1 2017-09-05 Marek Polacek <polacek@redhat.com>
4 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11 2017-09-05 Richard Biener <rguenther@suse.de>
13 PR tree-optimization/82084
14 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16 2017-09-05 Richard Biener <rguenther@suse.de>
18 PR tree-optimization/82102
19 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
21 2017-09-05 Martin Liska <mliska@suse.cz>
23 PR tree-optimization/82032
24 * tree-cfg.c (generate_range_test): New function.
25 * tree-cfg.h (generate_range_test): Declared here.
26 * tree-cfgcleanup.c (convert_single_case_switch): New function.
27 (cleanup_control_expr_graph): Use it.
28 * tree-switch-conversion.c (try_switch_expansion): Remove
30 (emit_case_nodes): Use generate_range_test.
32 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
35 * config/i386/i386.md (*<btsc><mode>_mask): Add
36 TARGET_USE_BT to insn constraint.
37 (*btr<mode>_mask): Ditto.
39 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
41 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
42 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
44 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
47 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
48 TARGET_NEON and TARGET_IWMMXT.
49 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
50 TARGET_NEON and TARGET_IWMMXT.
51 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
53 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
55 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
56 (ix86_rewrite_tls_address): Ditto.
57 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
58 (ix86_rewrite_tls_address_1): Ditto.
59 (ix86_rewrite_tls_address): Ditto.
60 * config/i386/predicates.md (tls_address_pattern): New predicate.
61 * config/i386/i386.md (TLS address splitter): New splitter.
63 2017-09-04 Richard Biener <rguenther@suse.de>
65 PR tree-optimization/82084
66 * fold-const.h (can_native_encode_string_p): Declare.
67 * fold-const.c (can_native_encode_string_p): Factor out from ...
68 (native_encode_string): ... here.
69 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
70 vectorizing stores from constants we later cannot handle.
72 2017-09-04 Marek Polacek <polacek@redhat.com>
75 * doc/invoke.texi: Update -Wtautological-compare documentation.
77 2017-09-04 Jeff Law <law@redhat.com>
79 PR tree-optimization/64910
80 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
81 swap the first and last operand if the last is a constant.
83 2017-09-04 Marek Polacek <polacek@redhat.com>
86 * convert.c (do_narrow): When sanitizing signed integer overflows,
87 bail out for signed types.
88 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
90 2017-09-04 Richard Biener <rguenther@suse.de>
92 PR tree-optimization/82060
93 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
94 Move devirtualization after stmt folding and before EH/AB/noreturn
95 cleanup to get the stmt refs canonicalized. Use a bool instead
96 of gimple_modified_p since that doesn't work for NOPs. Schedule
97 NOPs generated by folding for removal.
99 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
100 Alan Hayward <alan.hayward@arm.com>
101 David Sherwood <david.sherwood@arm.com>
103 * coretypes.h (pad_direction): New enum.
104 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
105 (FUNCTION_ARG_PADDING): Likewise.
106 * target.def (function_arg_padding): New hook.
107 * targhooks.h (default_function_arg_padding): Declare.
108 * targhooks.c (default_function_arg_padding): New function.
109 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
110 (TARGET_FUNCTION_ARG_PADDING): ...this.
111 * doc/tm.texi: Regenerate.
112 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
113 instead of direction.
114 (compute_argument_addresses): Likewise.
115 (load_register_parameters): Likewise.
116 (emit_library_call_value_1): Likewise.
117 (store_one_arg): Use targetm.calls.function_arg_padding instead
118 of FUNCTION_ARG_PADDING.
119 (must_pass_in_stack_var_size_or_pad): Likewise.
120 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
121 (emit_group_store): Likewise.
122 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
123 instead of FUNCTION_ARG_PADDING.
124 (emit_push_insn): Likewise, and propagate enum change throughout
126 * function.h (direction): Delete.
127 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
129 * function.c (assign_parm_find_stack_rtl): Likewise.
130 (assign_parm_setup_block_p): Likewise.
131 (assign_parm_setup_block): Likewise.
132 (gimplify_parameters): Likewise.
133 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
134 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
136 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
137 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
138 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
139 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
140 (aarch64_function_arg_padding): ...this new function.
141 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
142 (TARGET_FUNCTION_ARG_PADDING): Redefine.
143 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
144 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
145 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
146 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
147 (arm_pad_arg_upward): Replace with...
148 (arm_function_arg_padding): ...this new function.
149 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
151 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
152 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
153 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
154 (ia64_hpux_function_arg_padding): Replace with...
155 (ia64_function_arg_padding): ...this new function. Use pad_direction
156 instead of direction. Check for TARGET_HPUX.
157 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
158 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
159 (iq2000_function_arg_padding): New function.
160 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
161 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
162 (mips_function_arg_padding): ...this new function.
163 (mips_pad_reg_upward): Update accordingly.
164 (TARGET_FUNCTION_ARG_PADDING): Redefine.
165 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
166 targetm.calls.function_arg_padding.
167 (FUNCTION_ARG_PADDING): Delete.
168 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
169 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
170 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
171 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
172 (nios2_block_reg_padding): Return pad_direction instead of direction.
173 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
174 instead of direction.
175 (nios2_function_arg_padding): Likewise. Make static.
176 (TARGET_FUNCTION_ARG_PADDING): Redefine.
177 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
178 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
179 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
180 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
181 (pa_function_arg_padding): Make static. Return pad_direction instead
183 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
184 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
185 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
186 instead of direction. Use targetm.calls.function_arg_padding.
187 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
188 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
189 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
190 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
191 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
193 (function_arg_padding): Rename to...
194 (rs6000_function_arg_padding): ...this. Make static. Return
195 pad_direction instead of direction.
196 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
197 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
198 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
199 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
200 instead of direction. Use targetm.calls.function_arg_padding.
201 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
202 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
203 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
204 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
205 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
206 (function_arg_padding): Rename to...
207 (rs6000_function_arg_padding): ...this. Make static. Return
208 pad_direction instead of direction.
209 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
210 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
211 * config/s390/s390.c (s390_function_arg_padding): New function.
212 (TARGET_FUNCTION_ARG_PADDING): Redefine.
213 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
214 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
215 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
216 (function_arg_padding): Rename to...
217 (sparc_function_arg_padding): ...this. Make static. Return
218 pad_direction instead of direction.
219 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
220 * config/spu/spu.c (spu_function_arg_padding): New function.
221 (TARGET_FUNCTION_ARG_PADDING): Redefine.
222 * system.h (FUNCTION_ARG_PADDING): Poison.
224 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
225 Alan Hayward <alan.hayward@arm.com>
226 David Sherwood <david.sherwood@arm.com>
228 * target.def (modes_tieable_p): New hook.
229 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
230 (TARGET_MODES_TIEABLE_P): ...this.
231 * doc/tm.texi.in: Regenerate.
232 * hooks.h (hook_bool_mode_mode_true): Declare.
233 * hooks.c (hook_bool_mode_mode_true): New function.
234 * combine.c (subst): Use targetm.modes_tieable_p instead of
236 * dse.c (find_shift_sequence): Likewise.
237 * expmed.c (extract_low_bits): Likewise.
238 * lower-subreg.c: Include target.h.
239 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
241 * rtlanal.c (rtx_cost): Likewise.
242 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
243 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
244 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
245 (TARGET_MODES_TIEABLE_P): Redefine.
246 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
247 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
248 (TARGET_MODES_TIEABLE_P): Redefine.
249 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
250 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
251 (arc_modes_tieable_p): New function.
252 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
253 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
254 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
255 (arm_modes_tieable_p): Make static.
256 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
257 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
258 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
259 (TARGET_MODES_TIEABLE_P): Redefine.
260 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
261 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
262 (TARGET_MODES_TIEABLE_P): Redefine.
263 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
264 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
265 (cr16_modes_tieable_p): New function.
