype.
10006 (BE8_SPEC_FUNCTION): New define.
10007 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10008 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10009 (mlittle-endian): Similarly.
10010 (mbe8, mbe32): New options.
10011 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10012 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10014 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10016 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10018 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10020 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10021 (cleanup_tree_cfg_bb): Use it.
10022 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10024 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10026 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10029 * loop-doloop.c (add_test): Update profile.
10031 2017-07-03 Martin Liska <mliska@suse.cz>
10034 * sanopt.c (rewrite_usage_of_param): New function.
10035 (sanitize_rewrite_addressable_params): Likewise.
10036 (pass_sanopt::execute): Call rewrite_usage_of_param.
10038 2017-07-03 Richard Biener <rguenther@suse.de>
10040 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10041 back to using VIEW_CONVERT_EXPR.
10043 2017-07-03 Martin Liska <mliska@suse.cz>
10046 * doc/extend.texi: Document when a resolver function is
10047 generated for target_clones.
10049 2017-07-03 Martin Liska <mliska@suse.cz>
10051 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10052 and shadow memory of auto variables which are subject of
10053 use-after-scope sanitization.
10054 (asan_expand_mark_ifn): Add do set only when is_poison.
10056 2016-07-03 Richard Biener <rguenther@suse.de>
10058 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10060 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10061 (vectorizable_reduction): Perform reduction PHI creation when
10062 visiting a reduction PHI and adjust and simplify code generation
10063 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10064 (vect_transform_loop): Visit reduction PHIs.
10065 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10066 defs into the SLP tree.
10067 (vect_build_slp_tree): Reduction defs terminate the recursion.
10068 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10070 (vect_get_vec_defs_for_stmt_copy): Export.
10071 (vect_get_vec_defs): Likewise.
10072 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10074 (vect_get_vec_defs_for_stmt_copy): Declare.
10075 (vect_get_vec_defs): Likewise.
10077 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10079 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10080 parameter with a "loop" parameter and use it instead of the
10081 loop containing DR_STMT. Don't check simple_iv when doing
10082 BB analysis. Describe the two analysis modes in the comment.
10084 2017-07-03 Tom de Vries <tom@codesourcery.com>
10086 PR tree-optimization/69468
10087 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10088 (find_same_succ_bb): Handle ignore_edge_flags.
10090 2017-07-03 Tom de Vries <tom@codesourcery.com>
10092 PR tree-optimization/81192
10093 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10095 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10097 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10099 2017-07-03 Tom de Vries <tom@codesourcery.com>
10101 PR tree-optimization/81192
10102 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10103 BB_SAME_SUCC (bb) == NULL.
10105 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10107 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10110 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10112 * dumpfile.c: Include profile-count.h
10113 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10115 (insert_cond_bb): Update profile.
10116 * tree-cfg.h (insert_cond_bb): Update prototype.
10117 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10118 * tree-dump.c: Do not include tree-cfg.
10120 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10122 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10124 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10126 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10129 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10131 * tree-complex.c (expand_complex_div_wide): update profile.
10133 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10134 Alan Hayward <alan.hayward@arm.com>
10135 David Sherwood <david.sherwood@arm.com>
10137 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10138 (CORETYPES_H): New define.
10139 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10140 (insn-modes-inline.h, s-modes-inline-h): New rules.
10141 (generated_files): Add insn-modes-inline.h.
10142 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10143 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10144 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10145 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10146 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10147 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10148 (build/gencodes.o, build/genconditions.o): Likewise.
10149 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10150 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10151 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10152 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10153 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10154 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10155 * coretypes.h: Include everything up to real.h for generators.
10156 Include insn-modes.h first. Include wide-int-print.h after
10157 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10158 * machmode.h: Don't include insn-modes.h here.
10159 * function-tests.c: Remove includes of signop.h, machmode.h,
10160 double-int.h and wide-int.h.
10162 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10164 * optc-save-gen.awk: Likewise.
10165 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10166 * godump.c: Remove include of wide-int-print.h.
10167 * pretty-print.h: Likewise.
10168 * wide-int-print.cc: Likewise.
10169 * wide-int.cc: Likewise.
10170 * hash-map-tests.c: Remove include of signop.h.
10171 * hash-set-tests.c: Likewise.
10172 * rtl-tests.c: Likewise.
10173 * mkconfig.sh: Remove include of machmode.h.
10174 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10176 (emit_insn_modes_inline_h): ...this new function. Emit the code
10177 into an insn-modes-inline.h header file, adding appropriate
10178 include guards and end comments.
10179 (emit_insn_modes_c_header): Remove include of machmode.h.
10180 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10182 (main): Handle -i flag and call emit_insn_modes_inline_h when
10185 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10187 * tree-ssa-strlen.c (strinfo): Rename the length field to
10188 nonzero_chars. Add a full_string_p field.
10189 (compare_nonzero_chars, zero_length_string_p): New functions.
10190 (get_addr_stridx): Add an offset_out parameter.
10191 Use compare_nonzero_chars.
10192 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10193 (new_strinfo): Update after above changes to strinfo.
10194 (set_endptr_and_length): Set full_string_p.
10195 (get_string_length): Update after above changes to strinfo.
10196 (unshare_strinfo): Update call to new_strinfo.
10197 (maybe_invalidate): Likewise.
10198 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10199 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10200 as a uhwi instead of an shwi. Update after above changes to
10201 strinfo and new_strinfo.
10202 (zero_length_string): Assert that chainsi contains full strings.
10203 Use zero_length_string_p. Update call to new_strinfo.
10204 (adjust_related_strinfos): Update after above changes to strinfo.
10205 Copy full_string_p from origsi.
10206 (adjust_last_stmt): Use zero_length_string_p.
10207 (handle_builtin_strlen): Update after above changes to strinfo and
10208 new_strinfo. Install the lhs as the string length if the previous
10209 entry didn't describe a full string.
10210 (handle_builtin_strchr): Update after above changes to strinfo
10212 (handle_builtin_strcpy): Likewise.
10213 (handle_builtin_strcat): Likewise.
10214 (handle_builtin_malloc): Likewise.
10215 (handle_pointer_plus): Likewise.
10216 (handle_builtin_memcpy): Likewise. Track nonzero characters
10217 that aren't necessarily followed by a nul terminator.
10218 (handle_char_store): Likewise.
10220 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10222 PR tree-optimization/80769
10223 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10224 for malloc and calloc. Document the new invariant that all related
10225 strinfos have delayed lengths or none do.
10226 (verify_related_strinfos): Move earlier in file.
10227 (set_endptr_and_length): New function, split out from...
10228 (get_string_length): ...here. Also set the lengths of related
10230 (zero_length_string): Assert that chainsi has known (rather than
10232 (adjust_related_strinfos): Likewise.
10234 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10236 PR tree-optimization/81136
10237 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10238 assert that two references with the same misalignment have the same
10239 compile-time misalignment if those compile-time misalignments
10242 2017-07-01 Andi Kleen <ak@linux.intel.com>
10244 * print-tree.c (print_node): Print all attributes.
10246 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10248 * cfg.c (scale_bbs_frequencies): New function.
10249 * cfg.h (scale_bbs_frequencies): Declare it.
10250 * cfgloopanal.c (single_likely_exit): Cleanup.
10251 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10253 (scale_loop_profile): Likewise.
10254 (loop_version): Likewise.
10255 (create_empty_loop_on_edge): Update.
10256 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10257 scale_loop_frequencies, scale_loop_profile, loopify,
10258 loop_version): Update prototypes.
10259 * modulo-sched.c (sms_schedule): Update.
10260 * predict.c (unlikely_executed_edge_p): Also check probability.
10261 (probably_never_executed_edge_p): Fix typo.
10262 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10263 * tree-parloops.c (gen_parallel_loop): Update.
10264 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10265 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10266 * tree-ssa-loop-split.c (split_loop): Update.
10267 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10268 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10269 (vect_loop_versioning): Update.
10270 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10272 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10274 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10276 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10278 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10279 to keep profile consistent.
10281 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10283 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10284 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10285 * profile-count.h (max_safe_multiplier): Make unsigned.
10286 (profile_count::guessed_zero): New.
10288 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10290 * bb-reorder.c (fix_up_crossing_landing_pad,
10291 fix_crossing_conditional_branches): Use make_single_succ_edge
10292 to keep profile consistent.
10294 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10296 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10299 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10302 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10303 the right scopes, make sure cond_jump isn't preserved between multiple
10304 iterations. Search for fallthru edge whenever there are 3+ edges and
10305 use find_fallthru_edge for it.
10307 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10309 Patch by Alexander Monakov <amonakov@ispras.ru>
10310 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10311 probabilities consistently.
10313 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10315 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10316 emit_cmp_and_jump_insns.
10318 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10321 * tree-inline.c (expand_call_inline): Combine profile statuses.
10323 2017-06-30 Andrew Pinski <apinski@cavium.com>
10325 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10326 fold_stmt returned true.
10328 2017-06-30 Nathan Sidwell <nathan@acm.org>
10330 * ggc.h (empty_string): Delete.
10331 * cfgexpand.c (expand_asm_stmt): Use plain "".
10332 * optabs.c (expand_asm_memory_barrier): Likewise.
10333 * stringpool.c (empty_string): Delete.
10334 (digit_vector, digit_string): Delete.
10335 (ggc_alloc_string): Use plain "", don't optimize single digit
10336 strings. Use ggc_alloc_atomic.
10338 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10340 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10341 comparison set and one other set, use the cost of the non-comparison
10344 2017-06-30 Nathan Sidwell <nathan@acm.org>
10346 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10349 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10351 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10352 loops. Remove now unneeded calls to gimple_switch_set_label() that
10353 just set removed labels to NULL_TREE.
10355 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10357 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10358 (set_range_info): ...here. Only call set_range_info_raw if domain
10360 (set_nonzero_bits): Call set_range_info_raw.
10361 * tree-ssanames.h (set_range_info_raw): New.
10363 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10366 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10367 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10368 of nonimmediate_operand and <store_mask_constraint> instead of m
10369 for the input operand. For V8FI iterator, always split if input
10370 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10371 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10372 <store_mask_predicate> instead of register_operand and
10373 <store_mask_constraint> instead of v for the input operand. Make
10374 sure both operands aren't MEMs for if not <mask_applied>.
10376 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10378 * lto-wrapper.c (copy_file) Close both file descriptors before
10381 2017-06-30 Martin Liska <mliska@suse.cz>
10384 * multiple_target.c (create_dispatcher_calls): Make ifunc
10385 also for function that don't have calls or are not referenced.
10387 2017-06-30 Richard Biener <rguenther@suse.de>
10389 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10390 analyze the first scalar stmt. Move vector type computation
10391 for the BB case here from ...
10392 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10393 live operation processing in the SLP case properly.
10395 2017-06-30 Richard Biener <rguenther@suse.de>
10397 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10399 2017-06-30 Martin Liska <mliska@suse.cz>
10402 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10403 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10405 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10407 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10410 2017-06-30 Richard Biener <rguenther@suse.de>
10412 PR tree-optimization/81249
10413 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10414 condition reduction result to original scalar type.
10416 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10418 * profile-count.h (enum profile_quality): Fix typos and whitespace
10421 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10423 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10424 type for branch probabilities.
10426 2017-06-29 Julian Brown <julian@codesourcery.com>
10427 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10429 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10430 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10431 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10432 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10434 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10436 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10437 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10438 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10439 CC usage from generic code to here.
10440 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10441 CC usage into the target macros.
10443 2017-06-29 Maya Rashish <coypu@sdf.org>
10445 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10448 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10450 * arm/arm-builtins.c: Include profile-count.h
10451 * except.c (sjlj_emit_function_enter): Use
10452 profile_probability::unlikely.
10454 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10456 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10457 and tocrel_offset be pointer args rather than implicitly using
10459 (legitimate_constant_pool_address_p, rs6000_emit_move,
10460 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10461 tocrel_offset and use in toc_relative_expr_p call.
10462 (print_operand, print_operand_address): Use static tocrel_base_oac
10463 and tocrel_offset_oac.
10464 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10467 2017-06-29 Maya Rashish <coypu@sdf.org>
10469 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10471 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10473 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10474 objects, take into account only the alignment of 'op0' and 'mode1' if
10477 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10479 * ccmp.c (ccmp_tree_comparison_p): New function.
10480 (ccmp_candidate_p): Update to use above function.
10481 (get_compare_parts): New function.
10482 (expand_ccmp_next): Update to use new functions.
10483 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10485 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10486 take mode as argument.
10487 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10488 * expr.c (expand_expr_real_1): Pass mode as argument.
10490 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10492 * combine.c (combine_instructions): Print insns to dump_file, together
10495 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10497 * asan.c (asan_emit_stack_protection): Update.
10498 (create_cond_insert_point): Update.
10499 * auto-profile.c (afdo_propagate_circuit): Update.
10500 * basic-block.h (struct edge_def): Turn probability to
10501 profile_probability.
10502 (EDGE_FREQUENCY): Update.
10503 * bb-reorder.c (find_traces_1_round): Update.
10504 (better_edge_p): Update.
10505 (sanitize_hot_paths): Update.
10506 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10507 (make_single_succ_edge): Update.
10508 (check_bb_profile): Update.
10509 (dump_edge_info): Update.
10510 (update_bb_profile_for_threading): Update.
10511 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10512 probabilitycount to 0.
10513 * cfgbuild.c (compute_outgoing_frequencies): Update.
10514 * cfgcleanup.c (try_forward_edges): Update.
10515 (outgoing_edges_match): Update.
10516 (try_crossjump_to_edge): Update.
10517 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10518 (expand_gimple_tailcall): Update.
10519 (construct_init_block): Use make_single_succ_edge.
10520 (construct_exit_block): Use make_single_succ_edge.
10521 * cfghooks.c (verify_flow_info): Update.
10522 (redirect_edge_succ_nodup): Update.
10523 (split_edge): Update.
10524 (account_profile_record): Update.
10525 * cfgloopanal.c (single_likely_exit): Update.
10526 * cfgloopmanip.c (scale_loop_profile): Update.
10527 (set_zero_probability): Remove.
10528 (duplicate_loop_to_header_edge): Update.
10529 * cfgloopmanip.h (loop_version): Update prototype.
10530 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10531 (force_nonfallthru_and_redirect): Update.
10532 (update_br_prob_note): Update.
10533 (rtl_verify_edges): Update.
10534 (purge_dead_edges): Update.
10535 (rtl_lv_add_condition_to_bb): Update.
10536 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10537 * cgraphunit.c (init_lowered_empty_function): Update.
10538 (cgraph_node::expand_thunk): Update.
10539 * cilk-common.c: Include profile-count.h
10540 * dojump.c (inv): Remove.
10541 (jumpifnot): Update.
10542 (jumpifnot_1): Update.
10543 (do_jump_1): Update.
10545 (do_jump_by_parts_greater_rtx): Update.
10546 (do_compare_rtx_and_jump): Update.
10547 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10548 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10549 * dwarf2cfi.c: Include profile-count.h
10550 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10551 (sjlj_emit_dispatch_table): Likewise.
10552 * explow.c: Include profile-count.h
10553 * expmed.c (emit_store_flag_force): Update.
10554 (do_cmp_and_jump): Update.
10555 * expr.c (compare_by_pieces_d::generate): Update.
10556 (compare_by_pieces_d::finish_mode): Update.
10557 (emit_block_move_via_loop): Update.
10558 (store_expr_with_bounds): Update.
10559 (store_constructor): Update.
10560 (expand_expr_real_2): Update.
10561 (expand_expr_real_1): Update.
10562 * expr.h (try_casesi, try_tablejump): Update prototypes.
10563 * gimple-pretty-print.c (dump_probability): Update.
10564 (dump_profile): New.
10565 (dump_gimple_label): Update.
10566 (dump_gimple_bb_header): Update.
10567 * graph.c (draw_cfg_node_succ_edges): Update.
10568 * hsa-gen.c (convert_switch_statements): Update.
10569 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10570 (find_if_case_1): Update.
10571 (find_if_case_2): Update.
10572 * internal-fn.c (expand_arith_overflow_result_store): Update.
10573 (expand_addsub_overflow): Update.
10574 (expand_neg_overflow): Update.
10575 (expand_mul_overflow): Update.
10576 (expand_vector_ubsan_overflow): Update.
10577 * ipa-cp.c (good_cloning_opportunity_p): Update.
10578 * ipa-split.c (split_function): Use make_single_succ_edge.
10579 * ipa-utils.c (ipa_merge_profiles): Update.
10580 * loop-doloop.c (add_test): Update.
10581 (doloop_modify): Update.
10582 * loop-unroll.c (compare_and_jump_seq): Update.
10583 (unroll_loop_runtime_iterations): Update.
10584 * lra-constraints.c (lra_inheritance): Update.
10585 * lto-streamer-in.c (input_cfg): Update.
10586 * lto-streamer-out.c (output_cfg): Update.
10587 * mcf.c (adjust_cfg_counts): Update.
10588 * modulo-sched.c (sms_schedule): Update.
10589 * omp-expand.c (expand_omp_for_init_counts): Update.
10590 (extract_omp_for_update_vars): Update.
10591 (expand_omp_ordered_sink): Update.
10592 (expand_omp_for_ordered_loops): Update.
10593 (expand_omp_for_generic): Update.
10594 (expand_omp_for_static_nochunk): Update.
10595 (expand_omp_for_static_chunk): Update.
10596 (expand_cilk_for): Update.
10597 (expand_omp_simd): Update.
10598 (expand_omp_taskloop_for_outer): Update.
10599 (expand_omp_taskloop_for_inner): Update.
10600 * omp-simd-clone.c (simd_clone_adjust): Update.
10601 * optabs.c (expand_doubleword_shift): Update.
10602 (expand_abs): Update.
10603 (emit_cmp_and_jump_insn_1): Update.
10604 (expand_compare_and_swap_loop): Update.
10605 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10606 * predict.c (predictable_edge_p): Update.
10607 (edge_probability_reliable_p): Update.
10608 (set_even_probabilities): Update.
10609 (combine_predictions_for_insn): Update.
10610 (combine_predictions_for_bb): Update.
10611 (propagate_freq): Update.
10612 (estimate_bb_frequencies): Update.
10613 (force_edge_cold): Update.
10614 * profile-count.c (profile_count::dump): Add missing space into dump.
10615 (profile_count::debug): Add newline.
10616 (profile_count::differs_from_p): Explicitly convert to unsigned.
10617 (profile_count::stream_in): Update.
10618 (profile_probability::dump): New member function.
10619 (profile_probability::debug): New member function.
10620 (profile_probability::differs_from_p): New member function.
10621 (profile_probability::differs_lot_from_p): New member function.
10622 (profile_probability::stream_in): New member function.
10623 (profile_probability::stream_out): New member function.
10624 * profile-count.h (profile_count_quality): Rename to ...
10625 (profile_quality): ... this one.
10626 (profile_probability): New.
10627 (profile_count): Update.
10628 * profile.c (compute_branch_probabilities): Update.
10629 * recog.c (peep2_attempt): Update.
10630 * sched-ebb.c (schedule_ebbs): Update.
10631 * sched-rgn.c (find_single_block_region): Update.
10632 (compute_dom_prob_ps): Update.
10633 (schedule_region): Update.
10634 * sel-sched-ir.c (compute_succs_info): Update.
10635 * stmt.c (struct case_node): Update.
10636 (do_jump_if_equal): Update.
10637 (get_outgoing_edge_probs): Update.
10638 (conditional_probability): Update.
10639 (emit_case_dispatch_table): Update.
10640 (expand_case): Update.
10641 (expand_sjlj_dispatch_table): Update.
10642 (emit_case_nodes): Update.
10643 * targhooks.c: Update.
10644 * tracer.c (better_p): Update.
10645 (find_best_successor): Update.
10646 * trans-mem.c (expand_transaction): Update.
10647 * tree-call-cdce.c: Update.
10648 * tree-cfg.c (gimple_split_edge): Upate.
10649 (move_sese_region_to_fn): Upate.
10650 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10651 * tree-eh.c (lower_resx): Upate.
10652 (cleanup_empty_eh_move_lp): Upate.
10653 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10654 * tree-inline.c (copy_edges_for_bb): Update.
10655 (copy_cfg_body): Update.
10656 * tree-parloops.c (gen_parallel_loop): Update.
10657 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10658 (gimple_gen_time_profiler): Update.
10659 * tree-ssa-dce.c (remove_dead_stmt): Update.
10660 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10661 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10662 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10663 (unloop_loops): Update.
10664 (try_peel_loop): Update.
10665 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10666 * tree-ssa-loop-split.c (connect_loops): Update.
10667 (split_loop): Update.
10668 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10669 (hoist_guard): Update.
10670 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10671 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10672 (value_replacement): Update.
10673 * tree-ssa-reassoc.c (branch_fixup): Update.
10674 * tree-ssa-tail-merge.c (replace_block_by): Update.
10675 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10676 (create_edge_and_update_destination_phis): Update.
10677 (compute_path_counts): Update.
10678 (recompute_probabilities): Update.
10679 (update_joiner_offpath_counts): Update.
10680 (freqs_to_counts_path): Update.
10681 (duplicate_thread_path): Update.
10682 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10683 (struct switch_conv_info): Update.
10684 (gen_inbound_check): Update.
10685 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10686 (vect_do_peeling): Update.
10687 (vect_loop_versioning): Update.
10688 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10689 (optimize_mask_stores): Update.
10690 * ubsan.c (ubsan_expand_null_ifn): Update.
10691 * value-prof.c (gimple_divmod_fixed_value): Update.
10692 (gimple_divmod_fixed_value_transform): Update.
10693 (gimple_mod_pow2): Update.
10694 (gimple_mod_pow2_value_transform): Update.
10695 (gimple_mod_subtract): Update.
10696 (gimple_mod_subtract_transform): Update.
10697 (gimple_ic): Update.
10698 (gimple_stringop_fixed_value): Update.
10699 (gimple_stringops_transform): Update.
10700 * value-prof.h: Update.
10702 2017-06-29 Carl Love <cel@us.ibm.com>
10704 * config/rs6000/rs6000-c.c: Add support for built-in functions
10705 vector signed int vec_signed (vector float);
10706 vector signed long long vec_signed (vector double);
10707 vector signed int vec_signed2 (vector double, vector double);
10708 vector signed int vec_signede (vector double);
10709 vector signed int vec_signedo (vector double);
10710 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10711 instruction generator.
10712 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10713 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10714 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10716 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10717 vunsignede_v2df): Add define_expands.
10718 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10719 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10720 VEC_UNSIGNEDO): Add definitions.
10721 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10722 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10723 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10724 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10725 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10726 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10727 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10728 vec_unsignede, vec_unsignedo): Add builtin defines.
10729 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10731 * doc/extend.texi: Update the built-in documentation file for the
10732 new built-in functions.
10734 2017-06-29 Richard Biener <rguenther@suse.de>
10736 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10737 reduction chains to LOOP_VINFO_REDUCTIONS.
10738 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10739 SLP reductions after processing reduction chains.
10741 2017-06-29 Nathan Sidwell <nathan@acm.org>
10743 * builtins.c (fold_builtin_FUNCTION): Use
10744 lang_hooks.decl_printable_name.
10746 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10748 PR middle-end/81194
10749 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10750 with only one label.
10751 * stmt.c (expand_case): Assert NCASES is greater than one.
10753 2017-06-29 Richard Biener <rguenther@suse.de>
10755 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10757 (group_case_labels): Likewise.
10758 (find_taken_edge): Push sanity checking on val to workers...
10759 (find_taken_edge_cond_expr): ... here
10760 (find_taken_edge_switch_expr): ... and here, handle cases
10761 with just a default label.
10762 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10763 (group_case_labels): Likewise.
10764 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10765 group_case_labels does anything cleanup the CFG again.
10767 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10769 PR tree-optimization/81196
10770 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10771 exit condition comparing two IVs.
10773 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10775 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10776 profile to the dummy entry at the end of the list of architectures.
10777 * config/arm/arm-cpu-cdata.h: Regenerated.
10779 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10780 Michael Collison <michael.collison@arm.com>
10783 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10785 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10786 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10787 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10788 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10789 of shift when the shift amount is masked with constant equal to
10790 the size of the mode.
10791 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10794 2017-06-29 Martin Liska <mliska@suse.cz>
10796 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10798 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10801 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10804 (edge_growth_cache_entry::edge_growth_cache_entry): New
10806 (reset_edge_growth_cache): Update to use constructor.
10808 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10810 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10811 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10812 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10814 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10816 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10817 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10819 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10821 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10822 (*-linux-uclibc*): Add t-uclibc tmake_file.
10823 * config/t-musl: New.
10824 * config/t-uclibc: New.
10826 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10828 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10830 (gen_comm_data): Emit architectural setting of arch_prof.
10831 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10833 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10834 (armv8-m.base, armv8-m.main): Likewise.
10835 * arm-protos.h (arm_build_target): Add profile field.
10836 (arch_option): Likewise.
10837 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10839 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10840 arm_active_target.profile.
10842 2017-06-28 Richard Biener <rguenther@suse.de>
10844 PR middle-end/81227
10845 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10846 TYPE_OVERFLOW_WRAPS.
10847 * match.pd (negate_expr_p): Likewise.
10848 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10849 fold_build2, not fold_binary.
10851 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10853 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10854 Convert memory address to Pmode.
10855 (aarch64_print_operand): Assert MEM operands are always Pmode.
10857 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10860 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10861 Remove redundant if.
10862 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10863 * config/arm/aarch-common-protos.h
10864 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10865 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10867 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10870 * multiple_target.c (create_dispatcher_calls): Set the default
10871 clone to be static, not public.
10873 2017-06-28 Richard Biener <rguenther@suse.de>
10875 * tree-vect-loop.c (vectorizable_reduction): Move special
10876 cond reduction IV var creation ...
10877 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10878 parameter. Use STMT_VINFO_VECTYPE.
10879 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10882 2017-06-28 Martin Liska <mliska@suse.cz>
10885 * ipa-visibility.c (non_local_p): Handle visibility.
10887 2017-06-28 Martin Liska <mliska@suse.cz>
10890 * common.opt: Add IntegerRange to various options.
10891 * opt-functions.awk (integer_range_info): New function.
10892 * optc-gen.awk: Add integer_range_info to cl_options struct.
10893 * opts-common.c (decode_cmdline_option): Handle
10894 CL_ERR_INT_RANGE_ARG.
10895 (cmdline_handle_error): Likewise.
10896 * opts.c (print_filtered_help): Show valid interval in
10897 when --help is provided.
10898 * opts.h (struct cl_option): Add range_min and range_max fields.
10899 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10901 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10903 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10904 (x * C EQ/NE y * C): New transformation.
10906 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10908 * genmultilib (combination_space): Accept '+' in option names.
10910 2017-06-28 Martin Liska <mliska@suse.cz>
10913 * asan.c (instrument_derefs): Bail out inner references
10914 that are hard register variables.
10916 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10919 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10920 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10922 2017-06-28 Richard Biener <rguenther@suse.de>
10924 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10925 (vect_get_slp_defs): Adjust.
10926 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10928 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10930 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10931 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10932 (vectorizable_reduction): Adjust.
10933 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10935 (vect_get_slp_defs): Likewise.
10936 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10937 (vectorizable_bswap): Adjust.
10938 (vectorizable_call): Likewise.
10939 (vectorizable_conversion): Likewise.
10940 (vectorizable_assignment): Likewise.
10941 (vectorizable_shift): Likewise.
10942 (vectorizable_operation): Likewise.
10943 (vectorizable_store): Likewise.
10944 (vectorizable_condition): Likewise.
10945 (vectorizable_comparison): Likewise.
10947 2017-06-28 Michael Collison <michael.collison@arm.com>
10950 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10952 (arm_gen_movmemqi): Removed unused variable 'i'.
10953 Convert 'for' loop into 'while' loop.
10954 (arm_expand_prologue): Remove last unnecessary set of insn.
10955 (thumb_pop): Remove unused variable 'pushed_words'.
10956 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10958 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10960 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10961 * config/s390/s390-protos.h: Add prototype of
10962 s390_rel_address_ok_p.
10963 * config/s390/s390.c (s390_got_symbol): New function.
10964 (s390_rel_address_ok_p): New function.
10965 (legitimize_pic_address): Use s390_rel_address_ok_p.
10966 (s390_load_got): Use s390_got_symbol.
10967 (s390_option_override): Issue error if
10968 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10969 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10971 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10973 2017-06-27 Andrew Pinski <apinski@cavium.com>
10975 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10976 (X * copysign (1.0, X)): New pattern.
10977 (X * copysign (1.0, -X)): New pattern.
10978 (copysign (-1.0, CST)): New pattern.
10980 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10982 * genmultilib (combination_space): Remove variable.
10983 Validate reuse rules against regular expression for any sequence
10984 of multilib options in any order.
10986 2017-06-27 Michael Collison <michael.collison@arm.com>
10988 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10989 call aarch64_split_simd_combine.
10990 * (aarch64_combine_internal<mode>): Delete pattern.
10991 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10992 Allow register and subreg operands.
10994 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10996 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10997 specific need, just fallback on defaults.
10998 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11000 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11001 Olivier Hainque <hainque@adacore.com>
11003 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11005 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11006 targets. Pick a default if no particular attempt applied.
11007 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11010 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11012 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11013 (x86_64-wrs-vxworks7): Likewise.
11015 2017-06-27 Marek Polacek <polacek@redhat.com>
11018 * ubsan.c (instrument_null): Check get_base_address's result for null.
11020 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11022 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11024 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11026 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11027 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11028 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11029 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11030 New function types.
11031 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11032 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11033 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11034 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11035 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11036 BUILT_IN_FEUPDATEENV): New builtins.
11037 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11038 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11039 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11040 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11042 (builtin_structptr_types): Adjust size.
11043 * tree.c (builtin_structptr_types): Add four entries.
11045 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11046 Olivier Hainque <hainque@adacore.com>
11048 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11049 (TLS_SYM): New local macro, forcing reference to __tls__ on
11050 link command lines for VxWorks 7 RTPs, triggering initialization
11052 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11053 OS features TLS support, true for RTPs on VxWorks 7.
11054 * config/vxworks.c (vxworks_override_options): Setup emutls
11057 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11059 * predict.c (test_prediction_value_range): Use -1U instead of -1
11060 to avoid narrowing conversion warning.
11061 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11062 to avoid narrowing conversion warning.
11063 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11065 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11067 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11069 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11070 64bit configurations.
11071 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11072 (SIZE_TYPE): Likewise.
11073 * config/vxworks.c (vxworks_emutls_var_fields): Use
11074 long_unsigned_type_node instead of unsigned_type_node as the offset
11075 field type, which is "pointer" mode in emutls.c.
11077 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11080 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11082 PR middle-end/81207
11083 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11084 gimple_vuse copying separately from gimple_vdef copying.
11086 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11088 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11091 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11092 Olivier Hainque <hainque@adacore.com>
11094 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11095 for all vxworks7 targets.
11096 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11097 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11098 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11099 variations for VX6/VX7 and 32/64bits later on in ...
11100 (VXWORKS_LIB_SPEC): Leverage new macros.
11101 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11102 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11104 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11106 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11107 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11109 2017-06-26 Carl Love <cel@us.ibm.com>
11111 * config/rs6000/rs6000-c.c: Add support for built-in functions
11112 vector bool char vec_reve (vector bool char);
11113 vector signed char vec_reve (vector signed char);
11114 vector unsigned char vec_reve (vector unsigned char);
11115 vector bool int vec_reve (vector bool int);
11116 vector signed int vec_reve (vector signed int);
11117 vector unsigned int vec_reve (vector unsigned int);
11118 vector bool long long vec_reve (vector bool long long);
11119 vector signed long long vec_reve (vector signed long long);
11120 vector unsigned long long vec_reve (vector unsigned long long);
11121 vector bool short vec_reve (vector bool short);
11122 vector signed short vec_reve (vector signed short);
11123 vector double vec_reve (vector double);
11124 vector float vec_reve (vector float);
11125 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11126 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11127 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11128 (altivec_vreve): New pattern.
11129 * config/rs6000/altivec.h (vec_reve): New define.
11130 * doc/extend.texi (vec_rev): Update the built-in documentation file
11131 for the new built-in functions.
11133 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11135 PR tree-optimization/71815
11136 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11138 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11140 (slsr_process_phi): Likewise.
11141 (replace_uncond_cands_and_profitable_phis): Don't replace a
11142 multiply candidate with a stride of 1 (copy or cast).
11143 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11145 (lowest_cost_path): Likewise.
11146 (total_savings): Likewise.
11148 2017-06-26 Richard Biener <rguenther@suse.de>
11151 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11152 Use def_builtin_pure for all gather builtins.
11154 2017-06-26 Richard Biener <rguenther@suse.de>
11156 PR tree-optimization/81203
11157 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11158 non-dominating BBs.
11160 2017-06-26 Marek Polacek <polacek@redhat.com>
11163 * doc/invoke.texi: Document -Wmultistatement-macros.
11165 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11167 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11168 arm_neon_ok_no_float_abi effective target.
11170 2017-06-26 Richard Biener <rguenther@suse.de>
11172 PR tree-optimization/80928
11173 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11174 (copy_bbs): Set BB_DUPLICATED flag early.
11175 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11177 (execute_on_shrinking_pred): Likewise.
11178 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11179 BB_DUPLICATED blocks.
11180 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11181 iterate over all PHIs considering removal of *gsi.
11183 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11185 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11188 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11190 * config/rs6000/rs6000-string.c: (expand_block_clear,
11191 do_load_for_compare, select_block_compare_mode,
11192 compute_current_alignment, expand_block_compare,
11193 expand_strncmp_align_check, expand_strn_compare,
11194 expand_block_move, rs6000_output_load_multiple)
11195 Move functions related to string/block move/compare
11196 to a separate file.
11197 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11198 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11199 for this function which is now used in two files.
11200 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11201 * config.gcc: Add rs6000-string.o to extra_objs for
11202 targets powerpc*-*-* and rs6000*-*-*.
11204 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11207 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11208 32-bit, since indexed is not valid for DImode.
11209 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11210 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11211 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11212 (define_peephole2 for Altivec d-form store): Likewise.
11215 * multiple_target.c (create_dispatcher_calls): Only create the
11216 dispatcher call if the function is the default clone of a
11217 versioned function.
11219 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11221 PR middle-end/80902
11222 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11223 a call, force the call to not be a tail call.
11225 2017-06-23 Jeff Law <law@redhat.com>
11227 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11230 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11232 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11233 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11234 handling for early expansion of vector shifts (sl,sr,sra,rl).
11235 (builtin_function_type): Add vector shift right instructions
11236 to the unsigned argument list.
11238 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11240 rtl-optimizatoin/79286
11241 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11242 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11243 trap. PIC register plus a const unspec without offset can never trap.
11245 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11247 * tree.h (builtin_structptr_type): New type.
11248 (builtin_structptr_types): Declare new array.
11249 * tree.c (builtin_structptr_types): New array.
11250 (free_lang_data, build_common_tree_nodes): Use it.
11252 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11255 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11258 2017-06-22 Matt Turner <mattst88@gmail.com>
11260 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11261 Lake models to skylake case. Assume skylake for unknown
11262 models with clflushopt.
11264 2017-06-22 Jeff Law <law@redhat.com>
11266 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11267 frame sizes that do not satisfy aarch64_uimm12_shift.
11269 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11271 * profile-count.h (apply_probability,
11272 apply_scale, probability_in): Fix checks for zero.
11274 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11276 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11277 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11279 2016-06-22 Richard Biener <rguenther@suse.de>
11281 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11282 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11283 REDUC_MAX_EXPR support.
11284 (vectorizable_reduction): Likewise.
11285 (vect_create_epilog_for_reduction): Likewise.
11287 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11289 * match.pd (A / (1 << B) -> A >> B): New.
11290 * generic-match-head.c: Include optabs-tree.h.
11291 * gimple-match-head.c: Likewise.
11292 * optabs-tree.h (target_supports_op_p): New.
11293 * optabs-tree.c (target_supports_op_p): New.
11295 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11297 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11298 $gcc_cv_ld --help output.
11299 (gcc_cv_ld_demangle): Likewise.
11300 (gcc_cv_ld_eh_frame_hdr): Likewise.
11301 (gcc_cv_ld_pie): Likewise.
11302 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11303 (gcc_cv_ld_buildid): Likewise.
11304 (gcc_cv_ld_sysroot): Likewise.
11305 (ld_bndplt_support): Likewise.
11306 (ld_pushpopstate_support): Likewise.
11307 * configure: Regenerate.
11308 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11310 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11313 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11314 operands indexes to avoid gap between operands and match_dups.
11316 2017-06-21 Andrew Pinski <apinski@cavium.com>
11318 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11319 Increment Arith_shift and Arith_shift_reg by 1.
11320 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11322 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11323 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11324 (aarch64_strip_extend): Add new argument and test for it.
11325 (aarch64_cheap_mult_shift_p): New function.
11326 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11327 add a cost if it is true.
11328 Update calls to aarch64_strip_extend.
11329 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11331 2017-06-21 Andrew Pinski <apinski@cavium.com>
11333 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11335 (thunderxt88): Likewise.
11336 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11337 (thunderx_prefetch_tune): New variable.
11338 (thunderx2t99_prefetch_tune): Update for the correct values.
11339 (thunderxt88_tunings): New variable.
11340 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11341 generic_prefetch_tune.
11342 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11344 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11346 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11347 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11348 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11349 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11350 (aarch64_atomic_cas<mode>, GPI): Likewise.
11352 2017-06-21 Martin Liska <mliska@suse.cz>
11354 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11355 statements on cold and hot labels.
11356 * predict.c (tree_estimate_probability_bb): Remove the
11357 prediction from this place.
11359 2017-06-21 Martin Liska <mliska@suse.cz>
11361 PR tree-optimization/79489
11362 * gimplify.c (maybe_add_early_return_predict_stmt): New
11364 (gimplify_return_expr): Call the function.
11365 * predict.c (tree_estimate_probability_bb): Remove handling
11367 * predict.def: Update comment about early return predictor.
11368 * gimple-predict.h (is_gimple_predict): New function.
11369 * predict.def: Change default value of early return to 66.
11370 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11372 * passes.def: Put pass_strip_predict_hints to the beginning of
11375 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11377 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11378 FUNCTION_DECL declarations.
11379 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11381 (dwaf2out_decl): Likewise.
11382 * godump.c (go_early_global_decl): Skip call to the real debug hook
11383 for FUNCTION_DECL declarations.
11384 * passes.c (rest_of_decl_compilation): Skip call to the
11385 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11386 -fdump-go-spec is passed.
11388 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11390 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11391 Reorder for compactness.
11392 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11393 (def_builtin_pure, def_builtin_pure2): New functions.
11394 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11396 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11398 * match.pd (nop_convert): New predicate.
11399 ((A +- CST1) +- CST2): Allow some NOP conversions.
11401 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11404 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11405 with ctors/dtors if GOVD_SHARED is set.
11407 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11409 * config/aarch64/aarch64.md (movti_aarch64):
11410 Emit mov rather than orr.
11411 (movtf_aarch64): Likewise.
11412 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11413 Emit mov rather than orr.
11415 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11417 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11418 Swap alternatives, make integer dup more expensive.
11420 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11422 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11423 Return true for non-tls symbols.
11425 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11427 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11428 (cortex-a75): Likewise.
11429 (cortex-a75.cortex-a55): Likewise.
11430 * config/aarch64/aarch64-tune.md: Regenerate.
11431 * doc/invoke.texi (-mtune): Document new values for -mtune.
11433 2017-06-21 Tom de Vries <tom@codesourcery.com>
11435 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11436 stack_size feature.
11437 (Effective-Target Keywords, Other attributes): Suggest using
11438 dg-add-options stack_size feature to get stack limit in stack_size
11439 effective target documentation.
11441 2017-06-21 Julian Brown <julian@codesourcery.com>
11442 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11444 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11445 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11446 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11448 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11449 attribute type list for neon_multiply.
11450 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11451 attribute type list for neon_multiply.
11452 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11453 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11454 attribute type list for neon_multiply.
11455 * config/arm/types.md (crypto_pmull): Add.
11456 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11457 attribute type list.
11459 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11461 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11464 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11466 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11467 to make sure not to dereference a NULL cost_classes_ptr pointer.
11469 2017-06-20 Carl Love <cel@us.ibm.com>
11471 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11472 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11473 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11474 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11475 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11476 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11477 VMULOSW): New enum "unspec" values.
11478 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11479 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11480 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11481 altivec_vmulosw): New patterns.
11482 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11483 VMULOSW): Add definitions.
11485 2017-06-20 Julia Koval <julia.koval@intel.com>
11487 * config/i386/i386.c: Fix rounding expand for new pattern.
11488 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11490 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11492 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11493 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11495 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11497 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11500 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11502 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11503 architecture, then by implementer ID.
11504 * config/aarch64/aarch64-tune.md: Regenerate.
11506 2017-06-20 Richard Biener <rguenther@suse.de>
11508 PR middle-end/81097
11509 * fold-const.c (split_tree): Fold to type before negating.
11511 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11513 * diagnostic-show-locus.c
11514 (selftest::test_fixit_deletion_affecting_newline): New function.
11515 (selftest::diagnostic_show_locus_c_tests): Call it.
11517 2017-06-20 Andreas Schwab <schwab@suse.de>
11520 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11523 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11525 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11526 __ARM_FEATURE_COPROC according to support.
11528 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11530 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11531 Rewritten to avoid overflow for > 32-bit pointers.
11534 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11535 by removing enum keyword.
11536 (ubsan_type_descriptor): Likewise. Formatting fix.
11539 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11540 splitter): Require TARGET_SSE2 in the condition.
11542 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11545 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11546 for doing vector set of SFmode on ISA 3.0.
11547 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11548 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11550 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11551 SFmode value into a V4SF variable that was extracted from another
11552 V4SF variable without converting the element to double precision
11553 and back to single precision vector format.
11554 (vsx_insert_extract_v4sf_p9_2): Likewise.
11556 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11558 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11559 in UWHI to avoid undefined overflow.
11562 * ubsan.h (enum ubsan_encode_value_phase): New.
11563 (ubsan_encode_value): Change second argument to
11564 enum ubsan_encode_value_phase with default value of
11565 UBSAN_ENCODE_VALUE_GENERIC.
11566 * ubsan.c (ubsan_encode_value): Change second argument to
11567 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11568 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11569 create_tmp_var_raw instead of create_tmp_var and use a
11571 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11572 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11573 ubsan_encode_value callers.
11576 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11577 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11578 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11580 2017-06-19 Richard Biener <rguenther@suse.de>
11582 PR middle-end/81118
11583 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11584 estimates if we changed anything.
11586 2017-06-19 Richard Biener <rguenther@suse.de>
11588 PR tree-optimization/80887
11589 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11590 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11591 simplified lookups, then reset mprts_hook.
11592 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11594 (try_to_simplify): Likewise.
11596 2017-06-19 Martin Liska <mliska@suse.cz>
11599 * gimplify.c (gimplify_switch_expr):
11600 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11602 2017-06-19 Martin Liska <mliska@suse.cz>
11604 * doc/install.texi: Document that PGO runs in 4 stages.
11606 2017-06-19 Martin Liska <mliska@suse.cz>
11609 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11610 to dispatcher function name.
11611 * multiple_target.c (replace_function_decl): New function.
11612 (create_dispatcher_calls): Redirect both edges and references.
11614 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11616 * profile-count.c (profile_count::dump): Dump quality.
11617 (profile_count::differs_from_p): Update for unsigned val.
11618 * profile-count.h (profile_count_quality): New enum.
11619 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11621 2017-06-19 Richard Biener <rguenther@suse.de>
11623 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11624 struct function as arg.
11625 (estimate_numbers_of_iterations): Export overload with loop arg.
11626 (free_numbers_of_iterations_estimates_loop): Use an overload of
11627 free_numbers_of_iterations_estimates instead.
11628 * tree-cfg.c (remove_bb): Adjust.
11629 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11630 * tree-parloops.c (gen_parallel_loop): Likewise.
11631 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11633 (tree_unroll_loops_completely): Likewise.
11634 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11635 Use an overload instead and export.
11636 (estimated_loop_iterations): Adjust.
11637 (max_loop_iterations): Likewise.
11638 (likely_max_loop_iterations): Likewise.
11639 (estimate_numbers_of_iterations): Take struct function as arg
11641 (loop_exits_before_overflow): Adjust.
11642 (free_numbers_of_iterations_estimates_loop): Use an overload.
11643 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11644 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11646 2017-06-19 Richard Biener <rguenther@suse.de>
11649 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11650 RANGE_EXPR conservatively.
11652 2017-06-16 Carl Love <cel@us.ibm.com>
11654 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11655 definitions for vec_float, vec_float2, vec_floato,
11656 vec_floate built-ins.
11657 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11658 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11660 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11661 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11662 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11663 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11664 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11665 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11666 vec_floato): Add builtin defines.
11667 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11668 Update the built-in documentation file for the new built-in
11671 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11673 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11674 (mthumb): Mark as the negative of -marm.
11676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11678 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11680 (ARM Options, -mtune): Document that this accepts the same
11681 extension options as -mcpu.
11682 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11684 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11686 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11687 permitted extensions.
11689 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11691 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11692 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11693 (armv8-m.main): Add option +nodsp.
11694 * config/arm/arm-cpu-cdata.h: Regenerated.
11696 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11698 * config/arm/t-fuchsia: New file.
11699 * config.gcc (arm*-*-fuchsia*): Use it.
11701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11703 * config/arm/t-symbian: Rewrite for new option infrastructure.
11705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11707 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11708 (MULTILIB_REQUIRED): Likewise.
11710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11712 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11713 (MULTILIB_RESUE): Likewise.
11714 (MULTILIB_MATCHES): Likewise.
11715 (MULTLIB_REQUIRED): Likewise.
11717 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11719 * config/arm/t-rtems: Rewrite for new option framework.
11721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11723 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11724 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11725 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11726 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11727 * config/arm/t-multilib: ... here.
11728 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11729 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11730 armv7-a and armv8*-a when A-profile libraries have not been built.
11731 * config/arm/t-rmprofile: Rewrite.
11733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11735 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11736 with a backslash. Remove the backslash after substituting unescaped
11738 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11742 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11743 the driver to pass through the default setting of -mfloat-abi.
11744 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11746 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11747 (all_feat_combs): New rule.
11748 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11750 * config/arm/t-aprofile: Rewrite.
11752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11754 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11755 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11756 fpu is not specified by the user/command-line.
11757 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11758 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11759 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11760 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11761 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11762 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11766 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11767 * config/arm/t-arm-elf: Rewritten.
11769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11771 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11772 have some floating-point instructions.
11773 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11774 (TARGET_MAYBE_HARD_FLOAT): New macro.
11775 * config/arm/arm-builtins.c (arm_init_builtins): Use
11776 TARGET_MAYBE_HARD_FLOAT.
11777 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11781 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11782 (configargs.h): Include it.
11783 (arm_print_hint_for_fpu_option): New function.
11784 (arm_parse_fpu_option): New function.
11785 (candidate_extension): New class.
11786 (arm_canon_for_multilib): New function.
11787 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11788 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11789 (ARCH_CANONICAL_SPECS): New macro.
11790 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11794 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11795 are set after handling multilib fragments. Set target_cpu_default2
11798 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11800 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11802 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11803 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11805 (check_arch): Likewise.
11806 * config/arm/arm.c (arm_configure_build_target): Handle
11807 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11808 options in the default.
11810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11812 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11813 when an option is an alias of another.
11814 * config/arm/parsecpu.awk (optalias): New parser token.
11815 (gen_comm_data): Mark non-alias options as such. Emit entries
11816 for extension aliases.
11817 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11818 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11819 (armv6kz, armv6zk, armv6t2): Likewise.
11820 (armv7): Make vfpv3-d16 an alias.
11821 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11823 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11824 Sort in canonical order.
11825 (armv8-a): Sort in canonical order.
11826 (armv8.1-a, armv8.2-a): Likewise.
11827 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11829 (cortex-a9): Sort in canonical order.
11830 * config/arm/arm.c (selftests.h): Include it.
11831 (arm_test_cpu_arch_data): New function.
11832 (arm_run_self_tests): New function.
11833 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11834 (targetm): Move declaration to the end of the file.
11835 * arm-cpu-cdata.h: Regenerated.
11837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11839 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11840 call to target_mode_check describing the type of option passed.
11841 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11842 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11843 arm_parse_cpu_option_name to match parameters against list of
11845 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11846 arm_arch_core_flags data structure.
11847 * config/arm/arm-cpu_cdata.h: Regenerated.
11849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11851 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11853 (arm_print_hint_for_cpu_option): Likewise.
11854 (arm_print_hint_for_arch_option): Likewise.
11855 (arm_parse_cpu_option_name): Likewise.
11856 (arm_parse_arch_option_name): Likewise.
11857 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11858 of entries in the all_fpus list.
11859 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11860 (arm_parse_cpu_option_name): Declare.
11861 (arm_parse_arch_option_name): Declare.
11862 (arm_parse_option_features): Declare.
11863 (arm_intialize_isa): Declare.
11864 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11866 (gen_comm_data): ... here. Make definitions non-static.
11867 * config/arm/arm-cpu-data.h: Regenerated.
11868 * config/arm/arm-cpu-cdata.h: Regenerated.
11870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11872 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11873 (cpu_arch_extension): New structure.
11874 (cpu_arch_option, arch_option, cpu_option): New structures.
11875 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11876 architecture types.
11877 (gen_data): Generate new format data tables.
11878 * config/arm/arm.c (cpu_tune): New structure.
11879 (cpu_option, processors): Delete.
11880 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11881 (arm_print_hint_for_cpu_option): ... this and ...
11882 (arm_print_hint_for_arch_option): ... this.
11883 (arm_parse_arch_cpu_name): Delete. Replace with ...
11884 (arm_parse_cpu_option_name): ... this and ...
11885 (arm_parse_arch_option_name): ... this.
11886 (arm_unrecognized_feature): Change type of target parameter to
11888 (arm_parse_arch_cpu_features): Delete. Replace with ...
11889 (arm_parse_option_features): ... this.
11890 (arm_configure_build_target): Rework to use new configuration data
11892 (arm_print_tune_info): Rework for new configuration data tables.
11893 * config/arm/arm-cpu-data.h: Regenerated.
11894 * config/arm/arm-cpu.h: Regenerated.
11896 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11898 * Makefile.in (OBJS): Move sbitmap.o from here ...
11899 (OBJS-libcommon): ... to here.
11901 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11903 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11904 (ISA_ALL_CRYPTO): New macro.
11905 (ISA_ALL_SIMD): New macro
11906 (ISA_ALL_FP): New macro.
11907 * config/arm/arm.c (fpu_bitlist): Update initializer.
11908 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11910 (arm9e): Add fpu. Add option for nofp
11911 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11912 (arm926ej-s, arm1026ej-s): Likewise.
11913 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11914 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11915 neon-fp16, neon-vfpv4, nofp and nosimd.
11916 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11917 (cortex-a8): Add fpu. Add option for nofp.
11918 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11919 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11920 (cortex-r4f): Add fpu.
11921 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11922 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11924 (cortex-r8): Likewise.
11925 (cortex-m4): Add fpu. Add option for nofp.
11926 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11927 (cortex-a17.cortex-a7): Likewise.
11928 (cortex-a32): Add fpu. Add options for crypto and nofp.
11929 (cortex-a35, cortex-a53): Likewise.
11930 (cortex-a57): Add fpu. Add option for crypto.
11931 (cortex-a72, cortex-a73): Likewise.
11932 (exynos-m1): Likewise.
11933 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11934 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11935 (cortex-m33): Add fpu. Add option for nofp.
11936 * config/arm/arm-cpu-cdata.h: Regenerated
11937 * config/arm/arm-cpu-data.h: Regenerated.
11939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11941 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11942 (armv5te, armv5tej): Likewise.
11943 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11944 (armv7): Add options fp and vfpv3-d16.
11945 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11946 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11948 (armv7ve): Likewise.
11949 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11950 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11951 (armv8-a): Add nocrypto option.
11952 (armv8.1-a, armv8.2-a): Likewise.
11953 (armv8-m.main): add options fp, fp.dp and nofp.
11955 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11957 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11959 (armv8-a+crc): Delete.
11960 (armv8.1-a): Add options simd, crypto and nofp.
11961 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11962 (armv8.2-a+fp16): Delete.
11963 (armv8-m.main): Add option dsp.
11964 (armv8-m.main+dsp): Delete.
11965 (cortex-a8): Add fpu. Add nofp option.
11966 (cortex-a9): Add fpu. Add nofp and nosimd options.
11967 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11968 link to main cpu and architecture data structures.
11969 (gen_comm_data): Only put isa attributes from the main architecture
11971 (option): New statement for architecture and CPU entries.
11972 * arm.c (struct cpu_option): New structure.
11973 (struct processors): Add entry for options.
11974 (arm_unrecognized_feature): New function.
11975 (arm_parse_arch_cpu_name): Ignore any characters after the first
11977 (arm_parse_arch_cpu_feature): New function.
11978 (arm_configure_build_target): Separate out any CPU and architecture
11979 features and parse separately. Don't error out if -mfpu=auto is
11980 used with only an architecture string.
11981 (arm_print_asm_arch_directives): New function.
11982 (arm_file_start): Call it.
11983 * config/arm/arm-cpu-cdata.h: Regenerated.
11984 * config/arm/arm-cpu-data.h: Likewise.
11985 * config/arm/arm-tables.opt: Likewise.
11987 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11989 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11990 assembler when it is not -mfpu=auto.
11992 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11994 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11995 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11996 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11997 (ASM_CPU_SPEC): Rewrite.
11998 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11999 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12000 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12001 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12002 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12003 copied string is NUL-terminated. Also strip any characters prefixed
12005 (arm_rewrite_selected_arch): New function.
12006 (arm_rewrite_march): New function.
12008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12010 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12011 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12012 (march, mcpu, mtune): Convert to string-based options.
12013 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12014 (arm_parse_arch_cpu_name): New function.
12015 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12016 identify selected architecture or CPU.
12017 (arm_option_save): New function.
12018 (TARGET_OPTION_SAVE): Redefine.
12019 (arm_option_restore): Restore string options.