266 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
267 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
268 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
269 (TRULY_NOOP_TRUNCATION): Update comment.
270 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
271 (TRULY_NOOP_TRUNCATION): Update comment.
272 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
273 (frv_modes_tieable_p): New function.
274 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
275 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
276 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
277 (TARGET_MODES_TIEABLE_P): Redefine.
278 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
279 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
280 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
281 (TARGET_MODES_TIEABLE_P): Redefine.
282 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
283 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
284 (ia64_modes_tieable_p): New function.
285 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
286 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
287 (iq2000_modes_tieable_p): New function.
288 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
289 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
290 (lm32_modes_tieable_p): New function.
291 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
292 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
293 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
294 (TARGET_MODES_TIEABLE_P): Redefine.
295 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
296 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
297 (m32r_modes_tieable_p): New function.
298 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
299 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
300 (m68k_modes_tieable_p): New function.
301 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
302 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
303 (mcore_modes_tieable_p): New function.
304 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
305 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
307 (TARGET_MODES_TIEABLE_P): Redefine.
308 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
309 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
310 * config/mips/mips.c (mips_modes_tieable_p): Make static.
311 (TARGET_MODES_TIEABLE_P): Redefine.
312 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
313 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
314 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
315 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
316 (mn10300_modes_tieable_p): ...this and make static.
317 (TARGET_MODES_TIEABLE_P): Redefine.
318 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
319 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
320 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
321 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
322 (msp430_modes_tieable_p): Make static.
323 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
324 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
325 (TARGET_MODES_TIEABLE_P): Redefine.
326 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
327 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
328 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
329 (TARGET_MODES_TIEABLE_P): Redefine.
330 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
331 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
332 * config/pa/pa.c (pa_modes_tieable_p): Make static.
333 (TARGET_MODES_TIEABLE_P): Redefine.
334 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
335 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
336 (pdp11_modes_tieable_p): New function.
337 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
338 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
339 (rs6000_modes_tieable_p): New function.
340 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
341 * config/powerpcspe/powerpcspe.md: Update comment.
342 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
343 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
344 (TARGET_MODES_TIEABLE_P): Redefine.
345 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
346 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
347 (rl78_modes_tieable_p): New function.
348 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
349 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
350 (rs6000_modes_tieable_p): New function.
351 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
352 * config/rs6000/rs6000.md: Update comment.
353 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
354 * config/rx/rx.c (rx_modes_tieable_p): New function.
355 (TARGET_MODES_TIEABLE_P): Redefine.
356 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
357 * config/s390/s390.c (s390_modes_tieable_p): New function.
358 (TARGET_MODES_TIEABLE_P): Redefine.
359 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
360 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
361 (sh_modes_tieable_p): New function.
362 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
363 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
364 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
365 (sparc_modes_tieable_p): Make static.
366 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
367 * config/spu/spu.c (spu_modes_tieable_p): New function.
368 (TARGET_MODES_TIEABLE_P): Redefine.
369 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
370 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
371 (TARGET_MODES_TIEABLE_P): Redefine.
372 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
373 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
374 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
375 * config/v850/v850.c (v850_modes_tieable_p): New function.
376 (TARGET_MODES_TIEABLE_P): Redefine.
377 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
378 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
379 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
380 (visium_modes_tieable_p): New function.
381 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
382 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
383 (xtensa_modes_tieable_p): New function.
384 * system.h (MODES_TIEABLE_P): Poison.
386 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
387 Alan Hayward <alan.hayward@arm.com>
388 David Sherwood <david.sherwood@arm.com>
390 * target.def (hard_regno_mode_ok): New hook.
391 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
392 (TARGET_HARD_REGNO_MODE_OK): ...this.
393 * doc/tm.texi.in: Regenerate.
394 * hooks.h (hook_bool_uint_mode_true): Declare.
395 * hooks.c (hook_bool_uint_mode_true): New function.
396 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
398 * genpreds.c (write_insn_preds_c): Add an include of target.h.
399 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
400 instead of HARD_REGNO_MODE_OK.
401 * caller-save.c: Include target.h.
402 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
404 * combine.c (can_combine_p): Likewise.
405 (combinable_i3pat): Likewise.
406 (can_change_dest_mode): Likewise.
407 * expr.c (init_expr_target): Likewise.
408 (convert_move): Likewise.
409 (convert_modes): Likewise.
410 * ira.c (setup_prohibited_class_mode_regs): Likewise.
411 (setup_prohibited_mode_move_regs): Likewise.
412 * ira.h (target_ira): Likewise.
413 * lra-assigns.c (find_hard_regno_for_1): Likewise.
414 * lra-constraints.c (process_alt_operands): Likewise.
415 (split_reg): Likewise.
416 * recog.c (peep2_find_free_register): Likewise.
417 * ree.c (combine_reaching_defs): Likewise.
418 * regcprop.c (maybe_mode_change): Likewise.
419 * reginfo.c (init_reg_sets_1): Likewise.
420 (choose_hard_reg_mode): Likewise.
421 (simplifiable_subregs): Likewise.
422 * regrename.c (check_new_reg_p): Likewise.
423 * reload.c (find_valid_class): Likewise.
424 (find_valid_class_1): Likewise.
425 (reload_inner_reg_of_subreg): Likewise.
426 (push_reload): Likewise.
427 (combine_reloads): Likewise.
428 (find_dummy_reload): Likewise.
429 (find_reloads): Likewise.
430 * reload1.c (find_reg): Likewise.
431 (set_reload_reg): Likewise.
432 (allocate_reload_reg): Likewise.
433 (choose_reload_regs): Likewise.
434 (reload_adjust_reg_for_temp): Likewise.
435 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
436 (simplify_subreg_regno): Likewise.
437 * sel-sched.c (init_regs_for_mode): Likewise.
438 * varasm.c (make_decl_rtl): Likewise.
439 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
440 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
442 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
443 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
445 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
446 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
447 (TARGET_HARD_REGNO_MODE_OK): Redefine.
448 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
449 (arc_mode_class): Delete.
450 (HARD_REGNO_MODE_OK): Delete.
451 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
452 (arc_hard_regno_mode_ok): Rename old array to...
453 (arc_hard_regno_mode_ok_modes): ...this.
454 (arc_conditional_register_usage): Update accordingly.
455 (arc_mode_class): Make static.
456 (arc_hard_regno_mode_ok): New function.
457 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
458 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
459 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
460 (arm_hard_regno_mode_ok): Make static.
461 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
463 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
464 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
465 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
467 (TARGET_HARD_REGNO_MODE_OK): Redefine.
468 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
469 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
470 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
471 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
472 (TARGET_HARD_REGNO_MODE_OK): Redefine.
473 * config/bfin/predicates.md (valid_reg_operand): Use
474 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
475 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
476 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
477 (TARGET_HARD_REGNO_MODE_OK): Redefine.
478 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
479 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
480 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
481 (cr16_hard_regno_mode_ok): Make static and return a bool.
482 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
483 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
484 (cris_hard_regno_mode_ok): New function.
485 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
486 (epiphany_mode_class): Delete.
487 (HARD_REGNO_MODE_OK): Delete.
488 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
489 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
490 (hard_regno_mode_ok): Rename to...
491 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
492 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
493 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
495 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
496 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
497 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
498 (frv_hard_regno_mode_ok): Make static and return a bool.
499 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
501 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
502 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
503 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
504 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
506 (TARGET_HARD_REGNO_MODE_OK): Redefine.
507 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
508 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
509 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
511 (TARGET_HARD_REGNO_MODE_OK): Redefine.
512 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
513 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
514 (ia64_hard_regno_mode_ok): New function.
515 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
516 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
517 (iq2000_hard_regno_mode_ok): New function.
518 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
519 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
520 (lm32_hard_regno_mode_ok): New function.
521 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
522 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
523 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
524 instead of HARD_REGNO_MODE_OK.
525 (m32c_hard_regno_ok): Rename to...