12020 (arm_option_print): Print string options.
12022 2017-06-16 Martin Sebor <msebor@redhat.com>
12024 PR tree-optimization/80933
12025 PR tree-optimization/80934
12026 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12027 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12028 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12029 (gimple_fold_builtin): Call them.
12031 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12033 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12034 as unlikely; update profile.
12036 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12038 * predict.c (force_edge_cold): Handle declaring edges impossible
12041 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12043 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12045 (try_unroll_loop_completely): Fix reporting.
12047 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12049 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12051 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12054 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12055 if given a non-constant argument for an intrinsic which requires a
12058 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12060 * profile.c (compare_freqs): New function.
12061 (branch_prob): Sort edge list.
12062 (find_spanning_tree): Assume that the list is priority sorted.
12064 2017-06-16 Richard Biener <rguenther@suse.de>
12066 PR tree-optimization/81090
12067 * passes.def (pass_record_bounds): Remove.
12068 * tree-pass.h (make_pass_record_bounds): Likewise.
12069 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12070 make_pass_record_bounds): Likewise.
12071 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12072 not free niter estimates at the beginning but at the end.
12073 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12075 2017-06-16 Richard Biener <rguenther@suse.de>
12077 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12078 initializer to workaround ICE in host GCC 4.8.
12080 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12082 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12084 (clone_inlined_nodes): Update.
12086 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12088 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12089 prefetch settings, and enable prefetching by default at -O3.
12091 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12093 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12094 Set flag_prefetch_loop_arrays according to tuning data.
12096 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12098 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12099 New tune structure.
12100 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12101 [Unrelated to main purpose of the patch] Place the pointer field last
12102 to enable type checking errors when tune structure are wrongly merged.
12103 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12104 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12105 (thunderx2t99_prefetch_tune): New tune constants.
12106 (tune_params *_tunings): Update all tunings (no functional change).
12107 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12108 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12109 from tunings structures.
12111 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12114 * ubsan.c (instrument_null): Add T argument, use it instead
12115 of computing it based on IS_LHS.
12116 (instrument_object_size): Likewise.
12117 (pass_ubsan::execute): Adjust instrument_null and
12118 instrument_object_size callers to pass gimple_get_lhs or
12119 gimple_assign_rhs1 result to it. Use instrument_null instead of
12120 calling get_base_address and instrument_mem_ref. Handle
12121 aggregate call arguments for object-size sanitization.
12123 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12125 PR tree-optimization/81089
12126 * tree-vrp.c (is_masked_range_test): Validate operands of
12129 2017-06-15 Martin Sebor <msebor@redhat.com>
12132 * dumpfile.c (dump_register): Avoid calling memset to initialize
12133 a class with a default ctor.
12134 * gcc.c (struct compiler): Remove const qualification.
12135 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12136 * hash-table.h: Ditto.
12137 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12139 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12140 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12142 * params.h (struct param_info): Make struct members non-const.
12143 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12144 with default initialization.
12145 * vec.h (vec_copy_construct, vec_default_construct): New helper
12147 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12148 with vec_copy_construct.
12149 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12150 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12151 * doc/invoke.texi (-Wclass-memaccess): Document.
12153 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12155 * emit-rtl.h (is_leaf): Update comment about local
12156 register allocator.
12158 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12161 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12162 for a variable to have a section before checking if the section has a
12164 Set section to.persistent if persistent attribute is set.
12165 Warn if .persistent attribute is used on an automatic variable.
12167 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12169 PR rtl-optimization/80474
12170 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12172 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12174 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12177 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12179 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12180 (casesi): Emit bounds checking as RTL.
12181 (casesi_internal_mips16_<mode>): Remove bounds checking.
12183 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12185 * config/xtensa/xtensa.c (xtensa_option_override): Append
12186 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12187 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12188 xtensa_doloop_hooks): Define unconditionally.
12189 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12191 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12192 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12193 for it in xtensa_option_override.
12194 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12195 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12197 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12199 * doc/cppopts.texi: Document '-' special value to -MF.
12201 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12203 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12204 (cortex_a53_fconst): Likewise.
12205 (cortex_a53_fpmul): Likewise.
12206 (cortex_a53_f_load_64): Likewise.
12207 (cortex_a53_f_load_many): Likewise.
12208 (cortex_a53_advsimd_alu): Likewise.
12209 (cortex_a53_advsimd_alu_q): Likewise.
12210 (cortex_a53_advsimd_mul): Likewise.
12211 (cortex_a53_advsimd_mul_q): Likewise.
12212 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12213 Add missing fmul, r2f_cvt and fconst cases.
12215 2017-06-14 Richard Biener <rguenther@suse.de>
12217 PR middle-end/81088
12218 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12220 (fold_binary_loc): When associating do not treat pre-existing
12221 TREE_OVERFLOW on literal constants as a reason to allow
12222 TREE_OVERFLOW on associated literal constants.
12224 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12226 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12227 (MASK_FEATURES): New macro.
12228 * config/sparc/sparc.c (sparc_option_override): Remove the special
12229 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12231 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12233 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12234 a division of 0 if non-call exceptions are enabled.
12236 2017-06-14 Andrew Pinski <apinski@cavium.com>
12237 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12240 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12241 Improve vector initialization code gen for only variable case.
12243 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12245 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12247 2017-06-14 Richard Biener <rguenther@suse.de>
12249 PR tree-optimization/81083
12250 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12253 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12255 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12256 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12257 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12258 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12259 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12260 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12262 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12264 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12265 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12267 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12269 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12271 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12273 * config/rs6000/t-rtems: Don't handle SPE.
12275 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12277 * config/rs6000/t-linux: Don't handle SPE.
12279 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12281 * config/rs6000/eabispe.h: Delete file.
12283 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12285 * config/rs6000/t-spe: Delete file.
12287 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12289 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12290 (rs6000_legitimate_offset_address_p): Return false for anything in
12291 V2SImode or V2SFmode.
12293 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12295 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12296 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12297 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12299 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12300 (rs6000_legitimate_offset_address_p): Ditto.
12301 (rs6000_emit_move): Ditto.
12302 (rs6000_init_builtins): Remove V4HI_type_node.
12304 2017-06-13 Martin Liska <mliska@suse.cz>
12307 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12308 (gate_asan): Likewise.
12309 * asan.h (asan_no_sanitize_address_p): Remove the function.
12310 (sanitize_flags_p): New function.
12311 * builtins.def: Fix coding style.
12312 * common.opt: Use renamed enum value.
12313 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12314 * doc/extend.texi: Document no_sanitize attribute.
12315 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12316 to SANITIZE_UNDEFINED_NONDEFAULT.
12317 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12318 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12319 Use sanitize_flags_p.
12320 * gimplify.c (gimplify_function_tree): Likewise.
12321 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12322 * opts.c (parse_no_sanitize_attribute): New function.
12323 (common_handle_option): Use renamed enum value.
12324 * opts.h (parse_no_sanitize_attribute): Declare.
12325 * tree.c (sanitize_flags_p): New function.
12326 * tree.h: Declared here.
12327 * tsan.c: Use sanitize_flags_p.
12328 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12329 (instrument_mem_ref): Likewise.
12330 (instrument_bool_enum_load): Likewise.
12331 (do_ubsan_in_current_function): Remove the function.
12332 (pass_ubsan::execute): Use sanitize_flags_p.
12333 * ubsan.h: Remove do_ubsan_in_current_function
12334 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12335 (dump_function_to_file): Use it here.
12337 2017-06-13 Martin Jambor <mjambor@suse.cz>
12339 PR tree-optimization/80803
12340 PR tree-optimization/81063
12341 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12342 (propagate_subaccesses_across_link): Enqueue subtree whenever
12343 necessary instead of relying on the caller.
12345 2017-06-13 Martin Jambor <mjambor@suse.cz>
12347 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12348 that have a first_link.
12349 (sort_and_splice_var_accesses): Do not check first_link before
12351 (subtree_mark_written_and_enqueue): Likewise.
12352 (propagate_all_subaccesses): Likewise and do not stop at first
12353 parent with a first_link.
12355 2017-06-13 Martin Jambor <mjambor@suse.cz>
12357 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12360 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12362 * match.pd: New pattern.
12364 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12366 * tree-vrp.c (is_masked_range_test): New function.
12367 (register_edge_assert_for): Determine ranges for
12370 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12372 PR tree-optimization/67328
12373 * fold-const.c (maskable_range_p): New function.
12374 (build_range_check): Generate bittests if possible.
12376 2017-06-13 Martin Liska <mliska@suse.cz>
12378 * gimple-pretty-print.c (dump_probability): Add new argument.
12379 (dump_edge_probability): Dump both probability and count.
12380 (dump_gimple_label): Likewise.
12381 (dump_gimple_bb_header): Likewise.
12383 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12386 * config/avr/avr-devices.c: Fix indentation.
12387 * config/avr/gen-avr-mmcu-specs.c: Dito.
12389 2017-06-13 Richard Biener <rguenther@suse.de>
12391 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12392 instead get vector type from stmt_info.
12393 (vectorizable_reduction): Adjust. Remove dead code.
12395 2017-06-13 Richard Biener <rguenther@suse.de>
12397 PR middle-end/81065
12398 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12399 case of C * (x * C2 + C3).
12400 (fold_addr_of_array_ref_difference): Properly fold index difference.
12402 2017-06-12 David S. Miller <davem@davemloft.net>
12405 * config/sparc/sparc.md (return expander): Emit frame blockage if
12406 function uses alloca.
12408 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12410 * combine.c (make_field_assignment): Check len rather than the mode
12411 precision when calling force_to_mode.
12413 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12415 Support multilibs and devices that see flash in RAM address range.
12418 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12419 (avr_mcu_t) <flash_pm_offset>: New field.
12420 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12421 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12422 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12423 (AVR_TINY_PM_OFFSET): Remove macro.
12424 * config/avr/avr.opt (-mshort-calls): New option.
12425 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12426 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12427 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12428 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12429 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12430 instead of avr_arch->have_jmp_call.
12431 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12432 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12433 avr_arch->flash_pm_offset to define.
12434 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12435 new field flash_pm_offset. Add entry for avrxmega3.
12436 (avr_texinfo): Add entry for avrxmega3.
12437 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12438 attiny212, attiny214,
12439 attiny412, attiny414, attiny416, attiny417,
12440 attiny814, attiny816, attiny817,
12441 attiny1614, attiny1616, attiny1617,
12442 attiny3214, attiny3216, attiny3217.
12443 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12444 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12445 (avr_print_operand_address) [AVR_TINY]: Same.
12446 (avr_asm_init_sections) <readonly_data_section>: Only patch
12447 callback if avr_arch->flash_pm_offset = 0.
12448 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12449 for rodata if avr_arch->flash_pm_offset != 0.
12450 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12451 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12452 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12453 (m_options): Append opt_rcall.
12454 (m_dirnames): Append dir_rcall.
12455 * config/avr/t-multilib: Regenerate.
12457 * configure.ac [target=avr]: Check whether avrxmega3 default
12458 linker description file works as needed.
12459 * configure: Regenerate.
12460 * doc/avr-mmcu.texi: Regenerate.
12461 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12462 <__AVR_ARCH__>: Document avrxmega3 and 103.
12463 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12464 <__AVR_SHORT_CALLS__>: Document it.
12465 <__AVR_PM_BASE_ADDRESS__>: Document it.
12466 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12467 (AVR Variable Attributes) <progmem>: Document this is
12468 not needed for avrxmega3.
12469 (AVR Named Address Spaces) <__flash>: Dito.
12471 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12473 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12475 2017-06-12 Doug Rupp <rupp@adacore.com>
12477 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12478 Append vxworks-stdint.h to the tm_file list.
12479 * config/vxworks-stdint.h: New file.
12481 2017-06-12 Martin Liska <mliska@suse.cz>
12483 PR tree-optimization/81041
12484 * tree-profile.c (gimple_gen_ic_func_profiler):
12485 Create an extra BB in profile-generate
12486 (gimple_gen_time_profiler): Likewise.
12488 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12490 PR tree-optimization/81003
12491 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12492 (update_range_test): Use it instead of force_gimple_operand_gsi.
12494 2017-06-12 Richard Biener <rguenther@suse.de>
12496 PR tree-optimization/81053
12497 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12498 with backedge value not defined in loop. Simplify def stmt
12501 2017-06-11 Tom de Vries <tom@codesourcery.com>
12504 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12506 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12507 nvptx_cannot_force_const_mem.
12509 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12511 * opts.c (finish_options): Move test for flag_split_stack after
12512 it has been initialized.
12514 2017-06-11 Jason Merrill <jason@redhat.com>
12516 * tree.h (id_equal): New.
12517 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12518 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12519 instead of strcmp of IDENTIFIER_POINTER.
12521 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12523 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12524 (mark_all_inlined_calls_cdtor): Fix formating.
12525 (inline_transform): Rescale profile before inlining.
12527 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12529 * cgraph.h (cgraph_edge::clone): Update prototype.
12530 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12531 (cgraph_node::create_clone): Update.
12532 (cgraph_node::create_version_clone): Update.
12533 * tree-inline.c (copy_bb): Update.
12534 (expand_call_inline): Update.
12536 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12538 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12539 factored out from ...
12540 (rs6000_emit_prologue): ... here.
12542 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12544 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12545 factored out from ...
12546 (rs6000_emit_prologue): ... here.
12548 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12550 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12552 (handle_missing_profiles): Fix computation of tp_first_run.
12553 (counts_to_freqs): Do not touch freqs when count is 0.
12555 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12557 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12560 2017-06-10 Tom de Vries <tom@codesourcery.com>
12562 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12563 attributes): Document signal effective target.
12565 2017-06-10 Tom de Vries <tom@codesourcery.com>
12567 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12568 Document effective target stack_size.
12570 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12572 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12573 to the edit_context if they can be auto-applied.
12575 2017-06-9 Ian Lance Taylor <iant@golang.org>
12577 * opts.c (finish_options): If -fsplit-stack, disable implicit
12578 -forder-blocks-and-partition.
12579 * doc/invoke.texi (Optimize Options): Document that when using
12580 -fsplit-stack -forder-blocks-and-partition is not implicitly
12583 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12585 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12586 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12587 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12588 * builtins.def (abort, trap, unreachable): Declare cold.
12589 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12590 * tree-core.h (ECF_COLD): New.
12591 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12592 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12594 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12596 * predict.c (unlikely_executed_stmt_p): Cleanup.
12598 2017-06-09 Richard Biener <rguenther@suse.de>
12600 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12601 model if the ref is always written to.
12603 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12605 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12607 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12609 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12612 2017-06-09 Tom de Vries <tom@codesourcery.com>
12615 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12616 "target cannot support label values" when encountering LABEL_REF.
12618 2017-06-09 Martin Liska <mliska@suse.cz>
12620 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12621 (gimple_gen_ic_func_profiler): Emit direct comparison
12622 of __gcov_indirect_call_callee with NULL.
12623 (gimple_gen_time_profiler): Change probability from
12624 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12626 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12628 * profile.c (edge_gcov_counts): Turn to pointer.
12629 (compute_branch_probabilities, compute_branch_probabilities): Update.
12630 (branch_prob): Do not clear edge_gcov_count.
12631 * profile.h (edge_gcov_counts): Turn to pointer.
12632 (edge_gcov_count): Update.
12634 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12636 * gimple.h (gimple_check_failed): Mark cold.
12638 2017-06-09 Richard Biener <rguenther@suse.de>
12640 PR tree-optimization/66623
12641 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12642 refactor check_reduction into two parts, properly computing
12643 whether we have to check reduction validity for outer loop
12646 2017-06-09 Richard Biener <rguenther@suse.de>
12648 PR tree-optimization/79483
12649 * graphite-scop-detection.c (order): New global.
12650 (get_order): Compute bb to order mapping that satisfies code
12651 generation constraints.
12652 (cmp_pbbs): New helper.
12653 (build_scops): Start domwalk at entry block, sort generated
12656 2017-06-09 Richard Biener <rguenther@suse.de>
12658 PR middle-end/81007
12659 * ipa-polymorphic-call.c
12660 (ipa_polymorphic_call_context::restrict_to_inner_class):
12661 Skip FIELD_DECLs with error_mark_node type.
12662 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12665 2017-06-09 Martin Liska <mliska@suse.cz>
12667 * predict.c (struct branch_predictor): New struct.
12668 (test_prediction_value_range): New test.
12669 (predict_c_tests): New function.
12670 * selftest-run-tests.c (selftest::run_tests): Run the function.
12671 * selftest.h: Declare new tests.
12673 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12676 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12677 gen_add3_insn did not fail.
12678 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12679 r0, construct that number in a temporary reg and add that reg to r0.
12680 If asked to put the result in r0 as well, fail.
12682 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12684 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12685 for early expansion of vec_eqv.
12687 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12689 PR middle-end/81005
12690 * ubsan.c (instrument_null): Avoid pointless code temporary.
12691 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12693 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12697 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12700 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12703 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12705 * predict.c (unlikely_executed_edge_p): Move ahead.
12706 (probably_never_executed_edge_p): Use it.
12708 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12710 PR middle-end/79988
12711 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12712 gimple_call_builtin_p call.
12714 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12716 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12717 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12718 rtl_check_failed_type2, rtl_check_failed_code1,
12719 rtl_check_failed_code2, rtl_check_failed_code_mode,
12720 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12721 rtvec_check_failed_bounds, rtl_check_failed_flag,
12722 _fatal_insn_not_found, _fatal_insn): Likewise.
12723 * tree.h (tree_contains_struct_check_failed,
12724 tree_check_failed, tree_not_check_failed,
12725 tree_class_check_failed, tree_range_check_failed,
12726 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12727 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12728 tree_operand_check_failed, omp_clause_check_failed,
12729 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12732 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12734 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12735 flag_branch_probabilities.
12736 * ipa-inline.c (edge_badness): Likewise.
12737 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12738 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12739 * predict.c (maybe_hot_frequency_p): Likewise.
12740 (probably_never_executed): Likewise.
12741 * sched-ebb.c (schedule_ebbs): Likewise.
12742 * sched-rgn.c (find_single_block_region): Likewise.
12743 * tracer.c (tail_duplicate): Likewise.
12745 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12747 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12748 longer requires x_flag_profile_use.
12750 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12752 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12753 instead of flag_reorder_blocks_and_partition.
12754 * dbxout.c (dbxout_function_end): Likewise.
12755 * dwarf2out.c (gen_subprogram_die): Likewise.
12756 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12757 * hw-doloop.c (reorg_loops): Likewise.
12758 * varasm.c (assemble_start_function,
12759 assemble_end_function): Likewise.
12760 (decide_function_section): Do not check for
12761 flag_reorder_blocks_and_partition.
12763 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12765 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12767 (chkp_get_hard_register_fake_addr_expr): Ditto.
12768 (chkp_build_addr_expr): Add check for hard reg case.
12769 (chkp_parse_array_and_component_ref): Ditto.
12770 (chkp_find_bounds_1): Ditto.
12771 (chkp_process_stmt): Don't generate bounds store for
12774 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12776 * predict.c (maybe_hot_bb_p): Do not check profile status.
12777 (maybe_hot_edge_p): Likewise.
12778 (probably_never_executed): Check for zero counts even if profile
12780 (unlikely_executed_edge_p): New function.
12781 (unlikely_executed_stmt_p): New function.
12782 (unlikely_executed_bb_p): New function.
12783 (set_even_probabilities): Use unlikely predicates.
12784 (combine_predictions_for_bb): Likewise.
12785 (predict_paths_for_bb): Likewise.
12786 (predict_paths_leading_to_edge): Likewise.
12787 (determine_unlikely_bbs): New function.
12788 (estimate_bb_frequencies): Use it.
12789 (compute_function_frequency): Use zero counts even if profile is
12791 * profile-count.h: Fix typo.
12793 2017-08-08 Julia Koval <julia.koval@intel.com>
12795 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12796 _mm512_mask_cvtsepi16_storeu_epi8,
12797 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12798 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12799 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12800 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12801 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12802 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12803 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12804 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12805 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12806 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12807 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12808 __builtin_ia32_pmovuswb256mem_mask,
12809 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12810 __builtin_ia32_pmovwb512mem_mask): New builtins.
12812 2017-08-08 Julia Koval <julia.koval@intel.com>
12814 PR target/73350,80862
12815 * config/i386/subst.md (round): Fix round pattern.
12816 * config/i386/i386.c (ix86_erase_embedded_rounding):
12817 Fix erasing rounding for the fixed pattern.
12819 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12821 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12823 2017-06-08 Martin Liska <mliska@suse.cz>
12825 PR gcov-profile/80911
12826 * gcov.c (block_info::block_info): New constructor.
12828 2017-06-07 Carl Love <cel@us.ibm.com>
12830 * config/rs6000/rs6000-c: The return type of the following
12831 built-in functions was implemented as int not long long. Fix sign
12832 of return value for the unsigned version of vec_mulo and vec_mule.
12833 vector unsigned long long vec_bperm (vector unsigned long long,
12834 vector unsigned char)
12835 vector signed long long vec_mule (vector signed int,
12837 vector unsigned long long vec_mule (vector unsigned int,
12838 vector unsigned int)
12839 vector signed long long vec_mulo (vector signed int,
12841 vector unsigned long long vec_mulo (vector unsigned int,
12842 vector unsigned int)
12843 * doc/extend.texi: Fix the documentation for the built-in
12846 2017-06-07 Carl Love <cel@us.ibm.com>
12849 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12852 2017-06-07 Carl Love <cel@us.ibm.com>
12854 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12855 support, Generate doublehv for signed int/float for BE case only.
12857 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12859 * doc/invoke.texi (mcx16): Rewrite.
12861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12863 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12864 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12865 *mov<mode>_softfloat, and an anonymous splitter): Use
12866 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12868 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12870 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12872 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12873 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12874 (rs6000_debug_reg_global): Adjust.
12875 (rs6000_init_hard_regno_mode_ok): Adjust.
12876 (rs6000_dbx_register_number): Adjust.
12877 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12878 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12879 Remove SPE_ACC and SPEFSCR.
12880 (REG_ALLOC_ORDER): Ditto.
12881 (FRAME_POINTER_REGNUM): Change to 111.
12882 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12883 (REG_CLASS_NAMES): Ditto.
12884 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12885 (REGISTER_NAMES): Ditto.
12886 (ADDITIONAL_REG_NAMES): Ditto.
12887 (rs6000_reg_names): Ditto.
12888 * config/rs6000/rs6000.md: Renumber some register number
12891 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12893 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12895 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12896 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12898 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12899 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12900 Delete the SPE high registers.
12901 (REG_ALLOC_ORDER): Ditto.
12902 (enum reg_class): Remove SPE_HIGH_REGS.
12903 (REG_CLASS_NAMES): Ditto.
12904 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12905 (REGISTER_NAMES): Ditto.
12906 (rs6000_reg_names): Ditto.
12907 * doc/tm.texi.in: Remove SPE as example.
12908 * doc/tm.texi: Regenerate.
12910 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12912 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12913 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12914 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12915 * config/rs6000/rs6000.md (type): Remove "brinc".
12917 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12919 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12920 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12921 * config/rs6000/linuxspe.h: Delete file.
12922 * config/rs6000/rs6000.md: Don't include spe.md.
12923 * config/rs6000/spe.h: Delete file.
12924 * config/rs6000/spe.md: Delete file.
12925 * config/rs6000/t-rs6000: Remove spe.md.
12927 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12929 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12930 (reg_or_none500mem_operand): Delete.
12931 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12932 instead of reg_or_none500mem_operand.
12934 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12936 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12937 handling of SPE flags.
12938 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12940 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12942 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12944 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12945 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12946 paired_divv2sf3): Similar.
12947 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12948 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12949 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12951 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12952 Rename the paired_* instruction patterns.
12953 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12955 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12956 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12957 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12958 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12959 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12960 PAIRED_VECTOR_MODE.
12961 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12962 (spe_func_has_64bit_regs_p): Delete.
12963 (spe_expand_predicate_builtin): Delete.
12964 (spe_expand_evsel_builtin): Delete.
12965 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12966 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12967 (invalid_e500_subreg): Delete.
12968 (rs6000_legitimize_address): Always force_reg op2 as well, for
12969 paired single memory accesses.
12970 (rs6000_member_type_forces_blk): Delete.
12971 (rs6000_spe_function_arg): Delete.
12972 (rs6000_expand_unop_builtin): Delete SPE handling.
12973 (rs6000_expand_binop_builtin): Ditto.
12974 (spe_expand_stv_builtin): Delete.
12975 (bdesc_2arg_spe): Delete.
12976 (spe_expand_builtin): Delete.
12977 (spe_expand_predicate_builtin): Delete.
12978 (spe_expand_evsel_builtin): Delete.
12979 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12980 (spe_init_builtins): Delete.
12981 (spe_func_has_64bit_regs_p): Delete.
12982 (savres_routine_name): Delete "info" parameter. Adjust callers.
12983 (rs6000_emit_stack_reset): Ditto.
12984 (rs6000_dwarf_register_span): Delete.
12985 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12986 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12987 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12989 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12991 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12992 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12993 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12994 mulv2sf3, divv2sf3): Delete expanders.
12996 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12998 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13000 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13002 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13003 * config/rs6000/rs6000.c: Ditto.
13005 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13007 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13008 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13009 comparison_operator.
13011 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13013 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13014 * config/rs6000/rs6000.opt: Ditto.
13015 * config/rs6000/t-rtems: Ditto.
13017 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13019 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13020 TARGET_E500_SINGLE by 0, simplify.
13021 * config/rs6000/rs6000.c: Ditto.
13022 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13023 (spe_build_register_parallel): Delete.
13024 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13025 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13026 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13027 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13028 (E500_CONVERT): Delete.
13029 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13031 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13033 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13034 * config/rs6000/dfp.md: Ditto.
13035 (negdd2, *negdd2_fpr): Merge.
13036 (absdd2, *absdd2_fpr): Merge.
13037 (negtd2, *negtd2_fpr): Merge.
13038 (abstd2, *abstd2_fpr): Merge.
13039 * config/rs6000/e500.h: Delete file.
13040 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13041 TARGET_FPRS by 1 and simplify.
13042 * config/rs6000/rs6000-c.c: Ditto.
13043 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13044 TARGET_DF_SPE by 0.
13045 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13047 * config/rs6000/rs6000.md: Ditto.
13048 (floatdidf2, *floatdidf2_fpr): Merge.
13049 (move_from_CR_gt_bit): Delete.
13050 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13051 (E500_CR_IOR_COMPARE): Delete.
13052 (All patterns that require !TARGET_FPRS): Delete.
13053 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13055 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13057 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13059 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13061 * graphds.c (add_edge): Intitialize edge's attached data.
13062 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13063 pointer parameter. Call pointed function on each edge during
13064 graph traversing. Skip traversing the edge when the function
13066 (graphds_dfs, graphds_scc): Ditto.
13067 (for_each_edge): New parameter. Pass the new parameter to callback
13069 * graphds.h (skip_edge_callback): New function pointer type.
13070 (graphds_dfs, graphds_scc): New function pointer parameter.
13071 (graphds_edge_callback, for_each_edge): New parameter.
13073 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13075 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13076 out code checking if runtime alias check is possible to below ...
13077 Call the new function.
13078 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13079 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13081 2017-06-07 Marek Polacek <polacek@redhat.com>
13084 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13085 TYPE_OVERFLOW_WRAPS check.
13087 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13089 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13090 if versioning is required.
13091 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13092 peeling with the check for versioning.
13094 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13096 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13097 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13098 Set true to new parameter if new ssa variable is defined.
13099 (vect_gen_vector_loop_niters): Refactor. Set range information
13100 for the new vector loop bound variable.
13101 (vect_do_peeling): Ditto.
13103 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13105 * tree-affine.c (ssa.h): Include header file.
13106 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13107 has wrapping overflow behavior.
13109 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13111 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13113 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13115 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13116 (tree_to_aff_combination): ... here.
13118 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13120 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13121 reg_pressure model function.
13122 (ivopts_global_cost_for_size): Delete.
13123 (determine_set_costs, iv_ca_recount_cost): Call new model function
13124 ivopts_estimate_reg_pressure.
13126 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13128 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13129 expensive than udiv. Remove floating point cases from mod.
13131 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13133 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13134 Increase idiv cost.
13136 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13138 * config/aarch64/aarch64.md
13139 (copysignsf3): Fix mask generation.
13141 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13143 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13145 (class dump_manager): Add register_dumps method.
13146 * dumpfile.c: Include langhooks.h.
13147 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13148 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13149 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13150 (dump_manager::dump_register): Start with 512 entries instead of 32.
13151 (dump_manager::register_dumps): New method.
13152 * toplev.c (general_init): Instead of invoking register_dumps
13153 langhook, invoke register_dumps method on the dump manager.
13154 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13157 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13159 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13160 Say that # requires an associated define_split to exist, and that
13161 the define_split must be suitable for use after register allocation.
13163 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13165 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13166 (compute_outgoing_frequencies): Also initialize zero counts.
13167 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13168 around loops; preserve more of profile when nothing changes.
13170 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13172 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13174 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13175 * config/arm/arm-cpu-cdata.h: Regenerate.
13176 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13177 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13178 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13179 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13180 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13182 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13184 * config/arm/t-rmprofile: Likewise.
13185 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13187 2017-06-06 David S. Miller <davem@davemloft.net>
13190 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13191 blockage if function uses alloca.
13193 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13195 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13196 New "uid" fields to hold pretty-print IDs of group and ref.
13197 Memory references are now identified as <group_id>:<ref_id>
13198 instead of using [random] addresses.
13199 (dump_mem_details): Simplify, no functional change.
13200 (dump_mem_ref): Simplify and make output more concise.
13201 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13202 (find_or_create_group): Initialize group uid.
13203 (record_ref): Initialize ref uid. Improve debug output.
13204 (prune_group_by_reuse, should_issue_prefetch_p,)
13205 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13206 (mark_nontemporal_store, determine_loop_nest_reuse):
13207 Improve debug output.
13209 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13211 * dbgcnt.def (prefetch): New debug counter.
13212 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13213 (schedule_prefetches): Stop issueing prefetches if debug counter
13216 2017-06-06 Tom de Vries <tom@codesourcery.com>
13218 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13219 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13221 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13223 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13224 Use aarch64_reg_or_zero predicate for operand 4.
13225 (aarch64_compare_and_swap<mode> define_insn_and_split):
13226 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13227 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13229 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13231 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13232 (arm_compute_save_core_reg_mask): This.
13233 (thumb1_compute_save_reg_mask): Rename into ...
13234 (thumb1_compute_save_core_reg_mask): This.
13235 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13236 (arm_compute_frame_layout): Likewise.
13238 2017-06-06 Richard Biener <rguenther@suse.de>
13240 PR tree-optimization/80974
13241 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13242 keep or clear leaders SSA info.
13244 2017-06-06 Tom de Vries <tom@codesourcery.com>
13246 * config/nvptx/nvptx.c (split_mode_p): New function.
13247 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13249 2017-06-06 Tom de Vries <tom@codesourcery.com>
13251 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13253 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13256 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13259 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13261 * shrink-wrap.c (handle_simple_exit): Update profile.
13262 (try_shrink_wrapping): Upate profile.
13264 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13266 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13267 (tree_guess_outgoing_edge_probabilities): New.
13268 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13269 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13271 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13273 * ipa-split.c (split_function): Initialize return bb profile.
13275 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13277 * profile.c (compute_branch_probabilities): Also initialize
13278 EXIT_BLOCK profile.
13280 2017-06-06 Richard Biener <rguenther@suse.de>
13282 PR tree-optimization/80928
13283 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13284 (vect_analyze_loop_operations): Properly guard analysis for
13286 (vect_transform_loop): Likewise.
13287 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13288 (vect_model_induction_cost): Do not cost for pure SLP.
13289 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13290 of induction in inner loop vectorization.
13291 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13292 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13293 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13295 (vect_analyze_slp_cost_1): Cost induction.
13296 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13297 (vect_get_slp_vect_defs): Likewise.
13298 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13299 (vect_transform_stmt): Handle SLP reductions.
13300 * tree-vectorizer.h (vectorizable_induction): Adjust.
13302 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13304 * config/rs6000/rs6000.c (make_resolver_func): Update
13305 init_lowered_empty_function call.
13307 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13309 * doc/invoke.texi: Document the -fprofile-abs-path option.
13310 * common.opt (fprofile-abs-path): New option.
13311 * gcov-io.h (gcov_write_filename): Declare.
13312 * gcov-io.c (gcov_write_filename): New function.
13313 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13314 * profile.c (output_location): Likewise.
13316 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13318 * shring-wrap.c: Revert accidental commit.
13320 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13322 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13324 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13326 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13328 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13329 profile in callgraph edge.
13330 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13331 (apply_scale): Likewise.
13332 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13333 Also scale profile when inlining function with zero profile.
13334 (initialize_cfun): Update exit block profile even when it is zero.
13335 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13336 when profile is read.
13338 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13340 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13341 (CLONE_*): New constants to define the processors we can generate
13342 code for with the target_clone attribute.
13343 (rs6000_clone_map): New array to identify which clone processors
13344 the current program is running on.
13345 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13346 target_clone attribute.
13347 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13348 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13349 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13350 (cpu_expand_builtin): Add support for target_clone attribute.
13351 (rs6000_valid_attribute_p): Allow "default" attribute.
13352 (get_decl_name): New debug function to simplify printing the
13353 current function name in debugging statements.
13354 (rs6000_clone_priority): New functions to support the target_clone
13355 attribute, and be able to generate code to switch between ISA 2.05
13356 through ISA 3.0 (power6 through power9).
13357 (rs6000_compare_version_priority): Likewise.
13358 (rs6000_get_function_versions_dispatcher): Likewise.
13359 (make_resolver_func): Likewise.
13360 (add_condition_to_bb): Likewise.
13361 (dispatch_function_versions): Likewise.
13362 (rs6000_generate_version_dispatcher_body): Likewise.
13363 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13364 (fusion_gpr_load_p): Fix a spacing issue.
13365 * doc/extend.texi (Common Function Attributes): Document that the
13366 PowerPC supports the target_clone attribute.
13368 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13370 * config/arm/arm.h: explain F symbol found in description of ARM
13371 register allocation in its legend.
13373 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13375 * config/mips/frame-header-opt.c: Include profile-count.h.
13376 * config/riscv/riscv.c: Include profile-count.h
13378 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13380 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13382 (sm_set_flag_if_changed): Add bbs field.
13383 (execute_sm_if_changed_flag_set): Pass BBS.
13384 (execute_sm): Update.
13386 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13388 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13391 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13393 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13394 (peephole2): New peephole2 to emit the above.
13395 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13397 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13399 * config/aarch64/aarch64.c (define_peephole2 above
13400 *sub_<shift>_<mode>): New peephole.
13402 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13404 * config/i386/i386.c (make_resolver_func): Update.
13405 * Makefile.in: Add profile-count.h and profile-count.o
13406 * auto-profile.c (afdo_indirect_call): Update to new API.
13407 (afdo_set_bb_count): Update.
13408 (afdo_propagate_edge): Update.
13409 (afdo_propagate_circuit): Update.
13410 (afdo_calculate_branch_prob): Update.
13411 (afdo_annotate_cfg): Update.
13412 * basic-block.h: Include profile-count.h
13413 (struct edge_def): Turn count to profile_count.
13414 (struct basic_block_def): Likewie.
13415 (REG_BR_PROB_BASE): Move to profile-count.h
13416 (RDIV): Move to profile-count.h
13417 * bb-reorder.c (max_entry_count): Turn to profile_count.
13418 (find_traces): Update.
13419 (rotate_loop):Update.
13420 (connect_traces):Update.
13421 (sanitize_hot_paths):Update.
13422 * bt-load.c (migrate_btr_defs): Update.
13423 * cfg.c (RDIV): Remove.
13424 (init_flow): Use alloc_block.
13425 (alloc_block): Uninitialize count.
13426 (unchecked_make_edge): Uninitialize count.
13427 (check_bb_profile): Update.
13428 (dump_edge_info): Update.
13429 (dump_bb_info): Update.
13430 (update_bb_profile_for_threading): Update.
13431 (scale_bbs_frequencies_int): Update.
13432 (scale_bbs_frequencies_gcov_type): Update.
13433 (scale_bbs_frequencies_profile_count): New.
13434 * cfg.h (update_bb_profile_for_threading): Update.
13435 (scale_bbs_frequencies_profile_count): Declare.
13436 * cfgbuild.c (compute_outgoing_frequencies): Update.
13437 (find_many_sub_basic_blocks): Update.
13438 * cfgcleanup.c (try_forward_edges): Update.
13439 (try_crossjump_to_edge): Update.
13440 * cfgexpand.c (expand_gimple_tailcall): Update.
13441 (construct_exit_block): Update.
13442 * cfghooks.c (verify_flow_info): Update.
13443 (dump_bb_for_graph): Update.
13444 (split_edge): Update.
13445 (make_forwarder_block): Update.
13446 (duplicate_block): Update.
13447 (account_profile_record): Update.
13448 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13449 (get_estimated_loop_iterations): Update.
13450 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13451 (single_likely_exit): Update.
13452 * cfgloopmanip.c (scale_loop_profile): Update.
13454 (set_zero_probability): Update.
13455 (lv_adjust_loop_entry_edge): Update.
13456 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13457 (purge_dead_edges): Update.
13458 (rtl_account_profile_record): Update.
13459 * cgraph.c (cgraph_node::create): Uninitialize count.
13460 (symbol_table::create_edge): Uninitialize count.
13461 (cgraph_update_edges_for_call_stmt_node): Update.
13462 (cgraph_edge::dump_edge_flags): Update.
13463 (cgraph_node::dump): Update.
13464 (cgraph_edge::maybe_hot_p): Update.
13465 * cgraph.h: Include profile-count.h
13466 (create_clone), create_edge, create_indirect_edge): Update.
13467 (cgraph_node): Turn count to profile_count.
13468 (cgraph_edge0: Likewise.
13469 (make_speculative, clone): Update.
13470 (create_edge): Update.
13471 (init_lowered_empty_function): Update.
13472 * cgraphclones.c (cgraph_edge::clone): Update.
13473 (duplicate_thunk_for_node): Update.
13474 (cgraph_node::create_clone): Update.
13475 * cgraphunit.c (cgraph_node::analyze): Update.
13476 (cgraph_node::expand_thunk): Update.
13477 * final.c (dump_basic_block_info): Update.
13478 * gimple-streamer-in.c (input_bb): Update.
13479 * gimple-streamer-out.c (output_bb): Update.
13480 * graphite.c (print_global_statistics): Update.
13481 (print_graphite_scop_statistics): Update.
13482 * hsa-brig.c: Include basic-block.h.
13483 * hsa-dump.c: Include basic-block.h.
13484 * hsa-gen.c (T sum_slice): Update.
13485 (convert_switch_statements):Update.
13486 * hsa-regalloc.c: Include basic-block.h.
13487 * ipa-chkp.c (chkp_produce_thunks): Update.
13488 * ipa-cp.c (struct caller_statistics): Update.
13489 (init_caller_stats): Update.
13490 (gather_caller_stats): Update.
13491 (ipcp_cloning_candidate_p): Update.
13492 (good_cloning_opportunity_p): Update.
13493 (get_info_about_necessary_edges): Update.
13494 (dump_profile_updates): Update.
13495 (update_profiling_info): Update.
13496 (update_specialized_profile): Update.
13497 (perhaps_add_new_callers): Update.
13498 (decide_about_value): Update.
13499 (ipa_cp_c_finalize): Update.
13500 * ipa-devirt.c (struct odr_type_warn_count): Update.
13501 (struct decl_warn_count): Update.
13502 (struct final_warning_record): Update.
13503 (possible_polymorphic_call_targets): Update.
13504 (ipa_devirt): Update.
13505 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13506 * ipa-icf.c (sem_function::merge): Update.
13507 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13508 * ipa-inline.c (compute_uninlined_call_time): Update.
13509 (compute_inlined_call_time): Update.
13510 (want_inline_small_function_p): Update.
13511 (want_inline_self_recursive_call_p): Update.
13512 (edge_badness): Update.
13513 (lookup_recursive_calls): Update.
13514 (recursive_inlining): Update.
13515 (inline_small_functions): Update.
13516 (dump_overall_stats): Update.
13517 (dump_inline_stats): Update.
13518 * ipa-profile.c (ipa_profile_generate_summary): Update.
13519 (ipa_propagate_frequency): Update.
13520 (ipa_profile): Update.
13521 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13522 * ipa-utils.c (ipa_merge_profiles): Update.
13523 * loop-doloop.c (doloop_modify): Update.
13524 * loop-unroll.c (report_unroll): Update.
13525 (unroll_loop_runtime_iterations): Update.
13526 * lto-cgraph.c (lto_output_edge): Update.
13527 (lto_output_node): Update.
13528 (input_node): Update.
13529 (input_edge): Update.
13530 (merge_profile_summaries): Update.
13531 * lto-streamer-in.c (input_cfg): Update.
13532 * lto-streamer-out.c (output_cfg): Update.
13533 * mcf.c (create_fixup_graph): Update.
13534 (adjust_cfg_counts): Update.
13535 (sum_edge_counts): Update.
13536 * modulo-sched.c (sms_schedule): Update.
13537 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13538 * predict.c (maybe_hot_count_p): Update.
13539 (probably_never_executed): Update.
13540 (dump_prediction): Update.
13541 (combine_predictions_for_bb): Update.
13542 (propagate_freq): Update.
13543 (handle_missing_profiles): Update.
13544 (counts_to_freqs): Update.
13545 (rebuild_frequencies): Update.
13546 (force_edge_cold): Update.
13547 * predict.h: Include profile-count.h
13548 (maybe_hot_count_p, counts_to_freqs): UPdate.
13549 * print-rtl-function.c: Do not include cfg.h
13550 * print-rtl.c: Include basic-block.h
13551 * profile-count.c: New file.
13552 * profile-count.h: New file.
13553 * profile.c (is_edge_inconsistent): Update.
13554 (correct_negative_edge_counts): Update.
13555 (is_inconsistent): Update.
13556 (set_bb_counts): Update.
13557 (read_profile_edge_counts): Update.
13558 (compute_frequency_overlap): Update.
13559 (compute_branch_probabilities): Update; Initialize and deinitialize
13561 (branch_prob): Update.
13562 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13563 (edge_gcov_count): New.
13564 (bb_gcov_count): New.
13565 * shrink-wrap.c (try_shrink_wrapping): Update.
13566 * tracer.c (better_p): Update.
13567 * trans-mem.c (expand_transaction): Update.
13568 (ipa_tm_insert_irr_call): Update.
13569 (ipa_tm_insert_gettmclone_call): Update.
13570 * tree-call-cdce.c: Update.
13571 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13572 (gimple_duplicate_sese_tail): Update.
13573 (gimple_account_profile_record): Update.
13574 (execute_fixup_cfg): Update.
13575 * tree-inline.c (copy_bb): Update.
13576 (copy_edges_for_bb): Update.
13577 (initialize_cfun): Update.
13578 (freqs_to_counts): Update.
13579 (copy_cfg_body): Update.
13580 (expand_call_inline): Update.
13581 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13582 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13583 (try_unroll_loop_completely): Update.
13584 (try_peel_loop): Update.
13585 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13586 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13587 * tree-ssa-loop-split.c (connect_loops): Update.
13588 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13589 * tree-ssa-reassoc.c (branch_fixup): Update.
13590 * tree-ssa-tail-merge.c (replace_block_by): Update.
13591 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13592 (compute_path_counts): Update.
13593 (update_profile): Update.
13594 (recompute_probabilities): Update.
13595 (update_joiner_offpath_counts): Update.
13596 (estimated_freqs_path): Update.
13597 (freqs_to_counts_path): Update.
13598 (clear_counts_path): Update.
13599 (ssa_fix_duplicate_block_edges): Update.
13600 (duplicate_thread_path): Update.
13601 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13602 (struct switch_conv_info): Update.
13603 * tree-tailcall.c (decrease_profile): Update.
13604 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13605 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13606 * value-prof.c (check_counter): Update.
13607 (gimple_divmod_fixed_value): Update.
13608 (gimple_mod_pow2): Update.
13609 (gimple_mod_subtract): Update.
13610 (gimple_ic_transform): Update.
13611 (gimple_stringop_fixed_value): Update.
13612 * value-prof.h (gimple_ic): Update.
13614 2017-06-02 Carl Love <cel@us.ibm.com>
13616 * config/rs6000/rs6000-c: Add support for built-in functions
13617 vector double vec_doublee (vector signed int);
13618 vector double vec_doublee (vector unsigned int);
13619 vector double vec_doublee (vector float);
13620 vector double vec_doubleh (vector signed int);
13621 vector double vec_doubleh (vector unsigned int);
13622 vector double vec_doubleh (vector float);
13623 vector double vec_doublel (vector signed int);
13624 vector double vec_doublel (vector unsigned int);
13625 vector double vec_doublel (vector float);
13626 vector double vec_doubleo (vector signed int);
13627 vector double vec_doubleo (vector unsigned int);
13628 vector double vec_doubleo (vector float);.
13629 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13630 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13632 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13633 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13634 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13636 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13637 vec_doublel, vec_doubleh.
13638 * doc/extend.texi: Update the built-in documentation file for the
13639 new built-in functions.
13641 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13644 * ipa-inline-analysis.c (free_growth_caches): Set
13645 edge_removal_hook_holder to NULL after removing it.
13647 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13649 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13650 comparision with zero.
13652 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13653 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13654 for early expansion of vec_min and vec_max builtins.
13655 (builtin_function_type): Add min/max unsigned variants to those
13656 identified as having unsigned arguments.
13658 2017-06-02 Olivier Hainque <hainque@adacore.com>
13660 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13662 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13664 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13665 Use VALL_F16 iterator rather than VALL.
13667 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13669 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13670 Emit CBNZ inside loop when doing a strong exchange and comparing
13671 against zero. Generate the CC flags after the loop.
13673 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13675 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13676 (dl_section_ref): New.
13677 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13678 On AIX, append an expression to subtract the size of the
13679 section length to dl_section_ref.
13681 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13683 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13684 for early expansion of vector absolute builtins.
13686 2017-06-02 Richard Biener <rguenther@suse.de>
13688 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13689 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13691 2017-06-02 Richard Biener <rguenther@suse.de>
13693 PR tree-optimization/80948
13694 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13695 stmt order as well.
13697 2017-06-02 Richard Biener <rguenther@suse.de>
13699 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13701 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13702 for inductions as relevant.
13704 2017-06-02 Richard Biener <rguenther@suse.de>
13706 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13707 (vectorizable_induction): ... this. Remove dead code.
13709 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13711 * builtins. (expand_builtin_alloca): Remove second parameter and
13712 infer its value from the first parameter instead.
13713 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13715 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13717 PR rtl-optimization/80903
13718 * loop-doloop.c (add_test): Unshare sequence.
13720 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13722 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13724 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13726 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13728 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13729 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13730 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13731 (xlogue_layout::get_stub_rtx): Make static.
13732 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13733 (xlogue_layout::compute_stub_managed_regs): Rename to...
13734 (xlogue_layout::count_stub_managed_regs): ...this.
13735 (xlogue_layout::is_stub_managed_reg): New function.
13736 (xlogue_layout::m_stub_names): Rename to...
13737 (xlogue_layout::s_stub_names): ...this, make static.
13738 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13739 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13740 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13741 xlogue_layout::s_stub_names): Instantiate statics.
13742 (stub_managed_regs): Remove.
13743 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13744 (disable_call_ms2sysv_xlogues): Rename to...
13745 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13746 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13748 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13749 change after reload_completed.
13750 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13752 (ix86_expand_prologue): Likewise.
13753 (ix86_expand_epilogue): Likewise.
13754 (ix86_expand_split_stack_prologue): Likewise.
13755 (ix86_compute_frame_layout): Remove frame parameter ...
13756 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13757 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13759 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13760 (ix86_frame): Move from here ...
13761 * config/i386/i386.h (ix86_frame): ... to here.
13762 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13763 complete ix86_frame data structure instead. Remove some_ld_name.
13765 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13767 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13768 symbols that hold a DECL_VALUE_EXPR.
13770 2017-06-01 Martin Jambor <mjambor@suse.cz>
13772 PR tree-optimization/80898
13773 * tree-sra.c (process_subtree_disqualification): Removed.
13774 (disqualify_candidate): Do not acll
13775 process_subtree_disqualification.
13776 (subtree_mark_written_and_enqueue): New function.
13777 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13778 RHS has been disqualified and re-queue LHS if necessary. Apart
13779 from that, ignore disqualified RHS.
13781 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13783 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13784 address fetch for z10 or later.
13786 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13788 * config/arc/arc.md (tst_movb): Add guard when splitting.
13790 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13792 * config/arc/arc.c (arc_can_eliminate): Test against
13793 arc_frame_pointer_needed.
13795 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13797 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13798 to prevent store reordering.
13799 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13800 (type): Add block type.
13801 (stack_tie): Define special instruction to be used in
13804 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13806 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13807 constraint. It is not valid for the pattern.
13808 (noncommutative_binary_comparison): Likewise.
13810 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13812 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13815 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13817 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13818 be used by the reg-alloc.
13820 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13822 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13823 reg-alloc when having mul64 or mul32x16 instructions.
13824 (mulsidi3): Likewise.
13825 (umulsidi3): Likewise.
13826 (mulsi32x16): New pattern.
13827 (mulsi64): Likewise.
13828 (mulsidi64): Likewise.
13829 (umulsidi64): Likewise.
13830 (MUL32x16_REG): Define.
13831 (mul64_600): Use MUL32x16_REG.
13832 (mac64_600): Likewise.
13833 (umul64_600): Likewise.
13834 (umac64_600): Likewise.
13836 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13838 * config/arc/arc.md (mulsi3_700): Make it commutative.
13840 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13842 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13844 (*sign_extendsidi2_insn): Likewise for movstosw.
13846 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13848 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13849 the type of the input discriminant value. Convert the
13850 discriminant value of signedness vary.
13852 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13854 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13855 Add to -Wall section.
13857 2017-06-01 Richard Biener <rguenther@suse.de>
13859 PR middle-end/66313
13860 * fold-const.c (fold_plusminus_mult_expr): If the factored
13861 factor may be zero use a wrapping type for the inner operation.
13862 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13863 and handle moved defs.
13864 (process_assignment): Properly guard the unary op case. Return a
13865 tri-state indicating that moving the stmt before the call may allow
13866 to continue. Pass through to_move.
13867 (find_tail_calls): Handle moving unrelated defs before
13870 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13873 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13874 splitter result in the canonical way.
13876 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13878 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13879 also for 32bit target. Update insn attributes.
13880 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13882 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13884 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13885 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13886 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13887 (_mm_maskz_min_ss): New intrinsics.
13889 2017-05-31 Martin Liska <mliska@suse.cz>
13891 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13892 Change comment style to one we normally use.
13893 (vectorizable_reduction): Likewise.
13894 (vectorizable_induction): Likewise.
13895 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13896 (vectorizable_call): Likewise.
13897 (vectorizable_simd_clone_call): Likewise.
13898 (vectorizable_conversion): Likewise.
13899 (vectorizable_assignment): Likewise.
13900 (vectorizable_shift): Likewise.
13901 (vectorizable_operation): Likewise.
13902 (vectorizable_store): Likewise.
13903 (vectorizable_load): Likewise.
13904 * tree-vectorizer.h: Likewise.
13906 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13908 * passes.c (emergency_dump_function): New.
13909 * tree-pass.h (emergency_dump_function): Declare.
13910 * plugin.c (plugins_internal_error_function): Remove.
13911 * plugin.h (plugins_internal_error_function): Remove declaration.
13912 * toplev.c (internal_error_function): New static function. Use it...
13913 (general_init): ...here.
13915 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13917 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13918 (arc_rtx_costs): Add costs for new patterns.
13919 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13920 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13922 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13924 * tree-ssa-strlen.c (get_next_strinfo): New function.
13925 (get_stridx_plus_constant): Use it.
13926 (zero_length_string): Likewise.
13927 (adjust_related_strinfos): Likewise.
13928 (adjust_last_stmt): Likewise.
13930 2017-05-31 Richard Biener <rguenther@suse.de>
13933 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13934 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13936 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13938 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13939 loop_vinfo argument and use of dependence distance vectors.
13940 Check instead whether the two references differ only in their
13941 initial value and assume that they have the same alignment if the
13942 difference is a multiple of the vector alignment.
13943 (vect_analyze_data_refs_alignment): Update call accordingly.
13945 2017-05-31 Martin Liska <mliska@suse.cz>
13948 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13950 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13952 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13953 (create_intersect_range_checks): Move from ...
13954 * tree-data-ref.c (create_intersect_range_checks_index)
13955 (create_intersect_range_checks): ... to here.
13956 (create_runtime_alias_checks): New function factored from ...
13957 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13958 here. Call above function.
13959 * tree-data-ref.h (create_runtime_alias_checks): New function.
13961 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13963 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13964 segment length for dr_b and compute it in wide_int.
13966 2017-05-31 Richard Biener <rguenther@suse.de>
13968 PR tree-optimization/80906
13969 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13970 and pass through iv_map.
13971 (copy_bb_and_scalar_dependences): Adjust.
13972 (translate_pending_phi_nodes): Likewise.
13973 (copy_loop_close_phi_args): Handle code-generating IVs instead
13976 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13978 * diagnostic-color.c (color_dict): Add "type-diff".
13979 (parse_gcc_colors): Update comment.
13980 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13981 -fdiagnostics-show-template-tree and -fno-elide-type.
13982 (GCC_COLORS): Add type-diff to example.
13984 (-fdiagnostics-show-template-tree): New.
13985 (-fno-elide-type): New.
13986 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13987 the pp_format_decoder callback. Call any m_format_postprocessor's
13989 (pretty_printer::pretty_printer): Initialize
13990 m_format_postprocessor.
13991 (pretty_printer::~pretty_printer): Delete any
13992 m_format_postprocessor.
13993 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13994 (class format_postprocessor): New class.
13995 (struct pretty_printer::format_decoder): Document the new parameters.
13996 (struct pretty_printer::m_format_postprocessor): New field.
13997 * tree-diagnostic.c (default_tree_printer): Update for new
13998 bool and const char ** params.