526 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
527 (m32c_cannot_change_mode_class): Update accordingly.
528 (TARGET_HARD_REGNO_MODE_OK): Redefine.
529 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
530 (m32r_mode_class): Delete.
531 (HARD_REGNO_MODE_OK): Delete.
532 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
533 (m32r_hard_regno_mode_ok): Rename to...
534 (m32r_hard_regno_modes): ...this.
535 (m32r_mode_class): Make static.
536 (m32r_hard_regno_mode_ok): New function.
537 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
538 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
539 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
540 (m68k_hard_regno_mode_ok): Make static.
541 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
542 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
543 (mcore_hard_regno_mode_ok): New function.
544 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
545 (HARD_REGNO_MODE_OK): Delete.
546 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
548 (microblaze_hard_regno_mode_ok_p): ...this and make static.
549 (microblaze_hard_regno_mode_ok): New function.
550 (TARGET_HARD_REGNO_MODE_OK): Redefine.
551 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
552 (mips_hard_regno_mode_ok): Delete.
553 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
554 (mips_hard_regno_mode_ok_p): ...this and make static.
555 (mips_hard_regno_mode_ok_p): Rename to...
556 (mips_hard_regno_mode_ok_uncached): ...this.
557 (mips_hard_regno_mode_ok): New function.
558 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
559 of HARD_REGNO_MODE_OK.
560 (mips_option_override): Update after above name changes.
561 (TARGET_HARD_REGNO_MODE_OK): Redefine.
562 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
563 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
564 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
565 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
566 (TARGET_HARD_REGNO_MODE_OK): Redefine.
567 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
568 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
569 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
570 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
571 (msp430_hard_regno_mode_ok): Make static and return a bool.
572 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
573 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
574 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
577 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
578 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
579 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
580 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
581 (PA_HARD_REGNO_MODE_OK): ...this
582 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
583 (PA_HARD_REGNO_MODE_OK): ...this.
584 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
585 (pa_hard_regno_mode_ok): New function.
586 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
587 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
588 (pdp11_hard_regno_mode_ok): New function.
589 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
590 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
592 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
594 (TARGET_HARD_REGNO_MODE_OK): Redefine.
595 (rs6000_hard_regno_mode_ok): Rename to...
596 (rs6000_hard_regno_mode_ok_uncached): ...this.
597 (rs6000_init_hard_regno_mode_ok): Update accordingly.
598 (rs6000_hard_regno_mode_ok): New function.
599 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
600 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
601 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
602 (riscv_hard_regno_mode_ok): ...this and make static.
603 (TARGET_HARD_REGNO_MODE_OK): Redefine.
604 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
605 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
606 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
607 (rl78_hard_regno_mode_ok): Make static and return bool.
608 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
609 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
611 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
612 (TARGET_HARD_REGNO_MODE_OK): Redefine.
613 (rs6000_hard_regno_mode_ok): Rename to...
614 (rs6000_hard_regno_mode_ok_uncached): ...this.
615 (rs6000_init_hard_regno_mode_ok): Update accordingly.
616 (rs6000_hard_regno_mode_ok): New function.
617 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
618 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
620 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
621 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
622 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
624 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
625 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
626 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
627 (sh_hard_regno_mode_ok): Make static.
628 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
629 instead of HARD_REGNO_MODE_OK.
630 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
631 (sparc_mode_class): Delete.
632 (HARD_REGNO_MODE_OK): Delete.
633 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
634 (hard_regno_mode_classes): Make static.
635 (sparc_mode_class): Likewise.
636 (sparc_hard_regno_mode_ok): New function.
637 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
638 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
639 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
641 (TARGET_HARD_REGNO_MODE_OK): Redefine.
642 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
643 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
644 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
645 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
647 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
648 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
649 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
650 (visium_hard_regno_mode_ok): New function.
651 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
652 instead of HARD_REGNO_MODE_OK.
653 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
654 (HARD_REGNO_MODE_OK): Delete.
655 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
656 (xtensa_hard_regno_mode_ok_p): ...this and make static.
657 (xtensa_option_override): Update accordingly.
658 (TARGET_HARD_REGNO_MODE_OK): Redefine.
659 (xtensa_hard_regno_mode_ok): New function.
660 * system.h (HARD_REGNO_MODE_OK): Poison.
662 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
663 Alan Hayward <alan.hayward@arm.com>
664 David Sherwood <david.sherwood@arm.com>
666 * target.def (hard_regno_call_part_clobbered): New hook.
667 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
668 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
669 * doc/tm.texi: Regenerate.
670 * hooks.h (hook_bool_uint_mode_false): Declare.
671 * hooks.c (hook_bool_uint_mode_false): New function.
672 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
673 * cselib.c (cselib_process_insn): Use
674 targetm.hard_regno_call_part_clobbered instead of
675 HARD_REGNO_CALL_PART_CLOBBERED.
676 * ira-conflicts.c (ira_build_conflicts): Likewise.
677 * ira-costs.c (ira_tune_allocno_costs): Likewise.
678 * lra-constraints.c (need_for_call_save_p): Likewise.
679 * lra-lives.c: Include target.h.
680 (check_pseudos_live_through_calls): Use
681 targetm.hard_regno_call_part_clobbered instead of
682 HARD_REGNO_CALL_PART_CLOBBERED.
683 * regcprop.c: Include target.h.
684 (copyprop_hardreg_forward_1): Use
685 targetm.hard_regno_call_part_clobbered instead of
686 HARD_REGNO_CALL_PART_CLOBBERED.
687 * reginfo.c (choose_hard_reg_mode): Likewise.
688 * regrename.c (check_new_reg_p): Likewise.
689 * reload.c (find_equiv_reg): Likewise.
690 * reload1.c (emit_reload_insns): Likewise.
691 * sched-deps.c (deps_analyze_insn): Likewise.
692 * sel-sched.c (init_regs_for_mode): Likewise.
693 (mark_unavailable_hard_regs): Likewise.
694 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
695 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
696 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
699 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
700 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
702 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
704 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
705 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
706 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
708 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
709 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
710 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
712 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
713 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
715 * config/powerpcspe/powerpcspe.c
716 (rs6000_hard_regno_call_part_clobbered): New function.
717 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
718 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
719 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
721 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
722 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
723 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
725 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
726 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
727 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
729 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
730 Alan Hayward <alan.hayward@arm.com>
731 David Sherwood <david.sherwood@arm.com>
733 * rtl.h (subreg_memory_offset): Declare.
734 * emit-rtl.c (subreg_memory_offset): New function.
735 * expmed.c (store_bit_field_1): Use it.
736 * expr.c (undefined_operand_subword_p): Likewise.
737 * simplify-rtx.c (simplify_subreg): Likewise.
739 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
741 PR rtl-optimization/57448
744 * optabs.c (expand_atomic_load): Place compiler memory barriers if
745 using atomic_load pattern.
746 (expand_atomic_store): Likewise.
748 2017-09-04 Jakub Jelinek <jakub@redhat.com>
751 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
752 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
753 handling. Use replace_call_with_value with NULL instead of
754 gsi_replace, unlink_stmt_vdef and release_defs.
756 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
759 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
761 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
764 * rtl.h (emit_library_call_value_1): Declare.
765 (emit_library_call): Replace declaration with a series of overloads.
766 Remove the parameter count argument.
767 (emit_library_call_value): Likewise.
768 * calls.c (emit_library_call_value_1): Make global. Replace varargs
769 with an "rtx_mode_t *".
770 (emit_library_call_value): Delete.
771 (emit_library_call): Likewise.
772 * asan.c (asan_emit_stack_protection): Update calls accordingly.
773 (asan_emit_allocas_unpoison): Likewise.
774 * builtins.c (expand_builtin_powi): Likewise.
775 (expand_asan_emit_allocas_unpoison): Likewise.
776 * cfgexpand.c (expand_main_function): Likewise.
777 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
778 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
779 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
780 * config/arm/arm.c (arm_trampoline_init): Likewise.
781 (arm_call_tls_get_addr): Likewise.
782 (arm_expand_divmod_libfunc): Likewise.
783 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
784 (smulsi3_highpart): Likewise.
785 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
786 (c6x_expand_compare): Likewise.
787 (c6x_expand_movmem): Likewise.
788 * config/frv/frv.c (frv_trampoline_init): Likewise.
789 * config/i386/i386.c (ix86_trampoline_init): Likewise.
790 (ix86_expand_divmod_libfunc): Likewise.
791 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
792 (ia64_expand_compare): Likewise.
793 (ia64_profile_hook): Likewise.
794 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
795 (nonlocal_goto): Likewise.
796 (restore_stack_nonlocal): Likewise.
797 * config/m32r/m32r.c (block_move_call): Likewise.
798 (m32r_trampoline_init): Likewise.
799 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
800 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
801 (m68k_call_m68k_read_tp): Likewise.
802 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
803 (microblaze_expand_divide): Likewise.
804 * config/mips/mips.h (mips_args): Likewise.
805 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
806 (MIPS_ICACHE_SYNC): Likewise.
807 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
808 (nios2_trampoline_init): Likewise.
809 * config/pa/pa.c (hppa_tls_call): Likewise.
810 (pa_trampoline_init): Likewise.
811 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
812 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
813 (expand_strn_compare): Likewise.
814 (rs6000_generate_compare): Likewise.
815 (rs6000_expand_float128_convert): Likewise.
816 (output_profile_hook): Likewise.
817 (rs6000_trampoline_init): Likewise.
818 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
819 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
820 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
821 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
822 (rs6000_generate_compare): Likewise.
823 (rs6000_expand_float128_convert): Likewise.
824 (output_profile_hook): Likewise.
825 (rs6000_trampoline_init): Likewise.
826 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
827 * config/sh/sh.c (sh_trampoline_init): Likewise.
828 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
829 (sparc_emit_float_lib_cmp): Likewise.
830 (sparc32_initialize_trampoline): Likewise.
831 (sparc64_initialize_trampoline): Likewise.
832 (sparc_profile_hook): Likewise.
833 * config/spu/spu.c (ea_load_store): Likewise.
834 * config/spu/spu.md (floatunssidf2): Likewise.
835 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
836 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
837 * config/visium/visium.c (expand_block_move_4): Likewise.
838 (expand_block_move_2): Likewise.
839 (expand_block_move_1): Likewise.
840 (expand_block_set_4): Likewise.
841 (expand_block_set_2): Likewise.
842 (expand_block_set_1): Likewise.
843 (visium_trampoline_init): Likewise.
844 (visium_profile_hook): Likewise.
845 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
846 (xtensa_setup_frame_addresses): Likewise.
847 (xtensa_trampoline_init): Likewise.
848 * except.c (sjlj_emit_function_enter): Likewise.
849 (sjlj_emit_function_exit): Likewise.
850 * explow.c (allocate_dynamic_stack_space): Likewise.
851 (probe_stack_range): Likewise.
852 * expr.c (convert_mode_scalar): Likewise.
853 * optabs.c (expand_binop): Likewise.
854 (expand_twoval_binop_libfunc): Likewise.
855 (expand_unop): Likewise.
856 (prepare_cmp_insn): Likewise.
857 (prepare_float_lib_cmp): Likewise.
858 (expand_float): Likewise.
859 (expand_fix): Likewise.
860 (expand_fixed_convert): Likewise.
861 (maybe_emit_sync_lock_test_and_set): Likewise.
862 (expand_atomic_compare_and_swap): Likewise.
863 (expand_mem_thread_fence): Likewise.
864 (expand_atomic_fetch_op): Likewise.
866 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
868 * doc/generic.texi (OpenACC): Adjust URL.
869 * doc/invoke.texi (C Dialect Options): Ditto.
871 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
873 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
874 predicate for operand 1. Add (m,<S>) constraint.
875 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
876 Prevent memory operand 1 with register operand 2.
878 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
880 PR rtl-optimization/82024
881 * combine.c (try_combine): If the combination result is a PARALLEL,
882 and we only need to retain the SET in there that would be placed
883 at I2, check that we can place that at I3 instead, before doing so.
885 2017-09-01 Jakub Jelinek <jakub@redhat.com>
888 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
890 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
891 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
894 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
895 Jeff Law <law@redhat.com>
897 * varasm.c (bss_initializer_p): Do not put constants into .bss
898 (categorize_decl_for_section): Handle bss_initializer_p returning
899 false when DECL_INITIAL is NULL.
901 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
904 * config/s390/s390.c (s390_can_inline_p): New function.
906 2017-09-01 Jeff Law <law@redhat.com>
908 PR tree-optimization/82052
909 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
910 Always initialize the returned slot after a hash table miss
913 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
915 * config/s390/s390.md (mem_signal_fence): Remove.
916 * doc/md.texi (mem_signal_fence): Remove.
917 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
919 * target-insns.def (mem_signal_fence): Remove.
921 2017-09-01 Jakub Jelinek <jakub@redhat.com>
924 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
927 * asan.c (create_odr_indicator): Strip name encoding from assembler
928 name before appending it after __odr_asan_.
930 2017-09-01 Martin Liska <mliska@suse.cz>
932 PR tree-optimization/82059
933 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
934 frequency only when an edge is redirected.
936 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
938 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
939 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
940 (arc_conditional_register_usage): Remove ARC600 lp_count
942 (arc_file_start): Emit Tag_ARC_CPU_variation.
943 (arc_can_use_doloop_p): New conditions to use ZOLs.
944 (hwloop_fail): New function.
945 (hwloop_optimize): Likewise.
946 (hwloop_pattern_reg): Likewise.
947 (arc_doloop_hooks): New struct, to be used with reorg_loops.
948 (arc_reorg_loops): New function, calls reorg_loops.
949 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
950 (arc600_corereg_hazard): Remove ZOL checking, case handled by
952 (arc_loop_hazard): Remove function, functionality moved into
954 (arc_hazard): Remove arc_loop_hazard call.
955 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
956 into hwloop_optimize.
957 (arc_label_align): Remove ZOL handling.
958 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
959 * config/arc/arc.md (doloop_begin): Remove pattern.
960 (doloop_begin_i): Likewise.
961 (doloop_end_i): Likewise.
962 (doloop_fallback): Likewise.
963 (doloop_fallback_m): Likewise.
964 (doloop_end): Reimplement expand.
965 (arc_lp): New pattern for LP instruction.
966 (loop_end): New pattern.
967 (loop_fail): Likewise.
968 (decrement_and_branch_until_zero): Likewise.
969 * config/arc/arc.opt (mlpc-width): New option.
970 * doc/invoke.texi (mlpc-width): Document option.
972 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
974 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
975 (arc_ccfsm_advance): Fix checking for delay slots.
976 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
978 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
980 * config/arc/arc.md (movqi_insn): Add stores to save constant long
982 (movhi_insn): Update store instruction constraint which are saving
983 6-bit short immediates.
984 (movsi_insn): Consider also short scaled load operations.
985 (zero_extendhisi2_i): Use Usd constraint instead of T.
986 (extendhisi2_i): Add q constraint.
987 (arc_clzsi2): Add type and length attributes.
988 (arc_ctzsi2): Likewise.
989 * config/arc/constraints.md (Usc): Update constraint, the
990 assembler can parse two relocations for a single instruction.
992 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
994 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
995 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
997 2017-08-31 Olivier Hainque <hainque@adacore.com>
999 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1000 match as powerpc-wrs-vxworks*.
1002 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1004 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1005 register constraint for by-element operand.
1006 (aarch64_mls_elt_merge<mode>): Likewise.
1008 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1010 * config/arc/arc.c (arc_can_follow_jump): Check for short
1013 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1015 * config.gcc: Use g.opt for arc.