13999 * tree-diagnostic.h (default_tree_printer): Likewise.
14001 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14003 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14004 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14005 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14006 rs6000_gen_cell_microcode code.
14007 (rs6000_final_prescan_insn): Delete.
14008 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14009 "warn-cell-microcode" entries.
14010 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14011 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14012 throughout. Change cc_reg_not_micro_cr0_operand to
14013 cc_reg_not_cr0_operand throughout.
14014 (*extendhi<mode>2_noload): Delete.
14015 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14016 (mwarn-cell-microcode): Delete.
14017 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14018 -mgen-cell-microcode and -mwarn-cell-microcode.
14020 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14023 * config/i386/constraints.md (Yd): New constraint.
14025 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14026 and (?Yd, r) alternatives. Update insn attributes.
14027 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14028 and (?*Yd, r) alternatives. Update insn attributes.
14029 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14031 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14033 * gimplify.c (gimplify_modify_expr): Don't create a
14034 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14037 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14039 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14041 2017-05-30 Richard Biener <rguenther@suse.de>
14043 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14044 and reduc_def fields.
14045 (STMT_VINFO_REDUC_TYPE): New define.
14046 (STMT_VINFO_REDUC_DEF): Likewise.
14047 (vect_force_simple_reduction): Adjust prototype.
14048 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14049 (vect_is_simple_reduction): Remove check_reduction argument.
14050 (vect_force_simple_reduction): Adjust and set
14051 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14052 (vectorizable_reduction): Do not re-do reduction analysis
14053 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14054 * tree-parloops.c (gather_scalar_reductions): Adjust.
14056 2017-05-30 Richard Biener <rguenther@suse.de>
14058 PR middle-end/80901
14059 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14062 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14064 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14065 Introduce unknown_misalignment parameter and remove vf.
14066 (vect_peeling_hash_get_lowest_cost):
14067 Pass unknown_misalignment parameter.
14068 (vect_enhance_data_refs_alignment):
14069 Fix unsupportable data ref treatment.
14071 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14073 * tree-vect-data-refs.c (vect_get_data_access_cost):
14074 Workaround for SLP handling.
14075 (vect_enhance_data_refs_alignment):
14076 Compute costs for doing no peeling at all, compare to the best
14077 peeling costs so far and avoid peeling if cheaper.
14079 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14081 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14082 Return peeling info and set costs to zero for unlimited cost
14084 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14085 with unknown misalignment. Compute and costs for unknown
14086 misalignment, compare them to the costs for known misalignment
14087 and choose the cheapest for peeling.
14089 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14091 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14092 (vect_get_peeling_costs_all_drs): Create function.
14093 (vect_peeling_hash_get_lowest_cost):
14094 Use vect_get_peeling_costs_all_drs.
14095 (vect_peeling_supportable): Create function.
14096 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14098 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14100 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14101 DR_HAS_NEGATIVE_STEP.
14102 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14103 (vect_enhance_data_refs_alignment): Use.
14104 (vect_duplicate_ssa_name_ptr_info): Use.
14105 * tree-vectorizer.h (dr_misalignment): Use.
14106 (known_alignment_for_access_p): Use.
14108 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14111 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14113 (has_section_name): New function.
14115 2017-05-30 Martin Liska <mliska@suse.cz>
14118 * auto-profile.c (get_function_decl_from_block): Fix
14121 2017-05-30 Richard Biener <rguenther@suse.de>
14123 PR middle-end/80876
14124 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14126 2017-05-30 Martin Liska <mliska@suse.cz>
14128 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14129 * dumpfile.h (struct dump_file_info): Remove ctors.
14131 2017-05-30 Martin Liska <mliska@suse.cz>
14133 * predict.def: Fix GNU coding style.
14135 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14137 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14138 Mark 'to' argument with ATTRIBUTE_UNUSED.
14140 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14142 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14143 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14144 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14147 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14149 * doc/install.texi (Options specification): Restore entry of
14150 --enable-sjlj-exceptions.
14152 2017-05-27 Michael Eager <eager@eagercon.com>
14155 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14157 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14159 * config/microblaze/microblaze.h
14160 (FIXED_REGISTERS): Update in macro.
14161 (CALL_USED_REGISTERS): Update in macro.
14163 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14165 * doc/install.texi: Add links to macOS binary distributions.
14167 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14171 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14173 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14175 2017-05-26 Martin Liska <mliska@suse.cz>
14177 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14179 2017-05-26 Martin Liska <mliska@suse.cz>
14181 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14182 always leading ';; '.
14183 (dump_bb_info): Likewise.
14184 (brief_dump_cfg): Likewise.
14185 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14186 * dumpfile.c: Remove usage of TDF_VERBOSE.
14187 * dumpfile.h (enum dump_kind): Likewise.
14188 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14189 * print-tree.c (debug_verbose): Remove.
14190 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14191 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14192 * tree-diagnostic.c (default_tree_printer): Replace
14193 TDF_DIAGNOSTIC with TDF_SLIM.
14195 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14197 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14198 in parameter loop, rather than loop_vinfo.
14199 (create_intersect_range_checks): Ditto.
14200 (vect_create_cond_for_alias_checks): Update call to above functions.
14202 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14204 PR tree-optimization/80815
14205 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14206 for merging runtime alias checks. Handle negative DR_STEPs.
14208 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14210 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14212 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14213 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14214 out code pruning runtime alias checks.
14215 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14216 factored out from above.
14217 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14219 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14221 (prune_runtime_alias_test_list): New decalaration.
14223 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14225 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14226 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14227 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14228 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14229 (operator==, comp_dr_with_seg_len_pair): Ditto.
14230 (vect_prune_runtime_alias_test_list): Ditto.
14232 2017-05-26 Martin Liska <mliska@suse.cz>
14235 * params.def: Bound partial-inlining-entry-probability param.
14237 2017-05-26 Marek Polacek <polacek@redhat.com>
14240 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14243 2017-05-26 Richard Biener <rguenther@suse.de>
14245 PR tree-optimization/80842
14246 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14249 2017-05-26 Richard Biener <rguenther@suse.de>
14251 PR tree-optimization/80844
14252 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14254 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14256 * doc/md.texi (Machine Constraints): Update x86 family
14257 machine constraints section to match 'config/i386/constraints.md'.
14259 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14261 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14263 2017-05-25 Nathan Sidwell <nathan@acm.org>
14265 * doc/invoke.texi (--enable-languages): Update documentation.
14267 2017-05-25 Martin Liska <mliska@suse.cz>
14269 * dumpfile.c: Add TDF_FOLDING.
14270 * dumpfile.h (enum dump_kind): Likewise.
14271 * genmatch.c (dt_simplify::gen_1): Use it.
14273 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14275 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14277 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14279 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14280 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14282 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14284 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14285 * match.pd (X == C): Rewrite it here.
14286 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14287 with_certain_nonzero_bits2): New predicates.
14288 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14290 2017-05-24 Nathan Sidwell <nathan@acm.org>
14292 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14295 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14298 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14300 * config/powerpcspe: New port. Files are copied from the rs6000
14301 port, with "rs6000" in filenames replaced by "powerpcspe".
14303 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14305 PR rtl-optimization/80754
14306 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14308 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14310 * config/sparc/sparc.md (length): Return the correct value for -mflat
14311 sibcalls to match output_sibcall.
14313 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14317 * config/rs6000/rs6000.c (struct machine_function): Add new field
14319 (rs6000_get_separate_components): Init that field, use it.
14320 (rs6000_components_for_bb): Use the field.
14322 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14324 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14326 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14328 PR middle-end/80823
14329 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14331 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14334 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14335 address against address_operand predicate.
14336 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14338 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14340 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14341 parameters passed indirectly.
14343 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14345 * config/i386/i386.md (*movdi_internal): Remove SSE4
14346 alternative 18 (?r, *v). Update insn attributes.
14347 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14348 Update insn attributes.
14349 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14350 Update insn attributes.
14351 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14352 alternative 1 (r, v). Remove isa attribute.
14353 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14354 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14355 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14357 2017-05-23 Tom de Vries <tom@codesourcery.com>
14359 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14362 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14364 * cgraphunit.c (symbol_table::process_new_functions): Update.
14365 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14366 (inline_generate_summary): Rename to ...
14367 (ipa_fn_summary_generate): ... this one.
14368 (inline_read_summary): Rename to ...
14369 (ipa_fn_summary_read): ... this one.
14370 (inline_write_summary): Rename to ...
14371 (ipa_fn_summary_write): ... this one.
14372 (inline_free_summary): Rename to ...
14373 (ipa_free_fn_summary): ... this one.
14374 (pass_data_local_fn_summary, pass_local_fn_summary,
14375 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14376 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14377 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14378 make_pass_ipa_fn_summary): New.
14379 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14380 inline_write_summary, inline_free_summary): Remove.
14381 (ipa_free_fn_summary) : New.
14382 * ipa-inline.c (ipa_inline): Update.
14383 (pass_ipa_inline): Do not generate summaries.
14384 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14386 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14387 and add pass_ipa_fn_summary.
14388 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14390 (make_pass_inline_parameters): Remove.
14392 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14394 * omp-low.c (struct omp_context): Remove "default_kind" member.
14397 * omp-offload.c (execute_oacc_device_lower): Remove the
14398 parallelism dimensions function attributes for unparallelized
14399 OpenACC kernels constructs.
14401 2017-05-23 Martin Liska <mliska@suse.cz>
14403 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14405 (cgraph_edge::make_speculative): Likewise.
14406 (cgraph_edge::resolve_speculation): Likewise.
14407 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14408 (cgraph_node::dump): Likewise.
14409 * cgraph.h: Likewise.
14410 * cgraphunit.c (analyze_functions): Likewise.
14411 (symbol_table::compile): Likewise.
14412 * ipa-cp.c (print_all_lattices): Likewise.
14413 (determine_versionability): Likewise.
14414 (initialize_node_lattices): Likewise.
14415 (ipcp_verify_propagated_values): Likewise.
14416 (estimate_local_effects): Likewise.
14417 (update_profiling_info): Likewise.
14418 (create_specialized_node): Likewise.
14419 (perhaps_add_new_callers): Likewise.
14420 (decide_about_value): Likewise.
14421 (decide_whether_version_node): Likewise.
14422 (identify_dead_nodes): Likewise.
14423 (ipcp_store_bits_results): Likewise.
14424 * ipa-devirt.c (dump_targets): Likewise.
14425 (ipa_devirt): Likewise.
14426 * ipa-icf.c (sem_item::dump): Likewise.
14427 (sem_function::equals): Likewise.
14428 (sem_variable::equals): Likewise.
14429 (sem_item_optimizer::read_section): Likewise.
14430 (sem_item_optimizer::execute): Likewise.
14431 (congruence_class::dump): Likewise.
14432 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14433 (dump_inline_summary): Likewise.
14434 (estimate_node_size_and_time): Likewise.
14435 (inline_analyze_function): Likewise.
14436 * ipa-inline-transform.c (inline_call): Likewise.
14437 * ipa-inline.c (report_inline_failed_reason): Likewise.
14438 (want_early_inline_function_p): Likewise.
14439 (edge_badness): Likewise.
14440 (update_edge_key): Likewise.
14441 (inline_small_functions): Likewise.
14442 * ipa-profile.c (ipa_profile): Likewise.
14443 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14444 (ipa_make_edge_direct_to_target): Likewise.
14445 (remove_described_reference): Likewise.
14446 (ipa_impossible_devirt_target): Likewise.
14447 (propagate_controlled_uses): Likewise.
14448 (ipa_print_node_params): Likewise.
14449 (ipcp_transform_function): Likewise.
14450 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14451 (propagate_pure_const): Likewise.
14452 * ipa-reference.c (generate_summary): Likewise.
14453 (read_write_all_from_decl): Likewise.
14454 (propagate): Likewise.
14455 (ipa_reference_read_optimization_summary): Likewise.
14456 * ipa-utils.c (ipa_merge_profiles): Likewise.
14457 * ipa.c (walk_polymorphic_call_targets): Likewise.
14458 (symbol_table::remove_unreachable_nodes): Likewise.
14459 (ipa_single_use): Likewise.
14460 * passes.c (execute_todo): Likewise.
14461 * predict.c (drop_profile): Likewise.
14462 * symtab.c (symtab_node::get_dump_name): New function.
14463 (symtab_node::dump_name): Likewise.
14464 (symtab_node::dump_asm_name): Likewise.
14465 (symtab_node::dump_references): Likewise.
14466 (symtab_node::dump_referring): Likewise.
14467 (symtab_node::dump_base): Likewise.
14468 (symtab_node::debug_symtab): Likewise.
14469 * tree-sra.c (convert_callers_for_node): Likewise.
14470 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14471 * value-prof.c (init_node_map): Likewise.
14473 2017-05-23 Martin Liska <mliska@suse.cz>
14475 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14476 and symtab_node::debug_symtab to symbol_table::debug.
14477 * cgraphunit.c (analyze_functions): Use the renamed function.
14478 (symbol_table::compile): Likewise.
14479 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14480 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14481 * passes.c (execute_todo): Likewise.
14482 * symtab.c (symbol_table::dump): New function.
14483 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14485 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14487 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14488 that nonconst implies exec.
14490 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14492 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14493 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14494 (inline_edge_summary_vec): Turn into ...
14495 (ipa_call_summaries): ... this one.
14496 (redirect_to_unreachable, edge_set_predicate,
14497 evaluate_properties_for_edge, inline_summary_alloc,
14498 reset_ipa_call_summary, reset_inline_summary,
14499 inline_summary_t::duplicate): Update.
14500 (inline_edge_duplication_hook): Turn to ...
14501 (ipa_call_summary_t::duplicate): ... this one.
14502 (inline_edge_removal_hook): Turn to ...
14503 (ipa_call_summary_t::remove): ... this one.
14504 (dump_inline_edge_summary): Turn to ...
14505 (dump_ipa_call_summary): ... this one.
14506 (estimate_function_body_sizes): Update.
14507 (inline_update_callee_summaries): Update.
14508 (remap_edge_change_prob): Update.
14509 (remap_edge_summaries): Update.
14510 (inline_merge_summary): Update.
14511 (do_estimate_edge_time): Update.
14512 (inline_generate_summary): Update.
14513 (inline_read_section): Update.
14514 (inline_read_summary): Update.
14515 (inline_free_summary): Update.
14516 * ipa-inline.c (can_inline_edge_p): Update.
14517 (compute_inlined_call_time): Update.
14518 (want_inline_small_function_p): Update.
14519 (edge_badness): Update.
14520 (early_inliner): Update.
14521 * ipa-inline.h (inline_edge_summary): Turn to ...
14522 (ipa_call_summary): ... this one.
14523 (ipa_call_summary_t): New class.
14524 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14525 (ipa_call_summaries): New.
14526 (inline_edge_summary): Remove.
14527 (estimate_edge_growth): Update.
14528 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14529 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14530 * ipa-split.c (execute_split_functions): Update.
14531 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14533 2017-05-23 Tom de Vries <tom@codesourcery.com>
14535 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14536 attributes): Document rdrand effective target.
14538 2017-05-23 Tom de Vries <tom@codesourcery.com>
14540 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14541 attributes): Sort alphabetically.
14543 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14545 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14547 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14550 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14551 V2DF/V2DI splat into two separate patterns, one that handles
14552 registers, and the other that only handles memory. Drop support
14553 for splatting from a GPR on ISA 2.07 and then splitting the
14554 splat into direct move and splat.
14555 (vsx_splat_<mode>_reg): Likewise.
14556 (vsx_splat_<mode>_mem): Likewise.
14558 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14560 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14562 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14564 PR middle-end/80809
14565 * omp-low.c (finish_taskreg_remap): New function.
14566 (finish_taskreg_scan): If unit size of ctx->record_type
14567 is non-constant, unshare the size expression and replace
14568 decls in it with possible outer var refs.
14570 PR middle-end/80809
14571 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14572 GOVD_SHARED rather than GOVD_PRIVATE with it.
14573 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14574 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14576 PR middle-end/80853
14577 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14578 as last argument to build_outer_var_ref for pointer bases of array
14579 section reductions.
14581 2017-05-19 Martin Sebor <msebor@redhat.com>
14583 * print-tree.c (print_node): Print DECL_READ_P flag.
14585 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14587 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14588 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14589 * cgraph.c: Likewise.
14590 * cgraphunit.c: Likewise.
14591 * gengtype.c: Likewise.
14592 * ipa-cp.c: Likewise.
14593 * ipa-devirt.c: Likewise.
14594 * ipa-icf.c: Likewise.
14595 * ipa-predicate.c: Likewise.
14596 * ipa-profile.c: Likewise.
14597 * ipa-prop.c: Likewise.
14598 * ipa-split.c: Likewise.
14600 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14601 edge_predicate_pool, dump_inline_hints,
14602 inline_summary::account_size_time, redirect_to_unreachable,
14603 edge_set_predicate, set_hint_predicate,
14604 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14605 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14606 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14607 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14608 ipa_call_summary_t::remove, initialize_growth_caches,
14609 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14610 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14611 mark_modified, unmodified_parm_1, unmodified_parm,
14612 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14613 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14614 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14615 will_be_nonconstant_predicate, record_modified_bb_info,
14616 get_minimal_bb, record_modified, param_change_prob,
14617 phi_result_unknown_predicate, predicate_for_phi_result,
14618 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14619 estimate_function_body_sizes, compute_inline_parameters,
14620 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14621 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14622 inline_update_callee_summaries, remap_edge_change_prob,
14623 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14624 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14625 inline_analyze_function, inline_summary_t::insert,
14626 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14627 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14628 inline_free_summary): Move to ipa-fnsummary.h
14629 (predicate_t): Remove.
14630 * ipa-fnsummary.c: New file.
14631 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14632 (enum inline_hints_vals, inline_hints, agg_position_info,
14633 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14634 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14635 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14636 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14637 inline_read_summary, inline_write_summary, inline_free_summary,
14638 inline_analyze_function, initialize_inline_failed,
14639 inline_merge_summary, inline_update_overall_summary,
14640 compute_inline_parameters): Move to ipa-fnsummary.h
14641 * ipa-fnsummary.h: New file.
14642 * ipa-inline-transform.h: Include ipa-inline.h.
14643 * ipa-inline.c: LIkewise.
14645 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14647 * ipa-inline.c (edge_badness): Use inlined_time instead of
14648 inline_summaries->get.
14650 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14652 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14654 2017-05-22 Nathan Sidwell <nathan@acm.org>
14656 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14657 (fdump-lang): Document 'raw' option.
14658 * dumpfile.h (TDI_tu): Delete.
14659 * dumpfile.c (dump_files): Remove translation-unit.
14660 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14662 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14664 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14665 command option from $(AWK) call.
14666 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14667 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14668 [FORMAT]: Remove handling of variable.
14669 * config/avr/t-multilib: Regenerate.
14671 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14673 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14675 (dump_inline_summary): Do not print self_time.
14676 (estimate_function_body_sizes): Do not set self_time.
14677 (compute_inline_parameters): Likewise.
14678 (inline_read_section, inline_write_summary): Do not stream self_time.
14679 * ipa-inline.h (inline_summary): Drop self_time.
14681 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14683 * ipa-inline-analysis.c (account_size_time): Rename to ...
14684 (inline_summary::account_size_time): ... this one.
14685 (reset_ipa_call_summary): Turn to ...
14686 (ipa_call_summary::reset): ... this one.
14687 (reset_inline_summary): Turn to ...
14688 (inline_summary::reset): ... this one.
14689 (inline_summary_t::remove): Update.
14690 (inline_summary_t::duplicate): Update.
14691 (ipa_call_summary_t::remove): Update.
14692 (dump_inline_summary): Update.
14693 (estimate_function_body_sizes): Update.
14694 (compute_inline_parameters): Update.
14695 (estimate_node_size_and_time): Update.
14696 (inline_merge_summary): Update.
14697 (inline_update_overall_summary): Update.
14698 (inline_read_section): Update.
14699 (inline_write_summary): Update.
14700 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14701 add account_size_time and reset member functions.
14702 (ipa_call_summary): Add reset function.
14703 * ipa-predicate.h (predicate::operator &): Constify.
14705 2017-05-22 Richard Biener <rguenther@suse.de>
14707 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14709 2017-05-19 Jason Merrill <jason@redhat.com>
14711 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14713 2017-05-19 Marek Polacek <polacek@redhat.com>
14716 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14717 TYPE_OVERFLOW_WRAPS checks.
14719 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14721 * tree-core.h (enum omp_clause_default_kind): Add
14722 "OMP_CLAUSE_DEFAULT_PRESENT".
14723 * tree-pretty-print.c (dump_omp_clause): Handle it.
14724 * gimplify.c (enum gimplify_omp_var_data): Add
14725 "GOVD_MAP_FORCE_PRESENT".
14726 (gimplify_adjust_omp_clauses_1): Map it to
14727 "GOMP_MAP_FORCE_PRESENT".
14728 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14730 * gimplify.c (oacc_default_clause): Clarify.
14732 2017-05-19 Nathan Sidwell <nathan@acm.org>
14734 LANG_HOOK_REGISTER_DUMPS
14735 * toplev.c (general_init): Call register dump lang hook.
14736 * doc/invoke.texi: Document -fdump-lang option family.
14737 * dumpfile.c (dump_files): Remove class dump here.
14738 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14739 * dumpfile.h (tree_dump_index): Remove TDI_class.
14740 * langhooks-def.h (lhd_register_dumps): Declare.
14741 (LANG_HOOKS_REGISTER_DUMPS): Define.
14742 (LANG_HOOKS_INITIALIZER): Add it.
14743 * langhooks.c (lhd_register_dumps): Define.
14744 * langhooks.h (struct lang_hooks): Add register_dumps.
14746 2017-05-19 Nathan Sidwell <nathan@acm.org>
14748 * context.h (context::set_passes): New.
14749 * context.c (context::context): Do not create pass manager.
14750 * toplev.c (general_init): Create pass manager here.
14752 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14754 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14755 use this splitter if two add or or instructions would also work for
14756 the constant we want to generate.
14758 2017-05-19 Richard Biener <rguenther@suse.de>
14761 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14762 predicate evaluation.
14764 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14766 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14768 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14771 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14773 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14774 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14775 (inline_edge_summary_vec): Turn into ...
14776 (ipa_call_summaries): ... this one.
14777 (redirect_to_unreachable, edge_set_predicate,
14778 evaluate_properties_for_edge, inline_summary_alloc,
14779 reset_ipa_call_summary, reset_inline_summary,
14780 inline_summary_t::duplicate): Update.
14781 (inline_edge_duplication_hook): Turn to ...
14782 (ipa_call_summary_t::duplicate): ... this one.
14783 (inline_edge_removal_hook): Turn to ...
14784 (ipa_call_summary_t::remove): ... this one.
14785 (dump_inline_edge_summary): Turn to ...
14786 (dump_ipa_call_summary): ... this one.
14787 (estimate_function_body_sizes): Update.
14788 (inline_update_callee_summaries): Update.
14789 (remap_edge_change_prob): Update.
14790 (remap_edge_summaries): Update.
14791 (inline_merge_summary): Update.
14792 (do_estimate_edge_time): Update.
14793 (inline_generate_summary): Update.
14794 (inline_read_section): Update.
14795 (inline_read_summary): Update.
14796 (inline_free_summary): Update.
14797 * ipa-inline.c (can_inline_edge_p): Update.
14798 (compute_inlined_call_time): Update.
14799 (want_inline_small_function_p): Update.
14800 (edge_badness): Update.
14801 (early_inliner): Update.
14802 * ipa-inline.h (inline_edge_summary): Turn to ...
14803 (ipa_call_summary): ... this one.
14804 (ipa_call_summary_t): New class.
14805 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14806 (ipa_call_summaries): New.
14807 (inline_edge_summary): Remove.
14808 (estimate_edge_growth): Update.
14809 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14810 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14811 * ipa-split.c (execute_split_functions): Update.
14812 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14814 2017-05-19 Richard Biener <rguenther@suse.de>
14816 PR middle-end/80764
14817 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14819 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14821 * config/rs6000/rs6000.c (struct machine_function): Add field
14822 fpr_is_wrapped_separately.
14823 (rs6000_get_separate_components): Use 64 components. Handle the
14824 new FPR components.
14825 (rs6000_components_for_bb): Handle the FPR components.
14826 (rs6000_emit_prologue_components): Handle the FPR components.
14827 (rs6000_emit_epilogue_components): Handle the FPR components.
14828 (rs6000_set_handled_components): Handle the FPR components.
14829 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14830 that are already separately shrink-wrapped.
14831 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14832 that are already separately shrink-wrapped.
14834 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14837 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14840 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14841 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14842 cases where the register allocator uses a move and an offsettable
14843 memory operation to/from a FPR register on ISA 2.06/2.07.
14844 (define_peephole2 for Altivec d-form store): Likewise.
14846 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14849 * config/i386/mmx.md (*mov<mode>_internal): Enable
14850 alternatives 11, 12, 13 and 14 also for 32bit targets.
14851 Remove alternatives 15, 16, 17 and 18.
14852 * config/i386/sse.md (vec_concatv2di): Change
14853 alternative (!x, *y) to (x, ?!*Yn).
14855 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14857 * dumpfile.h (enum dump_kind): Remove stray comma.
14859 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14861 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14862 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14863 predicate::num_conditions
14864 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14865 (CHANGED): turn into predicate::changed.
14866 (agg_position_info): Move to ipa-predicate.h
14867 (add_condition, predicate::add_clause, predicate::operator &=,
14868 predicate::or_with, predicate::evaluate, predicate::probability,
14869 dump_condition, dump_clause, predicate::dump,
14870 predicate::remap_after_duplication, predicate::remap_after_inlining,
14871 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14872 (evaluate_conditions_for_known_args): Update.
14873 (set_cond_stmt_execution_predicate): Update.
14874 * ipa-inline.h: Include ipa-predicate.h
14875 (condition, inline_param_summary, conditions, agg_position_info,
14876 predicate): Move to ipa-predicate.h
14877 * ipa-predicate.c: New file.
14878 * ipa-predicate.h: New file.
14880 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14882 * final.c (leaf_function_p): Check we are not in a sequence.
14884 2017-05-18 Martin Liska <mliska@suse.cz>
14886 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14887 * dumpfile.c (dump_register): Use new enum dump_kind.
14888 (get_dump_file_name): Likewise.
14889 (dump_enable_all): Likewise.
14890 (dump_switch_p_1): Likewise.
14891 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14892 * dumpfile.h (enum dump_kind): New enum type.
14893 (struct dump_file_info): Create constructor and
14894 format fields and comments.
14895 * passes.c (pass_manager::register_one_dump_file):
14897 * statistics.c (statistics_early_init): Likewise.
14898 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14899 TDF_TREE with TDF_SLIM.
14900 (gather_memory_references_ref): Likewise.
14902 2017-05-18 Martin Liska <mliska@suse.cz>
14904 * vec.h (struct vnull): Use it.
14906 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14908 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14909 (true_predicate, false_predicate, true_predicate_p,
14910 false_predicate_p): Remove.
14911 (single_cond_predicate, not_inlined_predicate): Turn to member function
14913 (add_condition): Update.
14914 (add_clause): Turn to...
14915 (predicate::add_clause): ... this one; update; allow passing NULL
14917 (and_predicates): Turn to ...
14918 (predicate::operator &=): ... this one.
14919 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14920 (or_predicates): Turn to ...
14921 (predicate::or_with): ... this one.
14922 (evaluate_predicate): Turn to ...
14923 (predicate::evaluate): ... this one.
14924 (predicate_probability): Turn to ...
14925 (predicate::probability): ... this one.
14926 (dump_condition): Update.
14927 (dump_predicate): Turn to ...
14928 (predicate::dump): ... this one.
14929 (account_size_time): Update.
14930 (edge_set_predicate): Update.
14931 (set_hint_predicate): UPdate.
14932 (evaluate_conditions_for_known_args): Update.
14933 (evaluate_properties_for_edge): Update.
14934 (remap_predicate_after_duplication): Turn to...
14935 (predicate::remap_after_duplication): ... this one.
14936 (remap_hint_predicate_after_duplication): Update.
14937 (inline_summary_t::duplicate): UPdate.
14938 (dump_inline_edge_summary): Update.
14939 (dump_inline_summary): Update.
14940 (set_cond_stmt_execution_predicate): Update.
14941 (set_switch_stmt_execution_predicate): Update.
14942 (compute_bb_predicates): Update.
14943 (will_be_nonconstant_expr_predicate): Update.
14944 (will_be_nonconstant_predicate): Update.
14945 (phi_result_unknown_predicate): Update.
14946 (predicate_for_phi_result): Update.
14947 (array_index_predicate): Update.
14948 (estimate_function_body_sizes): Update.
14949 (estimate_node_size_and_time): Update.
14950 (estimate_ipcp_clone_size_and_time): Update.
14951 (remap_predicate): Rename to ...
14952 (predicate::remap_after_inlining): ... this one.
14953 (remap_hint_predicate): Update.
14954 (inline_merge_summary): Update.
14955 (inline_update_overall_summary): Update.
14956 (estimate_size_after_inlining): Update.
14957 (read_predicate): Rename to ...
14958 (predicate::stream_in): ... this one.
14959 (read_inline_edge_summary): Update.
14960 (write_predicate): Rename to ...
14961 (predicate::stream_out): ... this one.
14962 (write_inline_edge_summary): Update.
14963 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14964 (clause_t): Turn to uint32_t
14965 (predicate): Turn to class; implement constructor and operators
14967 (size_time_entry): Update.
14968 (inline_summary): Update.
14969 (inline_edge_summary): Update.
14971 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14973 * fold-const.c (fold_binary_loc): Move transformation...
14974 * match.pd (C - X CMP X): ... here.
14976 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14978 * config/sparc/sparc.c (sparc_option_override): Set function
14979 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14980 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14982 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14984 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14986 2017-05-18 Marek Polacek <polacek@redhat.com>
14989 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14990 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14991 gimple_assign_load_p.
14993 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14995 PR middle-end/80692
14996 * real.c (do_compare): Give decimal_do_compare preference over
14997 comparing just the signs.
14999 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15001 * doc/md.texi (Canonicalization of Instructions): Describe the
15002 canonical form of instructions that inherently set a condition
15005 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15007 PR middle-end/80775
15008 * tree-cfg.c: Move deletion of unreachable case statements to after
15009 the merging of consecutive case labels.
15011 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15013 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15014 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15015 restoring of callee-saved registers.
15017 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15019 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15020 * config/visium/visium.c (single_set_and_flags): Likewise.
15021 * config/visium/visium.md (Substitutions): Likewise.
15023 2017-05-17 Martin Liska <mliska@suse.cz>
15025 * cfg.c: Introduce dump_flags_t type and
15026 use it instead of int type.
15028 * cfghooks.c: Likewise.
15029 * cfghooks.h (struct cfg_hooks): Likewise.
15030 * cfgrtl.c: Likewise.
15031 * cfgrtl.h: Likewise.
15032 * cgraph.c (cgraph_node::get_body): Likewise.
15033 * coretypes.h: Likewise.
15034 * domwalk.c: Likewise.
15035 * domwalk.h: Likewise.
15036 * dumpfile.c (struct dump_option_value_info): Likewise.
15037 (dump_enable_all): Likewise.
15038 (dump_switch_p_1): Likewise.
15039 (opt_info_switch_p): Likewise.
15040 * dumpfile.h (enum tree_dump_index): Likewise.
15041 (struct dump_file_info): Likewise.
15042 * genemit.c: Likewise.
15043 * generic-match-head.c: Likewise.
15044 * gengtype.c (open_base_files): Likewise.
15045 * gimple-pretty-print.c: Likewise.
15046 * gimple-pretty-print.h: Likewise.
15047 * graph.c (print_graph_cfg): Likewise.
15048 * graphite-scop-detection.c (dot_all_sese): Likewise.
15049 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15050 * loop-unroll.c (report_unroll): Likewise.
15051 * passes.c (pass_manager::register_one_dump_file): Likewise.
15052 * print-tree.c: Likewise.
15053 * statistics.c: Likewise.
15054 * tree-cfg.c: Likewise.
15055 * tree-cfg.h: Likewise.
15056 * tree-dfa.c: Likewise.
15057 * tree-dfa.h: Likewise.
15058 * tree-dump.c (dump_function): Likewise.
15059 * tree-dump.h (struct dump_info): Likewise.
15060 * tree-pretty-print.c: Likewise.
15061 * tree-pretty-print.h: Likewise.
15062 * tree-ssa-live.c: Likewise.
15063 * tree-ssa-live.h: Likewise.
15064 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15065 * tree-vect-loop.c: Likewise.
15066 * tree-vect-slp.c: Likewise.
15068 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15069 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15071 PR tree-optimization/80457
15072 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15073 of all arguments to a statement as scalar_to_vec operations.
15074 (vectorizable_call): Adjust call to vect_model_simple_cost for
15076 (vectorizable_conversion): Likewise.
15077 (vectorizable_assignment): Likewise.
15078 (vectorizable_shift): Likewise.
15079 (vectorizable_operation): Likewise.
15080 (vectorizable_comparison): Likewise.
15081 (vect_is_simple_cond): Record the def types for operands.
15082 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15083 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15084 for statement argument count.
15086 2017-05-16 Carl Love <cel@us.ibm.com>
15088 * config/rs6000/rs6000-c: Add support for built-in functions
15089 vector unsigned long long vec_bperm (vector unsigned long long,
15090 vector unsigned char)
15091 vector signed long long vec_mule (vector signed int,
15093 vector unsigned long long vec_mule (vector unsigned int,
15094 vector unsigned int)
15095 vector signed long long vec_mulo (vector signed int,
15097 vector unsigned long long vec_mulo (vector unsigned int,
15098 vector unsigned int)
15099 vector signed char vec_sldw (vector signed char,
15100 vector signed char,
15102 vector unsigned char vec_sldw (vector unsigned char,
15103 vector unsigned char,
15105 vector signed short vec_sldw (vector signed short,
15106 vector signed short,
15108 vector unsigned short vec_sldw (vector unsigned short,
15109 vector unsigned short,
15111 vector signed int vec_sldw (vector signed int,
15114 vector unsigned int vec_sldw (vector unsigned int,
15115 vector unsigned int,
15117 vector signed long long vec_sldw (vector signed long long,
15118 vector signed long long,
15120 vector unsigned long long vec_sldw (vector unsigned long long,
15121 vector unsigned long long,
15123 * config/rs6000/rs6000-c: Add support for built-in functions
15124 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15125 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15126 * doc/extend.texi: Update the built-in documentation for the
15127 new built-in functions.
15129 2017-05-16 Marek Polacek <polacek@redhat.com>
15133 * tree.c (save_expr): Don't fold the expression.
15135 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15137 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15138 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15139 and (?*y,m). Update insn attributes.
15141 2017-05-16 Martin Liska <mliska@suse.cz>
15143 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15144 flags argument of print_gimple_stmt, print_gimple_expr,
15145 print_generic_stmt and print_generic_expr.
15146 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15147 * coretypes.h: Likewise.
15148 * except.c (dump_eh_tree): Likewise.
15149 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15150 * gimple-pretty-print.h: Likewise.
15151 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15152 (backprop::push_to_worklist): Likewise.
15153 (backprop::pop_from_worklist): Likewise.
15154 (backprop::process_use): Likewise.
15155 (backprop::intersect_uses): Likewise.
15156 (note_replacement): Likewise.
15157 * gimple-ssa-store-merging.c
15158 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15159 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15160 (pass_store_merging::execute): Likewise.
15161 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15162 (ssa_base_cand_dump_callback): Likewise.
15163 (dump_incr_vec): Likewise.
15164 (replace_refs): Likewise.
15165 (replace_mult_candidate): Likewise.
15166 (create_add_on_incoming_edge): Likewise.
15167 (create_phi_basis): Likewise.
15168 (insert_initializers): Likewise.
15169 (all_phi_incrs_profitable): Likewise.
15170 (introduce_cast_before_cand): Likewise.
15171 (replace_one_candidate): Likewise.
15172 * gimplify.c (gimplify_expr): Likewise.
15173 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15174 (set_rename): Likewise.
15175 (rename_uses): Likewise.
15176 (copy_loop_phi_nodes): Likewise.
15177 (add_close_phis_to_merge_points): Likewise.
15178 (copy_loop_close_phi_args): Likewise.
15179 (copy_cond_phi_args): Likewise.
15180 (graphite_copy_stmts_from_block): Likewise.
15181 (translate_pending_phi_nodes): Likewise.
15182 * graphite-poly.c (print_pdr): Likewise.
15183 (dump_gbb_cases): Likewise.
15184 (dump_gbb_conditions): Likewise.
15185 (print_scop_params): Likewise.
15186 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15187 (build_cross_bb_scalars_use): Likewise.
15188 (gather_bbs::before_dom_children): Likewise.
15189 * hsa-dump.c (dump_hsa_immed): Likewise.
15190 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15191 (get_replacement_map): Likewise.
15192 * ipa-inline-analysis.c (dump_condition): Likewise.
15193 (estimate_function_body_sizes): Likewise.
15194 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15195 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15196 * ipa-prop.c (ipa_dump_param): Likewise.
15197 (ipa_print_node_jump_functions_for_edge): Likewise.
15198 (ipa_modify_call_arguments): Likewise.
15199 (ipa_modify_expr): Likewise.
15200 (ipa_dump_param_adjustments): Likewise.
15201 (ipa_dump_agg_replacement_values): Likewise.
15202 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15203 * ipa-pure-const.c (check_stmt): Likewise.
15204 (pass_nothrow::execute): Likewise.
15205 * ipa-split.c (execute_split_functions): Likewise.
15206 * omp-offload.c (dump_oacc_loop_part): Likewise.
15207 (dump_oacc_loop): Likewise.
15208 * trans-mem.c (tm_log_emit): Likewise.
15209 (tm_memopt_accumulate_memops): Likewise.
15210 (dump_tm_memopt_set): Likewise.
15211 (dump_tm_memopt_transform): Likewise.
15212 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15213 (print_loop): Likewise.
15214 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15215 (chkp_gather_checks_info): Likewise.
15216 (chkp_get_check_result): Likewise.
15217 (chkp_remove_check_if_pass): Likewise.
15218 (chkp_use_outer_bounds_if_possible): Likewise.
15219 (chkp_reduce_bounds_lifetime): Likewise.
15220 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15221 (chkp_mark_completed_bounds): Likewise.
15222 (chkp_register_incomplete_bounds): Likewise.
15223 (chkp_mark_invalid_bounds): Likewise.
15224 (chkp_maybe_copy_and_register_bounds): Likewise.
15225 (chkp_build_returned_bound): Likewise.
15226 (chkp_get_bound_for_parm): Likewise.
15227 (chkp_build_bndldx): Likewise.
15228 (chkp_get_bounds_by_definition): Likewise.
15229 (chkp_generate_extern_var_bounds): Likewise.
15230 (chkp_get_bounds_for_decl_addr): Likewise.
15231 * tree-chrec.c (chrec_apply): Likewise.
15232 * tree-data-ref.c (dump_data_reference): Likewise.
15233 (dump_subscript): Likewise.
15234 (dump_data_dependence_relation): Likewise.
15235 (analyze_overlapping_iterations): Likewise.
15236 * tree-inline.c (expand_call_inline): Likewise.
15237 (tree_function_versioning): Likewise.
15238 * tree-into-ssa.c (dump_defs_stack): Likewise.
15239 (dump_currdefs): Likewise.
15240 (dump_names_replaced_by): Likewise.
15241 (dump_update_ssa): Likewise.
15242 (update_ssa): Likewise.
15243 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15244 * tree-parloops.c (build_new_reduction): Likewise.
15245 (try_create_reduction_list): Likewise.
15246 (ref_conflicts_with_region): Likewise.
15247 (oacc_entry_exit_ok_1): Likewise.
15248 (oacc_entry_exit_single_gang): Likewise.
15249 * tree-pretty-print.h: Likewise.
15250 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15251 (get_scalar_evolution): Likewise.
15252 (add_to_evolution): Likewise.
15253 (get_loop_exit_condition): Likewise.
15254 (analyze_evolution_in_loop): Likewise.
15255 (analyze_initial_condition): Likewise.
15256 (analyze_scalar_evolution): Likewise.
15257 (instantiate_scev): Likewise.
15258 (number_of_latch_executions): Likewise.
15259 (gather_chrec_stats): Likewise.
15260 (final_value_replacement_loop): Likewise.
15261 (scev_const_prop): Likewise.
15262 * tree-sra.c (dump_access): Likewise.
15263 (disqualify_candidate): Likewise.
15264 (create_access): Likewise.
15265 (reject): Likewise.
15266 (maybe_add_sra_candidate): Likewise.
15267 (create_access_replacement): Likewise.
15268 (analyze_access_subtree): Likewise.
15269 (analyze_all_variable_accesses): Likewise.
15270 (sra_modify_assign): Likewise.
15271 (initialize_constant_pool_replacements): Likewise.
15272 (find_param_candidates): Likewise.
15273 (decide_one_param_reduction): Likewise.
15274 (replace_removed_params_ssa_names): Likewise.
15275 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15276 * tree-ssa-copy.c (dump_copy_of): Likewise.
15277 (copy_prop_visit_cond_stmt): Likewise.
15278 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15279 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15280 (record_equivalences_from_stmt): Likewise.
15281 * tree-ssa-dse.c (compute_trims): Likewise.
15282 (delete_dead_call): Likewise.
15283 (delete_dead_assignment): Likewise.
15284 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15285 (forward_propagate_into_cond): Likewise.
15286 (pass_forwprop::execute): Likewise.
15287 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15288 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15290 (move_computations_worker): Likewise.
15291 (execute_sm): Likewise.
15292 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15293 (remove_exits_and_undefined_stmts): Likewise.
15294 (remove_redundant_iv_tests): Likewise.
15295 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15296 (adjust_iv_update_pos): Likewise.
15297 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15298 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15299 (value_replacement): Likewise.
15300 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15301 * tree-ssa-pre.c (print_pre_expr): Likewise.
15302 (get_representative_for): Likewise.
15303 (create_expression_by_pieces): Likewise.
15304 (insert_into_preds_of_block): Likewise.
15305 (eliminate_insert): Likewise.
15306 (eliminate_dom_walker::before_dom_children): Likewise.
15307 (eliminate): Likewise.
15308 (remove_dead_inserted_code): Likewise.
15309 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15310 * tree-ssa-reassoc.c (get_rank): Likewise.
15311 (eliminate_duplicate_pair): Likewise.
15312 (eliminate_plus_minus_pair): Likewise.
15313 (eliminate_not_pairs): Likewise.
15314 (undistribute_ops_list): Likewise.
15315 (eliminate_redundant_comparison): Likewise.
15316 (update_range_test): Likewise.
15317 (optimize_range_tests_var_bound): Likewise.
15318 (optimize_vec_cond_expr): Likewise.
15319 (rewrite_expr_tree): Likewise.
15320 (rewrite_expr_tree_parallel): Likewise.
15321 (linearize_expr): Likewise.
15322 (break_up_subtract): Likewise.
15323 (linearize_expr_tree): Likewise.
15324 (attempt_builtin_powi): Likewise.
15325 (attempt_builtin_copysign): Likewise.
15326 (transform_stmt_to_copy): Likewise.
15327 (transform_stmt_to_multiply): Likewise.
15328 (dump_ops_vector): Likewise.
15329 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15330 (print_scc): Likewise.
15331 (set_ssa_val_to): Likewise.
15332 (visit_reference_op_store): Likewise.
15333 (visit_use): Likewise.
15334 (sccvn_dom_walker::before_dom_children): Likewise.
15335 (run_scc_vn): Likewise.
15336 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15338 (expr_hash_elt::print): Likewise.
15339 (const_and_copies::pop_to_marker): Likewise.
15340 (const_and_copies::record_const_or_copy_raw): Likewise.
15341 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15342 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15343 (dump_predicates): Likewise.
15344 (find_uninit_use): Likewise.
15345 (warn_uninitialized_phi): Likewise.
15346 (pass_late_warn_uninitialized::execute): Likewise.
15347 * tree-ssa.c (verify_vssa): Likewise.
15348 (verify_ssa): Likewise.
15349 (maybe_optimize_var): Likewise.
15350 * tree-vrp.c (dump_value_range): Likewise.
15351 (dump_all_value_ranges): Likewise.
15352 (dump_asserts_for): Likewise.
15353 (register_edge_assert_for_2): Likewise.
15354 (vrp_visit_cond_stmt): Likewise.
15355 (vrp_visit_switch_stmt): Likewise.
15356 (vrp_visit_stmt): Likewise.
15357 (vrp_visit_phi_node): Likewise.
15358 (simplify_cond_using_ranges_1): Likewise.
15359 (fold_predicate_in): Likewise.
15360 (evrp_dom_walker::before_dom_children): Likewise.
15361 (evrp_dom_walker::push_value_range): Likewise.
15362 (evrp_dom_walker::pop_value_range): Likewise.
15363 (execute_early_vrp): Likewise.
15365 2017-05-16 Richard Biener <rguenther@suse.de>
15367 * dwarf2out.c (loc_list_from_tree_1): Do not create
15368 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15370 2017-05-16 Richard Biener <rguenther@suse.de>
15372 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15374 (note_variable_value_in_expr): If we resolved the decl ref
15375 do not push to the stack.
15377 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15379 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15380 operations in fast-math mode.
15381 (vaddq_f16): Likewise.
15382 (vmul_f16): Likewise.
15383 (vmulq_f16): Likewise.
15384 (vsub_f16): Likewise.
15385 (vsubq_f16): Likewise.
15386 * config/arm/neon.md (add<mode>3): New.
15388 (fma:<VH:mode>3): New. Also remove outdated comment.
15391 2017-05-16 Martin Liska <mliska@suse.cz>
15395 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15396 (odr_types_equivalent_p): Likewise.
15398 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15400 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15402 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15405 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15406 non-interunit SSE move alternatives with '?'.
15407 (zero-extendsidi peephole2): New peephole to skip intermediate
15408 general register in SSE zero-extend sequence.
15410 2017-05-15 Jeff Law <law@redhat.com>
15412 * reorg.c (relax_delay_slots): Create a new variable to hold
15413 the temporary target rather than clobbering TARGET_LABEL.
15415 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15416 missing argument to extract_bit_field call.
15417 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15419 2017-05-15 Martin Liska <mliska@suse.cz>
15422 * gcc.c (process_command): Do not allow empty argument of -o option.
15424 2017-05-15 Renlin Li <renlin.li@arm.com>
15426 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15427 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15428 * config/aarch64/constraints.md (Usf): Add long call check.
15429 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15430 (call_value): Likewise.
15431 (sibcall): Likewise.
15432 (sibcall_value): Likewise.
15434 (call_value_insn): New.
15435 (sibcall_insn): Update rtx pattern.
15436 (sibcall_value_insn): Likewise.
15437 (call_internal): Remove.
15438 (call_value_internal): Likewise.
15439 (sibcall_internal): Likewise.
15440 (sibcall_value_internal): Likewise.
15441 (call_reg): Likewise.
15442 (call_symbol): Likewise.
15443 (call_value_reg): Likewise.
15444 (call_value_symbol): Likewise.
15446 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15449 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15451 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15453 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15454 compatible with CCGOCmode and with CCZmode.
15456 2017-05-14 Martin Sebor <msebor@redhat.com>
15458 PR middle-end/77671
15459 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15460 (gimple_fold_builtin_snprintf): Same.
15461 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15462 (gimple_fold_builtin_snprintf): Same.
15463 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15464 of character types.
15465 (is_call_safe): New function.
15466 (try_substitute_return_value): Call it.
15467 (try_simplify_call): New function.
15468 (pass_sprintf_length::handle_gimple_call): Call it.
15470 2017-05-14 Martin Sebor <msebor@redhat.com>
15472 PR middle-end/80669
15473 * builtins.c (expand_builtin_stpncpy): Simplify.
15475 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15477 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15478 * config/i386/i386.h
15479 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15480 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15481 (struct machine_function): Add new members call_ms2sysv,
15482 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15483 (struct machine_frame_state): New fields sp_realigned and
15484 sp_realigned_offset.
15485 * config/i386/i386.c
15486 (enum xlogue_stub): New enum.
15487 (enum xlogue_stub_sets): New enum.
15488 (class xlogue_layout): New class.
15489 (struct ix86_frame): New fields stack_realign_allocate_offset,
15490 stack_realign_offset and outlined_save_offset. Modify comments to
15491 detail stack layout when using out-of-line stubs.
15492 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15493 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15494 -mcall-ms2sysv-xlogues.
15495 (stub_managed_regs): New static variable.
15496 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15497 registers managed by out-of-line stub.
15498 (disable_call_ms2sysv_xlogues): New function.
15499 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15500 m->call_ms2sysv when appropriate and compute frame layout for
15502 (sp_valid_at, fp_valid_at): New inline functions.
15503 (choose_basereg): New function.
15504 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15506 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15507 Use align parameter of choose_baseaddr to generated aligned SSE movs
15509 (pro_epilogue_adjust_stack): Modify to track
15510 machine_frame_state::sp_realigned.
15511 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15512 (ix86_nsaved_sseregs): Likewise.
15513 (ix86_emit_save_regs): Likewise.
15514 (ix86_emit_save_regs_using_mov): Likewise.
15515 (ix86_emit_save_sse_regs_using_mov): Likewise.
15516 (get_scratch_register_on_entry): Likewise.
15517 (gen_frame_set): New function.
15518 (gen_frame_load): Likewise.
15519 (gen_frame_store): Likewise.
15520 (emit_outlined_ms2sysv_save): Likewise.
15521 (emit_outlined_ms2sysv_restore): Likewise.
15522 (ix86_expand_prologue): Modify stack re-alignment code and call
15523 emit_outlined_ms2sysv_save when appropriate.
15524 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15525 parameter rtx_insn *insn, which allows the function to be used to only
15526 generate the notes.
15527 (ix86_expand_epilogue): Modify validity checks of frame and stack
15528 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15529 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15530 * config/i386/predicates.md
15531 (save_multiple): New predicate.
15532 (restore_multiple): Likewise.
15533 * config/i386/sse.md
15534 (save_multiple<mode>): New pattern.
15535 (save_multiple_realign<mode>): Likewise.
15536 (restore_multiple<mode>): Likewise.
15537 (restore_multiple_and_return<mode>): Likewise.
15538 (restore_multiple_leave_return<mode>): Likewise.
15539 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15541 2017-05-14 Julia Koval <julia.koval@intel.com>
15543 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15544 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15545 (__builtin_ia32_xsetbv): New builtins.
15546 * config/i386/i386.c (ix86_expand_special_args_builtin):
15548 (ix86_expand_builtin): Special expand for new intrinsics.
15549 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15550 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15551 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15553 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15555 * cfganal.c (inverted_post_order_compute): Change argument type
15557 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15558 * df-core.c (rest_of_handle_df_initialize): Adjust.
15559 (rest_of_handle_df_finish): Likewise.
15560 (df_analyze_1): Likewise.
15561 (df_analyze): Likewise.
15562 (loop_inverted_post_order_compute): Change argument to be a vec *.
15563 (df_analyze_loop): Adjust.
15564 (df_get_n_blocks): Likewise.
15565 (df_get_postorder): Likewise.
15566 * df.h (struct df_d): Change field to be a vec.
15567 * lcm.c (compute_laterin): Adjust.
15568 (compute_available): Likewise.
15569 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15570 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15571 * tree-ssa-pre.c (compute_antic): Likewise.
15573 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15575 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15576 (depth_first_search::depth_first_search): Change structure init
15577 function to this constructor.
15578 (depth_first_search::add_bb): Rename function to this member.
15579 (depth_first_search::execute): Likewise.
15580 (flow_dfs_compute_reverse_finish): Adjust.
15582 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15584 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15585 (longest_simple_path): Likewise.
15586 * shrink-wrap.c (spread_components): Likewise.
15587 (disqualify_problematic_components): Likewise.
15588 (emit_common_heads_for_components): Likewise.
15589 (emit_common_tails_for_components): Likewise.
15590 (insert_prologue_epilogue_for_components): Likewise.
15592 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15594 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15597 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15599 * df-core.c (df_set_blocks): Start using auto_bitmap.
15600 (df_compact_blocks): Likewise.
15601 * df-problems.c (df_rd_confluence_n): Likewise.
15602 * df-scan.c (df_insn_rescan_all): Likewise.
15603 (df_process_deferred_rescans): Likewise.
15604 (df_update_entry_block_defs): Likewise.
15605 (df_update_exit_block_uses): Likewise.
15606 (df_entry_block_bitmap_verify): Likewise.
15607 (df_exit_block_bitmap_verify): Likewise.
15608 (df_scan_verify): Likewise.
15609 * lra-constraints.c (lra_constraints): Likewise.
15610 (undo_optional_reloads): Likewise.
15611 (lra_undo_inheritance): Likewise.
15612 * lra-remat.c (calculate_gen_cands): Likewise.
15613 (do_remat): Likewise.
15614 * lra-spills.c (assign_spill_hard_regs): Likewise.
15615 (spill_pseudos): Likewise.
15616 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15617 (bitmap_set_subtract_values): Likewise.
15619 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15621 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15622 management with auto_bitmap.
15623 (fix_inter_tick): Likewise.
15624 (fix_recovery_deps): Likewise.
15625 * ira.c (add_store_equivs): Likewise.
15626 (find_moveable_pseudos): Likewise.
15627 (split_live_ranges_for_shrink_wrap): Likewise.
15628 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15629 (rtx_reuse_manager::seen_def_p): Likewise.
15630 (rtx_reuse_manager::set_seen_def): Likewise.
15631 * print-rtl.h (class rtx_reuse_manager): Likewise.
15633 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15635 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15637 (migrate_btr_def): Likewise.
15638 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15639 * df-core.c (loop_post_order_compute): Likewise.
15640 (loop_inverted_post_order_compute): Likewise.
15641 * hsa-common.h: Likewise.
15642 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15643 * init-regs.c (initialize_uninitialized_regs): Likewise.
15644 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15645 (inline_small_functions): Likewise.
15646 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15647 * ira.c (combine_and_move_insns): Likewise.
15648 (build_insn_chain): Likewise.
15649 * loop-invariant.c (find_invariants): Likewise.
15650 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15651 * predict.c (tree_predict_by_opcode): Likewise.
15652 (predict_paths_leading_to): Likewise.
15653 (predict_paths_leading_to_edge): Likewise.
15654 (estimate_loops_at_level): Likewise.
15655 (estimate_loops): Likewise.
15656 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15657 (spread_components): Likewise.
15658 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15659 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15660 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15661 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15662 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15663 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15664 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15665 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15666 (mark_threaded_blocks): Likewise.
15667 (thread_through_all_blocks): Likewise.
15668 * tree-ssa.c (verify_ssa): Likewise.
15669 (execute_update_addresses_taken): Likewise.
15670 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15672 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15674 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15676 (post_order_compute): Likewise.
15677 (inverted_post_order_compute): Likewise.
15678 (pre_and_rev_post_order_compute_fn): Likewise.
15680 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15682 * genrecog.c (int_set::int_set): Explicitly construct our
15683 auto_vec base class.
15684 * vec.h (auto_vec::auto_vec): New constructor.
15686 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15688 * bitmap.h (class auto_bitmap): New constructor taking
15689 bitmap_obstack * argument.
15691 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15693 * bitmap.h (class auto_bitmap): Change type of m_bits to
15694 bitmap_head, and adjust ctor / dtor and member operators.
15696 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15698 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15699 when returned register mode doesn't match original mode.
15701 2017-05-12 Jeff Law <law@redhat.com>
15702 Jakub Jelinek <jakub@redhat.com>
15704 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15705 we look for cc setter after the compare-elim changes.
15706 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15707 within the vector to match what compare-elim now expects.
15708 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15709 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15711 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15712 after the compare-elim changes.
15713 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15714 the vector to match what compare-elim now expects.
15715 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15716 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15717 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15718 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15719 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15721 * config/visium/visium.c (single_set_and_flags): Fix where
15722 we look for cc setter after the compare-elim changes.
15723 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15724 with the vector to match what compare-elim now expects.
15725 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15726 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15727 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15728 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15729 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15730 (neg<mode>2_insn_set_overflow): Likewise.
15732 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15734 PR middle-end/79794
15735 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15736 maybe_expand_insn call, set ops[0].target. If still set after call,
15737 set alt_rtl. Add extra arg to recursive calls.
15738 (extract_bit_field): Add alt_rtl argument. Pass to
15740 * expmed.h (extract_bit_field): Fix prototype.
15741 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15742 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15743 to extract_bit_field_calls.
15744 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15745 Pass alt_rtl to extract_bit_field calls.
15746 * calls.c (store_unaligned_arguments_into_psuedos)
15747 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15748 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15750 * optabs.h (struct expand_operand): Add target bitfield.
15752 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15754 * compare-elim.c (try_eliminate_compare): Canonicalize
15755 operation with embedded compare to
15756 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15757 (set (reg) (operation)].
15759 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15761 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15764 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15765 cost of adding a carry flag for ADC instruction.
15766 [case MINUS]: Ignore the cost of subtracting a carry flag
15767 for SBB instruction.
15769 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15771 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15773 * config/rs6000/bmiintrin.h: New file.
15774 * config/rs6000/bmi2intrin.h: New file.
15775 * config/rs6000/x86intrin.h: New file.
15777 2017-05-12 Jeff Law <law@redhat.com>
15779 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15782 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15784 PR middle-end/80707
15785 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15787 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15789 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15790 early expansion of vector divide builtins.
15791 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15792 builtins identified as having unsigned arguments.
15794 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15796 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15797 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15798 expansion of vector logical operations (and, andc, or, xor,
15801 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15803 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15804 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15806 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15808 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15809 early GIMPLE expansion of vector multiplies.
15811 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15813 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15814 TARGET_HAVE_MOVT conditional.
15815 (movt splitter): Likewise.
15817 2017-05-12 Richard Biener <rguenther@suse.de>
15819 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15820 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15821 Fold all stmts not inplace.
15823 2017-05-12 Richard Biener <rguenther@suse.de>
15825 PR tree-optimization/80713
15826 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15827 inserted_exprs bit for not removed stmts.
15829 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15831 PR middle-end/69921
15832 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15833 parallelized" attribute for parallelized OpenACC kernels.
15834 * omp-offload.c (execute_oacc_device_lower): Use it.
15836 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15837 Set "oacc kernels" attribute.
15838 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15839 parameter. Adjust all users.
15840 (oacc_fn_attrib_kernels_p): Remove function.
15841 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15842 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15843 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15844 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15845 assert "oacc kernels" attribute is set.
15847 2017-05-11 Carl Love <cel@us.ibm.com>
15849 * config/rs6000/rs6000-c: Add support for built-in functions
15850 vector unsigned char vec_popcnt (vector signed char)
15851 vector unsigned char vec_popcnt (vector unsigned char)
15852 vector unsigned short vec_popcnt (vector signed short)
15853 vector unsigned short vec_popcnt (vector unsigned short)
15854 vector unsigned int vec_popcnt (vector signed int)
15855 vector unsigned int vec_popcnt (vector unsigned int)
15856 vector unsigned long long vec_popcnt (vector signed long long)
15857 vector unsigned long long vec_popcnt (vector unsigned long long)
15858 vector signed long long vec_slo (vector signed long long,
15859 vector signed char)
15860 vector signed long long vec_slo (vector signed long long,
15861 vector unsigned char)
15862 vector unsigned long long vec_slo (vector unsigned long long,
15863 vector signed char)
15864 vector unsigned long long vec_slo (vector unsigned long long,
15865 vector unsigned char)
15866 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15867 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15868 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15869 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15870 * doc/extend.texi: Update the built-in documentation file for the
15871 new built-in functions.
15873 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15875 * attribs.h (sorted_attr_string): Move machine independent
15876 functions for target clone support from the i386 port to common
15877 code. Rename ix86_function_versions to common_function_versions.
15878 Rename make_name to make_unique_name.
15879 (common_function_versions): Likewise.
15880 (make_unique_name): Likewise.
15881 (make_dispatcher_decl): Likewise.
15882 (is_function_default_version): Likewise.
15883 * attribs.c (attr_strcmp): Likewise.
15884 (sorted_attr_string): Likewise.
15885 (common_function_versions): Likewise.
15886 (make_unique_name): Likewise.
15887 (make_dispatcher_decl): Likewise.
15888 (is_function_default_version): Likewise.
15889 * config/i386/i386.c (attr_strcmp): Likewise.
15890 (sorted_attr_string): Likewise.
15891 (ix86_function_versions): Likewise.
15892 (make_name): Likewise.
15893 (make_dispatcher_decl): Likewise.
15894 (is_function_default_version): Likewise.
15895 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15897 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15900 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15901 Account for direct move costs for vec_construct of integer
15904 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15907 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15908 (UNSPEC_STX_ATOMIC): Ditto.
15909 (loaddi_via_sse): New insn.
15910 (storedi_via_sse): Ditto.
15911 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15912 Update corresponding peephole2 patterns.
15913 (atomic_storedi_fpu): Ditto.
15915 2017-05-11 Julia Koval <julia.koval@intel.com>
15917 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15918 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15920 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15921 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15922 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15924 2017-05-11 Nathan Sidwell <nathan@acm.org>
15926 * graphite-poly.c: Include dumpfile.h.
15928 * dumpfle.h (dump_function): Declare here ...
15929 * tree-dump.h (dump_function): ... not here.
15930 * dumpfile.c: #include tree-cfg.h.
15931 (dump_function): Move here from ...
15932 * tree-dump.c (dump_function): ... here.
15933 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15934 * graphite-poly.c: Don't include tree-dump.h.
15935 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15936 * print-tree.c: Likewise.
15937 * stor-layout.c: Likewise.
15938 * tree-nested.c: Likewise.
15940 * dumpfile.c (dump_start): Use TDF_FLAGS.
15941 (dump_enable_all): Fix TDF_KIND check thinko.
15943 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15945 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15946 array entries to represent two legal parameterizations of the
15947 overloaded __builtin_cmpb function, as represented by the
15948 P6_OV_BUILTIN_CMPB constant.
15949 (altivec_resolve_overloaded_builtin): Add special case handling
15950 for the __builtin_cmpb function, as represented by the
15951 P6_OV_BUILTIN_CMPB constant.
15952 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15953 (BU_P6_64BIT_2): New macro.
15954 (BU_P6_OVERLOAD_2): New macro
15955 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15956 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15957 (CMPB): Add overload support to represent both 32-bit and 64-bit
15958 compare-bytes function.
15959 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15960 support for TARGET_CMPB.
15961 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15962 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15963 documentation of the __builtin_cmpb overloaded built-in function.
15965 2017-05-11 Richard Biener <rguenther@suse.de>
15967 PR tree-optimization/80705
15968 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15969 bases are not vectorizable.
15971 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15973 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15974 when counting register pressure.
15976 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15978 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15979 (iv_ca_more_deps): Renamed to ...
15980 (iv_ca_compare_deps): ... this.
15981 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15983 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15985 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15987 (determine_group_iv_costs): ... here.
15988 (find_inv_vars_cb): Record inv var if it's not recorded before.
15990 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15992 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15993 (get_shiftadd_cost): Ditto.
15995 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15997 * tree-ssa-address.c: Include header file.
15998 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16000 (add_to_parts): Refactor.
16001 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16002 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16005 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16007 PR tree-optimization/53090
16008 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16010 (extract_cond_operands): Detect condition with IV on both sides
16011 and return COMP_IV_EXPR_2.
16012 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16013 (rewrite_use_compare): Simplify by removing call to function
16014 extract_cond_operands.
16016 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16018 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16019 (extract_cond_operands): Detect condition comparing against non-
16020 invariant bound and return appropriate enum value.
16021 (find_interesting_uses_cond): Update use of extract_cond_operands.
16022 Handle its return value accordingly.
16023 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16027 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16028 nonlinear iv_use computation in loop invariant sensitive way.
16030 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16032 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16033 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16035 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16037 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16038 (dump_cand): Support iv_cand.inv_exprs.
16039 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16041 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16044 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16046 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16048 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16049 as local function. Include necessary header files.
16050 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16052 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16054 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16056 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16058 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16059 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16060 RSHIFT_EXPR and BIT_NOT_EXPR.
16062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16064 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16065 (adjust_setup_cost): New parameter supporting round up adjustment.
16066 (struct address_cost_data): Delete.
16067 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16068 (split_address_cost, ptr_difference_cost): Delete.
16069 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16070 (struct ainc_cost_data): New struct.
16071 (get_address_cost_ainc): New function.
16072 (get_address_cost, get_computation_cost): Reimplement.
16073 (determine_group_iv_cost_address): Record inv_expr for all uses of
16075 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16076 (iv_ca_has_deps): Reimplemented to ...
16077 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16079 (iv_ca_extend): Call iv_ca_more_deps.
16081 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16083 * tree-ssa-address.c (struct mem_address): Move to header file.
16084 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16085 * tree-ssa-address.h (struct mem_address): Move from C file.
16086 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16090 * tree-affine.h (aff_combination_type): New interface.
16091 (aff_combination_zero_p): Remove static.
16092 (aff_combination_const_p): New interface.
16093 (aff_combination_singleton_var_p): New interfaces.
16095 2017-05-11 Richard Biener <rguenther@suse.de>
16097 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16098 Skip unreachable blocks and destinations.
16099 (eliminate): Move stmt removal and fixup ...
16100 (fini_eliminate): ... here. Skip inserted exprs.
16101 (pass_pre::execute): Move fini_pre after fini_eliminate.
16102 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16103 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16104 PRE to get rid of dead code that has invalid SSA form and
16105 split critical edges again.
16107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16109 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16111 2017-05-11 Richard Biener <rguenther@suse.de>
16113 * passes.c (execute_function_todo): Verify loops if they are
16114 said to be up-to-date.
16115 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16116 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16118 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16121 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16122 handle calling assemble_external ourself.
16125 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16126 modes with zero size. Enhance comment.
16128 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16130 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16131 built-ins for vec_xl and vec_xst with short and char pointer
16134 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16136 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16137 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16138 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16139 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16140 (_mm_maskz_min_round_ss): New intrinsics.
16141 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16142 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16143 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16144 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16145 (__builtin_ia32_minss_mask_round): New builtins.
16146 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16147 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16148 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16150 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16151 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16153 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16156 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16158 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16159 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16160 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16161 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16162 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16163 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16164 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16165 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16166 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16167 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16168 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16169 (__builtin_ia32_mulss_mask_round): New builtins.
16170 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16171 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16172 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16174 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16175 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16177 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16180 2017-05-10 Julia Koval <julia.koval@intel.com>
16182 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16183 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16184 (_mm256_setr_m128i): New intrinsics.
16186 2017-05-10 Julia Koval <julia.koval@intel.com>
16188 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16189 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16190 (_mm_maskz_rcp14_ss): New intrinsics.
16191 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16192 (__builtin_ia32_rcp14ss_mask): New builtins.
16193 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16195 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16197 PR tree-optimization/51513
16198 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16199 (assert_unreachable_fallthru_edge_p): Use it.
16200 (group_case_labels_stmt): Likewise.
16201 * tree-cfg.h: Prototype it.
16202 * stmt.c: Include cfghooks.h and tree-cfg.h.
16203 (emit_case_dispatch_table) <gap_label>: New local variable.
16204 Use it to fill dispatch table gaps.
16205 Test for default_label before updating probabilities.
16206 (expand_case) <default_label>: Remove unneeded initialization.
16207 Test for unreachable default case statement and remove its edge.
16208 Set default_label accordingly.
16209 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16211 2017-05-10 Carl Love <cel@us.ibm.com>
16213 * config/rs6000/rs6000-c: Add support for built-in functions
16214 vector signed char vec_neg (vector signed char)
16215 vector signed short int vec_neg (vector short int)
16216 vector signed int vec_neg (vector signed int)
16217 vector signed long long vec_neg (vector signed long long)
16218 vector float vec_neg (vector float)
16219 vector double vec_neg (vector double)
16220 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16222 * config/rs6000/altivec.h: Add define for vec_neg
16223 * doc/extend.texi: Update the built-in documentation for the
16224 new built-in functions.
16226 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16228 PR tree-optimization/77644
16229 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16231 2017-05-10 Nathan Sidwell <nathan@acm.org>
16233 * dumpfile.h (TDI_lang_all): New.
16234 (TDF_KIND): New. Renumber others
16235 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16237 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16239 (get_dump_file_name): Adjust suffix generation.
16240 (dump_enable_all): Use TDF_KIND.
16241 * doc/invoke.texi (-fdump-lang-all): Document.
16243 * dumpfile.h: Tabify.
16245 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16248 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16249 Move member access before delete.
16251 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16253 * tree-inline.c (expand_call_inline): Split block at stmt
16256 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16259 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16260 are now unused after splitting mov{sf,sd}_hardfloat.
16261 (f32_lr2): Likewise.
16262 (f32_lm): Likewise.
16263 (f32_lm2): Likewise.
16264 (f32_li): Likewise.
16265 (f32_li2): Likewise.
16266 (f32_lv): Likewise.
16267 (f32_sr): Likewise.
16268 (f32_sr2): Likewise.
16269 (f32_sm): Likewise.
16270 (f32_sm2): Likewise.
16271 (f32_si): Likewise.
16272 (f32_si2): Likewise.
16273 (f32_sv): Likewise.
16274 (f32_dm): Likewise.
16275 (f32_vsx): Likewise.
16276 (f32_av): Likewise.
16277 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16278 For movsf, order stores so the VSX stores occur before the GPR
16279 store which encourages the register allocator to use a traditional
16280 FPR instead of a GPR. For movsd, order the stores so that the GPR
16281 store comes before the VSX stores to allow the power6 to work.
16282 This is due to the power6 not having a 32-bit integer store
16283 instruction from a FPR.
16284 (movsf_hardfloat): Likewise.
16285 (movsd_hardfloat): Likewise.
16287 2017-05-09 Martin Sebor <msebor@redhat.com>
16289 PR translation/80280
16290 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16293 PR translation/80280
16294 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16295 data member added in r247778.
16296 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16298 2017-05-09 Nathan Sidwell <nathan@acm.org>
16300 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16302 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16305 2017-05-09 Marek Polacek <polacek@redhat.com>
16307 * doc/invoke.texi: Fix typo.
16309 2017-05-09 Richard Biener <rguenther@suse.de>
16311 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16312 (vrp_val_is_min): Likewise.
16313 (set_value_range_to_value): Likewise.
16314 (set_value_range_to_nonnegative): Likewise.
16315 (gimple_assign_nonzero_p): Likewise.
16316 (gimple_stmt_nonzero_p): Likewise.
16317 (vrp_int_const_binop): Likewise. Remove unreachable case.
16318 (adjust_range_with_scev): Adjust comments.
16319 (compare_range_with_value): Likewise.
16320 (extract_range_from_phi_node): Likewise.
16321 (test_for_singularity): Likewise.
16323 2017-05-09 Richard Biener <rguenther@suse.de>
16325 * tree-vrp.c (get_single_symbol): Add assert that we don't
16326 get overflowed constants as invariant part.
16327 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16328 checks. Use wi::cmp instead of recursing for integer constants.
16329 (compare_values): Just ignore whether we assumed undefined
16330 overflow instead of failing the compare.
16331 (extract_range_for_var_from_comparison_expr): Add comment before the
16332 TREE_NO_WARNING sets.
16333 (test_for_singularity): Likewise.
16334 (extract_range_from_comparison): Do not disable optimization
16335 when we assumed undefined overflow.
16336 (extract_range_basic): Remove init of unused var.
16338 2017-05-09 Richard Biener <rguenther@suse.de>
16340 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16341 (extract_range_from_multiplicative_op_1): Adjust.
16342 (extract_range_from_binary_expr_1): Use int_const_binop.
16344 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16347 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16348 rs6000_store_data_bypass_p in seven define_bypass directives and
16349 in several comments.
16350 * config/rs6000/rs6000-protos.h: Add prototype for
16351 rs6000_store_data_bypass_p function.
16352 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16353 function implements slightly different (rs6000-specific) semantics
16354 than store_data_bypass_p, returning false rather than aborting
16355 with assertion error when arguments do not satisfy the
16356 requirements of store data bypass.
16357 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16358 rs6000_store_data_bypass_p.
16360 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16362 * config/xtensa/xtensa-protos.h
16363 (xtensa_initial_elimination_offset): New declaration.
16364 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16365 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16366 macro definition, add case for FRAME_POINTER_REGNUM when
16367 FRAME_GROWS_DOWNWARD.
16368 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16369 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16370 xtensa_initial_elimination_offset.
16372 2017-05-08 Nathan Sidwell <nathan@acm.org>
16374 * doc/invoke.texi: Alphabetize -fdump options.
16376 2017-05-08 Martin Sebor <msebor@redhat.com>
16378 PR translation/80280
16379 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16381 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16383 * target.def (compute_frame_layout): New optional target hook.
16384 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16385 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16386 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16388 * reload1.c (verify_initial_elim_offsets): Likewise.
16389 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16390 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16391 (arm_get_frame_offsets): Split up into this ...
16392 (arm_compute_frame_layout): ... and this function.
16394 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16396 * config/aarch64/constraints.md (Usa): New constraint.
16397 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16399 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16401 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16402 with_multilib_list after it has been checked.
16404 2017-05-08 Richard Biener <rguenther@suse.de>
16406 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16407 (bitmap_set_subtract_values): Likewise.
16409 2017-05-08 Richard Biener <rguenther@suse.de>
16411 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16412 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16414 (gimple_stmt_nonzero_warnv_p): Rename to ...
16415 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16417 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16418 (extract_range_basic): Adjust, do not disable propagation on
16419 strict overflow sensitive simplification.
16420 (vrp_visit_cond_stmt): Likewise.
16422 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16424 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16425 body size unconditionally.
16427 2017-05-07 Jeff Law <law@redhat.com>
16430 2017-05-06 Jeff Law <law@redhat.com>
16431 PR tree-optimization/78496
16432 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16435 PR tree-optimization/78496
16436 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16437 (simplify_stmt_using_ranges): Call it.
16438 (vrp_dom_walker::before_dom_children): Extract equivalences
16439 from an ASSERT_EXPR with an equality comparison against a
16442 2017-05-06 Jeff Law <law@redhat.com>
16444 PR tree-optimization/78496
16445 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16448 PR tree-optimization/78496
16449 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16450 (simplify_stmt_using_ranges): Call it.
16451 (vrp_dom_walker::before_dom_children): Extract equivalences
16452 from an ASSERT_EXPR with an equality comparison against a
16455 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16457 * lra-constraints.c (lra_copy_reg_equiv): New function.
16458 (split_reg): Use it to copy equivalence information from the
16459 original register to the spill register.
16461 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16463 PR rtl-optimization/75964
16464 * simplify-rtx.c (simplify_const_relational_operation): Remove
16465 invalid handling of comparisons of integer ABS.
16467 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16469 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16470 initialize to zero.
16471 (init_regs): Remove declaration.
16472 (function_arg_advance_32): Initialize error_p as boolean variable.
16474 2017-05-05 Nathan Sidwell <nathan@acm.org>
16476 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16477 lines. Use for (;;).
16479 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16481 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16482 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16483 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16484 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16485 VF=2 that require versioning.
16487 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16489 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16492 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16494 * diagnostic.h (diagnostic_override_option_index): Convert from
16495 macro to inline function.
16497 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16499 * diagnostic.c (last_module_changed_p): New function.
16500 (set_last_module): New function.
16501 (diagnostic_report_current_module): Convert macro usage to
16502 the above functions.
16503 * diagnostic.h (diagnostic_context::last_module): Strengthen
16504 from const line_map * to const line_map_ordinary *.
16505 (diagnostic_last_module_changed): Delete macro.
16506 (diagnostic_set_last_module): Delete macro.
16508 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16510 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16511 with diagnostic_report_diagnostic.
16512 (diagnostic_n_impl_richloc): Likewise.
16513 * diagnostic.h (report_diagnostic): Delete macro.
16514 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16515 with diagnostic_report_diagnostic.
16516 * substring-locations.c (format_warning_va): Likewise.
16518 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16520 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16521 save/restor of format_spec. Move option-printing code to...
16522 (print_option_information): ...this new function, and
16523 reimplement by simply printing to the pretty_printer,
16524 rather than appending to the format string.
16526 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16528 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16529 handling logic into...
16530 (update_effective_level_from_pragmas): ...this new function.
16532 2017-05-04 Andrew Waterman <andrew@sifive.com>
16534 * config/riscv/riscv.opt (mstrict-align): New option.
16535 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16536 (SLOW_UNALIGNED_ACCESS): Define.
16537 (riscv_slow_unaligned_access): Declare.
16538 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16540 (riscv_slow_unaligned_access): New variable.
16541 (rocket_tune_info): Set slow_unaligned_access to true.
16542 (optimize_size_tune_info): Set slow_unaligned_access to false.
16543 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16544 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16545 (riscv_option_override): Set riscv_slow_unaligned_access.
16546 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16548 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16550 * config/riscv/riscv.md: Unify indentation.
16552 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16557 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16559 (extendsi<mode>2): Add support for doing sign extension via
16560 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16561 don't have ISA 3.0 instructions.
16562 (extendsi<mode>2 splitter): Likewise.
16563 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16564 generate the normal insns since SImode can now go in vector
16565 registers. Disallow the special UNSPECs needed for previous
16566 machines to hide SImode being used. Add new insns
16567 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16568 (fix_trunc<mode>si2_stfiwx): Likewise.
16569 (fix_trunc<mode>si2_internal): Likewise.
16570 (fixuns_trunc<mode>si2): Likewise.
16571 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16572 (fctiw<u>z_<mode>_smallint): Likewise.
16573 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16574 of floating point to 32-bit integer from doing a direct move to
16575 the GPR registers to do a store.
16576 (fctiwz_<mode>): Break long line.
16578 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16580 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16581 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16582 (addr_list, addr_offset_valid_p): New.
16583 (split_address_groups): Check offset validity with above function.
16584 (gt-tree-ssa-loop-ivopts.h): Include header file.
16586 2017-05-05 Nathan Sidwell <nathan@acm.org>
16588 * config.gcc (arm*-*-*): Add missing 'fi'.
16590 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16592 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16593 included in -fopt-info does not matter.
16594 * doc/optinfo.texi (-fopt-info): Fix description of default
16595 behavour. Explicitly say order of options included in -fopt-info
16598 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16600 * config.gcc: Allow combinations of aprofile and rmprofile values for
16601 --with-multilib-list.
16602 * config/arm/t-multilib: New file.
16603 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16604 variables. Remove setting of ISA and floating-point ABI in
16605 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16606 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16607 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16608 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16610 * config/arm/t-rmprofile: Likewise except for the matches changes.
16611 * doc/install.texi (--with-multilib-list): Document the combination of
16612 aprofile and rmprofile values and warn about pitfalls in doing that.
16614 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16616 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16617 (movdi_aarch64): Likewise.
16619 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16621 PR tree-optimization/80632
16622 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16624 (build_arrays): Initialize it for virtual phis.
16625 (fix_phi_nodes): Use it for virtual phis.
16627 PR tree-optimization/80558
16628 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16629 [x, y] op z into [x op, y op z] for op & or | if conditions
16632 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16633 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16636 * config/arm/arm.md (use_literal_pool): Remove.
16637 (64-bit immediate split): No longer takes cost into consideration
16638 if arm_disable_literal_pool is enabled.
16639 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16640 used when arm_disable_literal_pool is enabled.
16641 (arm_max_const_double_inline_cost): Remove use of
16642 arm_disable_literal_pool.
16643 (push_minipool_fix): Add assert.
16644 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16645 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16646 (no_literal_pool_sf_immediate): New.
16648 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16650 PR tree-optimization/80613
16651 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16652 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16654 2017-05-05 Richard Biener <rguenther@suse.de>
16656 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16658 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16660 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16661 of this flag from insn conditions due to removal from r247495.
16663 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16665 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16667 (arm_early_store_addr_dep_ptr): Likewise.
16668 * config/arm/aarch-common-protos.h
16669 (arm_early_load_addr_dep_ptr): Add prototype.
16670 (arm_early_store_addr_dep_ptr): Likewise.
16671 * config/arm/cortex-a53.md: Add new bypasses.
16673 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16675 * tree.c (next_type_uid): Change type to unsigned.
16676 (type_hash_canon): Decrement back next_type_uid if
16677 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16678 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16681 2017-05-04 Martin Sebor <msebor@redhat.com>
16683 * builtins.c: Fix a trivial typo in a comment.
16685 PR middle-end/79234
16686 * builtins.c (check_sizes): Adjust to handle reading past the end.
16687 Avoid printing excessive upper bound of ranges. Use %E to print
16688 tree nodes instead of converting them to %wu.
16689 (expand_builtin_memchr): New function.
16690 (compute_dest_size): Rename...
16691 (compute_objsize): ...to this.
16692 (expand_builtin_memcpy): Adjust.
16693 (expand_builtin_mempcpy): Adjust.
16694 (expand_builtin_strcat): Adjust.
16695 (expand_builtin_strcpy): Adjust.
16696 (check_strncat_sizes): Adjust.
16697 (expand_builtin_strncat): Adjust.
16698 (expand_builtin_strncpy): Adjust and simplify.
16699 (expand_builtin_memset): Adjust.
16700 (expand_builtin_bzero): Adjust.
16701 (expand_builtin_memcmp): Adjust.
16702 (expand_builtin): Handle memcmp.
16703 (maybe_emit_chk_warning): Check strncat just once.
16705 2017-05-04 Martin Sebor <msebor@redhat.com>
16707 PR preprocessor/79214
16708 PR middle-end/79222
16709 PR middle-end/79223
16710 * builtins.c (check_sizes): Add inlining context and issue
16711 warnings even when -Wno-system-headers is set.
16712 (check_strncat_sizes): Same.
16713 (expand_builtin_strncat): Same.
16714 (expand_builtin_memmove): New function.
16715 (expand_builtin_stpncpy): Same.
16716 (expand_builtin): Handle memmove and stpncpy.
16718 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16720 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16721 which is not used any more.
16723 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16725 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16727 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16729 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16730 (cortexa53_tunings): Likewise.
16731 (cortexa57_tunings): Likewise.
16732 (cortexa72_tunings): Likewise.
16733 (cortexa73_tunings): Likewise.
16735 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16737 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16738 Set loop alignment to 8.
16740 2017-05-04 Martin Sebor <msebor@redhat.com>
16742 PR translation/80280
16743 * builtins.c (expand_builtin_object_size): Add missing quoting to
16744 %D and like directives.
16745 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16746 (hsa_type_for_tree_type): Same.
16747 (verify_function_arguments): Same.
16748 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16749 * varasm.c (get_section): Same.
16752 2017-05-04 Martin Sebor <msebor@redhat.com>
16754 PR translation/80280
16755 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16757 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16759 * config/aarch64/aarch64.c (generic_addrcost_table):
16760 Change HI/TI mode setting.
16762 2017-05-04 Martin Jambor <mjambor@suse.cz>
16764 PR tree-optimization/80622
16765 * tree-sra.c (comes_initialized_p): New function.
16766 (build_accesses_from_assign): Only set write lazily when
16767 comes_initialized_p is false.
16768 (analyze_access_subtree): Use comes_initialized_p.
16769 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16770 instead of testing for PARM_DECL.
16772 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16774 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16775 constraint on operand 0 to allow more general addressing modes.
16776 Adjust output template.
16777 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16779 * config/aarch64/aarch64-protos.h
16780 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16781 * config/aarch64/constraints.md (Dp): New address constraint.
16782 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16785 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16787 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16788 update use of estimate_ipcp_clone_size_and_time.
16789 (estimate_local_effects): Update use of
16790 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16791 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16792 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16793 Return nonspecialized time.
16795 2017-05-04 Richard Biener <rguenther@suse.de>
16797 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16798 for the last VUSE which def dominates the PHI. Directly call
16800 (get_continuation_for_phi_1): Remove.
16802 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16804 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16805 to explain the use of truncating division. Cap the number of
16806 iterations to the maximum given by nb_iterations_upper_bound,
16809 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16811 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16812 * configure: Regenerate.
16813 * config.in: Regenerate.
16814 * config/i386/driver-mingw32.c: new file.
16815 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16816 * config.host: Link driver-mingw32.o on MinGW host.
16817 * doc/install.texi: Document new --enable-mingw-wildcard configure
16820 2017-05-04 Marek Polacek <polacek@redhat.com>
16822 PR tree-optimization/80612
16823 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16825 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16826 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16828 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16829 (movt splitter): Likewise.
16830 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16831 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16832 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16833 block for Thumb-1 with MOVT.
16834 (thumb2_legitimate_address_p): Move code block ...
16835 (can_avoid_literal_pool_for_label_p): ... into this new function.
16836 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16838 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16839 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16840 "M-profile targets with the MOVT instruction".
16842 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16844 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16845 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16846 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16848 2017-05-04 Martin Liska <mliska@suse.cz>
16850 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16851 variable cond_code.
16853 2017-05-04 Richard Biener <rguenther@suse.de>
16855 * tree.c (array_at_struct_end_p): Handle arrays at struct
16856 end with flexarrays more conservatively. Refactor and treat
16857 arrays of arrays or aggregates more strict. Fix
16858 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16859 * tree.c (array_at_struct_end_p): Adjust prototype.
16860 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16861 * gimple-fold.c (get_range_strlen): Likewise.
16862 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16864 2017-05-04 Richard Biener <rguenther@suse.de>
16866 PR tree-optimization/31130
16867 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16869 (supports_overflow_infinity): Likewise.
16870 (is_negative_overflow_infinity): Likewise.
16871 (is_positive_overflow_infinity): Likewise.
16872 (is_overflow_infinity): Likewise.
16873 (stmt_overflow_infinity): Likewise.
16874 (overflow_infinity_range_p): Likewise.
16875 (usable_range_p): Remove as always returning true.
16876 (make_overflow_infinity): Remove.
16877 (negative_overflow_infinity): Likewise.
16878 (positive_overflow_infinity): Likewise.
16879 (avoid_overflow_infinity): Likewise.
16880 (set_value_range): Adjust accordingly.
16881 (set_value_range_to_nonnegative): Likewise, remove now unused
16882 overflow_infinity arg.
16883 (vrp_operand_equal_p): Adjust.
16884 (update_value_range): Likewise.
16885 (range_int_cst_singleton_p): Likewise.
16886 (operand_less_p): Likewise.
16887 (compare_values_warnv): Likewise.
16888 (extract_range_for_var_from_comparison_expr): Likewise.
16889 (vrp_int_const_binop): Likewise.
16890 (zero_nonzero_bits_from_vr): Likewise.
16891 (extract_range_from_multiplicative_op_1): Likewise.
16892 (extract_range_from_binary_expr_1): Likewise.
16893 (extract_range_from_unary_expr): Likewise.
16894 (extract_range_from_comparison): Likewise.
16895 (extract_range_basic): Likewise.
16896 (adjust_range_with_scev): Likewise.
16897 (compare_ranges): Likewise.
16898 (compare_range_with_value): Likewise.
16899 (dump_value_range): Likewise.
16900 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16902 (simplify_cond_using_ranges): Adjust.
16904 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16906 * brig-builtins.def: Added a builtin for class_f64.
16907 * builtin-types.def: Added a builtin type needed by class_f64.
16909 2017-05-03 Jason Merrill <jason@redhat.com>
16911 * timevar.def: Add TV_CONSTEXPR.
16913 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16915 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16917 2017-05-03 Martin Jambor <mjambor@suse.cz>
16919 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16920 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16921 * ipa-cp.c (ipcp_propagate_stage): Do not call
16922 ipa_update_after_lto_read.
16923 * ipa-inline.c (ipa_inline): Likewise.
16925 2017-05-03 Martin Jambor <mjambor@suse.cz>
16927 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16928 tag. Added a default constructor and a destructor.
16929 (ipa_edge_args_sum_t): New class;
16930 (ipa_edge_args_sum): Declare.
16931 (ipa_edge_args_vector): Remove declaration.
16932 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16933 (ipa_free_edge_args_substructures): Remove declaration.
16934 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16935 (ipa_edge_args_info_available_for_edge_p): Likewise.
16936 * ipa-prop.c (ipa_edge_args_vector): Removed.
16937 (edge_removal_hook_holder): Likewise.
16938 (edge_duplication_hook_holder): Likewise.
16939 (ipa_edge_args_sum): New variable.
16940 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16941 ipa_edge_args_vector.
16942 (ipa_free_edge_args_substructures): Likewise.
16943 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16944 ipa_edge_args_vector.
16945 (ipa_edge_removal_hook): Turned into method
16946 ipa_edge_args_sum_t::remove.
16947 (ipa_edge_duplication_hook): Turned into method
16948 ipa_edge_args_sum_t::duplicate.
16949 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16950 registering edge hooks.
16951 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16952 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16953 ipa_edge_args_sum instead of ipa_edge_args_vector.
16954 * ipa-profile.c (ipa_profile): Likewise.
16956 2017-05-03 Martin Jambor <mjambor@suse.cz>
16958 * symbol-summary.h (function_summary): New method exists.
16959 (function_summary::symtab_removal): Deallocate through release.
16960 (call_summary): New class.
16961 (gt_ggc_mx): New overload.
16962 (gt_pch_nx): Likewise.
16963 (gt_pch_nx): Likewise.
16965 2017-05-03 Jeff Law <law@redhat.com>
16967 PR tree-optimization/78496
16968 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16969 from simplify_cond_using_ranges. Split off code to walk
16970 backwards through casts into ...
16971 (simplify_cond_using_ranges_2): New function.
16972 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16973 (execute_vrp): After identifying jump threads, call
16974 simplify_cond_using_ranges_2.
16976 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16979 * ipa-inline.h (inline_summary): Add ctor.
16980 (create_ggc): Do not use ggc_cleared_alloc.
16982 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16983 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16985 * gcc.c (handle_braces): Support escaping in switch matching
16987 * doc/invoke.texi (Spec Files): Document it.
16988 Remove superfluous @code markup in items.
16990 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16992 * diagnostic-show-locus.c (struct column_range): New struct.
16993 (get_affected_columns): New function.
16994 (get_printed_columns): New function.
16995 (struct correction): New struct.
16996 (correction::ensure_capacity): New function.
16997 (correction::ensure_terminated): New function.
16998 (struct line_corrections): New struct.
16999 (line_corrections::~line_corrections): New dtor.
17000 (line_corrections::add_hint): New function.
17001 (layout::print_trailing_fixits): Reimplement in terms of the new
17003 (selftest::test_overlapped_fixit_printing): New function.
17004 (selftest::diagnostic_show_locus_c_tests): Call it.
17006 2017-05-03 Nathan Sidwell <nathan@acm.org>
17008 Canonicalize canonical type hashing
17009 * tree.h (type_hash_canon_hash): Declare.
17010 * tree.c (type_hash_list, attribute_hash_list): Move into
17011 type_hash_canon_hash.
17012 (build_type_attribute_qual_variant): Break out hash code calc into
17013 type_hash_canon_hash.
17014 (type_hash_canon_hash): New. Generic type hash computation.
17015 (build_range_type_1, build_array_type_1, build_function_type,
17016 build_method_type_directly, build_offset_type, build_complex_type,
17017 make_vector_type): Call it.
17019 2017-05-03 Richard Biener <rguenther@suse.de>
17021 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17022 When all DRs have unknown misaligned do not always peel
17023 when there is a store but apply the same costing model as if
17024 there were only loads.
17026 2017-05-03 Richard Biener <rguenther@suse.de>
17029 PR tree-optimization/80492
17030 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17031 compare_base_decls returning dont-know properly.
17033 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17035 * config/arm/iterators.md (CCSI): New mode iterator.
17036 (arch): New mode attribute.
17037 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17038 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17039 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17040 code iterator for success result mode.
17041 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17042 the corresponding new insn generators.
17044 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17047 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17048 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17050 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17052 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17053 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17054 (DDR_A): Wrap DDR argument in brackets.
17055 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17056 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17057 (DDR_REVERSED_P): Likewise.
17059 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17061 PR tree-optimization/79472
17062 * tree-switch-conversion.c (struct switch_conv_info): Add
17063 contiguous_range and default_case_nonstandard fields.
17064 (collect_switch_conv_info): Compute contiguous_range and
17065 default_case_nonstandard fields, don't clear final_bb if
17066 contiguous_range and only the default case doesn't have the required
17068 (check_all_empty_except_final): Set default_case_nonstandard instead
17069 of failing if contiguous_range and the default case doesn't have empty
17071 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17072 and only the default case doesn't have the required constants. Skip
17074 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17075 if default_case_nonstandard.
17076 (build_constructors): Build constant 1 just once. Assert that default
17077 values aren't inserted in between cases if contiguous_range. Skip
17079 (build_arrays): Skip virtual phis.
17080 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17081 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17082 Handle virtual phis.
17083 (gen_inbound_check): Handle default_case_nonstandard case.
17084 (process_switch): Adjust check_final_bb caller. Call
17085 gather_default_values with the first non-default case instead of
17086 default case if default_case_nonstandard.
17088 2017-05-02 Nathan Sidwell <nathan@acm.org>
17090 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17091 check. Fix formatting.
17093 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17095 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17096 errors when comparing specialized and unspecialized times.
17098 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17100 * diagnostic-show-locus.c
17101 (layout::should_print_annotation_line_p): Make private.
17102 (layout::print_annotation_line): Make private.
17103 (layout::annotation_line_showed_range_p): Make private.
17104 (layout::show_ruler): Make private.
17105 (layout::print_source_line): Make private. Pass in line and
17106 line_width, rather than calling location_get_source_line. Drop
17108 (layout::print_leading_fixits): New method.
17109 (layout::print_any_fixits): Rename to...
17110 (layout::print_trailing_fixits): ...this, and make private.
17111 Don't print newline fixits.
17112 (diagnostic_show_locus): Move logic for printing one row into...
17113 (layout::print_line): ...this new function. Move the
17114 location_get_source_line call and error-handling from
17115 print_source_line to here. Call print_leading_fixits, and rename
17116 print_any_fixits to print_trailing_fixits.
17117 (selftest::test_fixit_insert_containing_newline): Update now that
17118 newlines are partially supported.
17119 (selftest::test_fixit_insert_containing_newline_2): New test.
17120 (selftest::test_fixit_replace_containing_newline): Update comments.
17121 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17122 * edit-context.c (class added_line): New class.
17123 (class edited_line): Describe newline handling in comment.
17124 (edited_line::actually_edited_p): New method.
17125 (edited_line::print_content): Delete redundant decl.
17126 (edited_line::m_predecessors): New field.
17127 (edited_file::print_content): Call edited_line::print_content.
17128 (edited_file::print_diff): Update to support newlines.
17129 (edited_file::print_diff_hunk): Likewise.
17130 (edited_file::print_run_of_changed_lines): New function.
17131 (edited_file::print_diff_line): Convert to...
17132 (print_diff_line): ...this.
17133 (edited_file::get_effective_line_count): New function.
17134 (edited_line::edited_line): Initialize new field m_predecessors.
17135 (edited_line::~edited_line): Clean up m_predecessors.
17136 (edited_line::apply_fixit): Handle newlines.
17137 (edited_line::get_effective_line_count): New function.
17138 (edited_line::print_content): New function.
17139 (edited_line::print_diff_lines): New function.
17140 (selftest::test_applying_fixits_insert_containing_newline): New
17142 (selftest::test_applying_fixits_replace_containing_newline): New
17144 (selftest::insert_line): New function.
17145 (selftest::test_applying_fixits_multiple_lines): Add example of
17147 (selftest::edit_context_c_tests): Call the new tests.
17149 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17151 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17152 parameter cand. Update dump information.
17153 (get_computation_cost): Update uses.
17155 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17157 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17158 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17159 (get_computation_at, rewrite_use_address): Update use of
17160 get_computation_aff.
17162 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17164 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17165 (get_computation): Delete.
17166 (get_computation_cost): Implement like get_computation_cost_at.
17167 Use get_computation_at.
17168 (get_computation_cost_at): Delete.
17169 (rewrite_use_nonlinear_expr): Use get_computation_at.
17170 (rewrite_use_compare, remove_unused_ivs): Ditto.
17172 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17174 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17176 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17178 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17179 (ivopts_global_cost_for_size): Rename parameter and update uses.
17180 (iv_ca_recount_cost): Update uses.
17181 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17182 candidates seperately in n_invs and n_cands.
17183 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17185 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17187 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17188 (find_inv_vars_cb): New.
17189 (find_depends): Renamed to ...
17190 (find_inv_vars): ... this.
17191 (add_candidate_1, force_var_cost): Call find_inv_vars.
17192 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17194 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17196 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17197 inv_vars. Add inv_exprs.
17198 (struct iv_cand): Rename depends_on to inv_vars.
17199 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17200 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17201 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17202 (dump_cand): Dump inv_vars.
17203 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17204 (record_invariant, find_depends, add_candidate_1): Ditto.
17205 (set_group_iv_cost, force_var_cost): Ditto.
17206 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17207 (get_computation_cost_at, get_computation_cost): Ditto.
17208 (determine_group_iv_cost_generic): Ditto.
17209 (determine_group_iv_cost_address): Ditto.
17210 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17211 (determine_group_iv_costs): Ditto.
17212 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17213 (iv_ca_set_remove_invariants): Renamed to ...
17214 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17215 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17216 (iv_ca_set_add_invariants): Renamed to ...
17217 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17218 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17219 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17220 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17221 (create_new_ivs): Remove useless dump.
17223 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17225 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17227 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17228 (iv_ca_set_no_cp, create_new_iv): Ditto.
17230 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17232 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17234 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17236 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17237 function tree_check2.
17239 2017-05-02 Martin Liska <mliska@suse.cz>
17241 * doc/gcov.texi: Add missing preposition.
17242 * gcov.c (function_info::function_info): Properly fill up
17243 all member variables.
17245 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17247 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17249 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17251 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17253 2017-05-02 Martin Liska <mliska@suse.cz>
17256 * lto-streamer-in.c (lto_read_tree_1): Remove
17257 LTO_STREAMER_DEBUG.
17258 * lto-streamer.c (struct tree_hash_entry): Likewise.
17259 (struct tree_entry_hasher): Likewise.
17260 (tree_entry_hasher::hash): Likewise.
17261 (tree_entry_hasher::equal): Likewise.
17262 (lto_streamer_init): Likewise.
17263 (lto_orig_address_map): Likewise.
17264 (lto_orig_address_get): Likewise.
17265 (lto_orig_address_remove): Likewise.
17266 * lto-streamer.h: Likewise.
17267 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17268 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17270 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17272 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17273 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17274 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17275 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17276 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17277 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17278 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17279 (mm_maskz_sub_ss): New intrinsics.
17280 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17281 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17282 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17283 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17284 (__builtin_ia32_subss_mask_round): New builtins.
17285 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17286 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17287 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17289 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17290 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17292 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17295 2017-05-02 Martin Jambor <mjambor@suse.cz>
17297 PR tree-optimization/78687
17298 * tree-sra.c (access): New field parent.
17299 (process_subtree_disqualification): New function.
17300 (disqualify_candidate): Call it.
17301 (build_accesses_from_assign): Reset write flag if creating an
17303 (build_access_subtree): Fill in parent field and also prpagate
17304 down grp_write flag.
17305 (create_artificial_child_access): New parameter set_grp_write, set
17306 grp_write to its value.
17307 (propagate_subaccesses_across_link): Also propagate grp_write flag
17309 (propagate_all_subaccesses): Push the closest parent back to work
17310 queue if add_access_to_work_queue returned true.
17312 2017-05-02 Richard Biener <rguenther@suse.de>
17314 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17315 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17316 -fstrict-overflow documentation.
17317 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17318 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17319 flag_strict_overflow.
17320 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17321 * lto-opts.c (lto_write_options): Do not stream it.
17322 * lto-wrapper.c (merge_and_complain): Do not handle it.
17323 * opts.c (default_options_table): Do not set -fstrict-overflow.
17324 (finish_options): Likewise do not clear it when sanitizing.
17325 * simplify-rtx.c (simplify_const_relational_operation): Do not
17326 test flag_strict_overflow.
17328 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17330 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17331 using enabled attribute.
17332 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17333 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17334 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17335 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17336 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17337 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17338 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17339 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17340 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17341 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17343 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17345 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17347 2017-05-02 Richard Biener <rguenther@suse.de>
17349 PR tree-optimization/80591
17351 2017-04-10 Richard Biener <rguenther@suse.de>
17353 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17356 2017-05-02 Richard Biener <rguenther@suse.de>
17358 PR tree-optimization/80549
17359 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17360 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17361 headers if they do not have a preheader.
17363 2017-05-02 Martin Liska <mliska@suse.cz>
17366 * common.opt: Fix typo.
17367 * doc/invoke.texi: Likewise.
17369 2017-05-01 Jan Beulich <jbeulich@suse.com>
17371 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17372 swapping, add (x,x,m,x,n) alternative.
17374 2017-05-01 Nathan Sidwell <nathan@acm.org>
17376 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17377 unnecessary unadjusted_alignment check.
17379 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17382 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17384 * gimplify.c (gimplify_cilk_detach): New function.
17385 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17386 * tree-core.h: Document EXPR_CILK_SPAWN.
17387 * tree.h (EXPR_CILK_SPAWN): Define.
17389 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17391 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17392 to use new fixit_hint representation, using the "replace" logic.
17393 (get_line_span_for_fixit_hint): Likewise.
17394 (layout::print_any_fixits): Likewise.
17395 (selftest::test_one_liner_many_fixits): Rename to...
17396 (selftest::test_one_liner_many_fixits_1): ...this, and update
17397 comment and expected output to reflect that the multiple fix-it
17398 hints are now consolidated into one insertion.
17399 (selftest::test_one_liner_many_fixits_2): New test.
17400 (selftest::test_diagnostic_show_locus_one_liner): Update for
17402 (selftest::test_fixit_consolidation): Update for fix-it API
17404 * diagnostic.c (print_parseable_fixits): Likewise.
17405 * edit-context.c (edited_line::m_line_events): Convert from
17406 auto_vec <line_event *> to auto_vec <line_event>.
17407 (class line_event): Convert from abstract base class to a concrete
17408 class, taking over the role of replace_event.
17409 (class insert_event): Delete.
17410 (class replace_event): Rename to class line_event. Convert to
17412 (edit_context::add_fixits): Reimplement.
17413 (edit_context::apply_insert): Delete.
17414 (edit_context::apply_replace): Rename to...
17415 (edit_context::apply_fixit): ...this. Convert to half-open range.
17416 (edited_file::apply_insert): Delete.
17417 (edited_file::apply_replace): Rename to...
17418 (edited_file::apply_fixit): ...this.
17419 (edited_line::~edited_line): Drop deletion of events.
17420 (edited_line::apply_insert): Delete.
17421 (edited_line::apply_replace): Rename to...
17422 (edited_line::apply_fixit): ...this. Convert to half-open range.
17423 Update for change to type of m_line_events.
17424 * edit-context.h (edit_context::apply_insert): Delete.
17425 (edit_context::apply_replace): Rename to...
17426 (edit_context::apply_fixit): ...this.
17428 2017-05-01 Martin Sebor <msebor@redhat.com>
17430 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17433 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17436 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17437 __get_cpuid_max returns 0.
17438 (__get_cpuid_count): Ditto.
17440 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17442 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17443 replacement expression is another instance of one of its arguments.
17445 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17448 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17449 check for stack push/pop autoinc.
17450 * config/i386/i386.c (ix86_agi_dependent): Return false
17451 if the only reason why modified_in_p returned true is that
17452 addr is SP based and set_insn is a push or pop.
17454 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17456 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17459 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17462 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17463 (account_size_time): Use two predicates - exec_pred and
17465 (evaluate_conditions_for_known_args): Compute both clause and
17467 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17468 (inline_summary_t::duplicate): Update.
17469 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17471 (compute_inline_parameters): Likewise.
17472 (estimate_edge_size_and_time): Update caluclation of time.
17473 (estimate_node_size_and_time): Compute both time and nonspecialized
17475 (estimate_ipcp_clone_size_and_time): Update.
17476 (inline_merge_summary): Update.
17477 (do_estimate_edge_time): Update.
17478 (do_estimate_edge_size): Update.
17479 (do_estimate_edge_hints): Update.
17480 (inline_read_section, inline_write_summary): Stream both new predicates.
17481 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17483 (compute_inlined_call_time): Cleanup.
17484 (big_speedup_p): Update.
17485 (edge_badness): Update.
17486 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17487 (size_time_entry): Replace predicate by exec_predicate and
17488 nonconst_predicate.
17489 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17490 (estimate_edge_time): Return also nonspec_time.
17491 (reset_edge_growth_cache): Update.
17493 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17495 PR rtl-optimization/80491
17496 * ifcvt.c (noce_process_if_block): When looking for x setter
17497 with missing else_bb, don't check only the insn right before
17498 cond_earliest, but look for the last insn that x is modified in
17499 within the same bb.
17501 PR rtl-optimization/80491
17502 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17504 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17506 PR tree-optimization/80487
17507 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17509 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17511 PR tree-optimization/79697
17512 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17513 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17514 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17516 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17517 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17519 2017-04-28 Martin Sebor <msebor@redhat.com>
17521 PR tree-optimization/80523
17522 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17523 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17525 (maybe_warn, format_directive, parse_directive): Use new functions.
17526 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17528 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17530 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17532 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17534 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17535 target_header_dir): Set correctly.
17536 * configure: Regenerated.
17537 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17538 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17539 instead of SYSTEM_HEADER_DIR.
17541 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17543 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17544 (estimate_local_effects): Likewise.
17545 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17546 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17547 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17548 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17549 do_estimate_edge_time, estimate_edge_time): Likewise.
17550 * ipa-inline-analysis.c (estimate_node_size_and_time,
17551 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17552 (estimate_time_after_inlining): Remove.
17554 2017-04-28 Martin Liska <mliska@suse.cz>
17556 * doc/gcov.texi: Enhance documentation of gcov.
17558 2017-04-28 Martin Liska <mliska@suse.cz>
17560 * doc/gcov.texi: Sort options in alphabetic order.
17561 * doc/gcov-dump.texi: Likewise.
17562 * doc/gcov-tool.texi: Likewise.
17563 * gcov.c (print_usage): Likewise.
17564 * gcov-dump.c (print_usage): Likewise.
17565 * gcov-tool.c (print_merge_usage_message): Likewise.
17566 (print_rewrite_usage_message): Likewise.
17567 (print_overlap_usage_message): Likewise.
17569 2017-04-28 Martin Liska <mliska@suse.cz>
17571 PR gcov-profile/53915
17572 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17574 2017-04-28 Martin Liska <mliska@suse.cz>
17576 PR gcov-profile/79891
17577 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17578 is marked by compiler as living on a line.
17579 (get_cycles_count): Remove usage of the union.
17580 (output_intermediate_file): Likewise.
17581 (find_source): Fix GNU coding style.
17582 (accumulate_line_counts): Remove old non-all block mode.
17583 (output_lines): Remove usage of the union.
17584 * profile.c (output_location): Include all BBs, even if
17585 belonging to a same line (and file) as a previous BB.
17587 2017-04-28 Martin Liska <mliska@suse.cz>
17589 * gcov.c (process_args): Handle new argument 'w'.
17590 (read_graph_file): Assign ID to BBs.
17591 (output_branch_count): Display BB # if verbose flag is set.
17592 (output_lines): Likewise for arcs.
17593 (print_usage): Add '--verbose' option help.
17594 * doc/gcov.texi: Document --verbose (-w) option.
17596 2017-04-28 Martin Liska <mliska@suse.cz>
17598 * gcov.c (struct block_location_info): New struct.
17599 (process_file): Fill up the new structure.
17600 (read_graph_file): Replace usage of encoding by the newly added
17602 (add_line_counts): Likewise.
17603 (accumulate_line_counts): Remove usage of the union.
17604 (function_info::function_info): New function.
17605 (function_info::~function_info): Likewise.
17606 (process_file): Call delete instead of release_function.
17607 (release_function): Release the function.
17608 (release_structures): Call delete instead of release_function.
17609 (solve_flow_graph): Replace usage of num_blocks.
17610 (find_exception_blocks): Likewise.
17611 (output_lines): Fix GNU coding style.