1016 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1017 functionality moved to ...
1018 (legitimate_scaled_address_p): New function, ...here.
1019 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1020 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1021 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1023 (arc_override_options): Handle G option.
1024 (arc_output_pic_addr_const): Correct function definition.
1025 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1026 (arc_decl_anon_ns_mem_p): Delete.
1027 (arc_in_small_data_p): Overhaul this function to take into
1028 consideration the value given via G option.
1029 (arc_rewrite_small_data_1): Renamed and corrected old
1030 arc_rewrite_small_data function.
1031 (arc_rewrite_small_data): New function.
1032 (small_data_pattern): Don't use pic_offset_table_rtx.
1033 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1034 * config/arc/simdext.md (movmisalignv2hi): Use
1035 prepare_move_operands function.
1037 (movmisalign*): Likewise.
1038 * doc/invoke.texi (ARC options): Document -G option.
1040 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1042 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1044 * config/arc/arc.c (arc_print_operand): Output scalled address for
1045 sdata whenever is possible.
1046 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1047 load/stores are available.
1048 (compact_sda_memory_operand): Check for the alignment required by
1049 code density instructions.
1050 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1052 * config/arc/constraints.md (Usd): Update constraint.
1053 (Us0): New constraint.
1054 (Usc): Update constraint.
1056 2017-08-31 Richard Biener <rguenther@suse.de>
1059 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1062 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1064 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1065 Resize type_signature.
1067 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1068 Alan Hayward <alan.hayward@arm.com>
1069 David Sherwood <david.sherwood@arm.com>
1071 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1072 subregs whose inner modes can be stored in GPRs.
1073 (aarch64_classify_index): Likewise.
1075 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1076 Alan Hayward <alan.hayward@arm.com>
1077 David Sherwood <david.sherwood@arm.com>
1079 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1080 (V_INT_EQUIV): ...this.
1081 (v_cmp_result): Rename to...
1082 (v_int_equiv): ...this.
1083 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1084 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1085 (copysign<mode>3): Likewise.
1086 (aarch64_simd_bsl<mode>_internal): Likewise.
1087 (aarch64_simd_bsl<mode>): Likewise.
1088 (vec_cmp<mode><mode>): Likewise.
1089 (vcond<mode><mode>): Likewise.
1090 (vcond<v_cmp_mixed><mode>): Likewise.
1091 (vcondu<mode><v_cmp_mixed>): Likewise.
1092 (aarch64_cm<optab><mode>): Likewise.
1093 (aarch64_cmtst<mode>): Likewise.
1094 (aarch64_fac<optab><mode>): Likewise.
1095 (vec_perm_const<mode>): Likewise.
1096 (vcond_mask_<mode><v_cmp_result>): Rename to...
1097 (vcond_mask_<mode><v_int_equiv>): ...this.
1098 (vec_cmp<mode><v_cmp_result>): Rename to...
1099 (vec_cmp<mode><v_int_equiv>): ...this.
1101 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1102 Alan Hayward <alan.hayward@arm.com>
1103 David Sherwood <david.sherwood@arm.com>
1105 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1107 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1108 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1109 (UNSPEC_LD4_DREG): New unspecs.
1110 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1111 (aarch64_ld2<mode>_dreg_be): Replace with...
1112 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1114 (aarch64_ld3<mode>_dreg_le)
1115 (aarch64_ld3<mode>_dreg_be): Replace with...
1116 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1118 (aarch64_ld4<mode>_dreg_le)
1119 (aarch64_ld4<mode>_dreg_be): Replace with...
1120 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1123 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1125 PR tree-optimization/81987
1126 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1127 insert an initializer in a location not dominated by the stride
1130 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1132 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1133 on the entire header of the finally block in the fallthru case.
1135 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1137 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1139 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1141 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1142 rs6000_emit_move_from_cr and call renamed function.
1143 (rs6000_emit_prologue): Call renamed functions.
1144 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1145 movesi_from_cr, remove volatile CRs.
1147 2017-08-30 Jon Beniston <jon@beniston.com>
1148 Richard Biener <rguenther@suse.de>
1150 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1151 of VECTOR_MODE_P check.
1152 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1153 element vector types.
1155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1157 * df.h (df_read_modify_subreg_p): Remove in favor of...
1158 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1159 const_rtx instead of an rtx.
1160 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1161 * df-problems.c (df_word_lr_mark_ref): Likewise.
1162 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1163 (mark_pseudo_reg_dead): Likewise.
1164 (mark_ref_dead): Likewise.
1165 * reginfo.c (init_subregs_of_mode): Likewise.
1166 * sched-deps.c (sched_analyze_1): Likewise.
1167 * df-scan.c (df_def_record_1): Likewise.
1168 (df_uses_record): Likewise.
1169 (df_read_modify_subreg_p): Remove in favor of...
1170 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1171 const_rtx instead of an rtx.
1173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1174 Alan Hayward <alan.hayward@arm.com>
1175 David Sherwood <david.sherwood@arm.com>
1177 * rtl.h (partial_subreg_p): New function.
1178 * caller-save.c (save_call_clobbered_regs): Use it.
1179 * calls.c (expand_call): Likewise.
1180 * combine.c (combinable_i3pat): Likewise.
1181 (simplify_set): Likewise.
1182 (make_extraction): Likewise.
1183 (make_compound_operation_int): Likewise.
1184 (gen_lowpart_or_truncate): Likewise.
1185 (force_to_mode): Likewise.
1186 (make_field_assignment): Likewise.
1187 (reg_truncated_to_mode): Likewise.
1188 (record_truncated_value): Likewise.
1189 (move_deaths): Likewise.
1190 * cse.c (record_jump_cond): Likewise.
1191 (cse_insn): Likewise.
1192 * cselib.c (cselib_lookup_1): Likewise.
1193 * expmed.c (extract_bit_field_using_extv): Likewise.
1194 * function.c (assign_parm_setup_reg): Likewise.
1195 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1196 * ira-build.c (create_insn_allocnos): Likewise.
1197 * lra-coalesce.c (merge_pseudos): Likewise.
1198 * lra-constraints.c (match_reload): Likewise.
1199 (simplify_operand_subreg): Likewise.
1200 (curr_insn_transform): Likewise.
1201 * lra-lives.c (process_bb_lives): Likewise.
1202 * lra.c (new_insn_reg): Likewise.
1203 (lra_substitute_pseudo): Likewise.
1204 * regcprop.c (mode_change_ok): Likewise.
1205 (maybe_mode_change): Likewise.
1206 (copyprop_hardreg_forward_1): Likewise.
1207 * reload.c (push_reload): Likewise.
1208 (find_reloads): Likewise.
1209 (find_reloads_subreg_address): Likewise.
1210 * reload1.c (alter_reg): Likewise.
1211 (eliminate_regs_1): Likewise.
1212 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1214 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1216 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1217 back to if statements, including unpack.
1219 2017-08-30 Martin Liska <mliska@suse.cz>
1222 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1224 (func_checker::compare_asm_inputs_outputs): ... this function.
1225 (func_checker::compare_gimple_asm): Use the function to compare
1226 also ASM constrains.
1227 * ipa-icf-gimple.h: Rename the function.
1229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1230 Alan Hayward <alan.hayward@arm.com>
1231 David Sherwood <david.sherwood@arm.com>
1233 * coretypes.h (complex_mode): New type.
1234 * gdbhooks.py (build_pretty_printer): Handle it.
1235 * machmode.h (complex_mode): New class.
1236 (complex_mode::includes_p): New function.
1237 (is_complex_int_mode): Likewise.
1238 (is_complex_float_mode): Likewise.
1239 * genmodes.c (get_mode_class): Handle complex mode classes.
1240 * function.c (expand_function_end): Use is_complex_int_mode.