17613 2017-04-28 Martin Liska <mliska@suse.cz>
17616 * coverage.c (coverage_remove_note_file): New function.
17617 * coverage.h: Declare the function.
17618 * toplev.c (finalize): Clean if an error has been seen.
17620 2017-04-28 Martin Liska <mliska@suse.cz>
17622 PR gcov-profile/80031
17623 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17624 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17625 * gcov.c (read_graph_file): Read just number of blocks.
17626 * profile.c (branch_prob): Do not stream 0 flags per a basic
17629 2017-04-28 Martin Liska <mliska@suse.cz>
17631 * gcov-dump.c (tag_*): Add new argument to declarations.
17632 (dump_gcov_file): Likewise.
17633 (tag_blocks): Add and use new argument depth.
17634 (tag_arcs): Likewise.
17635 (tag_lines): Likewise.
17636 (tag_counters): Likewise.
17637 (tag_summary): Likewise.
17638 (dump_working_sets): Use depth to do a proper indentation.
17640 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17643 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17644 * symtab.c (symtab_node::debug_symtab): Move definition here.
17646 2017-04-28 Richard Biener <rguenther@suse.de>
17648 * lto-streamer.h (LTO_major_version): Bump to 7.
17650 2017-04-28 Richard Biener <rguenther@suse.de>
17652 * tree-vrp.c (assert_info): New struct.
17653 (add_assert_info): New helper.
17654 (register_edge_assert_for_2): Refactor to add asserts to a vector
17656 (register_edge_assert_for_1): Likewise.
17657 (register_edge_assert_for): Likewise.
17658 (finish_register_edge_assert_for): New helper actually registering
17659 asserts where live on edge.
17660 (find_conditional_asserts): Adjust.
17661 (find_switch_asserts): Likewise.
17662 (evrp_dom_walker::try_find_new_range): Generalize.
17663 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17665 2017-04-27 Marek Polacek <polacek@redhat.com>
17668 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17669 arg10 and arg11 to itype.
17671 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17673 * doc/extend.texi (Object Size Checking): Improve grammar.
17675 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17678 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17679 that the logic for permitting reciprocal estimates matches that
17682 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17685 * tree.c (type_cache_hasher::equal): Only compare
17686 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17687 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17688 non-aggregate element types.
17689 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17690 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17692 2017-04-27 Richard Biener <rguenther@suse.de>
17694 PR middle-end/80533
17695 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17696 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17697 keeping a reference to a trailing array.
17699 2017-04-27 Richard Biener <rguenther@suse.de>
17701 PR middle-end/80539
17702 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17703 being in loop-closed SSA form conservatively.
17704 (chrec_fold_multiply_poly_poly): Likewise.
17706 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17708 PR middle-end/79665
17709 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17710 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17712 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17715 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17716 (aarch64_function_arg_alignment): Return unsigned int again, but still
17717 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17718 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17719 Don't emit -Wpsabi note.
17720 (aarch64_function_arg_boundary): Likewise.
17721 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17724 2017-04-26 Nathan Sidwell <nathan@acm.org>
17726 * tree.h (crc32_unsigned_n): Declare.
17727 (crc32_unsigned, crc32_unsigned): Make inline.
17728 * tree.c (crc32_unsigned_bits): Replace with ...
17729 (crc32_unsigned_n): ... this.
17730 (crc32_unsigned, crc32_byte): Remove.
17731 (crc32_string): Remove unnecessary braces.
17733 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17735 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17736 * ipa-inline-analysis.c (MAX_TIME): Remove.
17737 (account_size_time): Use sreal for time.
17738 (dump_inline_summary): Update.
17739 (estimate_function_body_sizes): Update.
17740 (estimate_edge_size_and_time): Update.
17741 (estimate_calls_size_and_time): Update.
17742 (estimate_node_size_and_time): Update.
17743 (inline_merge_summary): Update.
17744 (inline_update_overall_summary): Update.
17745 (estimate_time_after_inlining): Update.
17746 (inline_read_section): Update.
17747 (inline_write_summary): Update.
17748 * ipa-inline.c (compute_uninlined_call_time): Update.
17749 (compute_inlined_call_time): Update.
17750 (recursive_inlining): Update.
17751 (inline_small_functions): Update.
17752 (dump_overall_stats): Update.
17753 * ipa-inline.h: Include sreal.h.
17754 (size_time_entry): Turn time to sreal.
17755 (inline_summary): Turn self_time nad time to sreal.
17757 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17759 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17761 (sreal::stream_out, sreal::stream_in): New.
17762 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17764 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17766 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17769 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17772 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17774 (dimode_scalar_chain::compute_convert_gain): Ditto.
17775 (dimode_scalar_chain::make_vector_copies): Ditto.
17776 (dimode_scalar_chain::convert_reg): Ditto.
17777 (dimode_scalar_chain::convert_insn): Ditto.
17778 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17779 (VI248_AVX512BW_1): New mode iterator.
17780 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17781 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17784 2017-04-25 Martin Sebor <msebor@redhat.com>
17786 PR tree-optimization/80497
17787 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17788 constants are representable in HOST_WIDE_INT.
17789 (parse_directive): Ditto.
17791 2017-04-25 Martin Sebor <msebor@redhat.com>
17794 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17795 (new_zero_array): Adjust signature.
17796 (dom_info::dom_init): Used unsigned rather that size_t.
17797 (dom_info::dom_info): Same.
17799 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17800 Jakub Jelinek <jakub@redhat.com>
17803 * config/arm/arm.c: Include gimple.h.
17804 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17805 returns negative, increment ncrn only if it returned positive.
17806 (arm_needs_doubleword_align): Return int instead of bool,
17807 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17808 members, but if there is any such non-FIELD_DECL
17809 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17810 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17811 returns negative, increment nregs only if it returned positive.
17812 (arm_setup_incoming_varargs): Likewise.
17813 (arm_function_arg_boundary): Emit -Wpsabi note if
17814 arm_needs_doubleword_align returns negative, return
17815 DOUBLEWORD_ALIGNMENT only if it returned positive.
17817 2017-04-25 Marek Polacek <polacek@redhat.com>
17820 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17821 first argument to type.
17823 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17826 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17827 type checks to test for compatibility instead of equality.
17829 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17830 Jakub Jelinek <jakub@redhat.com>
17833 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17835 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17836 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17837 the alignment computation, but return their maximum in warn_alignment.
17838 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17839 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17841 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17842 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17845 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17847 * config/arc/simdext.md (dmpyh): Fix typo.
17849 2017-04-25 Richard Biener <rguenther@suse.de>
17851 PR tree-optimization/80492
17852 * alias.c (compare_base_decls): Handle registers with asm
17853 specification conservatively.
17854 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17855 compare_base_decls returning dont-know properly.
17857 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17859 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17860 (legitimate_offset_address_p): New function.
17861 (arc_legitimate_address_p): Use above function.
17863 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17865 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17867 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17869 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17870 ACCH registers whenever they are available.
17872 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17874 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17875 double regs fix when not used.
17877 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17879 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17881 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17882 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17884 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17886 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17887 when emitting short ADD instructions.
17889 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17891 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17893 (cmpsi_cc_c_insn): Likewise.
17894 (cbranchsi4_scratch): Compute proper instruction length using
17895 compact_hreg_operand.
17896 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17898 2017-04-25 Richard Biener <rguenther@suse.de>
17900 PR middle-end/80509
17901 * passes.c (pass_manager::pass_manager): Initialize
17902 m_name_to_pass_map.
17904 2017-04-25 Richard Biener <rguenther@suse.de>
17906 PR tree-optimization/79201
17907 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17909 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17912 * config/s390/vector.md: Split MEM->GPR vector moves for
17913 non-s_operand addresses.
17915 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17918 * config/s390/predicates.md (reload_const_wide_int_operand): New
17920 * config/s390/s390.md ("movti"): Remove d/P alternative.
17921 ("movti_bigconst"): New pattern definition.
17923 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17926 * s390-protos.h (s390_expand_cs_hqi): Removed.
17927 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17928 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17929 modes as well as CCZ1mode and CCZmode.
17930 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17931 signature of s390_emit_compare_and_swap.
17932 (s390_expand_cs_hqi): Likewise, make static.
17933 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17934 compare-and-swap, in some cases.
17935 (s390_expand_cs): Wrapper function.
17936 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17938 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17939 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17940 patterns for small and large integers. Forbid symref memory operands.
17941 Move expander to s390.c. Require cc register.
17942 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17943 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17944 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17945 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17946 symref memory operands. Remove CC mode and call s390_match_ccmode
17948 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17950 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17952 * config/s390/s390.md (define_peephole2): New peephole to help
17953 combining the load-and-test pattern with volatile memory.
17955 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17957 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17958 with CCZmode for TARGET_Z196.
17960 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17962 PR rtl-optimization/80501
17963 * combine.c (make_compound_operation_int): Set subreg_code to SET
17964 even for AND with mask of the sign bit of mode.
17966 PR rtl-optimization/80500
17967 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17968 sum's initial value.
17970 2017-04-25 Julian Brown <julian@codesourcery.com>
17971 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17973 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17975 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17977 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17979 2017-04-25 Julian Brown <julian@codesourcery.com>
17980 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17982 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17983 (thunderx2t99_sha): New Reservation.
17985 2017-04-25 Julian Brown <julian@codesourcery.com>
17986 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17988 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17989 type for 1-element load.
17991 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17993 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17995 2017-04-24 Martin Jambor <mjambor@suse.cz>
17997 PR tree-optimization/80293
17998 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17999 char arrays not totally scalarizable if it is false.
18000 (analyze_all_variable_accesses): Pass correct value in the new
18001 parameter. Add a statistics counter.
18003 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18005 PR middle-end/79931
18006 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18008 2017-04-24 Richard Biener <rguenther@suse.de>
18010 PR tree-optimization/80494
18011 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18012 out for complex types.
18014 2017-04-24 Richard Biener <rguenther@suse.de>
18016 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18017 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18018 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18019 (DFS): Adjust and never fail.
18020 (sccvn_dom_walker::fail): Remove.
18021 (sccvn_dom_walker::before_dom_children): Adjust.
18022 (run_scc_vn): Likewise and never fail.
18023 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18024 (pass_fre::execute): Likewise.
18026 2017-04-24 Richard Biener <rguenther@suse.de>
18028 PR tree-optimization/79725
18029 * tree-ssa-sink.c (statement_sink_location): Return whether
18030 failure reason was zero uses. Move that check later.
18031 (sink_code_in_bb): Deal with zero uses by removing the stmt
18034 2017-04-24 Richard Biener <rguenther@suse.de>
18037 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18038 pointer-based references.
18040 2017-04-24 Richard Biener <rguenther@suse.de>
18043 * pass_manager.h (pass_manager::operator new): Remove.
18044 (pass_manager::operator delete): Likewise.
18045 * passes.c (pass_manager::operator new): Remove.
18046 (pass_manager::operator delete): Likewise.
18047 (pass_manager::pass_manager): Zero individual pass members.
18049 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18052 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18053 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18054 Check "XEXP (src, 1)" operand here.
18055 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18056 Check "XEXP (src, 1)" operand here.
18057 (dimode_scalar_chain::make_vector_copies): Detect count register
18058 of a shift instruction. Zero extend count register from QImode
18059 to DImode to satisfy vector shift pattern count operand predicate.
18060 Substitute vector shift count operand with a DImode copy.
18061 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18064 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18066 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18067 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18068 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18069 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18070 (UNSPEC_NOREX_MEM): Remove definition.
18072 2017-04-21 Richard Biener <rguenther@suse.de>
18074 PR tree-optimization/79547
18075 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18076 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18077 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18078 without any constraints.
18080 2017-04-21 Richard Biener <rguenther@suse.de>
18082 PR tree-optimization/78847
18083 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18085 2017-04-21 Richard Biener <rguenther@suse.de>
18087 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18088 (build_distinct_type_copy): Likewise.
18089 (build_variant_type_copy): Likewise.
18090 * tree.c (build_qualified_type): Pass down mem-stat info.
18091 (build_distinct_type_copy): Likewise.
18092 (build_variant_type_copy): Likewise.
18094 2017-04-21 Richard Biener <rguenther@suse.de>
18096 PR tree-optimization/80237
18097 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18099 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18100 for a simplified result.
18102 2016-04-21 Richard Biener <rguenther@suse.de>
18104 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18105 sth as strict as a simple_iv but a chrec without symbols and an
18106 operand defined in the loop we are peeling (and not some subloop).
18107 (propagate_constants_for_unrolling): Propagate all constants.
18109 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18112 * config/i386/i386.c (print_reg): Remove assert for disalowed
18113 regno values, call output_operand_lossage instead.
18115 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18118 * config/i386/constraints.md (Yc): New register constraint.
18119 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18120 Use Yc constraint for alternative 2 of operand 0. Remove
18121 preferred_for_speed attribute.
18123 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18125 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18126 lastprivate clauses in SIMT case.
18128 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18130 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18132 2017-04-20 Richard Biener <rguenther@suse.de>
18134 PR tree-optimization/57796
18135 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18136 as N scalar stores.
18137 (vect_model_load_cost): Cost gathers as N scalar loads.
18139 2017-04-20 Richard Biener <rguenther@suse.de>
18141 * ggc-page.c (ggc_allocated_p): Rename to ...
18142 (safe_lookup_page_table_entry): ... this and return the lookup
18144 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18146 2017-04-20 Richard Biener <rguenther@suse.de>
18148 PR tree-optimization/80453
18149 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18150 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18151 from the conditions.
18152 (vn_phi_eq): Pass them down.
18153 (vn_phi_lookup): Record them.
18154 (vn_phi_insert): Likewise.
18156 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18158 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18159 uninitialized variable warning to avoid buffer overrun.
18161 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18164 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18165 is suppressed for '{ 0 }' in C.
18167 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18169 * BASE-VER: Set to 8.0.0.
18171 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18173 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18174 priority .init_array and .fini_array section with SECTION_NOTYPE
18177 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18179 PR middle-end/80423
18180 * tree.h (build_array_type): Add typeless_storage default argument.
18181 * tree.c (type_cache_hasher::equal): Also compare
18182 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18183 (build_array_type): Add typeless_storage argument, set
18184 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18186 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18187 (build_array_type): Likewise. Add typeless_storage argument.
18189 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18190 Jakub Jelinek <jakub@redhat.com>
18192 PR tree-optimization/80426
18193 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18194 operation on symbolic operands, also compute the overflow for the
18195 invariant part when the operation degenerates into a negation.
18197 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18200 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18201 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18204 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18206 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18209 * config/avr/avr.c (tree.h): Include it.
18210 (cgraph.h): Include it.
18211 (avr_encode_section_info): Don't warn for uninitialized progmem
18212 variable if it's just an alias.
18214 2017-04-19 Richard Biener <rguenther@suse.de>
18217 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18218 when needed by AutoPGO.
18220 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18223 * doc/lto.texi: Remove an extra 'that'.
18225 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18227 PR rtl-optimization/80429
18228 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18229 are only used in debug insns.
18231 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18232 Vladimir Makarov <vmakarov@redhat.com>
18234 * config/sparc/predicates.md (input_operand): Add comment. Return
18235 true for any memory operand when LRA is in progress.
18236 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18238 2017-04-18 Jeff Law <law@redhat.com>
18241 * mips.md ({return,simple_return}_internal): Do not overwrite
18244 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18246 PR tree-optimization/80443
18247 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18248 instead of adding 1, subtract -1 and similarly instead of subtracting
18251 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18253 PR rtl-optimization/80357
18254 * haifa-sched.c (tmp_bitmap): New variable.
18255 (model_recompute): Handle duplicate use records.
18256 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18257 (free_global_sched_pressure_data): Free it.
18259 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18262 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18263 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18264 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18265 instead of SYSTEM_HEADER_DIR.
18267 2017-04-18 Jeff Law <law@redhat.com>
18269 PR middle-end/80422
18270 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18271 predecessors after walking up the insn chain.
18273 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18276 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18277 sizetype type into debug info.
18279 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18282 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18283 unneeded test for TARGET_UPPER_REGS_SF.
18284 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18286 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18289 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18290 instead of gsi_after_labels.
18292 2017-04-18 Jeff Law <law@redhat.com>
18294 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18298 2017-04-13 Jeff Law <law@redhat.com>
18299 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18300 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18302 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18305 * config/avr/avr.c (intl.h): Include it.
18306 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18308 2017-04-18 Martin Liska <mliska@suse.cz>
18310 PR gcov-profile/78783
18311 * gcov-tool.c (gcov_output_files): Validate that destination
18312 file is either removed by the tool or by a user.
18314 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18315 Guy Benyei <guybe@mellanox.com>
18317 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18318 block, and do not negate it, the stored id is already negative.
18320 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18322 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18324 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18327 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18328 masks of options that should be turned off if the VSX vector
18329 options are turned off.
18330 (OTHER_P8_VECTOR_MASKS): Likewise.
18331 (OTHER_VSX_VECTOR_MASKS): Likewise.
18332 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18333 rs6000_disable_incompatible_switches to validate no type switches
18335 (rs6000_incompatible_switch): New function to disallow turning on
18336 other vector options if -mno-vsx, -mno-power8-vector, or
18337 -mno-power9-vector are specified.
18339 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18341 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18343 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18345 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18346 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18347 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18348 (ARG_POINTER_CFA_OFFSET): Likewise.
18350 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18352 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18353 conditions to take advantage of various optimizations.
18355 2017-04-13 Jeff Law <law@redhat.com>
18357 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18358 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18359 (zero_extendsidi2_dext): Likewise.
18361 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18364 * fold-const.c (fold_ternary_loc): Revert
18365 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18368 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18370 PR rtl-optimization/80343
18371 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18372 new scratch pseudo.
18374 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18377 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18378 to ubsan_encode_value.
18380 2017-04-13 Jeff Law <law@redhat.com>
18382 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18383 appearing in DEBUG_INSNs.
18385 2017-04-13 Martin Liska <mliska@suse.cz>
18387 PR gcov-profile/80413
18388 * gcov-io.c (gcov_write_string): Copy to buffer just when
18389 allocated size is greater than zero.
18391 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18394 * dwarf2out.c (decls_for_scope): Ignore declarations of
18395 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18397 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18400 * ipa-visibility.c (non_local_p): Fix typos.
18401 (localize_node): When localizing symbol in same comdat group,
18402 dissolve the group only when we know external symbols are going
18404 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18406 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18408 PR tree-optimization/79390
18409 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18410 order does not result in usable sequence, retry with reversed operand
18416 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18417 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18418 op0 instead of fold_convert_loc (loc, type, arg0).
18420 2017-04-12 Jeff Law <law@redhat.com>
18422 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18423 has a delay slot in the generated code.
18425 * config/cris/cris.md (cris_preferred_reload_class): Return
18426 GENNONACR_REGS rather than GENERAL_REGS.
18428 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18431 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18432 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18433 signedness of the result type.
18435 2017-04-12 Richard Biener <rguenther@suse.de>
18436 Jeff Law <law@redhat.com>
18438 PR tree-optimization/80359
18439 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18440 trim stores to TARGET_MEM_REFs.
18442 2017-04-12 Richard Biener <rguenther@suse.de>
18444 PR tree-optimization/79390
18445 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18446 threading case even more.
18448 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18451 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18452 for quad_address_p for TImode, instead of just not indexed_address.
18454 2017-04-12 Richard Biener <rguenther@suse.de>
18455 Bernd Edlinger <bernd.edlinger@hotmail.de>
18457 PR middle-end/79671
18458 * alias.c (component_uses_parent_alias_set_from): Handle
18459 TYPE_TYPELESS_STORAGE.
18460 (get_alias_set): Likewise.
18461 * tree-core.h (tree_type_common): Add typeless_storage flag.
18462 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18463 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18464 for types containing members with TYPE_TYPELESS_STORAGE.
18465 (place_field): Likewise.
18466 (layout_type): Likewise for ARRAY_TYPE.
18467 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18468 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18469 TYPE_TYPELESS_STORAGE.
18470 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18472 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18475 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18476 first argument to type.
18478 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18482 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18483 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18484 (rs6000_expand_binop_builtin): Likewise.
18485 (rs6000_expand_ternop_builtin): Likewise; also add missing
18486 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18487 vshasigma built-ins.
18488 * doc/extend.texi: Document that vec_xxpermdi's third argument
18489 must be a constant.
18491 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18493 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18494 Use shift_const cost parameter when calculating gain of STV shifts.
18496 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18498 PR rtl-optimization/70478
18499 * lra-constraints.c (process_alt_operands): Check memory for
18500 disfavoring memory insn operand.
18502 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18504 PR middle-end/80100
18505 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18506 left shift in unsigned HOST_WIDE_INT type.
18508 PR rtl-optimization/80385
18509 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18510 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18513 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18514 if they have any depend clauses.
18516 2017-04-11 Martin Liska <mliska@suse.cz>
18519 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18520 * ipa-split.c (split_function): Create a local comdat symbol
18521 if caller is in a comdat group.
18523 2017-04-11 Martin Liska <mliska@suse.cz>
18526 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18529 2017-04-11 Martin Sebor <msebor@redhat.com>
18531 PR middle-end/80364
18532 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18533 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18535 (directive::set_width, directive::set_precision, format_character):
18537 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18540 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18543 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18544 conflict, set target->arch_name instead of target->cpu_name.
18546 2017-04-11 Richard Biener <rguenther@suse.de>
18548 PR tree-optimization/80374
18549 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18550 build_zero_cst, remove fold_convertible_p check again.
18552 2017-04-11 Martin Liska <mliska@suse.cz>
18555 * ubsan.c (instrument_object_size): Do not instrument register
18558 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18561 * config/i386/i386-builtin-types.def
18562 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18563 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18564 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18565 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18566 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18567 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18568 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18569 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18570 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18571 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18572 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18573 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18574 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18575 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18576 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18577 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18578 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18579 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18580 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18581 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18582 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18583 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18584 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18585 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18586 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18587 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18588 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18589 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18590 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18591 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18592 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18593 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18594 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18595 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18596 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18597 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18598 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18599 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18600 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18601 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18602 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18603 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18604 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18605 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18606 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18607 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18609 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18610 flag to second_arg_count, handle 4 argument function type _COUNT
18611 aliases, handle second_arg_count on second argument rather than last.
18613 2017-04-10 Jeff Law <law@redhat.com>
18615 PR tree-optimization/80374
18616 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18617 record anything if we can not convert integer_zero_node to the
18620 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18623 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18624 Enhance special handling given to the TARGET_P9_MINMAX option in
18625 relation to certain other options.
18627 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18629 PR tree-optimization/80153
18630 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18631 remove POINTER_PLUS_EXPR's base part directly, rather than through
18634 2017-04-10 Richard Biener <rguenther@suse.de>
18635 Bin Cheng <bin.cheng@arm.com>
18637 PR tree-optimization/80153
18638 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18639 the first element of pointer type aff_tree. Build result expr in
18641 (add_elt_to_tree): Convert to type unconditionally. Remove other
18642 fold_convert calls.
18643 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18644 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18646 2017-04-10 Richard Biener <rguenther@suse.de>
18648 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18651 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18653 PR rtl-optimization/70478
18654 * lra-constraints.c (curr_small_class_check): New.
18655 (update_and_check_small_class_inputs): New.
18656 (process_alt_operands): Update curr_small_class_check. Disfavor
18657 alternative insn memory operands. Check available regs for small
18660 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18663 * config/mips/mips.opt (-mvirt): Update description.
18664 * doc/invoke.texi (-mvirt): Likewise.
18666 2017-04-10 Richard Biener <rguenther@suse.de>
18668 PR middle-end/80362
18669 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18670 looking for NEGATE_EXPR in -A / -B to A / B folding.
18672 2017-04-10 Martin Liska <mliska@suse.cz>
18674 PR gcov-profile/80224
18675 * gcov.c (print_usage): Fix usage string.
18676 (get_gcov_intermediate_filename): Remove.
18677 (output_gcov_file): Use both for normal and intermediate format.
18678 (generate_results): Do not initialize special file for
18679 intermediate format.
18681 2017-04-10 Richard Biener <rguenther@suse.de>
18683 PR tree-optimization/80304
18684 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18687 2017-04-10 Nathan Sidwell <nathan@acm.org>
18690 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18691 (rs6000_init_builtins): Use it.
18693 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18695 * config/arm/arm.md (<mrc>): Add mode to SET source.
18696 (<mrrc>): Likewise.
18698 2017-04-10 Richard Biener <rguenther@suse.de>
18700 PR middle-end/80344
18701 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18703 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18706 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18707 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18708 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18709 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18710 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18711 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18712 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18713 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18714 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18715 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18716 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18717 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18718 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18719 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18720 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18721 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18722 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18723 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18724 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18725 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18726 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18727 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18728 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18730 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18732 PR rtl-optimization/70478
18733 * lra-constraints.c: Reverse the last patch.
18735 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18737 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18738 Add comment for WCHAR_T.
18740 2017-04-08 Martin Liska <mliska@suse.cz>
18743 2017-04-07 Martin Liska <mliska@suse.cz>
18746 * ipa-split.c (split_function): Add function part to a same comdat
18749 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18752 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18754 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18756 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18757 (vec_pairing): ...to this.
18758 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18759 (rs6000_sched_init): Adjust for name change.
18760 (struct rs6000_sched_context): Likewise.
18761 (rs6000_init_sched_context): Likewise.
18762 (rs6000_set_sched_context): Likewise.
18764 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18770 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18772 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18773 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18774 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18776 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18778 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18780 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18782 PR rtl-optimization/70703
18783 * ira-color.c (update_conflict_hard_regno_costs): Use
18784 int64_t instead of HOST_WIDE_INT.
18786 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18788 PR rtl-optimization/70478
18789 * lra-constraints.c (process_alt_operands): Disfavor alternative
18790 insn memory operands.
18792 2017-04-07 Jeff Law <law@redhat.com>
18794 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18795 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18797 2017-04-07 Martin Liska <mliska@suse.cz>
18800 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18801 Show error message instead of an ICE.
18803 2017-04-07 Martin Liska <mliska@suse.cz>
18806 * ipa-split.c (split_function): Add function part to a same comdat
18809 2017-04-07 Richard Biener <rguenther@suse.de>
18811 PR middle-end/80341
18812 * tree.c (get_unwidened): Also handle ! for_type case for
18814 * convert.c (do_narrow): Split out from ...
18815 (convert_to_integer_1): ... here. Do not pass final truncation
18816 type to get_unwidened for TRUNC_DIV_EXPR.
18818 2017-04-07 Richard Biener <rguenther@suse.de>
18820 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18822 (aff_combination_const): Adjust.
18823 (aff_combination_scale): Likewise.
18824 (aff_combination_add_elt): Likewise.
18825 (aff_combination_add_cst): Likewise.
18826 (aff_combination_convert): Likewise.
18827 (add_elt_to_tree): Likewise. Remove unused argument.
18828 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18830 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18832 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18834 * config/arm/arm.c (arm_default_short_enums): Use
18835 ARM_DEFAULT_SHORT_ENUMS.
18836 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18838 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18841 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18842 members with redundant out-of-class redeclaration.
18844 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18847 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18848 * config/i386/i386.md (*zero_extendsidi2):
18849 Add (?*x,*x) and (?*v,*v) alternatives.
18851 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18854 * config/i386/i386.c (ix86_expand_builtin)
18855 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18856 mode from insn data. Convert operands to insn operand mode.
18857 Copy operands that don't satisfy insn predicate to a register.
18859 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18861 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18864 2017-04-06 Richard Biener <rguenther@suse.de>
18866 PR tree-optimization/80334
18867 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18868 preserve alignment of accesses.
18870 2017-04-06 Richard Biener <rguenther@suse.de>
18872 PR tree-optimization/80262
18873 * tree-sra.c (build_ref_for_offset): Preserve address-space
18875 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18876 Drop useless address-space information on MEM_REF offsets.
18878 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18880 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18882 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18884 PR rtl-optimization/70703
18885 * ira-color.c (update_conflict_hard_regno_costs): Use
18886 HOST_WIDE_INT instead of long.
18888 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18891 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18892 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18894 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18895 for x86_64 target. Handle -m3dnowa option.
18897 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18899 PR rtl-optimization/70703
18900 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18901 (update_conflict_hard_regno_costs): Use long instead of unsigned
18902 arithmetic for cost calculation.
18904 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18905 Bernd Edlinger <bernd.edlinger@hotmail.de>
18908 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18911 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18914 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18915 ptr_mode with Pmode throughout.
18916 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18917 into probe_stack_range and use DImode.
18919 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18922 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18923 call_eh_return is true.
18925 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18927 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18928 Initialize last_match_fntype_index.
18930 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18933 * tree-nvr.c: Include internal-fn.h.
18934 (pass_return_slot::execute): Ignore internal calls without
18937 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18938 Richard Biener <rguenther@suse.de>
18941 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18942 captures used multiple times, except for the last use.
18943 * generic-match-head.c: Include gimplify.h.
18945 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18947 PR tree-optimization/79390
18948 * target.h (struct noce_if_info): Declare.
18949 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18950 * target.def (noce_conversion_profitable_p): New target hook.
18951 * ifcvt.h (struct noce_if_info): New type, moved from ...
18952 * ifcvt.c (struct noce_if_info): ... here.
18953 (noce_conversion_profitable_p): Renamed to ...
18954 (default_noce_conversion_profitable_p): ... this. No longer
18956 (noce_try_store_flag_constants, noce_try_addcc,
18957 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18958 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18959 instead of noce_conversion_profitable_p.
18960 * config/i386/i386.c: Include ifcvt.h.
18961 (ix86_option_override_internal): Don't override
18962 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18963 (ix86_noce_conversion_profitable_p): New function.
18964 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18965 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18966 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18967 * doc/tm.texi: Regenerated.
18969 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18971 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18974 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18977 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18978 instructions for small multiply cores.
18980 2017-04-04 Jeff Law <law@redhat.com>
18982 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18984 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18985 that are not set by the loop over the elements.
18987 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18990 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18991 int mode, convert_modes it to mode as unsigned, otherwise use
18992 lowpart_subreg to mode rather than SImode.
18993 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18994 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18995 Use DImode instead of SImode for the shift count operand.
18996 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18999 2017-04-04 Richard Biener <rguenther@suse.de>
19001 PR middle-end/80281
19002 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19003 arithmetic done for the negate or the plus. Simplify.
19004 (A - (-B) -> A + B): Likewise.
19005 * fold-const.c (split_tree): Make sure to not negate pointers.
19007 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19009 PR rtl-optimization/60818
19010 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19011 a compare of comparisons with the thing compared if this results
19012 in a different machine mode.
19014 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19016 * alias.c (base_alias_check): Fix typo in comment.
19017 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19018 * cgraphunit.c (symbol_table::compile): Likewise.
19019 * collect2.c (maybe_run_lto_and_relink): Likewise.
19020 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19021 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19022 * config/avr/avr.c (avr_map_op_t): Likewise.
19023 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19024 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19025 * config/epiphany/epiphany.md (movcc): Likewise.
19026 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19027 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19029 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19030 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19031 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19032 * config/sh/sh.c (sh_rtx_costs): Likewise.
19033 * fold-const.c (fold_truth_andor): Likewise.
19034 * genautomata.c (collapse_flag): Likewise.
19035 * gengtype.h (struct type::u::s): Likewise.
19036 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19037 * input.c (FORMAT_AMOUNT): Likewise.
19038 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19039 (known_aggs_to_agg_replacement_list): Likewise.
19040 * ipa-inline-analysis.c: Likewise.
19041 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19042 * ipa-polymorphic-call.c
19043 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19044 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19045 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19047 * modulo-sched.c (apply_reg_moves): Likewise.
19048 * omp-expand.c (build_omp_regions_1): Likewise.
19049 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19050 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19051 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19052 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19053 * value-prof.c: Likewise.
19054 * var-tracking.c (val_reset): Likewise.
19056 2017-04-03 Richard Biener <rguenther@suse.de>
19058 PR tree-optimization/80275
19059 * fold-const.c (split_address_to_core_and_offset): Handle
19062 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19064 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19065 descriptors is at least equal to that of functions.
19067 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19069 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19071 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19074 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19075 (mov<IMOD4:mode>): New expander.
19076 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19078 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19080 PR rtl-optimization/79405
19081 * fwprop.c (propagations_left): New variable.
19082 (forward_propagate_into): Decrement it.
19083 (fwprop_init): Initialize it.
19084 (fw_prop): If the variable has reached zero, stop propagating.
19085 (fwprop_addr): Ditto.
19087 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19090 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19091 a FUNCTION_DECL, pass it as decl instead of origin to
19094 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19096 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19099 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19102 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19103 TARGET_VSX_SMALL_INTEGER.
19105 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19107 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19108 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19110 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19112 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19113 extraction from odd-numbered MSA register.
19115 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19117 PR middle-end/80173
19118 * expmed.c (store_bit_field_1): Don't attempt to create
19119 a word subreg out of hard registers wider than word if they
19120 have HARD_REGNO_NREGS of 1 for their mode.
19122 PR middle-end/80163
19123 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19124 conversions to integer types wider than word and pointer.
19127 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19128 (rtx_equal_for_cselib_p): Pass 0 to it.
19129 * cselib.c (cselib_hasher::equal): Likewise.
19130 (rtx_equal_for_cselib_1): Add depth argument. If depth
19131 is 128, don't look up VALUE locs and punt. Increment
19132 depth in recursive calls when walking VALUE locs.
19134 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19136 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19137 (make_gcov_file_name): Use the canonical path name for generating
19139 (read_line): Fix handling of files with ascii null bytes.
19141 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19143 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19144 to initialise a vector register instead
19145 of using a const_int.
19147 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19149 PR translation/80189
19150 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19151 diagnostic messages.
19153 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19156 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19157 (dfp_diex_<mode>): Update mode of operand 1.
19158 * doc/extend.texi (dxex, dxexq): Document change to return type.
19159 (diex, diexq): Document change to argument type.
19161 2017-03-30 Martin Jambor <mjambor@suse.cz>
19164 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19165 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19166 it reflects the signature changes performed at the callee side.
19167 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19168 to cgraph_build_function_type_skip_args.
19169 (build_function_decl_skip_args): Adjust call to the above function.
19171 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19174 * config/i386/sse.md
19175 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19176 register as dest whenever it is a MEM not rtx_equal_p to the
19177 corresponding dup operand, and when forcing into reg move the
19178 reg into the memory afterwards.
19179 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19180 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19181 for the force_reg mode.
19182 (avx512vl_vextractf128<mode>): Use register as dest either
19183 always when a MEM, or when it is a MEM not rtx_equal_p to the
19184 corresponding dup operand, or even not when it is a CONST_VECTOR
19185 depending on the mode and lo vs. hi.
19186 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19188 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19189 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19190 Likewise. Require that operands[2] is even.
19191 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19192 Remove extraneous parens. Require that operands[2] is a multiple
19194 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19195 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19196 disallow memory then.
19198 2017-03-30 Richard Biener <rguenther@suse.de>
19200 PR tree-optimization/77498
19201 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19202 to non-constants over backedges.
19204 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19206 PR rtl-optimization/80233
19207 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19208 as last_combined_insn. Do not test for BARRIER_P separately.
19210 2017-03-29 Andreas Schwab <schwab@suse.de>
19213 * calls.c (prepare_call_address): Convert funexp to Pmode before
19214 copying to temp reg.
19216 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19218 PR tree-optimization/80158
19219 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19220 Handle possible future case of more than one alternate
19222 (replace_rhs_if_not_dup): Likewise.
19223 (replace_one_candidate): Likewise.
19225 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19227 PR rtl-optimization/80193
19228 * ira.c (ira): Do not check allocation for LRA.
19230 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19232 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19233 (nvptx_output_simt_exit): Declare.
19234 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19235 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19236 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19237 (nvptx_declare_function_name): ...here. Emit declaration of local
19238 memory space buffer for omp_simt_enter insn.
19239 (nvptx_output_unisimt_switch): New.
19240 (nvptx_output_softstack_switch): New.
19241 (nvptx_output_simt_enter): New.
19242 (nvptx_output_simt_exit): New.
19243 * config/nvptx/nvptx.h (struct machine_function): New fields
19244 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19245 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19246 (UNSPECV_SIMT_EXIT): Ditto.
19247 (omp_simt_enter_insn): New insn.
19248 (omp_simt_enter): New expansion.
19249 (omp_simt_exit): New insn.
19250 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19252 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19253 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19254 (expand_GOMP_SIMT_EXIT): New.
19255 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19256 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19257 (GOMP_SIMT_EXIT): Ditto.
19258 * target-insns.def (omp_simt_enter): New insn.
19259 (omp_simt_exit): Ditto.
19260 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19262 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19263 (lower_rec_input_clauses): Likewise.
19264 (lower_lastprivate_clauses): Handle SIMT privatization.
19266 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19267 (ompdevlow_adjust_simt_enter): New.
19268 (find_simtpriv_var_op): New.
19269 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19270 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19272 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19273 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19274 (copy_decl_for_dup_finish): Ditto.
19276 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19278 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19281 * config/i386/i386.c (ix86_option_override_internal): Always
19282 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19284 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19286 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19288 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19290 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19291 mark new edge's irreducible flag accordign to it.
19292 (vect_do_peeling): Check loop preheader edge's irreducible flag
19293 and pass it to function slpeel_add_loop_guard.
19295 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19297 PR tree-optimization/80218
19298 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19299 Update block frequencies and counts.
19301 2017-03-28 Richard Biener <rguenther@suse.de>
19303 PR tree-optimization/78644
19304 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19305 of a simplification result we may not use it at all.
19307 2017-03-28 Richard Biener <rguenther@suse.de>
19310 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19311 without arguments, generate default definition of a SSA name.
19313 2017-03-28 Richard Biener <rguenther@suse.de>
19315 PR middle-end/80222
19316 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19317 TYPE_REF_CAN_ALIAS_ALL references.
19318 * fold-const.c (fold_indirect_ref_1): Likewise.
19320 2017-03-28 Martin Liska <mliska@suse.cz>
19323 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19324 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19326 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19327 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19329 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19330 (EXTRA_SPECS): Define.
19331 (SUBTARGET_EXTRA_SPECS): Likewise.
19332 (SUBTARGET_CPP_SPEC): Likewise.
19333 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19334 SUBTARGET_EXTRA_SPECS.
19335 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19337 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19339 * config/arc/simdext.md (vst64_insn): Update pattern.
19340 (vld32wh_insn): Likewise.
19341 (vld32wl_insn): Likewise.
19342 (vld64_insn): Likewise.
19343 (vld32_insn): Likewise.
19345 2017-03-28 Marek Polacek <polacek@redhat.com>
19348 * fold-const.c (fold_comparison): Use protected_set_expr_location
19349 instead of SET_EXPR_LOCATION.
19351 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19353 * tree.c (add_expr): Avoid name lookup warning.
19355 2017-03-27 Jeff Law <law@redhat.com>
19357 PR tree-optimization/80216
19358 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19359 function name. Limit recursion depth.
19360 (record_temporary_equivalences): Corresponding changes.
19362 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19364 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19367 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19370 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19372 * cfgcleanup.c (reg_note_cfa_p): New array.
19373 (insns_have_identical_cfa_notes): New function.
19374 (old_insns_match_p): Don't cross-jump in between /f
19375 and non-/f instructions. If both i1 and i2 are frame related,
19376 verify all CFA notes, their order and content.
19378 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19381 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19382 HImode and SImode with zero extend to DImode to one insn.
19383 (bswap<mode>2_extenddi): Likewise.
19384 (bswapsi2_extenddi): Likewise.
19385 (bswaphi2_extendsi): Likewise.
19386 (bswaphi2): Combine bswap HImode and SImode into one insn.
19387 Separate memory insns from swapping register.
19388 (bswapsi2): Likewise.
19389 (bswap<mode>2): Likewise.
19390 (bswaphi2_internal): Delete, no longer used.
19391 (bswapsi2_internal): Likewise.
19392 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19393 store, and gpr<-gpr swap insns.
19394 (bswap<mode>2_store): Likewise.
19395 (bswaphi2_reg): Register only splitter, combine with the splitter.
19396 (bswaphi2 splitter): Likewise.
19397 (bswapsi2_reg): Likewise.
19398 (bswapsi2 splitter): Likewise.
19399 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19400 the insns into load, store, and register/register insns.
19401 (bswapdi2_ldbrx): Likewise.
19402 (bswapdi2_load): Likewise.
19403 (bswapdi2_store): Likewise.
19404 (bswapdi2_reg): Likewise.
19406 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19408 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19409 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19411 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19414 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19416 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19417 special handling for target option conflicts between dform
19418 options (-mpower9-dform, -mpower9-dform-vector,
19419 -mpower9-dform-scalar) and -mno-direct-move.
19421 2017-03-27 Richard Biener <rguenther@suse.de>
19423 PR tree-optimization/80181
19424 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19426 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19428 * config/arc/predicates.md (move_double_src_operand): Replace the
19429 call to move_double_src_operand with a call to address_operand.
19431 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19433 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19434 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19435 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19437 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19439 * config/arc/predicates.md (long_immediate_loadstore_operand):
19440 Consider scaled addresses cases.
19442 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19444 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19445 restored when in interrupt.
19446 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19447 doesn't have delay slot.
19449 2017-03-27 Richard Biener <rguenther@suse.de>
19452 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19453 inlined thunk clones.
19455 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19458 * asan.c (instrument_derefs): Copy over last operand from
19459 original COMPONENT_REF to the new COMPONENT_REF with
19460 DECL_BIT_FIELD_REPRESENTATIVE.
19461 * ubsan.c (instrument_object_size): Likewise.
19463 2017-03-27 Richard Biener <rguenther@suse.de>
19465 PR tree-optimization/80170
19466 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19467 sure DR/SCEV didnt fold in constants we do not see when looking
19468 at the reference base alignment.
19470 2017-03-27 Richard Biener <rguenther@suse.de>
19472 PR middle-end/80171
19473 * gimple-fold.c (fold_ctor_reference): Properly guard against
19474 NULL return value from canonicalize_constructor_val.
19476 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19479 * config/i386/i386.c (ix86_expand_builtin)
19480 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19481 flags reg setting and flags reg using instructions.
19482 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19483 clobbering instructions to zero extend op2.
19485 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19487 * doc/install.texi (Configuration) <--with-aix-soname>:
19488 Update link to AIX ld.
19490 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19492 PR rtl-optimization/80160
19493 PR rtl-optimization/80159
19494 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19495 reg_alternate_class into account.
19497 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19500 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19501 to consider in curr_insn_transform.
19503 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19505 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19506 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19507 and emit_mode_inner.
19509 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19511 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19512 argument to the overloaded builtin variants. Use the new flag to
19513 deprecate certain builtin variants.
19514 * config/s390/s390-builtin-types.def: Add new builtin types.
19515 * config/s390/s390-builtins.h: Support new flags field for
19516 overloaded builtins.
19517 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19518 (s390_macro_to_expand): Enable vector float data type.
19519 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19520 builtins by incrementing the __VEC__ version number.
19521 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19523 (s390_resolve_overloaded_builtin): Emit error messages depending
19524 on the builtin flags.
19525 * config/s390/s390.c (s390_expand_builtin): Support additional
19526 flags argument. Change error message to match the messages
19527 emitted in s390-c.c.
19528 * config/s390/s390.md: New UNSPEC_* constants.
19529 (op_type): Add new instruction types.
19530 * config/s390/vecintrin.h: Add new builtins and test data class
19532 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19533 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19534 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19535 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19536 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19537 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19539 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19540 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19541 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19542 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19544 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19545 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19546 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19547 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19549 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19550 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19551 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19552 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19553 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19554 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19555 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19557 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19558 ("vec_scatter_element<V_HW_4:mode>_DI")
19559 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19560 ("vec_fpint<mode>", "vflls")
19561 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19562 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19563 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19564 ("*vec_cmphe<mode>_cc"): ... these.
19566 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19567 mode constant instead of magic value.
19569 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19571 * config/s390/s390.c (s390_expand_vec_compare): Support other
19572 vector floating point modes than just V2DF.
19573 (s390_expand_vcond): Likewise.
19574 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19575 (s390_cannot_change_mode_class): Prevent mode changes between TF
19576 and V1TF in vector registers.
19577 * config/s390/s390.md (DF, SF): New mode attributes.
19578 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19579 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19580 SFmode support for VRs.
19581 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19583 (VFT, VF_HW): New mode iterators.
19584 (vw, sdx): New mode attributes.
19585 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19586 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19587 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19588 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19589 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19590 also the new vector floating point modes. Renaming to ...
19592 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19593 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19594 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19595 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19596 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19597 ("vec_unordered<mode>"): ... these.
19599 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19600 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19601 ("*vec_extendv2df"): New insn definitions.
19603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19605 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19606 ("mulditi3_2", "*muldi3_sign"): New patterns.
19607 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19608 rename the pattern definition.
19610 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19612 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19614 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19616 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19618 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19619 instruction if possible.
19620 * config/s390/vector.md (vec_halfnumelts): New mode
19622 ("*vec_vllezlf<mode>"): New pattern.
19624 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19626 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19627 ("popcountv4si2", "popcountv2di2"): Rename to ...
19628 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19629 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19631 ("popcount<mode>2_vxe"): New pattern.
19633 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19635 * common/config/s390/s390-common.c (processor_flags_table): Add
19637 * config.gcc: Add arch12.
19638 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19639 Default to arch12 for unknown CPU model numbers.
19640 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19641 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19642 PROCESSOR_max sanity check.
19643 * config/s390/s390-opts.h (enum processor_type): Add
19645 * config/s390/s390.c (processor_table): Add arch12.
19646 (s390_expand_builtin): Add check for B_VXE flag.
19647 (s390_issue_rate): Add PROCESSOR_ARCH12.
19648 (s390_get_sched_attrmask): Likewise.
19649 (s390_get_unit_mask): Likewise.
19650 (s390_sched_score): Enable z13 scheduling for arch12.
19651 (s390_sched_reorder): Likewise.
19652 (s390_sched_variable_issue): Likewise.
19653 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19655 (s390_tune_attr): Use z13 scheduling also for arch12.
19656 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19657 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19658 (TARGET_VXE_P): New macros.
19659 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19660 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19661 * config/s390/s390.opt: Add arch12 as processor_type.
19663 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19665 * config/s390/s390.md
19666 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19667 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19668 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19670 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19671 Rename expanders to ...
19673 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19674 ("fixuns_truncdddi2_emu"): ... these.
19676 ("fixuns_trunc<mode>si2_emu"): New expander.
19678 ("*fixuns_truncdfdi2_z13"): Rename to ...
19679 ("*fixuns_truncdfdi2_vx"): ... this.
19681 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19683 * config/s390/2964.md: Remove the single element vector compare
19684 instructions which are no longer used.
19685 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19687 (s390_canonicalize_comparison): Remove handling of DFmode
19689 (s390_expand_vec_compare_scalar): Remove function.
19690 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19691 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19693 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19695 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19697 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19699 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19700 will anyway by matched by mov<mode>_64dfp.
19702 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19704 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19705 vlef/vstef. Add missing operand to vleif.
19707 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19709 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19710 pair for all vector types with 64 bit elements.
19711 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19712 * config/s390/vector.md (V_HW_64): ... here.
19713 (V_128_NOSINGLE): New mode iterator.
19714 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19715 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19716 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19717 ("*vec_load_pairv2di"): Change to ...
19718 ("*vec_load_pair<mode>"): ... this one.
19720 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19722 * config/s390/constraints.md: Add comments.
19723 (jKK): Reject element sizes > 8 bytes.
19724 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19726 * config/s390/s390.md: Add the s_operand checks formerly in
19727 s390_split_ok_p to various splitters where they are still
19729 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19730 for 128 bit vectors. Plus two splitters.
19732 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19734 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19737 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19740 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19741 error if the boundary argument is not constant.
19743 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19745 PR rtl-optimization/80112
19746 * loop-doloop.c (doloop_condition_get): Don't check condition
19747 if cmp isn't SET with IF_THEN_ELSE src.
19749 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19751 PR tree-optimization/80158
19752 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19753 replacing a candidate statement, also replace it for the
19754 candidate's alternate interpretation.
19755 (replace_rhs_if_not_dup): Likewise.
19756 (replace_one_candidate): Likewise.
19758 2017-03-24 Richard Biener <rguenther@suse.de>
19760 PR tree-optimization/80167
19761 * graphite-isl-ast-to-gimple.c
19762 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19764 (translate_isl_ast_to_gimple::get_rename): Likewise.
19766 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19768 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19769 handling of certain combinations of target options, including the
19770 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19771 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19773 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19776 * config/arm/arm.md (*load_multiple): Add reload_completed to
19777 matching condition.
19779 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19780 Richard Biener <rguenth@suse.de>
19782 PR tree-optimization/79908
19783 PR tree-optimization/80136
19784 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19785 been cast away, gimplify_and_add suffices.
19787 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19789 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19791 2017-03-23 Richard Biener <rguenther@suse.de>
19793 PR tree-optimization/80032
19794 * gimplify.c (gimple_push_cleanup): Forced unconditional
19795 cleanups still have to go to the conditional_cleanups
19798 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19800 PR tree-optimization/80072
19801 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19803 (next_operand_entry_id): Change type to unsigned int.
19804 (sort_by_operand_rank): Make sure to return the right return value
19805 even if unsigned fields are bigger than INT_MAX.
19806 (struct oecount): Change cnt and id type to unsigned int.
19807 (oecount_hasher::equal): Formatting fix.
19808 (oecount_cmp): Make sure to return the right return value
19809 even if unsigned fields are bigger than INT_MAX.
19810 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19813 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19814 TREE_READONLY on result if writing it more than once.
19817 * doc/invoke.texi (-fsanitize=thread): Document that with
19818 -fnon-call-exceptions atomics are not able to throw
19822 * tsan.c: Include tree-eh.h.
19823 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19824 maybe_clean_or_replace_eh_stmt where needed.
19825 (instrument_memory_accesses): Add cfg_changed argument.
19826 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19827 if it returned true.
19828 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19830 PR rtl-optimization/63191
19831 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19832 wrapper function, moved the whole old content into ...
19833 (ix86_delegitimize_address_1): ... this. New inline function.
19834 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19835 true as last argument instead of ix86_delegitimize_address.
19837 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19839 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19840 cortexa57_branch_cost.
19842 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19844 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19846 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19849 * doc/md.texi (Constraints): Document wA constraint.
19850 * config/rs6000/constraints.md (wA): New.
19851 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19852 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19853 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19854 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19856 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19859 * gimplify.c (is_oacc_declared): New function.
19860 (oacc_default_clause): Use it to set default flags for acc declared
19861 variables inside parallel regions.
19862 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19863 declared variables.
19864 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19865 declare attribute to any decl as necessary.
19867 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19870 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19871 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19872 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19873 (arm_arch_lpae): This.
19874 * config/arm/arm.c (arm_arch7ve): Rename into ...
19875 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19876 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19879 2017-03-22 Martin Liska <mliska@suse.cz>
19882 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19883 error message instead of an ICE.
19885 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19887 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19889 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19891 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19893 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19896 2017-03-21 Martin Sebor <msebor@redhat.com>
19898 * doc/extend.texi: Use "cannot" instead of "can't."
19899 * doc/hostconfig.texi: Same.
19900 * doc/install.texi: Same.
19901 * doc/invoke.texi: Same.
19902 * doc/loop.texi: Same.
19903 * doc/md.texi: Same.
19904 * doc/objc.texi: Same.
19905 * doc/rtl.texi: Same.
19906 * doc/tm.texi: Same.
19907 * doc/tm.texi.in: Same.
19908 * doc/trouble.texi: Same.
19910 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19913 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19914 (collect_checksum_attributes): Set it.
19915 (die_checksum_ordered): Use it.
19917 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19919 PR tree-optimization/79908
19920 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19921 change: For a VA_ARG whose LHS has been cast away, use
19922 force_gimple_operand to construct the side effects.
19924 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19926 PR translation/80001
19927 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19928 more amenable to translation.
19929 (oacc_loop_auto_partitions): Likewise.
19931 2017-03-21 Marek Polacek <polacek@redhat.com>
19932 Martin Sebor <msebor@redhat.com>
19934 PR tree-optimization/80109
19935 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19936 on INTEGRAL_TYPE_P.
19938 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19939 Segher Boessenkool <segher@kernel.crashing.org>
19942 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19943 check reg_used_between_p between insn and one of succ or succ2
19944 depending on if succ is artificial insn not inserted into insn
19947 2017-03-21 Martin Liska <mliska@suse.cz>
19949 PR gcov-profile/80081
19950 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19951 * doc/gcc.texi: Include gcov-dump stuff.
19952 * doc/gcov-dump.texi: New file.
19954 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19956 PR rtl-optimization/79150
19957 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19958 conditional jump, if the jump is the last insn of the loop.
19960 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19961 Richard Biener <rguenth@suse.de>
19963 PR tree-optimization/79908
19964 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19965 been cast away, use force_gimple_operand to construct the side
19968 2017-03-21 Martin Liska <mliska@suse.cz>
19970 PR libfortran/79956
19971 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19974 2017-03-21 Brad Spengler <spender@grsecurity.net>
19977 * plugin.c (htab_hash_plugin): New function.
19978 (add_new_plugin): Use it and adjust.
19979 (parse_plugin_arg_opt): Adjust.
19980 (init_one_plugin): Likewise.
19982 2017-03-21 Richard Biener <rguenther@suse.de>
19984 PR tree-optimization/80032
19985 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19986 if set force the cleanup to happen unconditionally.
19987 (gimplify_target_expr): Push inserted clobbers with force_uncond
19988 to avoid them being removed by control-dependent DCE.
19990 2017-03-21 Richard Biener <rguenther@suse.de>
19992 PR tree-optimization/80122
19993 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19994 va_arg_pack_len when the inlined call stmt requires pack
19996 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19998 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20001 * tsan.c (instrument_builtin_call): If the memory model argument
20002 is not a constant, assume it is valid.
20005 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20008 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20010 PR rtl-optimization/79910
20011 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20012 if its dest is used by an insn before I2 (other than the combined
20013 insns themselves, which are properly handled already).
20015 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20018 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20020 * combine.c (record_used_regs): New static function.
20021 (try_combine): Handle situations where there is an additional
20022 instruction between I2 and I3 which needs to have a LOG_LINK
20026 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20028 * combine.c (try_combine): Delete redundant i1 test. Call
20029 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20031 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20034 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20035 alternatives 13/14.
20037 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20039 PR tree-optimization/80054
20040 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20041 the optimization if a PHI or any of its arguments is not dominated
20042 by the candidate's basis. Use gphi* rather than gimple* as
20044 (replace_profitable_candidates): Clean up a gimple* variable that
20045 should be a gphi* variable.
20047 2017-03-20 Martin Sebor <msebor@redhat.com>
20050 * doc/extend.texi (attribute constructor): Document present limitation.
20052 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20055 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20056 __POWER9_VECTOR__ #ifdef control, change template definition to
20057 use Power9-specific built-in function.
20058 (vec_any_eq): Likewise.
20059 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20060 to control outcomes from this test.
20061 (vector_ae_<mode>p): For VEC_F modes, likewise.
20063 2017-03-20 Ian Lance Taylor <iant@google.com>
20065 * config/i386/i386.c (ix86_function_regparm): Save an extra
20066 register for -fsplit-stack with DECL_STATIC_CHAIN.
20068 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20071 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20072 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20074 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20076 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20078 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20081 2017-03-20 Marek Polacek <polacek@redhat.com>
20084 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20086 2017-03-20 Richard Biener <rguenther@suse.de>
20088 PR tree-optimization/80113
20089 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20090 allocate extra SSA name for PHI def.
20091 (add_close_phis_to_outer_loops): Likewise.
20092 (add_close_phis_to_merge_points): Likewise.
20093 (copy_loop_close_phi_args): Likewise.
20094 (copy_cond_phi_nodes): Likewise.
20096 2017-03-20 Martin Liska <mliska@suse.cz>
20098 PR middle-end/79753
20099 * tree-chkp.c (chkp_build_returned_bound): Do not build
20100 returned bounds for a LHS that's not a BOUNDED_P type.
20102 2017-03-20 Martin Liska <mliska@suse.cz>
20106 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20107 COMPLEX_CST and VECTOR_CST.
20109 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20112 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20113 target operand. A new splitter adds the clobber statement in case
20114 the target operand is dead anyway.
20116 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20118 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20119 to age-old versions of binutils and glibc.
20121 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20123 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20125 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20127 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20129 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20131 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20132 requirement for binutils 2.13.
20134 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20136 * combine.c (try_combine): Delete redundant i1 test. Call
20137 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20139 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20141 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20142 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20144 <riscv64-*-elf>: Re-arrange section
20145 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20146 <riscv32-*-linux>: Likewise.
20147 <riscv64-*-elf>: Likewise
20148 <riscv64-*-linux>: Likewise.
20150 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20153 * aarch64.opt(verbose-cost-dump): Fix typo.
20155 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20158 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20159 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20161 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20163 * reload.c (find_reloads): When reloading a nonoffsettable address,
20164 use RELOAD_OTHER for it and its address reloads.
20166 PR rtl-optimization/79910
20167 * combine.c (record_used_regs): New static function.
20168 (try_combine): Handle situations where there is an additional
20169 instruction between I2 and I3 which needs to have a LOG_LINK
20172 2017-03-17 Jeff Law <law@redhat.com>
20174 PR tree-optimization/71437
20175 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20176 conditional in the hash table first.
20177 (vrp_dom_walker::before_dom_children): Extract condition from
20178 ASSERT_EXPR. Record condition, its inverion and any implied
20179 conditions as well.
20181 2017-03-17 Marek Polacek <polacek@redhat.com>
20182 Markus Trippelsdorf <markus@trippelsdorf.de>
20184 PR tree-optimization/80079
20185 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20188 2017-03-17 Richard Biener <rguenther@suse.de>
20190 PR middle-end/80075
20191 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20192 Properly verify the LHS before the RHS possibly claims to be
20194 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20197 2017-03-17 Martin Jambor <mjambor@suse.cz>
20199 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20200 (List of -O2 options): Likewise.
20201 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20204 2017-03-17 Tom de Vries <tom@codesourcery.com>
20206 * gcov-dump.c (print_usage): Print bug_report_url.
20208 2017-03-17 Richard Biener <rguenther@suse.de>
20210 PR middle-end/80050
20211 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20212 (parser::peek): Likewise.
20214 2017-03-17 Richard Biener <rguenther@suse.de>
20216 PR tree-optimization/80048
20217 * sese.c (free_sese_info): Properly release rename_map and
20218 copied_bb_map elements.
20220 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20222 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20223 Add linked-list forward and backlinks. Insert on
20224 construction, remove on destruction.
20225 (class pass_store_merging): Add m_stores_head field.
20226 (pass_store_merging::terminate_and_process_all_chains):
20227 Iterate over m_stores_head list.
20228 (pass_store_merging::terminate_all_aliasing_chains):
20230 (pass_store_merging::execute): Check for debug stmts first.
20231 Push new chains onto the m_stores_head stack.
20233 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20236 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20237 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20238 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20240 2017-03-16 Jeff Law <law@redhat.com>
20242 PR tree-optimization/71437
20243 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20244 member function. Implementation moved into after_dom_children
20245 member function and into the threader's thread_outgoing_edges
20247 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20248 some code into new thread_outgoing_edges.
20249 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20250 definition. Simplify marker handling (do it here). Assume we always
20251 have the available expression and the const/copies tables.
20252 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20254 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20255 * tree-vrp.c (equiv_stack): No longer file scoped.
20256 (vrp_dom_walker): New class.
20257 (vrp_dom_walker::before_dom_children): New member function.
20258 (vrp_dom_walker::after_dom_children): Likewise.
20259 (identify_jump_threads): Setup domwalker. Use it rather than
20260 walking edges in a random order by hand. Simplify setup/finalization.
20261 (finalize_jump_threads): Remove.
20262 (vrp_finalize): Do not call identify_jump_threads here.
20263 (execute_vrp): Do it here instead and call thread_through_all_blocks
20266 PR tree-optimization/71437
20267 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20269 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20271 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20272 (dom_opt_dom_walker::thread_across_edge): Remove
20273 handle_dominating_asserts argument. All callers changed.
20274 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20275 changes. Remove calls to lhs_of_dominating_assert. Other
20276 uses of handle_dominating_asserts turn into unconditional code
20277 (simplify_control_stmt_condition_1): Likewise.
20278 (simplify_control_stmt_condition): Likewise.
20279 (thread_through_normal_block, thread_across_edge): Likewise.
20280 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20281 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20282 object if it is not an SSA_NAME.
20283 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20284 before calling into the VRP specific simplifiers.
20285 (identify_jump_threads): Remove handle_dominating_asserts
20288 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20291 * tree-diagnostic.c (default_tree_printer): No longer static.
20292 * tree-diagnostic.h (default_tree_printer): New prototype.
20294 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20296 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20297 Change ins into fmov.
20299 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20301 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20302 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20303 Use h_con constraint for operand 1.
20304 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20305 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20307 2017-03-15 Jeff Law <law@redhat.com>
20309 PR tree-optimization/71437
20310 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20311 (record_temporary_equivalences): Use it.
20313 PR tree-optimization/71437
20314 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20315 tree-ssa-scopedtables.
20316 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20317 (record_conditions, record_cond, vuse_eq): Likewise.
20318 (record_edge_info): Adjust to API tweak of record_conditions.
20319 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20320 (record_temporary_equivalences, optimize_stmt): Likewise.
20321 (eliminate_redundant_computations): Likewise.
20322 (record_equivalences_from_stmt): Likewise.
20323 * tree-ssa-scopedtables.c: Include options.h and params.h.
20324 (vuse_eq): New function, moved from tree-ssa-dom.c
20325 (build_and_record_new_cond): Likewise.
20326 (record_conditions): Likewise. Accept vector of conditions rather
20327 than edge_equivalence structure for first argument.
20328 for the first argument.
20329 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20330 from tree-ssa-dom.c.
20331 (avail_exprs_stack::record_cond): Likewise.
20332 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20333 from tree-ssa-dom.c.
20334 (avail_exprs_stack): Add new member functions lookup_avail_expr
20336 (record_conditions): Declare.
20338 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20341 * lra-constraints.c (process_alt_operands): Increase reject for
20342 reloading an input/output operand.
20344 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20347 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20348 insns to convert from signed/unsigned char/short to IEEE 128-bit
20350 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20352 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20355 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20356 subreg of inner mode for values already in registers.
20358 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20360 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20361 iteration reg is used after the loop.
20363 2017-03-14 Martin Sebor <msebor@redhat.com>
20365 PR tree-optimization/79800
20366 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20367 precision in negative-positive range.
20368 (format_floating): Call non-const overload with adjusted precision.
20370 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20373 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20376 2017-03-14 Martin Sebor <msebor@redhat.com>
20378 PR middle-end/80020
20379 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20380 * builtins.def (aligned_alloc): Use it.
20383 * Makefile.in (GTFILES): Add calls.c.
20384 * calls.c: Include "gt-calls.h".
20386 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20388 PR rtl-optimization/79728
20389 * regs.h (struct target_regs): New field
20390 x_contains_allocatable_regs_of_mode.
20391 (contains_allocatable_regs_of_mode): New macro.
20392 * reginfo.c (init_reg_sets_1): Initialize it, and change
20393 contains_reg_of_mode so it includes global regs as well.
20394 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20395 rather than contains_regs_of_mode.
20397 2017-03-14 Martin Liska <mliska@suse.cz>
20399 * doc/invoke.texi: Document options that can't be combined with
20400 -fcheck-pointer-bounds.
20402 2017-03-14 Martin Liska <mliska@suse.cz>
20404 PR middle-end/79831
20405 * doc/invoke.texi (-Wchkp): Document the option.
20407 2017-03-14 Martin Liska <mliska@suse.cz>
20409 * Makefile.in: Install gcov-dump.
20411 2017-03-14 Martin Liska <mliska@suse.cz>
20413 * multiple_target.c (expand_target_clones): Bail out for
20414 an invalid attribute.
20416 2017-03-14 Richard Biener <rguenther@suse.de>
20418 * alias.c (struct alias_set_entry): Pack properly.
20419 * cfgloop.h (struct loop): Likewise.
20420 * cse.c (struct set): Likewise.
20421 * ipa-utils.c (struct searchc_env): Likewise.
20422 * loop-invariant.c (struct invariant): Likewise.
20423 * lra-remat.c (struct cand): Likewise.
20424 * recog.c (struct change_t): Likewise.
20425 * rtl.h (struct address_info): Likewise.
20426 * symbol-summary.h (function_summary): Likewise.
20427 * tree-loop-distribution.c (struct partition): Likewise.
20428 * tree-object-size.c (struct object_size_info): Likewise.
20429 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20430 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20431 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20432 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20433 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20434 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20435 (struct _stmt_vec_info): Likewise.
20437 2017-03-14 Martin Liska <mliska@suse.cz>
20440 * multiple_target.c (create_dispatcher_calls): Check that
20441 a target can create a function dispatcher.
20443 2017-03-14 Martin Liska <mliska@suse.cz>
20446 * multiple_target.c (expand_target_clones): Drop local.local
20447 flag for default implementation.
20449 2017-03-14 Richard Biener <rguenther@suse.de>
20451 PR tree-optimization/80030
20452 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20454 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20456 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20457 gcc_fallthrough() instead of __attribute__((fallthrough));
20459 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20461 * doc/gcc.texi: Remove "up" link to (DIR).
20462 * doc/gccint.texi: Ditto.
20464 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20466 * doc/install.texi (Specific) <avr>: Remove reference to
20469 2017-03-13 Jeff Law <law@redhat.com>
20471 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20472 attribute rather than comments.
20474 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20475 match_scratch operand is highest.
20477 2017-03-13 Martin Liska <mliska@suse.cz>
20479 PR middle-end/78339
20480 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20481 is a CHKP clone, use original declaration.
20483 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20485 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20486 (arc_conditional_register_usage): Use a different allocation order
20487 when optimizing for size.
20488 * common/config/arc/arc-common.c (arc_option_optimization_table):
20489 Section anchors default on when optimizing for size.
20491 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20493 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20495 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20497 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20498 * config/arc/arc.md (cpu_facility): Add cd variant.
20499 (*movqi_insn): Add code density variant.
20500 (*movhi_insn): Likewise.
20501 (*movqi_insn): Likewise.
20502 (*addsi3_mixed): Likewise.
20503 (subsi3_insn): Likewise.
20505 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20507 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20509 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20511 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20512 expressions with MINUS and UNARY ops.
20514 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20517 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20519 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20520 between vec_select and vector argument.
20521 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20522 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20523 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20524 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20525 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20526 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20528 2017-03-13 Richard Biener <rguenther@suse.de>
20531 * params.def (vect-max-peeling-for-alignment): Fix typo.
20533 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20535 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20536 issue that only occurred with binutils below 2.18.
20538 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20540 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20541 refer to binutils 2.11/2.12 minimum.
20543 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20545 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20546 ftp.kernel.org and simplify binutils requirement.
20548 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20550 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20552 (Optimize Options): Ditto. Also remove redundancy.
20554 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20556 PR translation/79848
20557 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20559 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20560 to G_ to avoid double translation.
20562 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20564 PR translation/79923
20565 * auto-profile.c (get_combined_location): Convert leading
20566 character of diagnostics to lower case and remove trailing period.
20567 (read_profile): Likewise for various diagnostics.
20568 * config/arm/arm.c (arm_option_override): Remove trailing period
20569 from various diagnostics.
20570 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20571 (msp430_expand_delay_cycles): Likewise.
20573 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20576 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20577 full command-line argument, rather than just "str".
20578 (aarch64_validate_march): Likewise.
20579 (aarch64_validate_mtune): Likewise.
20581 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20583 PR rtl-optimization/78911
20584 * lra-assigns.c (must_not_spill_p): New function.
20585 (spill_for): Use it.
20587 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20589 PR tree-optimization/79981
20590 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20591 ATOMIC_COMPARE_EXCHANGE ifn result.
20592 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20593 IFN_ATOMIC_COMPARE_EXCHANGE.
20595 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20598 * opts.c (parse_sanitizer_options): Add missing question mark to
20599 "did you mean" message.
20601 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20603 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20605 (VMULEUH_UNS): Likewise.
20606 (VMULOUB_UNS): Likewise.
20607 (VMULOUH_UNS): Likewise.
20608 * config/rs6000/rs6000.c (builtin_function_type): Remove
20609 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20611 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20614 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20615 x with result of extra_parsing_for_operand_code_0.
20616 (function_reader::extra_parsing_for_operand_code_0): Convert
20617 return type from void to rtx, returning x. When reading
20618 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20619 larger size containing struct block_symbol.
20621 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20623 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20624 -mfloat128-hardware without -m64.
20626 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20629 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20630 entries to the case statement that marks unsigned arguments to
20631 overloaded functions.
20633 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20635 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20636 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20638 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20641 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20642 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20644 2017-03-10 Martin Liska <mliska@suse.cz>
20648 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20649 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20652 2017-03-10 Olivier Hainque <hainque@adacore.com>
20654 * tree-switch-conversion (array_value_type): Start by resetting
20655 candidate type to it's main variant.
20657 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20659 PR rtl-optimization/79909
20660 * combine.c (try_combine): Use simplify_replace_rtx on individual
20661 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20662 of the whole CALL_INSN_FUNCTION_USAGE.
20664 PR tree-optimization/79972
20665 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20666 get_range_info on SSA_NAMEs. Formatting fixes.
20668 2017-03-10 Richard Biener <rguenther@suse.de>
20669 Jakub Jelinek <jakub@redhat.com>
20671 PR tree-optimization/77975
20672 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20673 edge to be constant.
20674 (get_val_for): For constant x return it. Formatting fix.
20675 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20676 would use the same bases as the current one.
20678 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20680 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20681 instead of vec_select for V1TImode.
20682 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20684 (VSX_LE_128): Add V1TI to this mode iterator.
20685 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20686 (*vsx_le_perm_store_<mode>): Likewise.
20687 (pre-reload splitter for VSX stores): Likewise.
20688 (post-reload splitter for VSX stores): Likewise.
20689 (*vsx_xxpermdi2_le_<mode>): Likewise.
20690 (*vsx_lxvd2x2_le_<mode>): Likewise.
20691 (*vsx_stxvd2x2_le_<mode>): Likewise.
20693 2017-03-09 Michael Eager <eager@eagercon.com>
20695 Correct failures with --enable-checking=yes,rtl.
20697 * config/microblaze/microblaze.c (microblaze_expand_shift):
20698 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20699 test for const0_rtx.
20700 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20701 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20703 2017-03-09 Richard Biener <rguenther@suse.de>
20705 PR tree-optimization/79977
20706 * graphite-scop-detection.c (scop_detection::merge_sese):
20707 Handle the case of extra exits to blocks dominating the entry.
20709 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20711 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20714 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20716 PR rtl-optimization/79949
20717 * lra-constraints.c (process_alt_operands): Check memory when
20718 trying to predict a cycle. Print about the overall increase.
20720 2017-03-09 Richard Biener <rguenther@suse.de>
20722 PR middle-end/79971
20723 * gimple-expr.c (useless_type_conversion_p): Preserve
20724 TYPE_SATURATING for fixed-point types.
20726 2017-03-09 Richard Biener <rguenther@suse.de>
20729 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20730 alignment of BLKmode params.
20732 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20735 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20736 (VALL_NO_V2Q): Likewise.
20738 * config/aarch64/aarch64-simd.md
20739 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20741 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20742 VALL_NO_V2Q mode iterator.
20743 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20745 2017-03-09 Martin Liska <mliska@suse.cz>
20747 PR tree-optimization/79631
20748 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20749 tree_int_cst_sign_bit just for INTEGER constants.
20751 2017-03-09 Martin Liska <mliska@suse.cz>
20755 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20758 2017-03-09 Marek Polacek <polacek@redhat.com>
20761 * tree.c (inchash::add_expr): Handle TREE_VEC.
20763 2017-03-09 Martin Liska <mliska@suse.cz>
20766 (chkp_narrow_size_and_offset): New function.
20767 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20768 (void chkp_parse_bit_field_ref): New function.
20769 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20770 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20772 2017-03-09 Martin Liska <mliska@suse.cz>
20775 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20776 (chkp_find_bounds_1): Remove gcc_unreachable.
20778 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20781 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20782 BUILT_IN_SYNC*, determine the access type from the size suffix and
20783 always build a MEM_REF with that type. Handle forgotten
20784 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20787 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20788 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20789 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20790 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20791 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20792 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20793 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20794 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20795 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20796 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20797 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20798 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20799 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20800 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20801 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20802 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20803 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20804 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20805 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20806 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20807 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20808 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20809 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20810 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20811 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20812 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20813 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20814 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20815 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20816 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20817 definitions outside of __OPTIMIZE__ guarded section.
20820 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20821 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20822 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20823 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20826 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20828 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20829 ("vfenez<mode>"): Add missing constraints.
20831 2017-03-08 Martin Sebor <msebor@redhat.com>
20834 * config/nds32/nds32.c (nds32_option_override):
20835 Fix misspelled diagnostic.
20837 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20840 * gimplify.c (gimplify_omp_for): Replace index var in outer
20841 taskloop statement with an artificial variable and add
20842 OMP_CLAUSE_PRIVATE clause for it.
20844 2017-03-08 Richard Biener <rguenther@suse.de>
20846 PR tree-optimization/79955
20847 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20848 for accesses that are completely outside of the variable.
20850 2017-03-08 Andrew Haley <aph@redhat.com>
20852 PR tree-optimization/79943
20853 * tree-ssa-loop-split.c (compute_new_first_bound): When
20854 calculating the new upper bound, (END-BEG) should be added, not
20857 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20859 * config/avr/avr.md (setmemhi): Make sure match_dup
20860 operand number comes before match_scratch.
20862 2017-03-08 Richard Biener <rguenther@suse.de>
20864 PR tree-optimization/79920
20865 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20866 with ncopies == 1 to ...
20867 (vect_transform_slp_perm_load): ... here. Properly compute
20868 all element loads by iterating VF times over the group. Do
20869 not handle ncopies (computed in a broken way) in
20870 vect_create_mask_and_perm.
20872 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20875 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20876 is a uniform vector, use uniform_vector_p return value instead of
20877 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20879 2017-03-07 Marek Polacek <polacek@redhat.com>
20881 PR middle-end/79809
20882 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20883 (alloca_call_type): Likewise.
20885 2017-03-07 Martin Liska <mliska@suse.cz>
20887 * gcov.c (process_args): Put comment to correct location.
20889 2017-03-07 Martin Liska <mliska@suse.cz>
20891 PR middle-end/68270
20892 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20893 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20894 (chkp_narrow_bounds_for_field): Likewise.
20895 (chkp_parse_array_and_component_ref): Pass one more argument to
20898 2017-03-07 Richard Biener <rguenther@suse.de>
20900 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20903 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20905 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20906 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20908 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20911 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20912 to end of description.
20913 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20915 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20917 PR rtl-optimization/79901
20918 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20920 (*avx512f_<code><mode>3<mask_name>): ... this.
20921 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20922 iterator instead of VI8_AVX2_AVX512BW.
20924 PR rtl-optimization/79901
20925 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20926 min/max expander, expand it using expand_vec_cond_expr.
20929 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20932 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20935 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20936 to void * for PCH reasons.
20937 * dwarf2out.c (output_loc_operands, output_die): Cast
20938 v.val_vec.array to unsigned char *.
20940 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20943 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20946 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20948 PR rtl-optimization/79571
20949 * lra-constraints.c (process_alt_operands): Calculate static
20950 reject and subtract it from overall when only addresses will be
20953 2017-03-06 Julia Koval <julia.koval@intel.com>
20956 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20957 incoming stack boundary to 128 for 64-bit targets.
20959 2017-03-06 Richard Biener <rguenther@suse.de>
20961 PR tree-optimization/79894
20962 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20963 to NULL after folding it.
20965 2017-03-06 Richard Biener <rguenther@suse.de>
20967 PR tree-optimization/79824
20968 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20969 check disabling peeling for gaps.
20971 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20973 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20974 attributes): Document gettimeofday.
20976 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20978 * config/s390/s390.c (s390_option_override_internal): Set
20979 PARAM_MIN_VECT_LOOP_BOUND
20981 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20983 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20984 * config/s390/s390.md: Likewise.
20986 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20989 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20990 (<avx2_avx512>_perm<mode>): Rename to ...
20991 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20993 (<avx512>_perm<mode>_mask): Rename to ...
20994 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20996 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20997 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20998 instead of VI8F_256_512.
20999 (avx512f_perm<mode>): New define_expand.
21000 (avx512f_perm<mode>_mask): Likewise.
21001 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21002 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21004 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21006 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21007 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21009 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21011 2017-03-06 Martin Liska <mliska@suse.cz>
21014 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21015 when having a SSA NAME w/o VAR_DECL assigned to it.
21017 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21019 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21020 msa_dpsub_<su>_d): Fix MODE for vec_select.
21022 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21024 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21026 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21028 2017-03-06 Richard Biener <rguenther@suse.de>
21030 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21031 * plugin.c (register_plugin_info): Likewise.
21032 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21034 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21036 * config/i386/sse.md (sse_storehps, sse_storelps,
21037 avx_<castmode><avxsizesuffix>_<castmode>,
21038 avx512f_<castmode><avxsizesuffix>_<castmode>,
21039 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21040 in condition that at least one operand is not a MEM.
21042 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21044 PR middle-end/79805
21045 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21046 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21048 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21049 gimple_call_nothrow_p flag based on whether original builtin can throw.
21050 If it can, emit following stmts on the fallthrough edge.
21051 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21052 don't create new bb if inserting just debug stmts on the edge, try to
21053 insert them on the fallthru bb or just reset debug stmts.
21055 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21058 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21059 restore recog_data (including the operand rtxes inside it) around
21060 the call to get_insn_template.
21062 2017-03-03 Martin Sebor <msebor@redhat.com>
21064 PR tree-optimization/79699
21065 * context.c (context::~context): Free MPFR caches to avoid
21066 a memory leak on program exit.
21068 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21070 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21071 Use wide_int::ulow () instead of .elt (0).
21073 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21075 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21076 (*pushxf): Limit oF constraint to 32bit targets and add oC
21077 constraint for 64bit targets.
21078 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21079 (*pushdf): Change rmF constraint to rmC.
21081 2017-03-03 Martin Liska <mliska@suse.cz>
21083 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21084 Remove unused variable.
21086 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21089 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21090 is a memory operand, increase num_memory.
21091 (ix86_expand_args_builtin): Likewise.
21093 2017-03-03 Jan Hubicka <jh@suse.cz>
21096 * ipa-devirt.c (maybe_record_node): Properly handle
21097 __cxa_pure_virtual visibility.
21099 2017-03-03 Martin Liska <mliska@suse.cz>
21101 PR tree-optimization/79803
21102 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21104 (pass_loop_prefetch::execute): Disabled optimization if an
21105 assumption about L1 cache size is not met.
21107 2017-03-03 Martin Liska <mliska@suse.cz>
21109 PR rtl-optimization/79574
21110 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21111 (hash_scan_set): Likewise.
21112 (dump_hash_table): Likewise.
21113 (hoist_code): Likewise.
21115 2017-03-03 Richard Biener <rguenther@suse.de>
21117 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21118 in place of uhigh ().
21119 (fixed_convert_from_real): Likewise.
21121 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21124 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21126 2017-03-03 Richard Biener <rguenther@suse.de>
21128 PR middle-end/79818
21129 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21130 TYPE_OVERFLOW_UNDEFINED check.
21132 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21134 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21136 (vector_ae_<mode>_p): Likewise.
21137 (vector_nez_<mode>_p): Likewise.
21138 (vector_ne_v2di_p): Likewise.
21139 (vector_ae_v2di_p): Likewise.
21140 (vector_ne_<mode>_p): Likewise.
21141 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21143 (vsx_tsqrt<mode>2_fe): Likewise.
21145 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21148 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21150 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21152 PR rtl-optimization/79780
21153 * cprop.c (one_cprop_pass): When second and further conditional trap
21154 in a single basic block is turned into an unconditional trap, turn it
21155 into a deleted note to avoid RTL verification failures.
21157 2017-03-02 Richard Biener <rguenther@suse.de>
21159 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21161 2017-03-02 Richard Biener <rguenther@suse.de>
21163 PR tree-optimization/79345
21165 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21166 param and abort the walk, returning -1 if it is hit.
21167 (walk_aliased_vdefs): Take a limit param and pass it on.
21168 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21169 defaulting to 0 and return a signed int.
21170 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21171 (check_defs): New helper.
21172 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21173 about uninitialized memory.
21174 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21175 bogus uninitialized warning.
21176 (fixed_convert_from_real): Likewise.
21178 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21180 PR tree-optimization/66768
21181 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21182 iv_use if base object can't be determined.
21184 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21186 PR tree-optimization/79345
21187 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21188 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21189 (get_pattern_stats): Initialize it.
21190 * genemit.c (gen_expand): Verify match_scratch numbers come after
21191 match_operand/match_dup numbers.
21192 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21193 match_scratch numbers.
21194 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21196 * config/s390/s390.md (trunctdsd2): Likewise.
21198 2017-03-02 Richard Biener <rguenther@suse.de>
21200 * wide-int.h (wide_int_storage::operator=): Implement in terms
21203 2017-03-02 Richard Biener <rguenther@suse.de>
21205 PR tree-optimization/79777
21206 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21207 the to insert expression to sth existing.
21209 2017-03-01 Martin Sebor <msebor@redhat.com>
21211 PR middle-end/79692
21212 * gimple-ssa-sprintf.c
21213 (directive::known_width_and_precision): New function.
21214 (format_integer): Use it.
21215 (get_mpfr_format_length): Consider the full range of precision
21216 when computing %g output with the # flag. Set the likely byte
21217 count to 3 rather than 1 when precision is indeterminate.
21218 (format_floating): Correct the lower bound of precision.
21220 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21222 * doc/invoke.texi: Document default code model for 64-bit Linux.
21224 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21227 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21228 udiv rather than div since input pattern is unsigned.
21230 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21232 * config/i386/i386.c (print_reg): Warn for values of
21233 unsupported size in integer register.
21235 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21238 * config/rs6000/predicates.md (current_file_function_operand): Do
21239 not allow self calls to be local if the function is replaceable.
21241 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21244 * config/rs6000/altivec.h (vec_ctz and others): Change the
21245 preprocessor macro that controls conditional compilation from
21246 _ARCH_PWR9 to __POWER9_VECTOR__.
21247 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21248 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21249 control (instead of _ARCH_PWR9 control) so that template
21250 definition uses power9-specific function.
21251 (vec_any_eq): Likewise.
21252 (vec_all_ne): Change macro definition to use a power9-specific
21253 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21254 _ARCH_PWR9 control).
21255 (vec_any_eq) Likewise.
21256 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21257 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21258 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21259 support for xvcmpnedp instruction.
21260 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21261 macro expansion so that Power9 implementation of vec_all_ne does
21262 not use the AltiVec predicate framework.
21263 (VCMPNEH_P): Likewise.
21264 (VCMPNEW_P): Likewise.
21265 (VCMPNED_P): Likewise.
21266 (VCMPNEFP_P): Likewise.
21267 (VCMPNEDP_P): Likewise.
21268 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21269 implementation of vec_any_eq to not use AltiVec predicate
21271 (VCMPAEH_P): Likewise.
21272 (VCMPAEW_P): Likewise.
21273 (VCMPAED_P): Likewise.
21274 (VCMPAEFP_P): Likewise.
21275 (VCMPAEDP_P): Likewise.
21276 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21277 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21278 not use the AltiVec predicate framework.
21279 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21280 of vec_any_eq to not use AltiVec predicate framework.
21281 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21282 support for predefined __POWER9_VECTOR__ macro to indicate that
21283 Power9 instruction selection is enabled.
21284 (altivec_overloaded_builtins): Remove extraneous
21285 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21286 function argument types RS6000_BTI_bool_V16QI and
21287 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21288 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21289 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21290 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21291 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21292 Power9 for implementations of vec_cmpne. Change the signature for
21293 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21294 (representing vec_all_ne) to remove the previously described first
21295 argument of type RS6000_BTI_INTSI, as this was an artifact of
21296 reliance on the AltiVec predicate framework, which is no longer
21297 used in the implementation of these functions. Add
21298 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21299 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21300 since, unlike the AltiVec predicate framework implementation, we
21301 do not share function descriptors between vec_alle and vec_anyeq.
21302 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21303 set of modes that receive special treatment even when
21304 TARGET_P9_VECTOR is true. The special treatment emits code that
21305 does not depend on Power9 instructions.
21306 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21307 define_expand to not rely on AltiVec predicate framework.
21308 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21310 (vector_ne_v2di_p): Change this define_expand to not rely on
21311 AltiVec predicate framework.
21312 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21314 (vector_ne_<mode>_p): Change this define_expand to not rely on
21315 AltiVec predicate framework.
21316 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21318 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21319 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21320 define_insn pattern.
21321 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21322 define_insn pattern because the xvcmpne<VSs>. instruction is not
21324 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21325 instruction is not supported.
21327 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21329 * config/nvptx/nvptx.c: Include intl.h.
21331 2017-03-01 Martin Jambor <mjambor@suse.cz>
21334 * ipa-prop.h (ipa_bits): Removed field known.
21335 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21336 to pointers. Adjusted their comments to warn about their sharing.
21337 (ipcp_transformation_summary): Change bits to a vector of pointers.
21338 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21339 (ipa_get_ipa_bits_for_value): Declare.
21340 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21341 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21342 (ipa_bits_hash_table): Likewise.
21343 (ipa_vr_ggc_hash_traits): Likewise.
21344 (ipa_vr_hash_table): Likewise.
21345 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21346 being pointers and vr_known being removed.
21347 (ipa_set_jf_unknown): Likewise.
21348 (ipa_get_ipa_bits_for_value): New function.
21349 (ipa_set_jfunc_bits): Likewise.
21350 (ipa_get_value_range): New overloaded functions.
21351 (ipa_set_jfunc_vr): Likewise.
21352 (ipa_compute_jump_functions_for_edge): Use the above functions to
21353 construct bits and vr parts of jump functions.
21354 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21355 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21357 (ipcp_grow_transformations_if_necessary): Also allocate
21358 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21360 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21361 them. Fix too long lines.
21362 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21363 vr_known being removed.
21364 (ipa_read_jump_function): Use new setter functions to construct bits
21365 and vr parts of jump functions or set them to NULL.
21366 (write_ipcp_transformation_info): Adjust for bits being pointers.
21367 (read_ipcp_transformation_info): Likewise.
21368 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21370 Include gt-ipa-prop.h.
21371 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21373 (ipcp_store_bits_results): Likewise.
21374 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21375 Do not write to existing jump functions but use a temporary instead.
21377 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21380 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21381 attempt to use its first operand as BIT_FIELD_REF base.
21383 2017-03-01 Richard Biener <rguenther@suse.de>
21385 PR middle-end/79721
21386 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21387 interpolating formula in wrapping arithmetic.
21388 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21390 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21392 PR tree-optimization/79734
21393 * tree-vect-generic.c (expand_vector_condition): Optimize
21394 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21395 Handle VEC_COND_EXPR where comparison has different inner width from
21396 type's inner width.
21398 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21400 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21401 markup, and similar issues. Remove @opindex entries for things
21402 that aren't options. Add missing -mmpy-option entries.
21404 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21406 PR tree-optimization/79737
21407 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21408 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21409 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21410 instead of byte_size. Formatting fix.
21411 (shift_bytes_in_array_right): Formatting fix.
21413 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21416 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21417 condition on optimize for the leaf function test.
21419 2017-02-28 Martin Liska <mliska@suse.cz>
21422 * read-rtl-function.c (function_reader::handle_unknown_directive):
21423 Bail out when one uses -flto.
21425 2017-02-28 Martin Liska <mliska@suse.cz>
21427 * common.opt: Replace space with tabular for options of <number>
21429 * config/i386/i386.opt: Show <number> value for
21430 -mlarge-data-threshold.
21431 * opts.c (print_filtered_help): Do not display number in hexadecimal
21434 2017-02-28 Martin Liska <mliska@suse.cz>
21436 * common.opt: Fix --help=option -Q for options which are of
21439 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21441 * config/i386/i386.c (print_reg): Error out for values
21442 of 8-bit size in invalid integer register.
21444 2017-02-28 Martin Sebor <msebor@redhat.com>
21446 PR tree-optimization/79691
21447 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21449 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21452 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21453 gcc_unreachable with output_operand_lossage.
21455 2017-02-28 Richard Biener <rguenther@suse.de>
21457 PR tree-optimization/79740
21458 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21460 (visit_nary_op): Insert the nary into the hashtable if we
21461 pattern-matched sth.
21462 * tree-ssa-pre.c (eliminate_insert): Robustify.
21464 2017-02-28 Richard Biener <rguenther@suse.de>
21466 PR middle-end/79731
21467 * fold-const.c (decode_field_reference): Reject out-of-bound
21470 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21472 * config/i386/i386.c: Include intl.h.
21473 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21474 instead of just cond ? "..." : "...".
21475 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21476 * coverage.c (read_counts_file): Likewise.
21477 * omp-offload.c: Include intl.h.
21478 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21479 of just cond ? "..." : "...".
21480 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21481 of just cond ? "..." : "...".
21483 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21486 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21488 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21489 'tune for' CPU name.
21490 * config/arm/arm-cpu-data.h: Regenerated.
21492 2017-02-28 Richard Biener <rguenther@suse.de>
21494 PR tree-optimization/79732
21495 * tree-inline.c (expand_call_inline): Do not shadow var.
21497 2017-02-28 Richard Biener <rguenther@suse.de>
21499 PR tree-optimization/79723
21500 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21501 address-space properly.
21503 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21505 * doc/optinfo.texi (Optimization groups): Fix option used for
21507 * doc/invoke.texi (-fopt-info): Document "omp".
21508 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21509 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21510 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21511 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21512 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21514 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21516 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21517 OPTGROUP_OMP with "omp".
21519 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21522 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21523 for arithmetic shift of unsigned V2DI.
21525 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21527 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21528 arc/linux.h headers.
21529 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21530 (LINK_SPEC): Likewise.
21531 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21532 (EXTRA_SPECS): Likewise.
21533 (STARTFILE_SPEC): Likewise.
21534 (ENDFILE_SPEC): Likewise.
21535 (LIB_SPEC): Likewise.
21536 (TARGET_SDATA_DEFAULT): Likewise.
21537 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21538 (MULTILIB_DEFAULTS): Likewise.
21539 (DWARF2_UNWIND_INFO): Likewise.
21540 * config/arc/big.h: New file.
21541 * config/arc/elf.h: Likewise.
21542 * config/arc/linux.h: Likewise.
21543 * config/arc/t-uClibc: Remove.
21545 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21547 PR tree-optimization/77536
21548 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21549 (tree_transform_and_unroll_loop): Use above function to compute the
21550 estimated niter of unrolled loop and use it when scaling profile.
21551 Also use count info rather than frequency if it's non-zero.
21552 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21553 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21554 (vect_transform_loop): Call above function.
21556 2017-02-27 Richard Biener <rguenther@suse.de>
21558 PR tree-optimization/45397
21559 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21560 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21561 (visit_nary_op): Add pattern matching for CSEing sign-changed
21562 or truncated operations with wider ones.
21564 2017-02-27 Richard Biener <rguenther@suse.de>
21566 PR tree-optimization/79690
21567 * tree-vect-stmts.c (vectorizable_store): Use vector type
21568 built from the DR with address-space.
21570 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21572 * doc/invoke.texi (Optimize Options): Refine the description
21573 of asan-use-after-return.
21575 2017-02-25 Alan Modra <amodra@gmail.com>
21577 PR rtl-optimization/79584
21578 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21579 base, not ad->base_term, the reg within base. Remove assertion
21580 that ad->base == ad->base_term. Replace gen_int_mode using
21581 bogus mode with const0_rtx.
21583 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21585 PR middle-end/79396
21586 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21587 FMA_EXPR like tcc_binary or tcc_unary.
21589 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21592 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21594 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21595 (output_loc_operands): Handle DW_OP_call_ref and
21596 DW_OP_GNU_variable_value.
21597 (struct variable_value_struct): New type.
21598 (struct variable_value_hasher): Likewise.
21599 (variable_value_hash): New variable.
21600 (string_types): Remove.
21601 (copy_loc_descr): New function.
21602 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21603 (prepend_loc_descr_to_each): New function.
21604 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21605 instead of add_loc_descr_to_each if the first argument is single
21606 location list and the second has multiple.
21607 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21608 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21609 when looking for variable value which doesn't have other location info.
21610 (loc_list_from_tree): Formatting fix.
21611 (gen_array_type_die): Simplify DW_AT_string_length handling.
21612 (adjust_string_types): Remove.
21613 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21614 string_types. Call resolve_variable_values.
21615 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21616 (resolve_addr_in_expr): Likewise. Add A argument.
21617 (copy_deref_exprloc): Remove deref argument. Adjust for the
21618 original expression being DW_OP_GNU_variable_value with optionally
21619 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21620 optionally after it.
21621 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21622 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21623 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21624 (variable_value_hasher::hash, variable_value_hasher::equal): New
21626 (resolve_variable_value_in_expr, resolve_variable_value,
21627 resolve_variable_values, note_variable_value_in_expr,
21628 note_variable_value): New functions.
21629 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21632 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21635 * opts.h (handle_generated_option): Add GENERATED_P argument.
21636 * opts-common.c (handle_option): Adjust function comment.
21637 (handle_generated_option): Add GENERATED_P argument, pass it to
21639 (control_warning_option): Pass false to handle_generated_option
21641 * opts.c (maybe_default_option): Pass true to handle_generated_option
21643 * optc-gen.awk: Likewise.
21645 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21647 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21648 a REG, look at the REG it is a SUBREG of.
21649 (splitter for cmpeqsi_t): Ditto.
21651 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21653 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21654 the special USEs with the pattern of the insn, not the insn itself.
21656 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21659 * doc/invoke.texi: Document -mload-store-pairs.
21661 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21662 Sandra Loosemore <sandra@codesourcery.com>
21664 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21665 argument isn't a CONST_INT.
21666 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21667 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21668 (nios2_validate_compare): Bypass alternate compare logic if *op2
21669 is not a CONST_INT.
21670 (ldstwm_operation_p): Return false if first_base is not a REG or
21671 if first_offset is not a CONST_INT.
21673 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21675 * config/cris/cris.md: Use correct operand in a define_peephole2.
21677 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21679 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21681 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21683 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21684 this_insn if it is an INSN or JUMP_INSN.
21685 (force_offsettable): Look at base, not at addr.
21686 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21687 on things that aren't necessarily CONST_INTs.
21689 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21691 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21692 -mfpmath=sse is the default also for x86-32 targets with SSE2
21693 instruction set when @option{-ffast-math} is enabled
21695 2017-02-24 Jeff Law <law@redhat.com>
21697 PR rtl-optimizatoin/79286
21698 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21701 2017-02-24 Richard Biener <rguenther@suse.de>
21703 PR tree-optimization/79389
21704 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21707 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21709 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21710 function comment to reflect reality.
21711 (loop_exits_before_overflow): Fix typo in function description.
21713 2017-02-24 Richard Biener <rguenther@suse.de>
21715 PR tree-optimization/79389
21716 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21717 properly that a threading opportunity exists. Detect conditional
21718 copy/constant propagation opportunities.
21720 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21722 * config/visium/visium.md (type): Add trap.
21723 (b): New mode attribute.
21724 (*btst): Rename into...
21725 (*btst<mode>): ...this and adjust.
21726 (*cbranchsi4_btst_insn): Rename into...
21727 (*cbranch<mode>4_btst_insn): ...this and adjust.
21728 (trap): New define_insn.
21730 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21732 PR tree-optimization/79389
21733 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21734 (noce_reversed_cond_code): New function.
21735 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21736 reversed_comparison_code. Formatting fix.
21737 (noce_try_store_flag): Test rev_cond != NULL in addition to
21738 reversed_comparison_code.
21739 (noce_try_store_flag_constants): Likewise.
21740 (noce_try_store_flag_mask): Likewise.
21741 (noce_try_addcc): Use rev_cond if non-NULL instead of
21742 reversed_comparison_code.
21743 (noce_try_cmove_arith): Likewise. Formatting fixes.
21744 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21745 (noce_find_if_block): Initialize rev_cond.
21746 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21747 instead of false as last argument never attempt to reverse it
21750 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21752 PR tree-optimization/79663
21753 * tree-predcom.c (combine_chains): Process refs in reverse order
21754 only for ZERO length chains, and add explaining comment.
21756 2017-02-23 Jeff Law <law@redhat.com>
21758 PR tree-optimization/79578
21759 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21760 in call to operand_equal_p.
21762 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21765 * config/i386/cpuid.h: Fix another undefined behavior.
21767 2017-02-23 Richard Biener <rguenther@suse.de>
21769 PR tree-optimization/79683
21770 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21771 vector types for data-refs.
21773 2017-02-23 Martin Liska <mliska@suse.cz>
21775 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21777 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21779 PR middle-end/79665
21780 * internal-fn.c (get_range_pos_neg): Moved to ...
21781 * tree.c (get_range_pos_neg): ... here. No longer static.
21782 * tree.h (get_range_pos_neg): New prototype.
21783 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21784 are known to be in between 0 and signed maximum inclusive, try to
21785 expand both unsigned and signed divmod and use the cheaper one from
21788 2017-02-22 Jeff Law <law@redhat.com>
21790 PR tree-optimization/79578
21791 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21792 to compare base operands.
21794 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21797 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21798 gpc_reg_operand instead of fpr_reg_operand.
21800 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21802 * config/mips/mips.c (mips_return_in_memory): Force FP
21803 vector types to be returned in memory for o32 ABI.
21805 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21807 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21808 instead of DW_TAG_member for static data member declarations and don't
21809 set no_linkage_name for static inline data members.
21810 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21813 2017-02-22 Martin Liska <mliska@suse.cz>
21815 * doc/invoke.texi: Replace inequality signs with square brackets
21818 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21820 PR tree-optimization/68644
21821 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21823 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21826 * lra-constraints.c (simplify_operand_subreg): Handle
21827 WORD_REGISTER_OPERATIONS targets.
21829 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21832 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21833 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21834 elimination by swapping fld*.
21836 2017-02-22 Richard Biener <rguenther@suse.de>
21838 PR tree-optimization/79673
21839 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21840 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21841 irrelevant address-space qualifiers and avoiding a
21842 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21844 2017-02-22 Richard Biener <rguenther@suse.de>
21846 PR tree-optimization/79666
21847 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21848 to not symbolically negate if that may introduce undefined
21851 2017-02-22 Martin Liska <mliska@suse.cz>
21854 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21855 * data-streamer-out.c (streamer_write_gcov_count_stream):
21857 * value-prof.c (stream_out_histogram_value): Make assert more
21858 precise based on type of counter.
21860 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21863 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21864 Use nonimmediate_operand instead of memory_operand for operand 1.
21865 (float-extend standard_x87sse_constant_load splitter): Ditto.
21867 2017-02-21 Jeff Law <law@redhat.com>
21869 PR tree-optimization/79621
21870 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21871 blocks with edges to themselves.
21873 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21876 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21877 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21878 Use gimple_call_builtin_p.
21881 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21882 on temporarily removed DEBUG_INSNs.
21884 PR tree-optimization/79649
21885 * tree-loop-distribution.c (classify_partition): Give up on
21886 non-generic address space loads/stores.
21888 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21890 * doc/loop.texi (Loop manipulation): Remove nonexistent
21891 tree_ssa_loop_version from the documentation.
21892 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21894 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21897 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21898 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21899 * config/rs6000/rs6000.c: Include except.h.
21900 (rs6000_expand_split_stack_prologue): Call
21901 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21903 2017-02-21 Martin Jambor <mjambor@suse.cz>
21906 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21907 have been analyzed.
21909 2017-02-21 Martin Jambor <mjambor@suse.cz>
21911 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21912 for backward compatibility only.
21913 * doc/invoke.texi (Option Summary): Remove all references to
21914 -fipa-cp-alignment.
21916 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21920 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21922 * lra-constraints.c (curr_insn_transform): Handle
21923 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21925 2017-02-21 Martin Liska <mliska@suse.cz>
21927 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21929 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21931 PR translation/79638
21932 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21934 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21937 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21938 (arm_function_ok_for_sibcall): Return false for an indirect call by
21939 descriptor if all the argument registers are used.
21940 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21941 alignment of the function.
21943 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21945 PR tree-optimization/61441
21946 * simplify-rtx.c (simplify_const_unary_operation): For
21947 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21948 the sNaN unmodified.
21950 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21952 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21953 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21954 instead of SYSTEM_HEADER_DIR.
21956 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21957 Martin Liška <mliska@suse.cz>
21959 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21960 Fix typos and grammar, use active voice, and clarify.
21962 2017-02-20 Marek Polacek <polacek@redhat.com>
21964 PR middle-end/79537
21965 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21968 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21970 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21973 * config/i386/i386.c (ix86_expand_builtin): Handle
21974 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21975 ix86_builtins_isa[fcode].isa as a requirement of those
21976 flags and any other flag in the bitmask.
21977 (ix86_init_mmx_sse_builtins): Use 0 instead of
21978 ~OPTION_MASK_ISA_64BIT as mask.
21979 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21980 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21981 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21982 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21984 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21987 * lra-constraints.c (split_reg): Check requested split mode
21988 is supported by the register.
21990 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21992 * lra-constraints.c (simplify_operand_subreg): Remove early
21995 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21998 * lra-constraints.c (curr_insn_transform): Tighten condition
21999 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22001 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22004 * lra-constraints.c (curr_insn_transform): Handle
22005 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22007 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22010 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22012 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22014 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22017 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22020 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22022 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22025 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22027 * final.c (last_columnnum, override_columnnum): New variables.
22028 (final_start_function): Set last_columnnum, pass it to begin_prologue
22029 hook and pass 0 to dwarf2out_begin_prologue.
22030 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22031 to source_line debug hook.
22032 (notice_source_line): Compute last_columnnum and for debug_column_info
22033 return true on column changes.
22034 * debug.h (struct gcc_debug_hooks): Add column argument to
22035 source_line and begin_prologue hooks.
22036 (debug_nothing_int_charstar_int_bool): Remove prototype.
22037 (debug_nothing_int_int_charstar,
22038 debug_nothing_int_int_charstar_int_bool): New prototypes.
22039 (dwarf2out_begin_prologue): Add column argument.
22040 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22041 begin_prologue hooks.
22042 (debug_nothing_int_charstar_int_bool): Remove.
22043 (debug_nothing_int_int_charstar,
22044 debug_nothing_int_int_charstar_int_bool): New functions.
22045 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22046 through to dwarf2out_source_line.
22047 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22048 (dwarf2out_source_line): Add column argument, emit it if requested.
22049 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22051 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22052 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22053 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22054 through to dwarf2out_begin_prologue.
22055 (vmsdbgout_source_line): Add column argument, pass it through to
22056 dwarf2out_source_line.
22057 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22058 dbxout_source_line caller.
22059 (dbxout_source_line): Add column argument.
22061 * common.opt (gno-column-info, gcolumn-info): New options.
22062 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22063 (check_die): Also test for multiple DW_AT_decl_column attributes.
22064 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22065 DW_AT_decl_column if requested.
22066 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22068 (gen_variable_die): Likewise.
22069 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22070 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22073 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22074 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22075 (ix86_handle_option): Handle OPT_m3dnowa.
22076 * doc/invoke.texi (-m3dnowa): Document.
22077 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22078 -m3dnowa instead of -m3dnow -march=athlon.
22081 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22082 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22084 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22087 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22088 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22089 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22090 generator for vsx_xxpermdi_<mode>_be.
22091 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22092 force big-endian semantics.
22093 (vsx_xxpermdi_<mode>_be): New define_expand with same
22094 implementation as previous version of vsx_xxpermdi_<mode>.
22096 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22098 PR tree-optimization/79327
22099 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22100 variable, its initialization and use.
22102 2017-02-17 Julia Koval <julia.koval@intel.com>
22104 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22105 (OPTION_MASK_ISA_PKU_UNSET): New.
22106 (ix86_handle_option): Handle -mrdpid.
22107 * config/i386/cpuid.h (bit_RDPID): New.
22108 * config/i386/driver-i386.c (host_detect_local_cpu):
22109 Detect RDPID feature.
22110 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22111 * config/i386/i386-c.c (ix86_target_macros_internal):
22113 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22114 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22115 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22116 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22117 * config/i386/i386.md (define_insn "rdpid"): New.
22118 * config/i386/i386.opt Add -mrdpid.
22119 * config/i386/immintrin.h (_rdpid_u32): New.
22121 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22123 PR rtl-optimization/79541
22124 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22125 instead of transforming it into USE.
22127 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22129 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22130 If HONOR_SNANS (SFmode) force the input to a register.
22131 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22132 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22133 an frsp or similar insn.
22135 2017-02-17 Martin Liska <mliska@suse.cz>
22137 PR rtl-optimization/79577
22138 * params.def (selsched-max-sched-times): Increase minimum to 1.
22140 2017-02-17 Martin Liska <mliska@suse.cz>
22142 PR rtl-optimization/79574
22143 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22145 2017-02-17 Martin Liska <mliska@suse.cz>
22147 PR tree-optimization/79529
22148 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22149 ssa_defined_default_def_p to handle cases which are implicitly
22151 * tree-ssa.c (ssa_defined_default_def_p): New function.
22152 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22153 which are implicitly defined.
22154 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22156 2017-02-17 Richard Biener <rguenther@suse.de>
22158 PR middle-end/79576
22159 * params.def (max-ssa-name-query-depth): Limit to 10.
22161 2017-02-17 Richard Biener <rguenther@suse.de>
22163 PR tree-optimization/79552
22164 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22167 2017-02-17 Richard Biener <rguenther@suse.de>
22170 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22172 2017-02-17 Marek Polacek <polacek@redhat.com>
22174 PR middle-end/79536
22175 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22176 (fold_negate_expr): New wrapper.
22178 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22180 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22181 Correct terminology and de-emphasize pre-standard behavior.
22183 2017-02-16 Alan Modra <amodra@gmail.com>
22185 PR rtl-optimization/79286
22186 * ira.c (def_dominates_uses): New function.
22187 (update_equiv_regs): Don't create an equivalence for insns that
22188 may trap where the register def does not dominate the use.
22190 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22192 PR rtl-optimization/78127
22193 * lra.c (lra): Call lra_eliminate before finish the loop after
22196 2017-02-16 Richard Biener <rguenther@suse.de>
22198 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22200 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22201 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22202 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22203 (isl_val_int_from_wi): New function.
22204 (extract_affine_gmp): Rename to ...
22205 (extract_affine_wi): ... this, take a widest_int.
22206 (extract_affine_int): Just wrap extract_affine_wi.
22207 (add_param_constraints): Use isl_val_int_from_wi.
22208 (add_loop_constraints): Likewise, and extract_affine_wi.
22210 2017-02-15 Jeff Law <law@redhat.com>
22212 PR middle-end/79521
22213 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22214 ira_init_register_move_cost_if_necessary.
22216 2017-02-15 Martin Sebor <msebor@redhat.com>
22218 PR middle-end/32003
22219 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22220 removed in a prior commit.
22222 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22224 PR tree-optimization/79347
22225 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22226 counters during peeling.
22228 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22230 * Makefile.in (site.exp): Remove "set ISLVER".
22232 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22235 * real.c (real_from_integer): Call real_convert even for decimal.
22237 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22240 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22242 2017-02-14 Andrew Pinski <apinski@cavium.com>
22244 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22245 cores and change the partno/implementer to be correct.
22246 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22247 the 'B" as the implementer.
22248 * config/aarch64/aarch64-tune.md: Regenerate.
22250 2017-02-14 Carl Love <cel@us.ibm.com>
22252 * config/rs6000/rs6000.c: Add case statement entry to make the
22253 xvcvuxdsp built-in argument unsigned.
22254 * config/rs6000/vsx.md: Fix the source and return operand types so they
22255 match the instruction definitions from the ISA document. Fix typo
22256 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22259 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22262 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22263 member early_clobber_alts.