1242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1243 Alan Hayward <alan.hayward@arm.com>
1244 David Sherwood <david.sherwood@arm.com>
1246 * coretypes.h (scalar_mode_pod): New typedef.
1247 * gdbhooks.py (build_pretty_printer): Handle it.
1248 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1249 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1250 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1251 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1255 Alan Hayward <alan.hayward@arm.com>
1256 David Sherwood <david.sherwood@arm.com>
1258 * machmode.h (mode_for_vector): Take a scalar_mode instead
1260 * stor-layout.c (mode_for_vector): Likewise.
1261 * explow.c (promote_mode): Use as_a <scalar_mode>.
1262 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1265 Alan Hayward <alan.hayward@arm.com>
1266 David Sherwood <david.sherwood@arm.com>
1268 * target.def (preferred_simd_mode): Take a scalar_mode
1269 instead of a machine_mode.
1270 * targhooks.h (default_preferred_simd_mode): Likewise.
1271 * targhooks.c (default_preferred_simd_mode): Likewise.
1272 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1273 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1274 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1275 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1276 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1277 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1278 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1279 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1281 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1282 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1283 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1284 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1285 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1286 * doc/tm.texi: Regenerate.
1287 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1291 Alan Hayward <alan.hayward@arm.com>
1292 David Sherwood <david.sherwood@arm.com>
1294 * target.def (scalar_mode_supported_p): Take a scalar_mode
1295 instead of a machine_mode.
1296 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1297 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1298 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1299 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1300 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1301 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1302 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1303 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1304 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1305 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1306 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1307 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1308 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1309 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1311 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1312 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1313 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1314 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1315 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1317 * doc/tm.texi: Regenerate.
1319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1320 Alan Hayward <alan.hayward@arm.com>
1321 David Sherwood <david.sherwood@arm.com>
1323 * coretypes.h (opt_scalar_mode): New typedef.
1324 * gdbhooks.py (build_pretty_printers): Handle it.
1325 * machmode.h (mode_iterator::get_2xwider): Add overload for
1327 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1329 * expr.c (convert_mode_scalar): Likewise.
1330 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1331 * optabs.c (expand_float): Likewise.
1332 (expand_fix): Likewise.
1333 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * optabs.c (expand_float): Explicitly check for scalars before
1340 using a branching expansion.
1341 (expand_fix): Likewise.
1343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1344 Alan Hayward <alan.hayward@arm.com>
1345 David Sherwood <david.sherwood@arm.com>
1347 * expr.c (convert_mode): Split scalar handling out into...
1348 (convert_mode_scalar): ...this new function. Treat the modes
1351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1352 Alan Hayward <alan.hayward@arm.com>
1353 David Sherwood <david.sherwood@arm.com>
1355 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1357 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1364 rather than a machine_mode.
1365 (fixed_from_string): Likewise.
1366 (fixed_convert): Likewise.
1367 (fixed_convert_from_int): Likewise.
1368 (fixed_convert_from_real): Likewise.
1369 (real_convert_from_fixed): Likewise.
1370 * fixed-value.c (fixed_from_double_int): Likewise.
1371 (fixed_from_string): Likewise.
1372 (fixed_convert): Likewise.
1373 (fixed_convert_from_int): Likewise.
1374 (fixed_convert_from_real): Likewise.
1375 (real_convert_from_fixed): Likewise.
1376 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1379 Alan Hayward <alan.hayward@arm.com>
1380 David Sherwood <david.sherwood@arm.com>
1382 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1383 of separate mode class checks. Do not allow vector modes here.
1384 (immed_wide_int_const): Use as_a <scalar_mode>.
1385 * explow.c (trunc_int_for_mode): Likewise.
1386 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1387 (wi::shwi): Likewise.
1388 (wi::min_value): Likewise.
1389 (wi::max_value): Likewise.
1390 * dwarf2out.c (loc_descriptor): Likewise.
1391 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1395 Alan Hayward <alan.hayward@arm.com>
1396 David Sherwood <david.sherwood@arm.com>
1398 * tree.h (SCALAR_TYPE_MODE): New macro.
1399 * expr.c (expand_expr_addr_expr_1): Use it.
1400 (expand_expr_real_2): Likewise.
1401 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1402 (fold_convert_const_fixed_from_int): Likewise.
1403 (fold_convert_const_fixed_from_real): Likewise.
1404 (native_encode_fixed): Likewise
1405 (native_encode_complex): Likewise
1406 (native_encode_vector): Likewise.
1407 (native_interpret_fixed): Likewise.
1408 (native_interpret_real): Likewise.
1409 (native_interpret_complex): Likewise.
1410 (native_interpret_vector): Likewise.
1411 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1412 (simd_clone_adjust_argument_types): Likewise.
1413 (simd_clone_init_simd_arrays): Likewise.
1414 (simd_clone_adjust): Likewise.
1415 * stor-layout.c (layout_type): Likewise.
1416 * tree.c (build_minus_one_cst): Likewise.
1417 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1418 * tree-inline.c (estimate_move_cost): Likewise.
1419 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1420 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1421 (vectorizable_reduction): Likewise.
1422 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1423 (vect_recog_mixed_size_cond_pattern): Likewise.
1424 (check_bool_pattern): Likewise.
1425 (adjust_bool_pattern): Likewise.
1426 (search_type_for_mask_1): Likewise.
1427 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1428 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1429 (vectorizable_load): Likewise.
1430 (vectorizable_store): Likewise.
1431 * ubsan.c (ubsan_encode_value): Likewise.
1432 * varasm.c (output_constant): Likewise.
1434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1435 Alan Hayward <alan.hayward@arm.com>
1436 David Sherwood <david.sherwood@arm.com>
1438 * coretypes.h (scalar_mode): New class.
1439 * machmode.h (scalar_mode): Likewise.
1440 (scalar_mode::includes_p): New function.
1441 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1442 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1443 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1444 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1445 * expmed.c (store_bit_field_1): Likewise.
1446 (extract_bit_field_1): Likewise.
1447 * expr.c (write_complex_part): Likewise.
1448 (read_complex_part): Likewise.
1449 (emit_move_complex_push): Likewise.
1450 (expand_expr_real_2): Likewise.
1451 * function.c (assign_parm_setup_reg): Likewise.
1452 (assign_parms_unsplit_complex): Likewise.
1453 * optabs.c (expand_binop): Likewise.
1454 * rtlanal.c (subreg_get_info): Likewise.
1455 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1456 * varasm.c (output_constant_pool_2): Likewise.
1458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1459 Alan Hayward <alan.hayward@arm.com>
1460 David Sherwood <david.sherwood@arm.com>
1462 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1464 (expmed_mult_highpart_optab): Likewise.
1465 (expmed_mult_highpart): Likewise.
1467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hayward@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1472 instead of a machine_mode.
1473 (builtin_memset_read_str): Likewise.
1474 * builtins.c (c_readstr): Likewise.
1475 (builtin_memcpy_read_str): Likewise.
1476 (builtin_strncpy_read_str): Likewise.
1477 (builtin_memset_read_str): Likewise.
1478 (builtin_memset_gen_str): Likewise.
1479 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1480 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1481 instead of a machine_mode.
1482 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1484 (make_extraction): Likewise.
1485 (try_widen_shift_mode): Take and return scalar_int_modes instead
1487 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1488 a scalar_int_mode instead of a machine_mode.
1489 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1490 (avr_addr_space_pointer_mode): Likewise.
1491 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1492 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1493 (msp430_unwind_word_mode): Likewise.
1494 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1495 (spu_addr_space_pointer_mode): Likewise.
1496 (spu_addr_space_address_mode): Likewise.
1497 (spu_libgcc_cmp_return_mode): Likewise.
1498 (spu_libgcc_shift_count_mode): Likewise.
1499 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1500 (rl78_addr_space_pointer_mode): Likewise.
1501 (fl78_unwind_word_mode): Likewise.
1502 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1504 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1505 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1506 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1507 (mips_valid_pointer_mode): Likewise.
1508 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1509 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1510 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1512 (ft32_addr_space_address_mode): Likewise.
1513 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1514 scalar_int_mode instead of a machine_mode.
1515 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1517 (m32c_addr_space_address_mode): Likewise.
1518 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1519 (rs6000_eh_return_filter_mode): Likewise.
1520 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1521 (rs6000_eh_return_filter_mode): Likewise.
1522 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1523 (s390_libgcc_shift_count_mode): Likewise.
1524 (s390_unwind_word_mode): Likewise.
1525 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1527 * target.def (mode_rep_extended): Likewise.
1528 (valid_pointer_mode): Likewise.
1529 (addr_space.valid_pointer_mode): Likewise.
1530 (eh_return_filter_mode): Return a scalar_int_mode rather than
1532 (libgcc_cmp_return_mode): Likewise.
1533 (libgcc_shift_count_mode): Likewise.
1534 (unwind_word_mode): Likewise.
1535 (addr_space.pointer_mode): Likewise.
1536 (addr_space.address_mode): Likewise.
1537 * doc/tm.texi: Regenerate.
1538 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1540 (do_jump): Use scalar_int_mode for local variables.
1541 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1542 rather than a machine_mode.
1543 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1544 (scompare_loc_descriptor_wide): Likewise.
1545 (scompare_loc_descriptor_narrow): Likewise.
1546 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1548 * except.c (sjlj_emit_dispatch_table): Likewise.
1549 (expand_builtin_eh_copy_values): Likewise.
1550 * explow.c (convert_memory_address_addr_space_1): Likewise.
1551 Take a scalar_int_mode rather than a machine_mode.
1552 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1553 than a machine_mode.
1554 (memory_address_addr_space): Use scalar_int_mode for local variables.
1555 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1556 rather than a machine_mode.
1557 * expmed.c (mask_rtx): Likewise.
1558 (init_expmed_one_conv): Likewise.
1559 (expand_mult_highpart_adjust): Likewise.
1560 (extract_high_half): Likewise.
1561 (expmed_mult_highpart_optab): Likewise.
1562 (expmed_mult_highpart): Likewise.
1563 (expand_smod_pow2): Likewise.
1564 (expand_sdiv_pow2): Likewise.
1565 (emit_store_flag_int): Likewise.
1566 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
1568 (extract_low_bits): Likewise.
1569 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
1571 * expr.c (pieces_addr::adjust): Likewise.
1572 (can_store_by_pieces): Likewise.
1573 (store_by_pieces): Likewise.
1574 (clear_by_pieces_1): Likewise.
1575 (expand_expr_addr_expr_1): Likewise.
1576 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
1577 (expand_expr_real_1): Likewise.
1578 (try_casesi): Likewise.
1579 * final.c (shorten_branches): Likewise.
1580 * fold-const.c (fold_convert_const_int_from_fixed): Change the
1581 type of "mode" to machine_mode.
1582 * internal-fn.c (expand_arith_overflow_result_store): Take a
1583 scalar_int_mode rather than a machine_mode.
1584 (expand_mul_overflow): Use scalar_int_mode for local variables.
1585 * loop-doloop.c (doloop_modify): Likewise.
1586 (doloop_optimize): Likewise.
1587 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
1588 than a machine_mode.
1589 (expand_doubleword_shift_condmove): Likewise.
1590 (expand_doubleword_shift): Likewise.
1591 (expand_doubleword_clz): Likewise.
1592 (expand_doubleword_popcount): Likewise.
1593 (expand_doubleword_parity): Likewise.
1594 (expand_absneg_bit): Use scalar_int_mode for local variables.
1595 (prepare_float_lib_cmp): Likewise.
1596 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
1597 rather than a machine_mode.
1598 (convert_memory_address_addr_space): Likewise.
1599 (get_mode_bounds): Likewise.
1600 (get_address_mode): Return a scalar_int_mode rather than a
1602 * rtlanal.c (get_address_mode): Likewise.
1603 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
1604 than a machine_mode.
1605 * targhooks.c (default_mode_rep_extended): Likewise.
1606 (default_valid_pointer_mode): Likewise.
1607 (default_addr_space_valid_pointer_mode): Likewise.
1608 (default_eh_return_filter_mode): Return a scalar_int_mode rather
1609 than a machine_mode.
1610 (default_libgcc_cmp_return_mode): Likewise.
1611 (default_libgcc_shift_count_mode): Likewise.
1612 (default_unwind_word_mode): Likewise.
1613 (default_addr_space_pointer_mode): Likewise.
1614 (default_addr_space_address_mode): Likewise.
1615 * targhooks.h (default_eh_return_filter_mode): Likewise.
1616 (default_libgcc_cmp_return_mode): Likewise.
1617 (default_libgcc_shift_count_mode): Likewise.
1618 (default_unwind_word_mode): Likewise.
1619 (default_addr_space_pointer_mode): Likewise.
1620 (default_addr_space_address_mode): Likewise.
1621 (default_mode_rep_extended): Take a scalar_int_mode rather than
1623 (default_valid_pointer_mode): Likewise.
1624 (default_addr_space_valid_pointer_mode): Likewise.
1625 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
1627 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
1628 rather than a machine_mode.
1629 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
1630 for local variables.
1631 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1632 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
1633 than a machine_mode.
1635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1636 Alan Hayward <alan.hayward@arm.com>
1637 David Sherwood <david.sherwood@arm.com>
1639 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
1640 the mode argument to scalar_int_mode.
1641 (do_jump_by_parts_zero_rtx): Likewise.
1642 (do_jump_by_parts_equality_rtx): Likewise.
1643 (do_jump_by_parts_greater): Take a mode argument.
1644 (do_jump_by_parts_equality): Likewise.
1645 (do_jump_1): Update calls accordingly.
1647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1648 Alan Hayward <alan.hayward@arm.com>
1649 David Sherwood <david.sherwood@arm.com>
1651 * is-a.h (safe_dyn_cast): New function.
1652 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
1653 (jump_table_for_label): Likewise.
1654 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
1655 instead of an rtx_insn *.
1656 (shorten_branches): Use dyn_cast instead of LABEL_P and
1657 JUMP_TABLE_DATA_P. Use jump_table_for_label and
1658 rtx_jump_table_data::get_data_mode.
1659 (final_scan_insn): Likewise.
1661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1662 Alan Hayward <alan.hayward@arm.com>
1663 David Sherwood <david.sherwood@arm.com>
1665 * combine.c (try_combine): Use is_a <scalar_int_mode> when
1666 trying to combine a full-register integer set with a subreg
1669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1670 Alan Hayward <alan.hayward@arm.com>
1671 David Sherwood <david.sherwood@arm.com>
1673 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
1674 that is always either address_mode or pointer_mode.
1676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1677 Alan Hayward <alan.hayward@arm.com>
1678 David Sherwood <david.sherwood@arm.com>
1680 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
1681 when the two are known to be equal.
1683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1684 Alan Hayward <alan.hayward@arm.com>
1685 David Sherwood <david.sherwood@arm.com>
1687 * simplify-rtx.c (simplify_const_unary_operation): Use
1688 is_a <scalar_int_mode> instead of checking for a nonzero
1689 precision. Forcibly convert op_mode to a scalar_int_mode
1690 in that case. More clearly differentiate the operand and
1691 result modes and use the former when deciding what the value
1692 of a count-bits operation should be. Use is_int_mode instead
1693 of checking for a MODE_INT. Remove redundant check for whether
1694 this mode has a zero precision.
1696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1697 Alan Hayward <alan.hayward@arm.com>
1698 David Sherwood <david.sherwood@arm.com>
1700 * optabs.c (widen_leading): Change the type of the mode argument
1701 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
1702 (widen_bswap): Likewise.
1703 (expand_parity): Likewise.
1704 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
1705 (expand_ffs): Likewise.