22264 * lra-lives.c (reg_early_clobber_p): New.
22265 (process_bb_lives): Use it.
22266 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22267 (debug_operand_data): Initialize early_clobber_alts.
22268 (setup_operand_alternative): Set up early_clobber_alts.
22269 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22270 alternatives to new_insn_reg.
22271 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22273 (lra_update_insn_regno_info): Pass the new arg.
22275 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22277 PR middle-end/79505
22278 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22279 (new_oacc_loop_raw): Don't clear already cleared fields.
22282 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22283 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22284 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22286 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22289 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22291 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22294 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22295 the extra instruction to the right place to store 128-bit constant
22298 2017-02-14 Martin Sebor <msebor@redhat.com>
22300 PR middle-end/79448
22301 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22302 warning for strings of unknown length.
22304 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22306 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22308 2017-02-14 Jeff Law <law@redhat.com>
22311 * ira-costs.c (scan_one_insn): Initialize register move costs
22312 for pseudos seen in USE/CLOBBER insns.
22314 PR tree-optimization/79095
22315 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22316 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22317 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22318 if the operands are known to be not equal, then the resulting range
22320 (intersect_ranges): If the new range is ~[0,0] and the old range is
22321 wide, then prefer ~[0,0].
22322 * tree-vrp.c (overflow_comparison_p_1): New function.
22323 (overflow_comparison_p): New function.
22324 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22325 if NAME is used in an overflow test.
22326 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22327 overflow check that can be expressed as an equality test, then adjust
22328 ops to be that equality test.
22330 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22332 * config/s390/s390-builtin-types.def: Remove flags argument.
22333 * config/s390/s390.c (s390_init_builtins): Likewise.
22335 2017-02-14 Martin Liska <mliska@suse.cz>
22337 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22338 vector. Fix trailing white spaces.
22340 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22342 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22345 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22347 PR rtl-optimization/68664
22348 * config/arm/arm.c (arm_sched_can_speculate_insn):
22349 New function. Declare prototype.
22350 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22352 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22354 PR rtl-optimization/68664
22355 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22357 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22359 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22361 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22362 max skip bytes for function, loop and jump.
22364 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22366 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22367 ABS_EXPR for gimple dump.
22369 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22372 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22374 PR tree-optimization/79408
22375 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22376 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22377 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22378 also if rhs1 is INTEGER_CST.
22380 2017-02-14 Richard Biener <rguenther@suse.de>
22382 PR middle-end/79432
22383 * tree-into-ssa.c (insert_phi_nodes): When the function can
22384 have abnormal edges rewrite SSA names with broken use-def
22385 dominance out of SSA and register them for PHI insertion.
22387 2017-02-13 Martin Sebor <msebor@redhat.com>
22389 PR middle-end/79496
22390 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22391 clearing info.nowrite flag when snprintf size argument is a range.
22393 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22395 * cprop.c (cprop_jump): Add missing space in string literal.
22396 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22397 (get_constraint_for_component_ref): Likewise.
22398 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22399 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22400 * lra-constraints.c (process_alt_operands): Likewise.
22401 * ipa-inline.c (inline_small_functions): Likewise.
22402 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22403 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22404 * trans-mem.c (diagnose_tm_1_op): Likewise.
22405 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22406 (grid_parallel_clauses_gridifiable): Likewise.
22408 * config/nvptx/mkoffload.c (process): Add space in between
22411 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22412 "MOD4_SSE_REGS" and "ALL_REGS".
22414 * spellcheck.c (test_data): Add , in between "foo" and "food".
22416 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22419 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22420 boundary crossing check and subsequent code generation agree.
22422 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22424 * config/aarch64/aarch64.c (has_memory_op): Delete.
22425 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22428 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22430 PR rtl-optimization/79388
22431 PR rtl-optimization/79450
22432 * combine.c (distribute_notes): When removing TEM_INSN for which
22433 corresponding dest has last value recorded, invalidate that last
22436 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22438 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22439 of explicit '@'. Add missing assembly comment marker on branch costs
22442 2017-02-13 Nathan Sidwell <nathan@acm.org>
22444 * gengtype-lex.l (<in_struct>): Add '/'.
22446 2017-02-13 Martin Liska <mliska@suse.cz>
22449 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22451 2017-02-13 Richard Biener <rguenther@suse.de>
22453 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22455 * configure: Re-generate.
22456 * config.in: Likewise.
22457 * graphite-dependences.c: Simplify as if
22458 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22459 * graphite-isl-ast-to-gimple.c: Likewise.
22460 * graphite-optimize-isl.c: Likewise.
22461 * graphite-poly.c: Likewise.
22462 * graphite-sese-to-poly.c: Likewise.
22463 * graphite.h: Likewise.
22464 * toplev.c: Include isl/version.h and use isl_version () for
22465 printing the ISL version.
22466 * doc/install.texi: Update ISL requirement.
22468 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22470 * doc/standards.texi (Standards): Update reference to
22473 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22475 * doc/extend.texi (Named Address Spaces): sourceware.org now
22477 * doc/install.texi (Binaries): Ditto.
22480 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22482 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22483 standard terminology "stringize"/"stringizing" throughout.
22484 * doc/cppinternals.texi: Likewise.
22486 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22488 * doc/extend.texi: Fix some spelling mistakes and typos.
22489 * doc/invoke.texi: Likewise.
22491 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22494 * params.def (inline-min-speedup) Change from 10 to 8.
22496 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22498 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22501 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22504 * ipa-inline-analysis.c (get_minimal_bb): New function.
22505 (record_modified): Use it.
22506 (remap_edge_change_prob): Handle also ancestor functions.
22508 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22510 * doc/contrib.texi (Contributors): Remove broken link into
22511 the Mauve CVS repository.
22513 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22515 PR middle-end/79454
22516 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22517 result computation whenever lhs doesn't have vector mode, not
22518 just when it has BLKmode.
22520 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22522 * doc/makefile.texi (profiledbootstrap): Refer to the
22523 installation instructions only in textual form.
22525 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22528 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22530 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22532 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22533 (Specific): Update mingw-w64 reference.
22535 (Specific): Remove broken link to Renesas RX processor.
22537 2017-02-10 Richard Biener <rguenther@suse.de>
22539 * toplev.c (process_options): Do not mention obsolete graphite
22540 options when printing sorry message about missing graphite support.
22541 Mention -floop-nest-optimize.
22543 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22545 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22546 (vtst_p16): Likewise.
22547 (vtstq_p8): Likewise.
22548 (vtstq_p16): Likewise.
22550 (vtstq_p64): Likewise.
22551 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22552 (vset_lane_p64): New.
22553 (vsetq_lane_p64): New.
22555 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22557 PR tree-optimization/79411
22558 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22559 stmt operands are SSA_NAMEs used in abnormal phis.
22560 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22563 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22566 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22569 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22571 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22573 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22575 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22579 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22580 not arbitrary TREE_CONSTANT.
22583 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22584 "omp declare target link" attribute unless is_global_var.
22585 * omp-offload.c (find_link_var_op): Likewise.
22587 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22588 Chung-Lin Tang <cltang@codesourcery.com>
22590 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22592 (gimplify_adjust_omp_clauses): Don't delete TILE.
22593 (gimplify_omp_for): Deal with TILE.
22594 * internal-fn.c (expand_GOACC_TILE): New function.
22595 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22597 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22598 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22600 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22601 avoid DIV for outermost collapse var.
22602 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22603 Remove out of date comments, fix whitespace.
22604 * omp-general.c (omp_extract_for_data): Deal with tiling.
22605 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22606 adjust OLF_DIM_BASE value.
22607 (struct omp_for_data): Add tiling field.
22608 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22609 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22610 for auto loops. Remove default auto determining, moved to
22611 oacc_loop_fixed_partitions.
22612 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22613 stmts, add e_mask field.
22614 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22615 (oacc_thread_numbers): Use oacc_dim_call.
22616 (oacc_xform_tile): New.
22617 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22618 (finish_oacc_loop): Adjust for ifns vector.
22619 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22620 add case for GOACC_TILE fns.
22621 (oacc_loop_xform_loop): Delete.
22622 (oacc_loop_process): Iterate over call list directly, and add
22623 handling for GOACC_TILE fns.
22624 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22626 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22627 vector partitioning to outer loops. Assign 2 partitions to loops
22628 when available. Add TILE handling.
22629 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22630 (execite_oacc_device_lower): Process GOACC_TILE fns,
22631 ignore unknown specs.
22632 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22633 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22634 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22636 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22638 * configure.ac (ACX_BUGURL): Update.
22639 * configure: Regenerate.
22641 2017-02-09 Richard Biener <rguenther@suse.de>
22643 PR tree-optimization/69823
22644 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22645 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22647 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22649 * config/arc/arc-c.def: Add __NPS400__ definition.
22650 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22651 (TARGET_NPS400): Define.
22653 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22655 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22657 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22658 pointer, arch_info.
22659 (arc_cpu_types): Fill the arch_info field with a pointer into the
22660 arc_arch_types table.
22661 (arc_selected_cpu): Declare.
22662 * config/arc/arc.c (arc_selected_cpu): Make global.
22663 (arc_selected_arch): Delete.
22664 (arc_base_cpu): Delete.
22665 (arc_override_options): Remove references to deleted variables,
22666 update access to arch information.
22667 (ARC_OPT): Update access to arch information.
22668 (ARC_OPTX): Likewise.
22669 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22670 (TARGET_ARC600): Update access to arch information.
22671 (TARGET_ARC601): Likewise.
22672 (TARGET_ARC700): Likewise.
22673 (TARGET_EM): Likewise.
22674 (TARGET_HS): Likewise.
22675 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22678 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22681 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22682 condition/operands for integer GE/LE/GEU/LEU operations.
22684 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22686 PR translation/79397
22687 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22690 2017-02-08 Martin Jambor <mjambor@suse.cz>
22693 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22694 whether allocation happened.
22695 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22696 nothing was allocated.
22698 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22700 PR tree-optimization/79408
22701 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22702 constant, but SSA_NAME with a known integer range, use the minimum
22703 of that range instead of op1 to determine if modulo can be replaced
22704 with its first operand.
22706 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22708 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22710 2017-02-08 Richard Biener <rguenther@suse.de>
22712 PR tree-optimization/71824
22713 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22714 Check all loops contained in the merged region.
22716 2017-02-07 Andrew Pinski <apinski@cavium.com>
22718 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22720 2017-02-07 Andrew Pinski <apinski@cavium.com>
22722 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22723 (thunderxt88): Likewise.
22724 (thunderxt81): Disable LSE and change v8.1 to v8.
22725 (thunderxt83): Likewise.
22727 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22728 Richard Biener <rguenther@suse.de>
22730 PR middle-end/79399
22731 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22732 type from int to size_t.
22733 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22735 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22737 PR rtl-optimization/79386
22738 * cprop.c (bypass_conditional_jumps): Initialize
22739 bypass_last_basic_block already before splitting bbs after
22740 unconditional traps...
22741 (bypass_conditional_jumps): ... rather than here.
22744 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22745 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22746 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22747 fix -masm=intel patterns.
22749 2017-02-07 Richard Biener <rguenther@suse.de>
22751 PR tree-optimization/79256
22752 PR middle-end/79278
22753 * builtins.c (get_object_alignment_2): Use min_align_of_type
22754 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22755 and ADJUST_FIELD_ALIGN.
22757 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22759 * doc/tm.texi: Regenerate.
22760 * stor-layout.c (layout_decl): Adjust.
22761 (update_alignment_for_field): Likewise.
22762 (place_field): Likewise.
22763 (min_align_of_type): Likewise.
22764 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22765 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22766 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22767 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22768 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22769 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22770 * config/i386/i386.c (x86_field_alignment): Likewise.
22771 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22772 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22773 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22774 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22775 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22776 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22780 2017-01-30 Richard Biener <rguenther@suse.de>
22782 PR tree-optimization/79256
22783 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22784 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22787 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22789 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22790 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22791 builtins to SImode and emit a zero-extend, if necessary.
22793 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22795 * docs/invoke.texi (RISC-V Options): Alphabetize.
22797 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22799 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22802 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22804 * config/riscv/riscv.c: New file.
22805 * common/config/riscv/riscv-common.c: Likewise.
22806 * config.gcc: Likewise.
22807 * config/riscv/constraints.md: Likewise.
22808 * config/riscv/elf.h: Likewise.
22809 * config/riscv/generic.md: Likewise.
22810 * config/riscv/linux.h: Likewise.
22811 * config/riscv/multilib-generator: Likewise.
22812 * config/riscv/peephole.md: Likewise.
22813 * config/riscv/pic.md: Likewise.
22814 * config/riscv/predicates.md: Likewise.
22815 * config/riscv/riscv-builtins.c: Likewise.
22816 * config/riscv/riscv-c.c: Likewise.
22817 * config/riscv/riscv-ftypes.def: Likewise.
22818 * config/riscv/riscv-modes.def: Likewise.
22819 * config/riscv/riscv-opts.h: Likewise.
22820 * config/riscv/riscv-protos.h: Likewise.
22821 * config/riscv/riscv.h: Likewise.
22822 * config/riscv/riscv.md: Likewise.
22823 * config/riscv/riscv.opt: Likewise.
22824 * config/riscv/sync.md: Likewise.
22825 * config/riscv/t-elf-multilib: Likewise.
22826 * config/riscv/t-linux: Likewise.
22827 * config/riscv/t-linux-multilib: Likewise.
22828 * config/riscv/t-riscv: Likewise.
22829 * configure.ac: Likewise.
22830 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22831 Waterman as RISC-V maintainers.
22832 * doc/install.texi: Add RISC-V entries.
22833 * doc/invoke.texi: Add RISC-V options section.
22834 * doc/md.texi: Add RISC-V constraints section.
22835 * configure: Regenerated.
22837 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22840 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22841 false values to be constant vectors with all 0 or all 1 bits set.
22842 (vcondu<mode><mode>): Likewise.
22843 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22845 (fpmask_comparison_operator): Update comment.
22846 (vecint_comparison_operator): New predicate.
22847 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22848 vector conditionals when the true and false values are constant
22849 vectors with all 0 bits or all 1 bits set.
22851 2017-02-06 Martin Sebor <msebor@redhat.com>
22853 PR tree-optimization/79376
22854 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22856 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22858 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22859 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22860 to simplify split condition.
22862 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22864 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22865 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22868 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22870 PR rtl-optimization/68664
22871 * target.def (can_speculate_insn): New hook.
22872 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22873 * doc/tm.texi: Regenerate.
22874 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22875 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22876 (rs6000_sched_can_speculate_insn): New function.
22878 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22880 PR tree-optimization/79284
22881 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22882 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22883 vectorizable_mask_load_store, vectorizable_operation,
22884 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22885 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22886 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22887 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22888 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22889 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22890 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22891 is_gimple_assign (stmt). Replace another such test with
22892 is_gimple_assign (stmt).
22894 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22897 * config/avr/avr.c (rtl-iter.h): Include it.
22898 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22899 (avr_legitimate_combined_insn): ...and implementation.
22901 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22903 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22904 * config/s390/s390.c (s390_const_operand_ok)
22905 (s390_canonicalize_comparison, s390_extract_part)
22906 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22907 (s390_contiguous_bitmask_p, s390_rtx_costs)
22908 (legitimize_pic_address): Likewise.
22909 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22910 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22911 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22912 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22913 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22915 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22917 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22918 REGNO($0) == REGNO($1).
22920 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22922 * config/s390/linux.h(SIZE_TYPE): Add comment.
22924 2017-02-06 Julian Brown <julian@codesourcery.com>
22925 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22926 Virendra Pathak <virendra.pathak@broadcom.com>
22928 * config/aarch64/aarch64-cores.def: Change the scheduler
22930 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22931 * config/aarch64/thunderx2t99.md: New file.
22933 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22935 * doc/standards.texi (Go Language): Update link to language
22938 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22940 * tree-eh.c (lower_resx): Sanitize profile.
22941 (cleanup_empty_eh_move_lp): Likewise.
22943 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22946 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22948 * cfgloopmanip.h (loop_version): Update prototype.
22949 * modulo-sched.c (sms_schedule): Update call of loop_version.
22950 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22951 * tree-parloops.c (gen_parallel_loop): Likewise.
22952 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22953 * tree-ssa-loop-split.c (split_loop): Likewise.
22954 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22955 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22957 2017-02-05 Martin Liska <mliska@suse.cz>
22960 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22962 (print_operand_address): Initialize a struct to zero.
22964 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22966 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22967 garbage collector only in textual form.
22969 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22971 * doc/extend.texi (x86 specific memory model extensions for
22972 transactional memory): Simplify a phrase.
22974 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22977 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22978 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22979 (atomic_storedi_1): Likewise.
22981 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22983 PR tree-optimization/79338
22984 * tree-parloops.c (gather_scalar_reductions): Don't call
22985 vect_analyze_loop_form for loop->inner before destroying loop's
22988 2017-02-03 Martin Sebor <msebor@redhat.com>
22990 PR tree-optimization/79327
22991 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22992 when precision has resulted in leading zeros.
22993 (format_integer): Adjust the likely counter to assume an unknown
22994 argument that may be zero is non-zero.
22996 2017-02-03 Jason Merrill <jason@redhat.com>
22999 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23000 avoid copying non-taken branch.
23002 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23004 PR tree-optimization/79340
23005 * tree-vect-loop.c (vectorizable_reduction): Release
23006 vec_defs elements after safe_splicing them into other vectors.
23009 PR tree-optimization/79327
23010 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23011 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23013 (format_integer): Use wide_int_to_tree instead of build_int_cst
23014 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23015 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23016 of shortest and longest sequence.
23018 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23020 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23021 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23023 2017-02-03 Walter Lee <walt@tilera.com>
23026 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23027 after initial stackframe link reg save.
23028 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23030 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23033 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23034 wu for stxssp alternative.
23036 2017-02-03 Martin Sebor <msebor@redhat.com>
23038 PR tree-optimization/79352
23039 * gimple-fold.c (get_range_strlen): Add argument.
23040 (get_range_strlen): Change return type to bool.
23041 (get_maxval_strlen): Pass in a dummy argument.
23042 * gimple-fold.h (get_range_strlen): Change return type to bool.
23043 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23044 * tree.h (array_at_struct_end_p): Add argument.
23045 * tree.c (array_at_struct_end_p): Handle it.
23047 2017-02-03 Martin Liska <mliska@suse.cz>
23050 * multiple_target.c (create_dispatcher_calls): Redirect edge
23051 from a caller of a dispatcher.
23052 (expand_target_clones): Make the clones local.
23053 (ipa_target_clone): Do both target clones and resolvers.
23054 (ipa_dispatcher_calls): Remove the pass.
23055 (pass_dispatcher_calls::gate): Likewise.
23056 (make_pass_dispatcher_calls): Likewise.
23057 * passes.def (pass_target_clone): Put as very first IPA early
23060 2017-02-03 Martin Liska <mliska@suse.cz>
23062 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23063 in case of a function with ifunc attribute.
23065 2017-02-03 Martin Liska <mliska@suse.cz>
23067 * cgraph.c (cgraph_node::dump): Dump function version info.
23068 * symtab.c (symtab_node::dump_base): Add missing new line.
23070 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23072 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23073 (ifcombine_ifandif): Use it.
23075 2017-02-03 Martin Liska <mliska@suse.cz>
23077 * doc/invoke.texi: Document default value for
23078 use-after-scope-direct-emission-threshold.
23080 2017-02-03 Martin Liska <mliska@suse.cz>
23082 PR tree-optimization/79339
23083 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23084 (format_floating): Likewise.
23086 2017-02-03 Martin Liska <mliska@suse.cz>
23089 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23091 (ipa_node_params_t::remove): Likewise.
23092 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23093 initialization from removed ipa_node_params_t::insert.
23094 (ipa_node_params::~ipa_node_params): Move from removed
23095 ipa_node_params_t::release.
23096 * symbol-summary.h (symbol_summary::m_released): New member.
23097 Do not release a summary twice. Do not allow to call finalizer
23098 for types of a summary that live in GGC memory.
23100 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23102 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23105 2017-02-02 Martin Sebor <msebor@redhat.com>
23107 PR middle-end/79275
23108 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23109 (format_string): Tighten up the range of output for non-constant
23110 strings and correct the expected range for wide non-constant strings.
23112 2017-02-02 Martin Sebor <msebor@redhat.com>
23114 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23116 PR middle-end/32003
23117 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23119 (-fdump-tree-@var): Add to index and document how to come up
23120 with pass-specific option and dump file names.
23121 (-fdump-passes): Clarify where to look for output.
23123 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23125 PR middle-end/77445
23126 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23127 statistics of the analyzed path; allow threading for speed when
23128 any of BBs along the path are optimized for speed.
23130 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23132 PR middle-end/78468
23133 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23134 settings of the virtual registers.
23137 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23139 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23140 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23143 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23145 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23146 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23148 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23150 * config/s390/s390.md: Add missing comments with the expanded
23152 * config/s390/vector.md: Likewise.
23153 * config/s390/vx-builtins.md: Likewise.
23155 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23158 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23159 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23160 conditions on a single line.
23162 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23164 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23165 __S390_VX__ to __VX__.
23167 2017-02-01 Andrew Pinski <apinski@cavium.com>
23169 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23170 stmt_info to record_stmt_cost.
23171 (vect_get_known_peeling_cost): Pass stmt_info if known to
23173 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23174 cpu_vector_cost field into
23175 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23176 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23177 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23178 splitting of scalar_stmt_cost and vec_stmt_cost.
23179 (thunderx_vector_cost): Likewise.
23180 (cortexa57_vector_cost): LIkewise.
23181 (exynosm1_vector_cost): Likewise.
23182 (xgene1_vector_cost): Likewise.
23183 (thunderx2t99_vector_cost): Improve after the splitting of the two
23185 (aarch64_builtin_vectorization_cost): Update for the splitting of
23186 scalar_stmt_cost and vec_stmt_cost.
23188 2017-02-01 Torvald Riegel <triegel@redhat.com>
23189 Richard Henderson <rth@redhat.com>
23191 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23192 conditional on existance of a fast atomic load.
23193 * optabs-query.c (can_atomic_load_p): New function.
23194 * optabs-query.h (can_atomic_load_p): Declare it.
23195 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23196 no fast atomic load is available for the particular size of access.
23197 (expand_atomic_compare_and_swap): Likewise.
23198 (expand_atomic_load): Likewise.
23199 (expand_atomic_store): Likewise.
23200 (expand_atomic_fetch_op): Likewise.
23201 * testsuite/lib/target-supports.exp
23202 (check_effective_target_sync_int_128): Remove x86 because it provides
23203 no fast atomic load.
23204 (check_effective_target_sync_int_128_runtime): Likewise.
23206 2017-02-01 Richard Biener <rguenther@suse.de>
23208 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23209 (graphite_transform_loops): Provide opt-info for optimized nests.
23210 * tree-parloop.c (parallelize_loops): Provide opt-info for
23211 parallelized loops.
23213 2017-02-01 Richard Biener <rguenther@suse.de>
23215 PR middle-end/79315
23216 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23217 was not set before.
23219 2017-02-01 Richard Biener <rguenther@suse.de>
23221 PR tree-optimization/71824
23222 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23223 Verify the loops are valid in the merged SESE region.
23224 (scop_detection::can_represent_loop_1): Check analyzing the
23225 evolution of the number of iterations in the region succeeds.
23227 2017-01-31 Ian Lance Taylor <iant@golang.org>
23229 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23230 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23232 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23234 PR preprocessor/79210
23235 * input.c (get_substring_ranges_for_loc): Replace line_width
23236 assertion with error-handling.
23238 2017-01-31 Richard Biener <rguenther@suse.de>
23240 PR tree-optimization/77318
23241 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23242 (create_pw_aff_from_tree): Take loop parameter.
23243 (add_condition_to_pbb): Pass loop of the condition to
23244 create_pw_aff_from_tree.
23246 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23248 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23249 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23251 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23255 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23257 (convert_int_to_float128): Likewise.
23258 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23259 (convert_int_to_float128): Likewise.
23260 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23261 (UNSPEC_IEEE128_CONVERT): Likewise.
23262 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23263 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23264 Use local variables for IBM extended format.
23265 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23266 (fix_trunc<mode>si2_fprs): Likewise.
23267 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23268 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23269 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23270 to know that we can now have integers of all sizes in vector
23272 (fix<uns>_<mode>di2_hw): Likewise.
23273 (float<uns>_<mode>si2_hw): Likewise.
23274 (fix_<mode>si2_hw): Likewise.
23275 (fixuns_<mode>si2_hw): Likewise.
23276 (float<uns>_<mode>di2_hw): Likewise.
23277 (float_<mode>di2_hw): Likewise.
23278 (float_<mode>si2_hw): Likewise.
23279 (floatuns_<mode>di2_hw): Likewise.
23280 (floatuns_<mode>si2_hw): Likewise.
23281 (xscvqp<su>wz_<mode>): Delete, no longer used.
23282 (xscvqp<su>dz_<mode>): Likewise.
23283 (xscv<su>dqp_<mode>): Likewise.
23284 (ieee128_mfvsrd_64bit): Likewise.
23285 (ieee128_mfvsrd_32bit): Likewise.
23286 (ieee128_mfvsrwz): Likewise.
23287 (ieee128_mtvsrw): Likewise.
23288 (ieee128_mtvsrd_64bit): Likewise.
23289 (ieee128_mtvsrd_32bit): Likewise.
23291 2017-01-31 Martin Liska <mliska@suse.cz>
23294 * ipa-prop.c (ipa_free_all_node_params): Call release method
23295 instead of ~sumbol_summary to not to trigger double times
23298 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23300 PR tree-optimization/71691
23301 * bitmap.h (class auto_bitmap): New.
23302 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23303 is_maybe_undefined instead of ssa_undefined_value_p.
23305 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23307 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23308 __S390_ARCH_LEVEL__ to __ARCH__.
23310 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23312 PR tree-optimization/79267
23313 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23314 if should_remove_lhs_p is true.
23316 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23319 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23320 (add_alignment_attribute): New.
23321 (base_type_die): Add alignment attribute.
23322 (subrange_type_die): Likewise.
23323 (modified_type_die): Likewise.
23324 (gen_array_type_die): Likewise.
23325 (gen_descr_array_type_die: Likewise.
23326 (gen_enumeration_type_die): Likewise.
23327 (gen_subprogram_die): Likewise.
23328 (gen_variable_die): Likewise.
23329 (gen_field_die): Likewise.
23330 (gen_ptr_to_mbr_type_die): Likewise.
23331 (gen_struct_or_union_type_die): Likewise.
23332 (gen_subroutine_type_die): Likewise.
23333 (gen_typedef_die): Likewise.
23334 (base_type_cmp): Compare alignment attribute.
23336 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23339 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23340 (setb_unsigned) New pattern for setb with CCUNS.
23341 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23342 subfc./subfe sequence to avoid overflow problems. Generate a
23343 shorter sequence with cmpld/setb for power9.
23344 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23345 for generating subfc. instruction.
23346 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23347 now uses this instruction.
23349 2017-01-30 Ian Lance Taylor <iant@google.com>
23352 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23353 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23355 2017-01-30 Martin Sebor <msebor@redhat.com>
23357 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23358 Move constant to the right of a relational operator.
23359 (get_mpfr_format_length, format_character, format_string): Ditto.
23360 (should_warn_p, maybe_warn): Same.
23362 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23364 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23367 * asan.c (get_translation_unit_decl): Remove function.
23368 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23370 2017-01-30 Martin Liska <mliska@suse.cz>
23372 PR gcov-profile/79259
23373 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23374 -fprofile-generate.
23376 2017-01-30 Martin Liska <mliska@suse.cz>
23379 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23380 Initialize variables with NULL value.
23382 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23385 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23387 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23389 2017-01-30 Richard Biener <rguenther@suse.de>
23391 PR tree-optimization/79276
23392 * tree-vrp.c (process_assert_insertions): Properly adjust common
23393 when removing a duplicate.
23395 * gcc.dg/torture/pr79276.c: New testcase.
23397 2017-01-30 Richard Biener <rguenther@suse.de>
23399 PR tree-optimization/79256
23400 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23401 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23403 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23405 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23408 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23409 ("*r<noxa>sbg_<mode>_sll_bitmask")
23410 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23411 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23412 Use contiguous_bitmask_nowrap_operand.
23414 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23417 * config/rs6000/altivec.h (vec_xl): Revise #define.
23418 (vec_xst): Likewise.
23420 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23422 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23424 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23426 PR rtl-optimization/79194
23427 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23428 traps before call to bypass_conditional_jumps.
23430 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23432 PR tree-optimization/71374
23433 * lra-constraints.c (check_conflict_input_operands): New.
23434 (match_reload): Use it.
23436 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23439 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23440 account to calculate conflict_set.
23442 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23444 PR rtl-optimization/78559
23445 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23446 other_insn in combine.
23448 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23450 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23451 uint16_type_node for BT_UINT16.
23453 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23455 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23456 "RTL Tests" to menu.
23457 (GIMPLE Tests): New node.
23458 (RTL Tests): New node.
23460 2017-01-27 Richard Biener <rguenther@suse.de>
23462 PR tree-optimization/79245
23463 * tree-loop-distribution.c (distribute_loop): Apply cost
23464 modeling also to detected patterns.
23466 2017-01-27 Richard Biener <rguenther@suse.de>
23468 PR tree-optimization/71433
23469 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23470 (compare_assert_loc): New function.
23471 (process_assert_insertions): Sort and optimize assert locations
23472 to remove duplicates and push down identical assertions on
23473 edges to their destination block.
23475 2017-01-27 Richard Biener <rguenther@suse.de>
23477 PR tree-optimization/79244
23478 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23479 out SSA names even if abnormal.
23481 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23483 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23484 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23485 instead of MPFR_RNDN.
23487 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23490 * arm.c (arm_option_override): Don't call build_target_option_node
23491 until after doing all option overrides.
23492 (arm_valid_target_attribute_tree): Likewise.
23494 2017-01-27 Martin Liska <mliska@suse.cz>
23496 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23497 for {cd}tors and C++ {cd}tors.
23499 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23501 * config/s390/s390.md ("*setmem_long_and")
23502 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23504 2017-01-26 Martin Sebor <msebor@redhat.com>
23506 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23509 2017-01-26 Martin Sebor <msebor@redhat.com>
23511 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23512 HAVE_DFmode before using XFmode or DFmode.
23513 (parse_directive): Avoid using the z length modifier to avoid
23514 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23516 PR middle-end/78703
23517 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23518 to accept adjustment as an array.
23519 (get_int_range): New function.
23520 (struct directive): Make width and prec arrays.
23521 (directive::set_width, directive::set_precision): Call get_int_range.
23522 (format_integer, format_floating): Handle width and precision ranges.
23523 (format_string, parse_directive): Same.
23525 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23528 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23529 comdat_type_p set, just clone them, but keep the children in the
23533 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23534 which have direct callers with -fvar-tracking-assignments enabled
23536 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23537 inside of type units.
23539 2017-01-26 Martin Sebor <msebor@redhat.com>
23541 PR middle-end/78703
23542 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23544 (struct format_result): Replace number_chars, number_chars_min,
23545 and number_chars_max with a single member of struct result_range.
23547 (format_result::operator+=): Adjust.
23548 (struct fmtresult): Remove bounded. Handle likely and unlikely
23550 (fmtresult::adjust_for_width_or_precision): New function.
23551 (fmtresult:type_max_digits): New function.
23552 (bytes_remaining): Handle likely and unlikely counters.
23553 (min_bytes_remaining): Remove.
23554 (format_percent): Simplify.
23555 (format_integer, format_floating): Set likely and unlikely counters.
23556 (get_string_length, format_character, format_string): Same.
23557 (format_plain, should_warn_p): New function.
23558 (maybe_warn): Call should_warn_p. Update diagnostic messages
23559 and handle those for all directives, including plain strings.
23560 (format_directive): Handle likely and unlikely counters.
23561 Remove unnecessary quoting from diagnostics. Add an informational
23563 (add_bytes): Remove.
23564 (pass_sprintf_length::compute_format_length): Simplify.
23565 (try_substitute_return_value): Handle likely and unlikely counters.
23567 2017-01-26 Carl Love <cel@us.ibm.com>
23569 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23570 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23572 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23575 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23576 endianess for subregs into account.
23577 * lra-constraints.c (lra_constraints): Do risky transformations
23578 always on the first iteration.
23579 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23580 last_call_used_reg_set.
23581 (process_bb_lives): Define and use last_call_used_reg_set.
23582 * lra.c (lra): Always continue after lra_constraints on the first
23585 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23587 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23589 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23591 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23593 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23594 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23595 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23596 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23597 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23598 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23599 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23600 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23601 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23603 2017-01-26 Marek Polacek <polacek@redhat.com>
23606 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23607 for the third operand.
23609 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23611 PR middle-end/79236
23612 * omp-low.c (struct omp_context): Add simt_stmt field.
23613 (scan_omp_for): Return omp_context *.
23614 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23615 context to the _simt_ SIMD stmt.
23616 (lower_omp_for): For combined SIMD with sibling _simt_
23617 SIMD, make sure to use the same decls in _looptemp_
23618 clauses as in the sibling.
23620 2017-01-26 David Sherwood <david.sherwood@arm.com>
23622 PR middle-end/79212
23623 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23626 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23629 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23630 emit fld b; fld a; if possible.
23632 * brig-builtins.def: Update copyright years.
23633 * config/arm/arm_acle_builtins.def: Update copyright years.
23635 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23638 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23639 constraint instead of o for the stxsd instruction.
23641 2017-01-25 Carl Love <cel@us.ibm.com>
23643 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23644 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23646 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23648 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23650 2017-01-25 Richard Biener <rguenther@suse.de>
23652 PR tree-optimization/69264
23653 * target.def (vector_alignment_reachable): Improve documentation.
23654 * doc/tm.texi: Regenerate.
23655 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23657 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23658 earlier changes with respect to TYPE_USER_ALIGN.
23659 (vector_alignment_reachable_p): Likewise. Improve dumping.
23661 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23664 * config/arm/arm.md (xordi3): Force constant operand into a register
23667 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23669 * doc/invoke.texi (-fstore-merging): Correct default optimization
23670 levels at which it is enabled.
23671 (-O): Move -fstore-merging from list to...
23674 2017-01-25 Richard Biener <rguenther@suse.de>
23677 * omp-expand.c: Include debug.h.
23678 (expand_omp_taskreg): Make sure to generate early debug before
23679 outlining anything from a function.
23680 (expand_omp_target): Likewise.
23681 (grid_expand_target_grid_body): Likewise.
23683 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23686 * asan.c (get_translation_unit_decl): New function.
23687 (asan_add_global): Extract modules file name from globals
23688 TRANSLATION_UNIT_DECL name.
23690 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23693 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23694 for long calls with APCS frame and VFP.
23696 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23698 * cfg.c (original_copy_tables_initialized_p): New function.
23699 * cfg.h (original_copy_tables_initialized_p): New decl.
23700 * cfgrtl.c (relink_block_chain): Guard the call to
23701 free_original_copy_tables with a call to
23702 original_copy_tables_initialized_p.
23703 * cgraph.h (symtab_node::native_rtl_p): New decl.
23704 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23705 (symtab_node::needed_p): Don't assert for early assembly output
23706 for __RTL functions.
23707 (cgraph_node::finalize_function): Set "force_output" for __RTL
23709 (cgraph_node::analyze): Bail out early for __RTL functions.
23710 (analyze_functions): Update assertion to support __RTL functions.
23711 (cgraph_node::expand): Bail out early for __RTL functions.
23712 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23714 * function.h (struct function): Update comment for field
23716 * gimple-expr.c: Include "tree-pass.h".
23717 (gimple_has_body_p): Return false for __RTL functions.
23718 * Makefile.in (OBJS): Add run-rtl-passes.o.
23719 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23721 (gcc::pass_manager::get_clean_slate): New accessor.
23722 * passes.c: Include "insn-addr.h".
23723 (should_skip_pass_p): Add logging. Update logic for running
23724 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23725 property-provider override so it is only done for gimple passes.
23727 (skip_pass): New function.
23728 (execute_one_pass): Call skip_pass when skipping passes.
23729 * read-md.c (md_reader::read_char): Support filtering
23730 the input to a subset of line numbers.
23731 (md_reader::md_reader): Initialize fields
23732 m_first_line and m_last_line.
23733 (md_reader::read_file_fragment): New function.
23734 * read-md.h (md_reader::read_file_fragment): New decl.
23735 (md_reader::m_first_line): New field.
23736 (md_reader::m_last_line): New field.
23737 * read-rtl-function.c (function_reader::create_function): Only
23738 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23739 curr_properties. Set DECL_INITIAL to a dummy block.
23740 (read_rtl_function_body_from_file_range): New function.
23741 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23743 * run-rtl-passes.c: New file.
23744 * run-rtl-passes.h: New file.
23746 2017-01-24 Jeff Law <law@redhat.com>
23748 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23751 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23753 PR tree-optimization/79159
23754 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23755 (record_nonwrapping_iv): Improve boundary using above function if no
23756 value range information.
23758 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23759 Martin Jambor <mjambor@suse.cz>
23761 * brig-builtins.def: New file.
23762 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23763 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23764 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23765 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23766 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23767 * builtin-types.def (BT_INT8): New.
23768 (BT_INT16): Likewise.
23769 (BT_UINT8): Likewise.
23770 (BT_UINT16): Likewise.
23771 (BT_FN_ULONG): Likewise.
23772 (BT_FN_UINT_INT): Likewise.
23773 (BT_FN_UINT_ULONG): Likewise.
23774 (BT_FN_UINT_LONG): Likewise.
23775 (BT_FN_UINT_PTR): Likewise.
23776 (BT_FN_ULONG_PTR): Likewise.
23777 (BT_FN_INT8_FLOAT): Likewise.
23778 (BT_FN_INT16_FLOAT): Likewise.
23779 (BT_FN_UINT32_FLOAT): Likewise.
23780 (BT_FN_UINT16_FLOAT): Likewise.
23781 (BT_FN_UINT8_FLOAT): Likewise.
23782 (BT_FN_UINT64_FLOAT): Likewise.
23783 (BT_FN_UINT16_UINT32): Likewise.
23784 (BT_FN_UINT32_UINT16): Likewise.
23785 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23786 (BT_FN_INT_PTR_INT): Likewise.
23787 (BT_FN_UINT_PTR_UINT): Likewise.
23788 (BT_FN_LONG_PTR_LONG): Likewise.
23789 (BT_FN_ULONG_PTR_ULONG): Likewise.
23790 (BT_FN_VOID_UINT64_UINT64): Likewise.
23791 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23792 (BT_FN_INT8_INT8_INT8): Likewise.
23793 (BT_FN_INT16_INT16_INT16): Likewise.
23794 (BT_FN_INT_INT_INT): Likewise.
23795 (BT_FN_UINT_FLOAT_UINT): Likewise.
23796 (BT_FN_FLOAT_UINT_UINT): Likewise.
23797 (BT_FN_ULONG_UINT_UINT): Likewise.
23798 (BT_FN_ULONG_UINT_PTR): Likewise.
23799 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23800 (BT_FN_UINT_UINT_UINT): Likewise.
23801 (BT_FN_VOID_UINT_PTR): Likewise.
23802 (BT_FN_UINT_UINT_PTR: Likewise.
23803 (BT_FN_UINT32_UINT64_PTR): Likewise.
23804 (BT_FN_INT_INT_UINT_UINT): Likewise.
23805 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23806 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23807 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23808 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23809 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23810 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23811 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23812 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23813 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23814 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23815 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23816 * doc/frontends.texi: List BRIG FE.
23817 * doc/install.texi (Testing): Add BRIG tesring requirements.
23818 * doc/invoke.texi (Overall Options): Mention BRIG.
23819 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23821 2017-01-24 Richard Biener <rguenther@suse.de>
23823 PR translation/79208
23824 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23826 2017-01-24 Martin Jambor <mjambor@suse.cz>
23829 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23830 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23831 and known_contexts.
23833 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23835 PR middle-end/79123
23836 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23837 casts from signed to unsigned really don't have a range.
23839 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23841 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23842 GMP_RNDx for compatiblity.
23844 2017-01-24 Martin Liska <mliska@suse.cz>
23847 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23848 that would prevent us to call alloca with -1 as argument.
23850 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23852 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23853 Avoid -Wformat-security warning.
23855 2017-01-23 Andrew Pinski <apinski@cavium.com>
23857 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23860 2017-01-23 Martin Sebor <msebor@redhat.com>
23862 PR middle-end/78703
23863 * gimple-ssa-sprintf.c (warn_level): New global.
23864 (format_integer): Use it here and throughout the rest of the file.
23865 Use the same switch to compute sign as base.
23866 (maybe_warn): New function.
23867 (format_directive): Factor out warnings into maybe_warn.
23868 Add debugging output. Use warn_level.
23869 (add_bytes): Use warn_level.
23870 (pass_sprintf_length::compute_format_length): Add debugging output.
23871 (try_substitute_return_value): Same.
23872 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23874 PR middle-end/78703
23875 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23876 (struct fmtresult, format_integer, format_floating): Adjust.
23877 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23878 (get_string_length, format_string,format_directive): Same.
23879 (pass_sprintf_length::compute_format_length): Same.
23880 (try_substitute_return_value): Simplify slightly.
23882 PR middle-end/78703
23883 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23884 (fmtresult::operator+=): Outlined.
23885 (struct fmtresult): Add ctors.
23886 (struct conversion_spec): Rename...
23887 (struct directive): ...to this. Add and remove data members.
23888 (directive::set_width, directive::set_precision): New functions.
23889 (format_percent): Use fmtresult ctor.
23890 (get_width_and_precision): Remove.
23891 (format_integer): Make naming changes. Avoid computing width and
23893 (format_floating): Same. Adjust indentation.
23894 (format_character, format_none): New functions.
23895 (format_string): Moved character handling to format_character.
23896 (format_directive): Remove arguments, change return type.
23897 (parse_directive): New function.
23898 (pass_sprintf_length::compute_format_length): Move directive
23899 parsing to parse_directive.
23901 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23903 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23904 (assign_assembler_name_if_needed): ... this.
23905 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23906 (assign_assembler_name_if_needed): ... this.
23907 (free_lang_data_in_cgraph): Adjust callers.
23908 * cgraphunit.c (cgraph_node::analyze): Likewise.
23909 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23912 2017-01-23 Richard Biener <rguenther@suse.de>
23914 PR tree-optimization/79088
23915 PR tree-optimization/79188
23916 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23917 resetting loop bounds after last path deletion. Reset loop
23918 bounds of the target loop, make code match the comments.
23919 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23920 Make sure loops need no fixups.
23922 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23924 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23925 exponent support with double type for first argument.
23926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23927 type returned by __builtin_vec_extract_sig,
23928 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23929 functions from "vector int" to "vector unsigned int" or from
23930 "vector long long int" to "vector unsigned long long int".
23931 Changed type returned by __builtin_vec_extract_exp,
23932 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23933 functions from "vector int" to "vector unsigned int" or from
23934 "vector long long int" to "vector unsigned long long int".
23935 Changed return type of __builtin_vec_test_data_class,
23936 __builtin_vec_test_data_class_sp, and
23937 __builtin_vec_test_data_class_dp from "vector int" to
23938 "vector bool int" or from "vector long long int" to "vector bool
23939 long long int" and changed second argument type from "unsigned
23940 int" to "int". Added new overloaded function forms "vector float
23941 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23942 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23943 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23944 double, vector unsigned long long int)" and "vector double
23945 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23946 long int)". Changed return type of
23947 __builtin_scalar_test_data_class and
23948 __builtin_scalar_test_data_class_sp and
23949 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23950 int" and changed second argument from "unsigned int" to "int".
23951 Changed type returned by __builtin_scalar_test_neg,
23952 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23953 from "int" to "bool int". Added new overloaded function form
23954 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23955 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23956 exponent double-precision with floating point first argument.
23957 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23958 documentation of scalar_test_data_class, scalar_test_neg,
23959 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23960 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23961 vec_test_data_class built-in functions to reflect refinements in
23962 their type signatures.
23964 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23966 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23968 (aarch64_elf_asm_destructor): Likewise.
23970 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23972 PR rtl-optimization/78634
23973 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23974 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23975 * ifcvt.c (noce_try_cmove): Add missing cost check.
23977 PR rtl-optimization/71724
23978 * combine.c (if_then_else_cond): Look for situations where it is
23979 beneficial to undo the work of one of the recursive calls.
23981 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23983 PR tree-optimization/70754
23984 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23985 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23986 combined stmt before it if not NULL.
23987 (combine_chains): Process refs reversely and compute dominance point
23990 2017-01-23 Martin Liska <mliska@suse.cz>
23992 PR tree-optimization/79196
23993 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23994 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23996 (strlen_optimize_stmt): Call the renamed function.
23998 2017-01-23 Michael Matz <matz@suse.de>
24000 PR tree-optimization/78384
24001 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24003 2017-01-23 Richard Biener <rguenther@suse.de>
24005 PR tree-optimization/79186
24006 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24007 both incoming edges before moving an assert.
24009 2017-01-23 Martin Jambor <mjambor@suse.cz>
24011 * ipa-prop.c (load_from_param_1): Removed.
24012 (load_from_unmodified_param): Bits from load_from_param_1 put back
24014 (load_from_param): Removed.
24015 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24016 with stmt. Reverted back to use of load_from_unmodified_param.
24018 2017-01-23 Martin Jambor <mjambor@suse.cz>
24021 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24022 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24023 field a pointer to garbage collected vector, mark lattices and
24024 ipcp_orig_node with GTY((skip)).
24025 (ipa_get_param_count): Adjust to descriptors being a pointer.
24026 (ipa_get_param): Likewise.
24027 (ipa_get_type): Likewise.
24028 (ipa_get_param_move_cost): Likewise.
24029 (ipa_set_param_used): Likewise.
24030 (ipa_get_controlled_uses): Likewise.
24031 (ipa_set_controlled_uses): Likewise.
24032 (ipa_is_param_used): Likewise.
24033 (ipa_node_params_t): Move into garbage collector. New methods insert
24035 (ipa_node_params_sum): Annotate wth GTY(()).
24036 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24038 (ipa_load_from_parm_agg): Adjust declaration.
24039 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24040 * ipa-profile.c (ipa_profile): Likewise.
24041 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24042 (ipa_populate_param_decls): Make descriptors parameter garbage
24044 (ipa_dump_param): Adjust to descriptors being a pointer.
24045 (ipa_alloc_node_params): Likewise.
24046 (ipa_initialize_node_params): Likewise.
24047 (load_from_param_1): Make descriptors parameter garbage collected.
24048 (load_from_unmodified_param): Likewise.
24049 (load_from_param): Likewise.
24050 (ipa_load_from_parm_agg): Likewise.
24051 (ipa_node_params::~ipa_node_params): Removed.
24052 (ipa_free_all_node_params): Remove call to delete operator.
24053 (ipa_node_params_t::insert): New.
24054 (ipa_node_params_t::remove): Likewise.
24055 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24056 copy known_csts and known_contexts vectors.
24057 (ipa_read_node_info): Adjust to descriptors being a pointer.
24058 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24060 (ipcp_transform_function): Make descriptors variable garbage
24063 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24065 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24066 * config/i386/avx512dqintrin.h: Ditto.
24067 * config/i386/avx512fintrin.h: Ditto.
24068 * config/i386/i386.c: Handle new builtins.
24069 * config/i386/i386-builtin.def: Add new builtins.
24070 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24071 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24073 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24074 Martin Liska <mliska@suse.cz>
24076 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24077 * asan.c (asan_expand_poison_ifn): Support stores and use
24078 appropriate ASAN report function.
24079 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24080 * internal-fn.def (ASAN_POISON_USE): Declare.
24081 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24082 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24083 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24084 ASAN_POISON calls w/o LHS.
24085 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24086 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24087 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24088 * gimplify.c (asan_poison_variables): Add attribute
24089 use_after_scope_memory to variables that really needs to live
24091 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24092 having the attribute.
24094 2017-01-23 Martin Liska <mliska@suse.cz>
24096 * asan.c (create_asan_shadow_var): New function.
24097 (asan_expand_poison_ifn): Likewise.
24098 * asan.h (asan_expand_poison_ifn): New declaration.
24099 * internal-fn.c (expand_ASAN_POISON): Likewise.
24100 * internal-fn.def (ASAN_POISON): New builtin.
24101 * sanopt.c (pass_sanopt::execute): Expand
24102 asan_expand_poison_ifn.
24103 * tree-inline.c (copy_decl_for_dup_finish): Make function
24105 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24106 * tree-ssa.c (is_asan_mark_p): New function.
24107 (execute_update_addresses_taken): Rewrite local variables
24108 (identified just by use-after-scope as addressable) into SSA.
24110 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24112 * doc/install.texi (Specific): opensource.apple.com uses https
24113 now. Remove trailing slash.
24115 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24117 * README.Portability: Remove note on an Irix compatibility issue.
24119 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24121 * gcov.c (INCLUDE_ALGORITHM): Define.
24122 (INCLUDE_VECTOR): Define.
24123 No longer include <vector> and <algorithm> directly.
24125 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24127 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24129 * doc/invoke.texi (Code Gen Options): Ditto.
24131 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24134 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24136 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24138 rtl-optimization/79125
24139 * cprop.c (local_cprop_pass): Handle cases where we make an
24140 unconditional trap.
24142 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24146 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24147 read from, for big endian.
24149 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24151 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24152 register pauth builtins for LP64 only.
24154 2017-01-20 Marek Polacek <polacek@redhat.com>
24157 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24160 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24162 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24164 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24165 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24166 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24168 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24171 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24172 in big-endian mode when they are not a single duplicated value.
24174 2017-01-20 Richard Biener <rguenther@suse.de>
24176 * BASE-VER: Bump to 7.0.1.
24178 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24180 * omp-low.c (omplow_simd_context): New struct. Use it...
24181 (lower_rec_simd_input_clauses): ...here and...
24182 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24183 references to idx, lane, max_vf, is_simt.
24185 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24187 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24190 2017-01-20 Martin Jambor <mjambor@suse.cz>
24192 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24193 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24195 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24196 (GTFILES): Rename hsa.c to hsa-common.c.
24197 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24198 * hsa-dump.c: Likewise.
24199 * hsa-gen.c: Likewise.
24200 * hsa-regalloc.c: Likewise.
24201 * ipa-hsa.c: Likewise.
24202 * omp-expand.c: Likewise.
24203 * omp-low.c: Likewise.
24204 * toplev.c: Likewise.
24206 2017-01-20 Marek Polacek <polacek@redhat.com>
24209 * doc/invoke.texi: Document -Wduplicated-branches.
24210 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24211 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24212 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24213 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24214 return 0 only when not OEP_LEXICOGRAPHIC.
24215 (fold_build_cleanup_point_expr): Use the expression
24216 location when building CLEANUP_POINT_EXPR.
24217 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24218 * tree.c (add_expr): Handle error_mark_node.
24220 2017-01-20 Martin Liska <mliska@suse.cz>
24223 * tree-profile.c (init_ic_make_global_vars): Do not call
24225 (gimple_init_gcov_profiler): Likewise.
24227 2017-01-20 Martin Liska <mliska@suse.cz>
24230 * cgraph.h (maybe_create_reference): Remove argument and
24232 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24234 * ipa-cp.c (create_specialized_node): Likewise.
24235 * symtab.c (symtab_node::maybe_create_reference): Handle
24236 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24238 2017-01-20 Martin Liska <mliska@suse.cz>
24240 * read-rtl-function.c (function_reader::create_function): Use
24241 build_decl instread of build_decl_stat.
24243 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24245 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24246 * config/i386/avx512dqintrin.h: Ditto.
24247 * config/i386/avx512fintrin.h: Ditto.
24248 * config/i386/i386-builtin-types.def: Add new types.
24249 * config/i386/i386.c: Handle new types.
24250 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24251 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24252 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24253 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24254 (__builtin_ia32_kshiftridi): New.
24255 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24257 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24261 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24262 define to rs6000_init_stack_protect_guard.
24263 (rs6000_init_stack_protect_guard): New function.
24265 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24266 Yunqiang Su <yunqiang.su@imgtec.com>
24268 * config.gcc (supported_defaults): Add madd4.
24269 (with_madd4): Add validation.
24270 (all_defaults): Add madd4.
24271 * config/mips/mips.opt (mmadd4): New option.
24272 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24274 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24276 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24277 (ISA_HAS_FUSED_MADD4): Likewise.
24278 * doc/invoke.texi (-mmadd4): Document the new option.
24279 * doc/install.texi (--with-madd4): Document the new option.
24281 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24283 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24284 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24285 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24286 (aarch64_init_pauth_hint_builtins): New.
24287 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24288 (aarch64_expand_builtin): Expand new builtins.
24290 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24292 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24293 * combine-stack-adj.c (no_unhandled_cfa): Handle
24294 REG_CFA_TOGGLE_RA_MANGLE.
24295 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24296 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24297 info for return address signing.
24298 (aarch64_expand_epilogue): Likewise.
24300 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24302 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24303 * config/aarch64/aarch64-protos.h
24304 (aarch64_return_address_signing_enabled): New declaration.
24305 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24307 (aarch64_expand_prologue): Sign return address before it's pushed onto
24309 (aarch64_expand_epilogue): Authenticate return address fetched from
24311 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24312 (aarch64_attributes): New function attributes for "sign-return-address".
24313 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24314 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24315 ("*do_return"): Generate combined instructions according to key index.
24316 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24317 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24319 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24320 * config/aarch64/aarch64.opt (msign-return-address=): New.
24321 * doc/extend.texi (AArch64 Function Attributes): Documents
24322 "sign-return-address=".
24323 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24325 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24327 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24328 overall option summary.
24330 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24332 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24333 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24334 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24335 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24337 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24339 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24340 -mpower9-minmax by default for -mcpu=power9.
24341 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24342 128-bit floating point.
24344 2017-01-20 Alan Modra <amodra@gmail.com>
24346 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24347 optimizing for size.
24349 2017-01-20 Alan Modra <amodra@gmail.com>
24352 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24353 for strcmp and strncmp from corresponding builtin decl.
24355 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24357 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24358 instead of i386/rtems-64.h.
24359 * config/i386/rtems-64.h: Remove.
24361 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24365 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24367 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24369 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24371 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24372 Change int to HOST_WIDE_INT.
24373 * config/aarch64/aarch64-protos.h
24374 (aarch64_simd_gen_const_vector_dup): Likewise.
24375 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24377 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24379 * langhooks-def.h (lhd_type_for_size): New decl.
24380 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24381 * langhooks.c (lhd_type_for_size): New function, taken from
24384 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24386 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24387 define_bypass for CR latency.
24388 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24389 (power9-alu2): Add define_bypass for CR latency.
24391 (power9-mul): Update insn latency.
24392 (power9-mul-compare): Update insn latency, bypass latency and remove
24395 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24397 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24399 * config/aarch64/aarch64.md
24400 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24401 aarch64_nopcrelative_literal_loads.
24402 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24404 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24406 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24407 TARGET_LOONGSON_3A.
24408 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24410 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24413 * config.gcc (supported_defaults): Add lxc1-sxc1.
24414 (with_lxc1_sxc1): Add validation.
24415 (all_defaults): Add lxc1-sxc1.
24416 * config/mips/mips.opt (mlxc1-sxc1): New option.
24417 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24419 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24420 __mips_no_lxc1_sxc1.
24421 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24422 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24423 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24425 2017-01-19 Richard Biener <rguenther@suse.de>
24427 PR tree-optimization/72488
24428 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24429 sure to restore SSA info.
24430 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24432 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24434 PR rtl-optimization/79121
24435 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24436 of the inner type when shifting an extended value.
24438 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24441 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24442 interposable aliases.
24444 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24447 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24448 Use the evmergelohi instruction.
24449 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24450 (mov_sitf_e500_subreg8_2_be): Likewise.
24451 (mov_sitf_e500_subreg12_2_le): Likewise.
24452 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24453 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24454 (mov_sitf_e500_subreg8_2_le): Likewise.
24455 (mov_sitf_e500_subreg12_2_be): Likewise.
24457 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24459 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24460 attribute from vecsimple to vecperm.
24461 (altivec_vbpermq2): Likewise.
24463 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24466 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24468 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24469 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24470 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24471 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24472 case where N arg is SIZE_MAX.
24473 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24474 (cmpstrsi): Add pattern.
24476 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24478 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24479 __builtin_vec_revb builtins.
24480 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24481 built-in functions to support generation of the ISA 3.0 XXBR<x>
24482 vector byte reverse instructions.
24483 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24484 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24485 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24486 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24487 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24488 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24489 (P9V_BUILTIN_VEC_REVB): Likewise.
24490 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24491 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24492 (p9_xxbrq_v16qi): Likewise.
24493 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24494 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24495 (p9_xxbrh_v8hi): Likewise.
24496 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24497 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24498 vec_revb built-in functions.
24500 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24502 PR rtl-optimization/78952
24503 * config/i386/i386.md (any_extract): New code iterator.
24504 (*insvqi_2): Use any_extract for source operand.
24505 (*insvqi_3): Use any_shiftrt for source operand.
24507 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24509 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24511 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24513 2017-01-18 Matthias Klose <doko@ubuntu.com>
24515 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24517 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24519 * config/rs6000/altivec.h (vec_bperm): Change #define.
24520 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24521 (altivec_vbpermq2): New define_insn.
24522 (altivec_vbpermd): Likewise.
24523 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24524 function interface.
24525 (VBPERMD): Likewise.
24526 (VBPERM): New polymorphic function interface.
24527 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24528 Add entries for P9V_BUILTIN_VEC_VBPERM.
24529 * doc/extend.texi: Add interfaces for vec_bperm.
24531 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24533 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24534 first letter of error messages.
24535 (s390_resolve_overloaded_builtin): Likewise.
24536 * config/s390/s390.c (s390_expand_builtin): Likewise.
24537 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24538 (s390_valid_target_attribute_inner_p): Likewise.
24539 * config/s390/s390.md ("tabort"): Likewise.
24541 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24543 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24544 (ISA_AVOID_DIV_HILO): New macro.
24545 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24546 (ISA_HAS_DDIV): Likewise.
24548 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24550 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24552 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24554 * doc/invoke.texi (fabi-version): Spelling fix.
24556 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24559 * doc/invoke.texi (fabi-version): Mention mangling fix for
24562 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24565 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24567 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24570 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24571 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24573 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24575 (stack_protect_test): Ditto.
24576 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24577 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24579 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24580 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24581 -mstack-protector-guard-offset=.
24582 (RS/6000 and PowerPC Options): Ditto.
24584 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24586 * config/i386/i386.h (MASK_CLASS_P): New define.
24587 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24588 there are no registers from different register sets also when
24589 mask registers are used. Update function comment.
24590 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24591 to (*k/*r) and (*k/*km) alternatives.
24593 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24595 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24596 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24597 (EH_RETURN_HANDLER_RTX): New define.
24598 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24599 Force frame pointer in EH return functions.
24600 (aarch64_expand_epilogue): Add barrier for eh_return.
24601 (aarch64_final_eh_return_addr): Remove.
24602 (aarch64_eh_return_handler_rtx): New function.
24603 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24605 (aarch64_eh_return_handler_rtx): New prototype.
24607 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24609 * config/rs6000/altivec.h (vec_rlmi): New #define.
24610 (vec_vrlnm): Likewise.
24611 (vec_rlnm): Likewise.
24612 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24613 (UNSPEC_VRLNM): Likewise.
24614 (VIlong): New mode iterator.
24615 (altivec_vrl<VI_char>mi): New define_insn.
24616 (altivec_vrl<VI_char>nm): Likewise.
24617 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24619 (VRLDNM): Likewise.
24620 (RLNM): New polymorphic function entry.
24621 (VRLWMI): New monomorphic function entry.
24622 (VRLDMI): Likewise.
24623 (RLMI): New polymorphic function entry.
24624 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24625 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24626 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24629 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24632 * dwarf2out.c (field_byte_offset): Restore the
24633 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24634 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24635 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24638 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24641 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24643 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24646 * dwarf2out.c (add_data_member_location_attribute): For constant
24647 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24648 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24649 DW_AT_byte_size attributes.
24651 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24653 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24654 after forcing to constant memory when the code model is medium.
24656 2017-01-17 Julia Koval <julia.koval@intel.com>
24659 * config/i386/avx512fintrin.h
24660 (_mm512_i32gather_ps): Change __addr type to void const*.
24661 (_mm512_mask_i32gather_ps): Ditto.
24662 (_mm512_i32gather_pd): Ditto.
24663 (_mm512_mask_i32gather_pd): Ditto.
24664 (_mm512_i64gather_ps): Ditto.
24665 (_mm512_mask_i64gather_ps): Ditto.
24666 (_mm512_i64gather_pd): Ditto.
24667 (_mm512_mask_i64gather_pd): Ditto.
24668 (_mm512_i32gather_epi32): Ditto.
24669 (_mm512_mask_i32gather_epi32): Ditto.
24670 (_mm512_i32gather_epi64): Ditto.
24671 (_mm512_mask_i32gather_epi64): Ditto.
24672 (_mm512_i64gather_epi32): Ditto.
24673 (_mm512_mask_i64gather_epi32): Ditto.
24674 (_mm512_i64gather_epi64): Ditto.
24675 (_mm512_mask_i64gather_epi64): Ditto.
24676 (_mm512_i32scatter_ps): Change __addr type to void*.
24677 (_mm512_mask_i32scatter_ps): Ditto.
24678 (_mm512_i32scatter_pd): Ditto.
24679 (_mm512_mask_i32scatter_pd): Ditto.
24680 (_mm512_i64scatter_ps): Ditto.
24681 (_mm512_mask_i64scatter_ps): Ditto.
24682 (_mm512_i64scatter_pd): Ditto.
24683 (_mm512_mask_i64scatter_pd): Ditto.
24684 (_mm512_i32scatter_epi32): Ditto.
24685 (_mm512_mask_i32scatter_epi32): Ditto.
24686 (_mm512_i32scatter_epi64): Ditto.
24687 (_mm512_mask_i32scatter_epi64): Ditto.
24688 (_mm512_i64scatter_epi32): Ditto.
24689 (_mm512_mask_i64scatter_epi32): Ditto.
24690 (_mm512_i64scatter_epi64): Ditto.
24691 (_mm512_mask_i64scatter_epi64): Ditto.
24692 * config/i386/avx512pfintrin.h
24693 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24694 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24695 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24696 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24697 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24698 (_mm512_prefetch_i32scatter_ps): Ditto.
24699 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24700 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24701 (_mm512_prefetch_i64scatter_pd): Ditto.
24702 (_mm512_prefetch_i64scatter_ps): Ditto.
24703 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24704 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24705 * config/i386/avx512vlintrin.h
24706 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24707 (_mm_mmask_i32gather_ps): Ditto.
24708 (_mm256_mmask_i32gather_pd): Ditto.
24709 (_mm_mmask_i32gather_pd): Ditto.
24710 (_mm256_mmask_i64gather_ps): Ditto.
24711 (_mm_mmask_i64gather_ps): Ditto.
24712 (_mm256_mmask_i64gather_pd): Ditto.
24713 (_mm_mmask_i64gather_pd): Ditto.
24714 (_mm256_mmask_i32gather_epi32): Ditto.
24715 (_mm_mmask_i32gather_epi32): Ditto.
24716 (_mm256_mmask_i32gather_epi64): Ditto.
24717 (_mm_mmask_i32gather_epi64): Ditto.
24718 (_mm256_mmask_i64gather_epi32): Ditto.
24719 (_mm_mmask_i64gather_epi32): Ditto.
24720 (_mm256_mmask_i64gather_epi64): Ditto.
24721 (_mm_mmask_i64gather_epi64): Ditto.
24722 (_mm256_i32scatter_ps): Change __addr type to void*.
24723 (_mm256_mask_i32scatter_ps): Ditto.
24724 (_mm_i32scatter_ps): Ditto.
24725 (_mm_mask_i32scatter_ps): Ditto.
24726 (_mm256_i32scatter_pd): Ditto.
24727 (_mm256_mask_i32scatter_pd): Ditto.
24728 (_mm_i32scatter_pd): Ditto.
24729 (_mm_mask_i32scatter_pd): Ditto.
24730 (_mm256_i64scatter_ps): Ditto.
24731 (_mm256_mask_i64scatter_ps): Ditto.
24732 (_mm_i64scatter_ps): Ditto.
24733 (_mm_mask_i64scatter_ps): Ditto.
24734 (_mm256_i64scatter_pd): Ditto.
24735 (_mm256_mask_i64scatter_pd): Ditto.
24736 (_mm_i64scatter_pd): Ditto.
24737 (_mm_mask_i64scatter_pd): Ditto.
24738 (_mm256_i32scatter_epi32): Ditto.
24739 (_mm256_mask_i32scatter_epi32): Ditto.
24740 (_mm_i32scatter_epi32): Ditto.
24741 (_mm_mask_i32scatter_epi32): Ditto.
24742 (_mm256_i32scatter_epi64): Ditto.
24743 (_mm256_mask_i32scatter_epi64): Ditto.
24744 (_mm_i32scatter_epi64): Ditto.
24745 (_mm_mask_i32scatter_epi64): Ditto.
24746 (_mm256_i64scatter_epi32): Ditto.
24747 (_mm256_mask_i64scatter_epi32): Ditto.
24748 (_mm_i64scatter_epi32): Ditto.
24749 (_mm_mask_i64scatter_epi32): Ditto.
24750 (_mm256_i64scatter_epi64): Ditto.
24751 (_mm256_mask_i64scatter_epi64): Ditto.
24752 (_mm_i64scatter_epi64): Ditto.
24753 (_mm_mask_i64scatter_epi64): Ditto.
24754 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24755 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24756 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24757 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24758 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24759 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24760 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24761 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24762 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24763 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24764 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24765 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24766 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24767 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24768 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24769 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24770 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24771 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24772 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24773 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24774 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24775 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24776 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24777 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24778 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24779 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24780 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24781 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24782 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24783 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24784 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24785 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24786 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24787 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24788 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24789 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24790 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24791 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24792 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24793 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24794 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24795 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24796 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24797 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24798 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24799 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24800 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24801 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24802 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24803 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24804 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24805 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24806 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24807 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24808 definitions accordingly.
24810 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24811 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24814 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24817 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24820 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24821 hard regs for inner regno.
24823 2017-01-17 Martin Liska <mliska@suse.cz>
24826 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24827 assumption and add comment.
24829 2017-01-17 Nathan Sidwell <nathan@acm.org>
24831 * ipa-visibility.c (localize_node): New function, broken out of ...
24832 (function_and_variable_visibility): ... here. Call it.
24834 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24836 PR middle-end/77445
24837 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24838 correctly set frequency of oudgoing edge.
24839 (duplicate_thread_path): Fix profile updating.
24841 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24844 * configure.ac: Add GCC_BASE_VER.
24845 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24846 version from BASE-VER file.
24847 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24848 (gcc.o): Depend on $(BASEVER).
24849 * common.opt (dumpfullversion): New option.
24850 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24851 * doc/invoke.texi: Document -dumpfullversion.
24852 * doc/install.texi: Document --with-gcc-major-version-only.
24853 * configure: Regenerated.
24855 2017-01-17 Richard Biener <rguenther@suse.de>
24857 PR tree-optimization/71433
24858 * tree-vrp.c (register_new_assert_for): Merge same asserts
24859 on all incoming edges.
24860 (process_assert_insertions_for): Handle insertions at the
24863 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24865 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24866 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24868 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24871 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24874 2017-01-17 Alan Modra <amodra@gmail.com>
24877 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24878 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24879 symbolic stack limit when pic.
24881 2017-01-16 Martin Sebor <msebor@redhat.com>
24883 PR tree-optimization/78608
24884 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24886 2017-01-16 Jeff Law <law@redhat.com>
24889 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24890 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24891 for several include directories that may be relative to sysroot.
24892 * config/i386/x-mingw32 (gplus_includedir): Define.
24893 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24894 (native_system_includedir): Likewise.
24895 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24896 override if TARGET_SYSTEM_ROOT is defined.
24897 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24899 PR tree-optimization/79090
24900 PR tree-optimization/33562
24901 PR tree-optimization/61912
24902 PR tree-optimization/77485
24903 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24904 and computed trims into the dump file.
24906 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24908 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24910 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24913 * gimplify.c (gimplify_init_constructor): If want_value and
24914 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24918 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24919 sequence. Formatting fixes.
24920 (doloop_optimize): Formatting fixes.
24923 * gcc.c (debug_level_greater_than_spec_func): New function.
24924 (static_spec_functions): Add debug-level-gt spec function.
24925 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24927 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24928 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24929 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24930 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24931 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24932 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24934 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24936 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24937 QImode fixups to general and mask registers only.
24939 2017-01-16 Carl Love <cel@us.ibm.com>
24941 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24942 for built-in functions
24943 vector signed char vec_nabs (vector signed char)
24944 vector signed short vec_nabs (vector signed short)
24945 vector signed int vec_nabs (vector signed int)
24946 vector signed long long vec_nabs (vector signed long long)
24947 vector float vec_nabs (vector float)
24948 vector double vec_nabs (vector double)
24949 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24951 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24952 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24953 * doc/extend.texi: Update the documentation file for the new built-in
24956 2017-01-16 Martin Sebor <msebor@redhat.com>
24958 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24961 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24963 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24964 UNSPEC_VSX__XXSPLTD to require special splat handling.
24966 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24969 * system.h: Poison strndup.
24971 2017-01-16 Alan Modra <amodra@gmail.com>
24974 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24977 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24979 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24981 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24983 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24984 call recog here. Assert that INSN_CODE (insn) is non-negative.
24986 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24989 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24991 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24992 in the currently scheduled RTL fragment.
24994 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24996 PR rtl-optimization/78751
24997 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25000 2017-01-14 Jeff Law <law@redhat.com>
25002 PR tree-optimization/79090
25003 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25004 variable length stores.
25005 (compute_trims): Delete dead assignment to *trim_tail.
25006 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25009 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25011 PR rtl-optimization/78626
25012 PR rtl-optimization/78727
25013 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25014 of a block, and split such blocks after everything else is finished.
25016 2017-01-14 Alan Modra <amodra@gmail.com>
25019 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25020 target legitimate_combined_insn.
25021 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25022 (rs6000_legitimate_combined_insn): New function.
25023 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25025 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25026 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25027 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25029 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25031 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25033 2017-01-13 Jeff Law <law@redhat.com>
25035 PR tree-optimization/33562
25036 PR tree-optimization/61912
25037 PR tree-optimization/77485
25038 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25040 (delete_dead_assignment): Likewise.
25041 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25042 statement to delete_dead_call and delete_dead_assignment.
25044 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25047 * substring-locations.c (format_warning_va): Strengthen case 1 so
25048 that both endpoints of the substring must be within the format
25049 range for just the substring to be printed.
25051 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25053 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25054 * config/i386/i386.c (ix86_target_string): Add missing options
25055 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25056 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25057 flags_other and ix86_target_other to flags2_other. Display unknown
25059 (ix86_valid_target_attribute_inner_p): Add missing options and
25060 reorder options by implied ISAs, as in ix86_target_string.
25062 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25064 * hash-table.h (hash_table::too_empty_p): New function.
25065 (hash_table::expand): Use it.
25066 (hash_table::traverse): Likewise.
25067 (hash_table::empty_slot): Use sizeof (value_type) instead of
25068 sizeof (PTR) to convert bytes to elements. Shrink the table
25069 if the current size is excessive for the current number of
25072 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25074 * ira-costs.c (record_reg_classes): Break from the inner loop
25075 early once alt_fail is known to be true. Update outer loop
25076 handling accordingly.
25078 2017-01-13 Jeff Law <law@redhat.com>
25080 * tree-ssa-dse.c (decrement_count): New function.
25081 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25082 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25083 when we know the partially dead statement is a mem* function.
25085 PR tree-optimization/61912
25086 PR tree-optimization/77485
25087 * tree-ssa-dse.c: Include expr.h.
25088 (maybe_trim_constructor_store): New function.
25089 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25091 PR tree-optimization/33562
25092 PR tree-optimization/61912
25093 PR tree-optimization/77485
25094 * doc/invoke.texi: Document new dse-max-object-size param.
25095 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25096 * tree-ssa-dse.c: Include params.h.
25097 (dse_store_status): New enum.
25098 (initialize_ao_ref_for_dse): New, partially extracted from
25100 (valid_ao_ref_for_dse, normalize_ref): New.
25101 (setup_live_bytes_from_ref, compute_trims): Likewise.
25102 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25103 (maybe_trim_partially_dead_store): Likewise.
25104 (maybe_trim_complex_store): Likewise.
25105 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25106 Track what bytes live from the original store. Return tri-state
25107 for dead, partially dead or live.
25108 (dse_dom_walker): Add constructor, destructor and new private members.
25109 (delete_dead_call, delete_dead_assignment): New extracted from
25111 (dse_optimize_stmt): Make a member of dse_dom_walker.
25112 Use initialize_ao_ref_for_dse.
25114 PR tree-optimization/33562
25115 PR tree-optimization/61912
25116 PR tree-optimization/77485
25117 * sbitmap.h (bitmap_count_bits): Prototype.
25118 (bitmap_clear_range, bitmap_set_range): Likewise.
25119 * sbitmap.c (bitmap_clear_range): New function.
25120 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25122 2017-01-13 Martin Liska <mliska@suse.cz>
25125 * function.c (set_cfun): Add new argument force.
25126 * function.h (set_cfun): Likewise.
25127 * ipa-inline-transform.c (inline_call): Use the function when
25128 strict alising from is dropped for function we inline to.
25130 2017-01-13 Richard Biener <rguenther@suse.de>
25132 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25133 for dumping GIMPLE INTEGER_CSTs.
25135 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25137 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25138 to 201112L since C++17.
25140 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25143 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25144 if -fsanitize=kernel-address is present.
25146 2017-01-13 Richard Biener <rguenther@suse.de>
25148 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25149 as _Literal ( type ) number in case usual suffixes do not
25150 preserve all information.
25152 2017-01-13 Richard Biener <rguenther@suse.de>
25154 PR tree-optimization/77283
25155 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25156 and ssa-iterators.h.
25157 (is_feasible_trace): Implement a cost model based on joiner
25160 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25163 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25164 char or short to __float128/_Float128 directly.
25166 2017-01-12 Martin Sebor <msebor@redhat.com>
25168 to -Wformat-overflow.
25169 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25170 (min_bytes_remaining): Same.
25171 (get_string_length): Same.
25172 (format_string): Same.
25173 (format_directive): Same.
25175 (pass_sprintf_length::handle_gimple_call): Same.
25177 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25179 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25180 info.nowrite calls with no lhs that can't throw. Return bool
25181 whether gsi_remove has been called or not.
25182 (pass_sprintf_length::handle_gimple_call): Return bool whether
25183 try_substitute_return_value called gsi_remove. Formatting fix.
25184 (pass_sprintf_length::execute): Don't use gsi_remove if
25185 handle_gimple_call returned true.
25188 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25189 be removed due to side-effects, don't remove following barrier nor
25190 turn the successor edge into fallthru edge.
25192 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25195 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25196 element-reversing loads and stores as not swappable.
25198 2017-01-12 Nathan Sidwell <nathan@acm.org>
25199 Nicolai Stange <nicstange@gmail.com>
25201 * combine.c (try_combine): Don't ignore result of overlap checking
25202 loop. Combine overlap & asm check into single loop.
25204 2017-01-12 Richard Biener <rguenther@suse.de>
25206 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25207 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25209 2017-01-12 Richard Biener <rguenther@suse.de>
25211 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25212 and TS_TARGET_OPTION directly derive from TS_BASE.
25213 * tree-core.h (tree_optimization_option): Derive from tree_base.
25214 (tree_target_option): Likewise.
25216 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25218 * config/i386/i386.c (memory_address_length): Increase len
25219 only when rip_relative_addr_p returns false.
25221 2017-01-11 Julia Koval <julia.koval@intel.com>
25223 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25224 (OPTION_MASK_ISA_SGX_SET): New.
25225 (ix86_handle_option): Handle OPT_msgx.
25226 * config.gcc: Added sgxintrin.h.
25227 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25228 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25229 * config/i386/i386.c (ix86_target_string): Add -msgx.
25231 (ix86_option_override_internal): Handle new options.
25232 (ix86_valid_target_attribute_inner_p): Add sgx.
25233 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25234 * config/i386/i386.opt: Add msgx.
25235 * config/i386/sgxintrin.h: New file.
25236 * config/i386/x86intrin.h: Add sgxintrin.h.
25238 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25241 * fold-const.c (maybe_nonzero_address): Return 1 for function
25243 (tree_single_nonzero_warnv_p): Don't handle function local objects
25247 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25250 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25253 * opts.c: Include "spellcheck.h"
25254 (struct string_fragment): New struct.
25255 (struct edit_distance_traits<const string_fragment &>): New
25257 (get_closest_sanitizer_option): New function.
25258 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25260 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25262 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25264 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25265 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25266 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25267 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25268 for initial die_offset if dwarf_split_debug_info.
25269 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25270 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25272 (output_skeleton_debug_sections): Formatting fix. Use
25273 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25274 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25276 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25278 * config/arm/cortex-a53.md: Add bypasses for
25279 cortex_a53_r2f_cvt.
25280 (cortex_a53_r2f): Only use for transfers.
25281 (cortex_a53_f2r): Likewise.
25282 (cortex_a53_r2f_cvt): Add reservation for conversions.
25283 (cortex_a53_f2r_cvt): Likewise.
25285 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25287 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25288 to all inlined functions, change static to extern.
25290 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25293 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25295 (arm_assemble_integer): Likewise.
25297 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25299 * config.gcc: Use new awk script to check CPU, FPU and architecture
25300 parameters for --with-... options.
25301 * config/arm/parsecpu.awk: New file
25302 * config/arm/arm-cpus.in: New file.
25303 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25305 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25307 * config/arm/t-arm: Update dependency rules.
25308 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25309 of processing .def files.
25310 * config/arm/genopt.sh: Deleted.
25311 * config/arm/gentune.sh: Deleted.
25312 * config/arm/arm-cores.def: Deleted.
25313 * config/arm/arm-arches.def: Deleted.
25314 * config/arm/arm-fpus.def: Deleted.
25315 * config/arm/arm-tune.md: Regenerated.
25316 * config/arm/arm-tables.opt: Regenerated.
25317 * config/arm/arm-cpu.h: New generated file.
25318 * config/arm/arm-cpu-data.h: New generated file.
25319 * config/arm/arm-cpu-cdata.h: New generated file.
25321 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25324 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25326 (input_varpool_node): Unpack dynamically_initialized bit.
25328 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25330 PR rtl-optimization/79032
25331 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25332 the alignment of the adjusted memory reference against that of MODE,
25333 instead of the alignment of the original memory reference.
25335 2017-01-11 Martin Jambor <mjambor@suse.cz>
25337 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25339 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25340 decorated functions.
25342 2017-01-11 Richard Biener <rguenther@suse.de>
25344 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25345 set range/nonnull info for PHI results. Do not set it on
25346 stmts marked for removal.
25348 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25350 * expr.c (store_field): In the bitfield case, fetch the return value
25351 from the registers before applying a single big-endian adjustment.
25352 Always do a final load for a BLKmode value not larger than a word.
25354 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25357 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25358 that we correctly handle column numbers greater than
25359 LINE_MAP_MAX_COLUMN_NUMBER.
25361 2017-01-10 Martin Sebor <msebor@redhat.com>
25363 PR middle-end/78245
25364 * gimple-ssa-sprintf.c (get_destination_size): Call
25365 {init,fini}object_sizes.
25366 * tree-object-size.c (addr_object_size): Adjust.
25367 (pass_through_call): Adjust.
25368 (pass_object_sizes::execute): Adjust.
25369 * tree-object-size.h (fini_object_sizes): Declare.
25371 2017-01-10 Martin Sebor <msebor@redhat.com>
25373 PR tree-optimization/78775
25374 * builtins.c (get_size_range): Move...
25375 * calls.c: ...to here.
25376 (alloc_max_size): Accept zero argument.
25377 (operand_signed_p): Remove.
25378 (maybe_warn_alloc_args_overflow): Call get_size_range.
25379 * calls.h (get_size_range): Declare.
25381 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25383 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25384 from TI's devices.csv file as of September 2016.
25385 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25387 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25389 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25390 * doc/invoke.texi: Likewise.
25391 * doc/md.texi: Likewise.
25392 * doc/objc.texi: Likewise.
25394 2017-01-10 Joshua Conner <joshconner@google.com>
25396 * config/arm/fuchsia-elf.h: New file.
25397 * config/fuchsia.h: New file.
25398 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25399 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25401 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25403 2016-01-10 Richard Biener <rguenther@suse.de>
25405 PR tree-optimization/79034
25406 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25407 Propagate out degenerate PHIs in the joiner.
25409 2017-01-10 Martin Liska <mliska@suse.cz>
25411 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25412 (sort_congruence_classes_by_decl_uid): Likewise.
25413 (sort_congruence_class_groups_by_decl_uid): Likewise.
25414 (sem_item_optimizer::merge_classes): Sort class, groups in these
25415 classes and members in the groups by DECL_UID of declarations.
25416 This would make merge operations stable.
25418 2017-01-10 Martin Liska <mliska@suse.cz>
25420 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25421 usage of m_classes_vec.
25422 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25423 (sem_item_optimizer::get_group_by_hash): Likewise.
25424 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25425 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25426 (sem_item_optimizer::verify_classes): Likewise.
25427 (sem_item_optimizer::process_cong_reduction): Likewise.
25428 (sem_item_optimizer::dump_cong_classes): Likewise.
25429 (sem_item_optimizer::merge_classes): Likewise.
25430 * ipa-icf.h (congruence_class_hash): Rename from
25431 congruence_class_group_hash. Remove declaration of m_classes_vec.
25433 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25435 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25436 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25437 * config.gcc: Add avx512vpopcntdqintrin.h.
25438 * config/i386/avx512vpopcntdqintrin.h: New.
25439 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25440 * config/i386/i386-builtin-types.def: Add new types.
25441 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25442 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25443 __builtin_ia32_vpopcountq_v8di_mask): New.
25444 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25445 __AVX512VPOPCNTDQ__.
25446 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25447 (PTA_AVX512VPOPCNTDQ): Define.
25448 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25449 TARGET_AVX512VPOPCNTDQ_P): Define.
25450 * config/i386/i386.opt: Add mavx512vpopcntdq.
25451 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25452 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25454 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25456 PR middle-end/77484
25457 * predict.def (PRED_CALL): Set to 67.
25459 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25461 * expr.c (store_field): In the bitfield case, if the value comes from
25462 a function call and is of an aggregate type returned in registers, do
25463 not modify the field mode; extract the value in all cases if the mode
25464 is BLKmode and the size is not larger than a word.
25466 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25469 * config/i386/cpuid.h: Fix undefined behavior.
25471 2017-01-04 Jeff Law <law@redhat.com>
25473 PR tree-optimization/79007
25474 PR tree-optimization/67955
25475 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25476 conservative for pt.null when flag_non_call_exceptions is on.
25478 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25480 PR translation/79019
25481 PR translation/79020
25482 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25483 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25484 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25486 * config/avr/avr.opt (maccumulate-args): Likewise.
25487 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25488 * common.opt (freport-bug): Likewise.
25489 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25490 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25491 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25492 translatable string.
25493 * config/i386/i386.c (function_value_32): Likewise.
25494 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25495 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25497 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25498 * common/config/msp430/msp430-common.c (msp430_handle_option):
25500 * symtab.c (symtab_node::verify_base): Likewise.
25501 * opts.c (set_debug_level): Likewise.
25502 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25503 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25504 missing whitespace to translatable strings.
25505 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25506 * config/sh/superh.h: Likewise.
25507 * config/i386/xopintrin.h: Likewise.
25508 * config/i386/znver1.md: Likewise.
25509 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25510 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25511 * double-int.h (struct double_int): Likewise.
25512 * double-int.c (div_and_round_double): Likewise.
25513 * wide-int.cc: Likewise.
25514 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25515 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25516 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25517 (crossjumps_occurred): ... this.
25518 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25521 PR tree-optimization/78899
25522 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25523 returning bool return struct loop *, NULL for failure and the new
25525 (versionable_outer_loop_p): Don't version outer loop if it has
25526 dont_vectorized bit set.
25527 (tree_if_conversion): When versioning outer loop, ensure
25528 tree_if_conversion is performed also on the inner loop of the
25529 non-vectorizable outer loop copy.
25530 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25531 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25532 prevent vectorization of it.
25533 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25534 the outer loop vectorization of the non-scalar version is attempted
25535 before vectorization of the inner loop in scalar version. If
25536 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25537 vectorization of its inner loop.
25538 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25539 has 2 inner loops, rename also on edges from bb whose single pred
25540 is outer_loop->header. Fix typo in function comment.
25542 2017-01-09 Martin Sebor <msebor@redhat.com>
25545 * asan.c (asan_emit_stack_protection): Increase local buffer size
25546 to avoid snprintf truncation warning.
25548 2017-01-09 Andrew Pinski <apinski@cavium.com>
25550 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25551 to reference thunderx2t99 for the tuning structure
25552 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25554 (thunderx2t99_extra_costs): This.
25555 * config/aarch64/aarch64-tune.md: Regenerate.
25556 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25557 (vulcan_addrcost_table): This.
25558 (vulcan_regmove_cost): Rename to ...
25559 (thunderx2t99_regmove_cost): This.
25560 (vulcan_vector_cost): Rename to ...
25561 (thunderx2t99_vector_cost): this.
25562 (vulcan_branch_cost): Rename to ...
25563 (thunderx2t99_branch_cost): This.
25564 (vulcan_tunings): Rename to ...
25565 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25566 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25568 2017-01-09 Martin Jambor <mjambor@suse.cz>
25572 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25573 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25574 (propagate_vr_accross_jump_function): Use the above function for all
25575 value range computations for pass-through jump functions and type
25576 converasion from explicit value range values.
25577 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25578 parameters from TYPE_ARG_TYPES.
25579 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25580 (ipa_write_node_info): Stream type of the actual argument.
25581 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25583 2017-01-09 Martin Liska <mliska@suse.cz>
25586 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25587 (lookup_compiler): Do not show error message with have_E.
25589 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25591 PR tree-optimization/78938
25592 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25593 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25594 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25595 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25598 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25600 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25603 2017-01-09 Richard Biener <rguenther@suse.de>
25605 PR tree-optimization/78997
25606 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25607 name condition properly.
25609 2017-01-09 Richard Biener <rguenther@suse.de>
25612 * dwarf2out.c (is_cxx): New overload with context.
25613 (is_naming_typedef_decl): Use it.
25615 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25617 * invoke.texi (Option Summary): Correct spacing in option lists
25618 and add line breaks to fix over-long lines.
25620 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25622 PR middle-end/17660
25624 * extend.texi (Common Variable Attributes): Add xref to GCC
25625 Internals manual to explain mode attribute keywords.
25627 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25630 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25631 and Preprocessor Options.
25632 (Options for Linking): Document -pthread here....
25633 (RS/6000 and PowerPC Options): ...not here.
25634 (Solaris 2 Options): ...or here.
25635 * doc/cppopts.texi: Document -pthread.
25637 2017-01-08 Martin Sebor <msebor@redhat.com>
25639 PR middle-end/77708
25640 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25641 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25642 New member functions.
25643 (format_directive): Used them.
25645 (pass_sprintf_length::handle_gimple_call): Same.
25646 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25647 to avoid truncation for any argument.
25648 (extract_affine_mul): Same.
25649 * tree.c (get_file_function_name): Same.
25651 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25653 PR middle-end/77484
25654 * predict.def (PRED_INDIR_CALL): Set to 86.
25656 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25658 PR preprocessor/54124
25659 * doc/cppopts.texi: Reformat -d subtable to list the full name
25660 of the options. Add cross-reference to the docs for the general
25661 compiler -d options.
25662 * doc/invoke.texi (Developer Options): Add cross-reference to the
25663 preprocessor-specific -d option documentation.
25665 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25667 PR preprocessor/13498
25668 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25669 redudant material, and reflect new command-line options.
25670 (System Headers): Likewise.
25672 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25674 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25675 -isystem, and -idirafter. Copy-edit.
25676 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25677 default for -ftrack-macro-expansion. Delete obsolete and
25678 badly-formatted implementation details about -fdebug-cpp output.
25679 * doc/cppwarnopts.texi: Copy-edit.
25681 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25684 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25685 that the transition from a max line width >= 1<<10 to narrower
25686 lines works correctly.
25688 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25690 * doc/options.texi (PerFunction): New.
25691 * opt-functions.awk (switch_flags): Map both Optimization and
25692 PerFunction to CL_OPTIMIZATION.
25693 * opth-gen.awk: Test for PerFunction flag along with
25695 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25696 it only when the latter is present. Skip those that don't in
25697 the hash function generator.
25698 * common.opt (fvar-tracking): Mark as PerFunction instead of
25700 (fvar-tracking-assignments): Likewise.
25701 (fvar-tracking-assignments-toggle): Likewise.
25702 (fvar-tracking-uninit): Likewise.
25704 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25706 PR translation/79018
25707 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25710 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25713 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25714 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25715 TARGET_LONG_JUMP_TABLE_OFFSETS.
25716 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25717 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25718 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25719 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25720 * config/m68k/m68k.md (tablejump expander): Likewise.
25721 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25722 TARGET_LONG_JUMP_TABLE_OFFSETS.
25723 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25724 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25726 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25727 David Holsgrove <david.holsgrove@xilinx.com>
25729 * common/config/microblaze/microblaze-common.c
25730 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25731 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25733 * config/microblaze/microblaze.c (microblaze_must_save_register)
25734 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25736 (microblaze_eh_return): New function.
25737 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25738 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25739 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25740 * config/microblaze/microblaze.md (eh_return): New pattern.
25742 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25744 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25745 GCC_DIAGNOSTIC_STRINGIFY): Define.
25747 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25749 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25751 * config/arm/arm.md (<mcrr>): New.
25753 * config/arm/arm.c (arm_arch5te): New.
25754 (arm_option_override): Set arm_arch5te.
25755 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25757 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25758 (arm_mcrr_qualifiers): ... this. New.
25759 (MRRC_QUALIFIERS): Define to...
25760 (arm_mrrc_qualifiers): ... this. New.
25761 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25763 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25764 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25765 (MRRCI, mrrc, MRRC): New.
25766 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25767 VUNSPEC_MRRC2): New.
25769 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25771 * config/arm/arm.md (<mcr>): New.
25773 * config/arm/arm.c (arm_coproc_builtin_available): Add
25774 support for mcr, mrc, mcr2 and mrc2.
25775 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25776 (arm_mcr_qualifiers): ... this. New.
25777 (MRC_QUALIFIERS): Define to ...
25778 (arm_mrc_qualifiers): ... this. New.
25779 (MCR_QUALIFIERS): Define to ...
25780 (arm_mcr_qualifiers): ... this. New.
25781 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25783 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25784 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25785 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25786 VUNSPEC_MRC2): New.
25788 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25790 * config/arm/arm.md (*ldc): New.
25794 * config/arm/arm.c (arm_coproc_builtin_available): Add
25795 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25796 (arm_coproc_ldc_stc_legitimate_address): New.
25797 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25798 'qualifier_const_pointer'.
25799 (LDC_QUALIFIERS): Define to...
25800 (arm_ldc_qualifiers): ... this. New.
25801 (STC_QUALIFIERS): Define to...
25802 (arm_stc_qualifiers): ... this. New.
25803 * config/arm/arm-protos.h
25804 (arm_coproc_ldc_stc_legitimate_address): New.
25805 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25806 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25807 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25808 stc2, stcl, stc2l): New.
25809 * config/arm/constraints.md (Uz): New.
25810 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25811 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25812 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25813 VUNSPEC_STC2L): New.
25815 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25817 * config/arm/arm.md (<cdp>): New.
25818 * config/arm/arm.c (neon_const_bounds): Rename this ...
25819 (arm_const_bounds): ... this.
25820 (arm_coproc_builtin_available): New.
25821 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25822 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25823 (CDP_QUALIFIERS): Define to...
25824 (arm_cdp_qualifiers): ... this. New.
25826 (arm_expand_builtin_args): Add case for 6 arguments.
25827 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25828 (arm_const_bounds): ... this.
25829 (arm_coproc_builtin_available): New.
25830 * config/arm/arm_acle.h (__arm_cdp): New.
25832 * config/arm/arm_acle_builtins.def (cdp): New.
25834 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25835 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25836 'arm_const_bounds'.
25837 * config/arm/types.md (coproc): New.
25838 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25839 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25840 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25841 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25843 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25845 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25846 (UBINOP_QUALIFIERS): New.
25848 (acle_builtin_data): New. Change comment.
25849 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25850 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25851 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25852 arm_acle_builtins.def.
25853 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25854 (arm_init_acle_builtins): New.
25855 (CRC32_BUILTIN): Remove.
25856 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25857 crc32cb, crc32ch and crc32cw.
25858 (arm_init_crc32_builtins): Remove.
25859 (arm_init_builtins): Use arm_init_acle_builtins rather
25860 than arm_init_crc32_builtins.
25861 (arm_expand_acle_builtin): New.
25862 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25863 * config/arm/arm_acle_builtins.def: New.
25865 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25867 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25868 (arm_builtin_datum): ... this.
25869 (arm_init_neon_builtin): Rename to ...
25870 (arm_init_builtin): ... this. Add a new parameters PREFIX
25871 and USE_SIG_IN_NAME.
25872 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25873 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25874 'arm_builtin_datum'.
25875 (arm_init_vfp_builtins): Likewise.
25876 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25877 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25878 (arm_expand_neon_args): Rename to ...
25879 (arm_expand_builtin_args): ... this. Rename builtin_arg
25880 enum values and differentiate between ARG_BUILTIN_MEMORY
25881 and ARG_BUILTIN_NEON_MEMORY.
25882 (arm_expand_neon_builtin_1): Rename to ...
25883 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25884 values, arm_expand_builtin_args and add bool parameter NEON.
25885 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25886 (arm_expand_vfp_builtin): Likewise.
25887 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25889 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25891 PR middle-end/77484
25892 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25893 * predict.c (tree_estimate_probability_bb): Reverse direction of
25894 polymorphic call predictor.
25896 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25898 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25899 (determine_pass_name_match): ...this new function and...
25900 (should_skip_pass_p): ...this new function.
25902 2017-01-06 Nathan Sidwell <nathan@acm.org>
25904 * ipa-visibility.c (function_and_variable_visibility): Reformat
25905 comments and long lines. Remove extrneous if.
25906 * symtab.c (symtab_node::make_decl_local): Fix code format.
25907 (symtab_node::set_section_for_node): Fix comment typo.
25909 2017-01-06 Martin Liska <mliska@suse.cz>
25912 * lra-constraints.c: Rename invariant to lra_invariant.
25913 * predict.c (set_even_probabilities): Initialize e to NULL.
25915 2017-01-05 Martin Sebor <msebor@redhat.com>
25917 PR tree-optimization/78910
25918 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25919 (format_integer): Correct off-by-one error in the handling
25920 of precision with negative numbers in signed conversions..
25922 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25924 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25926 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25928 PR tree-optimization/71016
25929 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25930 factor_out_conditional_conversion. Formatting fix.
25931 (factor_out_conditional_conversion): Add cond_stmt argument.
25932 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25933 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25936 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25938 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25939 read-rtl-function.o, and selftest-rtl.o.
25940 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25941 (selftest::aarch64_test_loading_full_dump): New function.
25942 (selftest::aarch64_run_selftests): New function.
25943 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25944 selftest::aarch64_run_selftests.
25945 * config/i386/i386.c
25946 (selftest::ix86_test_loading_dump_fragment_1): New function.
25947 (selftest::ix86_test_loading_call_insn): New function.
25948 (selftest::ix86_test_loading_full_dump): New function.
25949 (selftest::ix86_test_loading_unspec): New function.
25950 (selftest::ix86_run_selftests): Call the new functions.
25951 * emit-rtl.c (maybe_set_max_label_num): New function.
25952 * emit-rtl.h (maybe_set_max_label_num): New decl.
25953 * function.c (instantiate_decls): Guard call to
25954 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25955 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25957 * gensupport.c (gen_reader::gen_reader): Pass "false"
25958 for new "compact" param of rtx_reader.
25959 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25960 rather than an empty string for NULL strings.
25961 * read-md.c: Potentially include config.h rather than bconfig.h.
25962 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25963 (have_error): New global, copied from errors.c.
25964 (md_reader::read_name): Rename to...
25965 (md_reader::read_name_1): ...this, adding "out_loc" param,
25966 and converting "missing name or number" to returning false, rather
25968 (md_reader::read_name): Reimplement in terms of read_name_1.
25969 (md_reader::read_name_or_nil): New function.
25970 (md_reader::read_string): Handle "(nil)" by returning NULL.
25971 (md_reader::md_reader): Add new param "compact".
25972 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25973 (md_reader::read_file): New method.
25974 * read-md.h (md_reader::md_reader): Add new param "compact".
25975 (md_reader::read_file): New method.
25976 (md_reader::is_compact): New accessor.
25977 (md_reader::read_name): Convert return type from void to file_location.
25978 (md_reader::read_name_or_nil): New decl.
25979 (md_reader::read_name_1): New decl.
25980 (md_reader::m_compact): New field.
25981 (noop_reader::noop_reader): Pass "false" for new "compact" param
25983 (rtx_reader::rtx_reader): Add new "compact" param.
25984 (rtx_reader::read_rtx_operand): Make virtual and convert return
25985 type from void to rtx.
25986 (rtx_reader::read_until): New decl.
25987 (rtx_reader::handle_any_trailing_information): New virtual function.
25988 (rtx_reader::postprocess): New virtual function.
25989 (rtx_reader::finalize_string): New virtual function.
25990 (rtx_reader::m_in_call_function_usage): New field.
25991 (rtx_reader::m_reuse_rtx_by_id): New field.
25992 * read-rtl-function.c: New file.
25993 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25994 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25995 (selftest::verify_three_block_rtl_cfg): New decl.
25996 * read-rtl-function.h: New file.
25997 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25998 For host, include function.h, memmodel.h, and emit-rtl.h.
25999 (one_time_initialization): New function.
26000 (struct compact_insn_name): New struct.
26001 (compact_insn_names): New array.
26002 (find_code): Handle insn codes in compact dumps.
26003 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26004 (bind_subst_iter_and_attr): Likewise.
26005 (add_condition_to_string): Likewise.
26006 (add_condition_to_rtx): Likewise.
26007 (apply_attribute_uses): Likewise.
26008 (add_current_iterators): Likewise.
26009 (apply_iterators): Likewise.
26010 (initialize_iterators): Guard usage of apply_subst_iterator with
26011 #ifdef GENERATOR_FILE.
26012 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26013 (md_reader::read_mapping): Likewise.
26014 (add_define_attr_for_define_subst): Likewise.
26015 (add_define_subst_attr): Likewise.
26016 (read_subst_mapping): Likewise.
26017 (check_code_iterator): Likewise.
26018 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26020 (one_time_initialization): New function.
26021 (rtx_reader::read_until): New method.
26022 (read_flags): New function.
26023 (parse_reg_note_name): New function.
26024 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26025 Handle reuse_rtx ids.
26026 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26027 Add parsing support for RTL dumps, mirroring the special-cases in
26028 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26029 values, and calling handle_any_trailing_information.
26030 (rtx_reader::read_rtx_operand): Convert return type from void
26031 to rtx, returning return_rtx. Handle case 'e'. Call
26032 finalize_string on XSTR and XTMPL fields.
26033 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26034 "(nil)" values were omitted. Call the postprocess vfunc on the
26036 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26037 class ctor. Initialize m_in_call_function_usage. Call
26038 one_time_initialization.
26039 * rtl-tests.c (selftest::test_uncond_jump): Call
26040 set_new_first_and_last_insn.
26041 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26042 * selftest-rtl.c: New file.
26043 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26044 (selftest::get_insn_by_uid): New decl.
26045 * selftest-run-tests.c (selftest::run_tests): Call
26046 read_rtl_function_c_tests.
26047 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26048 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26051 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26053 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26054 operands in a special way. Assert that pos+len <= mode precision.
26056 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26058 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26059 3 argument Alias with unlimited for the negative form.
26060 (fno-vect-cost-model): Removed.
26062 2017-01-05 Martin Liska <mliska@suse.cz>
26064 * hsa-gen.c (gen_hsa_divmod): New function.
26065 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26067 2017-01-05 Martin Liska <mliska@suse.cz>
26070 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26073 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26075 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26076 small constant length operands.
26078 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26080 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26081 between loop iterations.
26083 2017-01-05 Martin Liska <mliska@suse.cz>
26086 * gimplify.c (gimplify_decl_expr): Compare to
26087 asan_poisoned_variables instread of checking flags.
26088 (gimplify_target_expr): Likewise.
26089 (gimplify_expr): Likewise.
26090 (gimplify_function_tree): Conditionally initialize
26091 asan_poisoned_variables.
26093 2017-01-04 Jeff Law <law@redhat.com>
26095 PR tree-optimizatin/78812
26096 * rtl.h (contains_mem_rtx_p): Prototype.
26097 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26098 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26099 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26100 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26101 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26103 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26105 * input.c (assert_char_at_range): Default-initialize actual_range.
26107 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26109 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26110 to match the caller.
26112 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26114 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26115 insns after final jump in test to emit dummy move.
26117 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26119 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26120 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26122 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26124 * multiple_target.c (create_dispatcher_calls): Init e_next.
26125 * tree-ssa-loop-split.c (split_loop): Init border.
26126 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26129 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26134 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26135 (altivec_register_operand): Do not return true if the operand
26136 contains a SUBREG mixing SImode and SFmode.
26137 (vsx_register_operand): Likewise.
26138 (vsx_reg_sfsubreg_ok): New predicate.
26139 (vfloat_operand): Do not return true if the operand contains a
26140 SUBREG mixing SImode and SFmode.
26141 (vint_operand): Likewise.
26142 (vlogical_operand): Likewise.
26143 (gpc_reg_operand): Likewise.
26144 (int_reg_operand): Likewise.
26145 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26146 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26147 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26149 (rs6000_emit_move_si_sf_subreg): New helper function.
26150 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26151 fixup SUBREGs involving SImode and SFmode.
26152 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26153 numbers for the new peephole2 optimization.
26154 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26155 the GLIBC math library that do AND/IOR/XOR operations on single
26156 precision floating point.
26157 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26158 target macros to say whether we need to avoid SUBREGs mixing
26160 (TARGET_ALLOW_SF_SUBREG): Likewise.
26161 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26162 (UNSPEC_SI_FROM_SF): Likewise.
26163 (iorxor): Change spacing.
26164 (and_ior_xor): New iterator for AND, IOR, and XOR.
26165 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26166 (movdi_from_sf_zero_ext): Likewise.
26167 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26168 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26169 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26170 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26171 (fms<mode>4): Likewise.
26172 (fnma<mode>4): Likewise.
26173 (fnms<mode>4): Likewise.
26174 (nfma<mode>4): Likewise.
26175 (nfms<mode>4): Likewise.
26177 2017-01-04 Marek Polacek <polacek@redhat.com>
26180 * doc/invoke.texi: Document -Wpointer-compare.
26182 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26184 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26187 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26188 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26191 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26194 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26195 documentation of the powerpc_popcntb_ok attribute.
26196 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26197 code to issue warning messages if a requested CPU configuration is
26198 not supported by the binary (assembler and loader) toolchain.
26199 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26200 made to define a built-in function that has been disabled.
26201 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26202 made to define a built-in function that has been disabled.
26203 (altivec_init_builtins): Add comment explaining why definition
26204 of the DST built-in functions is not preceded by an assertion
26205 check. Add assertions to prevent ICE if attempts are made to
26206 define an altivec predicate or an abs* built-in function that has
26208 (htm_init_builtins): Add comment explaining why definition of the
26209 htm built-in functions is not preceded by an assertion check.
26211 2017-01-04 Jeff Law <law@redhat.com>
26213 PR tree-optimizatin/67955
26214 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26215 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26216 the points-to solution does not include pt_null. Use DECL_PT_UID
26219 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26221 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26222 Use gen_int_mode instead of gen_lopwart for const_int operands.
26224 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26226 PR tree-optimization/71563
26227 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26228 out of range value - has low bits known to be zero.
26230 2017-01-04 Alan Modra <amodra@gmail.com>
26232 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26233 * configure: Regenerate.
26234 * config.in: Regenerate.
26236 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26239 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26240 a substring of the message, but strcmp with the whole message. Ifdef
26241 ENABLE_NLS, translate the message first using dgettext.
26243 2017-01-03 Jeff Law <law@redhat.com>
26245 PR tree-optimizatin/78856
26246 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26247 (mark_threaded_blocks): Remove code to truncate thread paths that
26248 cross multiple loop headers. Instead invalidate the cached loop
26249 iteration information and handle case of a thread path walking
26250 into an irreducible region.
26252 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26255 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26256 assertions. Add support for doing the signbit if the IEEE 128-bit
26257 floating point value is in a GPR.
26258 * config/rs6000/rs6000.md (Fsignbit): Delete.
26259 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26260 Update the length attribute if the value is in a GPR.
26261 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26262 the sign or zero extension instruction, since the value is always 0/1.
26263 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26266 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26267 extracting SImode to a GPR register so that we can generate a
26268 store, limit the vector to be in a traditional Altivec register
26269 for the vextuwrx instruction.
26271 2017-01-03 Ian Lance Taylor <iant@google.com>
26273 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26275 2017-01-03 Martin Sebor <msebor@redhat.com>
26277 PR tree-optimization/78696
26278 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26279 precision. Use MPFR for %f for greater fidelity. Correct handling
26281 (pass_sprintf_length::compute_format_length): Set width and precision
26282 specified by asrerisk to void_node for vararg functions.
26283 (try_substitute_return_value): Adjust dump output.
26285 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26287 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26289 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26291 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26292 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26293 -mlra/-mno-lra was passed to the compiler.
26295 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26297 PR rtl-optimization/65618
26298 * emit-rtl.c (try_split): Move initialization of "before" and
26299 "after" to just before the call to emit_insn_after_setloc.
26301 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26303 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26305 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26307 * dwarf2out.c (gen_enumeration_type_die): When
26308 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26310 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26312 PR tree-optimization/78965
26313 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26314 Change first argument from const call_info & to call_info &. For %n
26315 set info.nowrite to false.
26317 PR middle-end/78901
26318 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26319 possibly throwing calls.
26321 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26322 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26323 and fns handling, rather than in a separate case SSA_NAME.
26325 2017-01-02 Jeff Law <law@redhat.com>
26327 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26328 fixes for first_period and second_period variables.
26330 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26333 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26334 (*insvqi_1): New insn pattern.
26335 (*insvqi_1_mem_rex64): Ditto.
26336 (*insvqi_2): Ditto.
26337 (*insvqi_3): Rename from *insvqi.
26339 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26341 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26343 * doc/cfg.texi (Edges): Remove reference to Java.
26344 (Maintaining the CFG): Ditto.
26346 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26348 PR middle-end/77674
26349 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26350 transparent aliases.
26352 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26354 PR middle-end/77484
26355 * predict.def (PRED_CALL): Update hitrate.
26356 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26357 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26358 into direct/indirect/polymorphic variants.
26360 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26362 Update copyright years.
26364 * gcc.c (process_command): Update copyright notice dates.
26365 * gcov-dump.c (print_version): Ditto.
26366 * gcov.c (print_version): Ditto.
26367 * gcov-tool.c (print_version): Ditto.
26368 * gengtype.c (create_file): Ditto.
26369 * doc/cpp.texi: Bump @copying's copyright year.
26370 * doc/cppinternals.texi: Ditto.
26371 * doc/gcc.texi: Ditto.
26372 * doc/gccint.texi: Ditto.
26373 * doc/gcov.texi: Ditto.
26374 * doc/install.texi: Ditto.
26375 * doc/invoke.texi: Ditto.
26377 Copyright (C) 2017 Free Software Foundation, Inc.
26379 Copying and distribution of this file, with or without modification,
26380 are permitted in any medium without royalty provided the copyright
26381 notice and this notice are preserved.