1706 (epand_unop): Check for scalar integer modes before calling the
1709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1710 Alan Hayward <alan.hayward@arm.com>
1711 David Sherwood <david.sherwood@arm.com>
1713 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
1715 (expand_expr_real_1): Update call accordingly.
1717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1718 Alan Hayward <alan.hayward@arm.com>
1719 David Sherwood <david.sherwood@arm.com>
1721 * expmed.c (store_bit_field_using_insv): Add op0_mode and
1722 value_mode arguments. Use scalar_int_mode internally.
1723 (store_bit_field_1): Rename the new integer mode from imode
1724 to op0_mode and use it instead of GET_MODE (op0). Update calls
1725 to store_split_bit_field, store_bit_field_using_insv and
1726 store_fixed_bit_field.
1727 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1728 Use scalar_int_mode internally. Use a bit count rather than a mode
1729 when calculating the largest bit size for get_best_mode.
1730 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1731 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1732 Remove assertion that OP0 has a scalar integer mode.
1733 (store_split_bit_field): Add op0_mode and value_mode arguments.
1734 Update calls to extract_fixed_bit_field.
1735 (extract_bit_field_using_extv): Add an op0_mode argument.
1736 Use scalar_int_mode internally.
1737 (extract_bit_field_1): Rename the new integer mode from imode to
1738 op0_mode and use it instead of GET_MODE (op0). Update calls to
1739 extract_split_bit_field, extract_bit_field_using_extv and
1740 extract_fixed_bit_field.
1741 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1742 to extract_split_bit_field and extract_fixed_bit_field_1.
1743 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1744 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1746 (extract_split_bit_field): Add an op0_mode argument. Update call
1747 to extract_fixed_bit_field.
1749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1750 Alan Hayward <alan.hayward@arm.com>
1751 David Sherwood <david.sherwood@arm.com>
1753 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1754 * explow.c (hard_function_value): Likewise.
1755 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1756 convert_to_mode call outside the loop.
1757 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1758 for the mode iterator. Require the mode specified by max_pieces
1760 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
1762 (copy_blkmode_to_reg): Likewise.
1763 (set_storage_via_setmem): Likewise.
1764 * optabs.c (prepare_cmp_insn): Likewise.
1765 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1766 * stor-layout.c (finish_bitfield_representative): Likewise.
1768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1769 Alan Hayward <alan.hayward@arm.com>
1770 David Sherwood <david.sherwood@arm.com>
1772 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
1773 * expr.c (convert_move): Use them.
1774 (convert_modes): Likewise.
1775 (store_expr_with_bounds): Likewise.
1777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1778 Alan Hayward <alan.hayward@arm.com>
1779 David Sherwood <david.sherwood@arm.com>
1781 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
1782 parameter for the mode of "x". Remove the "known_x", "known_mode"
1783 and "known_ret" arguments. Change the type of the mode argument
1785 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
1786 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
1787 (reg_num_sign_bit_copies_for_combine): Likewise.
1788 * rtlanal.c (nonzero_bits1): Likewise.
1789 (num_sign_bit_copies1): Likewise.
1790 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
1791 (reg_num_sign_bit_copies_general): Likewise.
1792 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
1793 (reg_nonzero_bits_general): Likewise.
1795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1796 Alan Hayward <alan.hayward@arm.com>
1797 David Sherwood <david.sherwood@arm.com>
1799 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
1800 than in subroutines. Return 1 for non-integer modes.
1801 (cached_num_sign_bit_copies): Change the type of the mode parameter
1803 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
1804 classes. Handle CONST_INT_P first and then check whether X also
1805 has a scalar integer mode. Check the same thing for inner registers
1806 of a SUBREG and for values that are being extended or truncated.
1808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1809 Alan Hayward <alan.hayward@arm.com>
1810 David Sherwood <david.sherwood@arm.com>
1812 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
1813 in subroutines. Return the mode mask for non-integer modes.
1814 (cached_nonzero_bits): Change the type of the mode parameter
1816 (nonzero_bits1): Likewise. Remove early exit for other mode
1817 classes. Handle CONST_INT_P first and then check whether X
1818 also has a scalar integer mode.
1820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1821 Alan Hayward <alan.hayward@arm.com>
1822 David Sherwood <david.sherwood@arm.com>
1824 * expr.c (widest_int_mode_for_size): Make the comment match the code.
1825 Return a scalar_int_mode and assert that the size is greater than
1827 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
1828 (op_by_pieces_d::op_by_pieces_d): Likewise.
1829 (op_by_pieces_d::run): Likewise.
1830 (can_store_by_pieces): Likewise.
1832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1833 Alan Hayward <alan.hayward@arm.com>
1834 David Sherwood <david.sherwood@arm.com>
1836 * combine.c (extract_left_shift): Add a mode argument and update
1838 (make_compound_operation_int): Change the type of the mode parameter
1839 to scalar_int_mode and update the call to extract_left_shift.
1841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1842 Alan Hayward <alan.hayward@arm.com>
1843 David Sherwood <david.sherwood@arm.com>
1845 * combine.c (simplify_and_const_int): Change the type of the mode
1846 parameter to scalar_int_mode.
1847 (simplify_and_const_int_1): Likewise. Update recursive call.
1849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1850 Alan Hayward <alan.hayward@arm.com>
1851 David Sherwood <david.sherwood@arm.com>
1853 * combine.c (simplify_compare_const): Check that the mode is a
1854 scalar_int_mode (rather than VOIDmode) before testing its
1856 (simplify_comparison): Move COMPARISON_P handling out of the
1857 loop and restrict the latter part of the loop to scalar_int_modes.
1858 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
1859 and when considering SUBREG_REGs. Use is_int_mode instead of
1860 checking GET_MODE_CLASS against MODE_INT.
1862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1863 Alan Hayward <alan.hayward@arm.com>
1864 David Sherwood <david.sherwood@arm.com>
1866 * combine.c (try_widen_shift_mode): Move check for equal modes to...
1867 (simplify_shift_const_1): ...here. Use scalar_int_mode for
1868 shift_unit_mode and for modes involved in scalar shifts.
1870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1871 Alan Hayward <alan.hayward@arm.com>
1872 David Sherwood <david.sherwood@arm.com>
1874 * combine.c (force_int_to_mode): New function, split out from...
1875 (force_to_mode): ...here. Keep xmode up-to-date and use it
1876 instead of GET_MODE (x).
1878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1879 Alan Hayward <alan.hayward@arm.com>
1880 David Sherwood <david.sherwood@arm.com>
1882 * optabs-query.h (extraction_insn::struct_mode): Change type to
1883 opt_scalar_int_mode and update comment.
1884 (extraction_insn::field_mode): Change type to scalar_int_mode.
1885 (extraction_insn::pos_mode): Likewise.
1886 * combine.c (make_extraction): Update accordingly.
1887 * optabs-query.c (get_traditional_extraction_insn): Likewise.
1888 (get_optab_extraction_insn): Likewise.
1889 * recog.c (simplify_while_replacing): Likewise.
1890 * expmed.c (narrow_bit_field_mem): Change the type of the mode
1891 parameter to opt_scalar_int_mode.
1893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1894 Alan Hayward <alan.hayward@arm.com>
1895 David Sherwood <david.sherwood@arm.com>
1897 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
1898 to a scalar_int_mode instead of a machine_mode.
1899 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
1900 (get_best_mode): Return a boolean and use a pointer argument to store
1901 the selected mode. Replace the limit mode parameter with a bit limit.
1902 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
1903 for the values returned by bit_field_mode_iterator::next_mode.
1904 (store_bit_field): Update call to get_best_mode.
1905 (store_fixed_bit_field): Likewise.
1906 (extract_fixed_bit_field): Likewise.
1907 * expr.c (optimize_bitfield_assignment_op): Likewise.
1908 * fold-const.c (optimize_bit_field_compare): Likewise.
1909 (fold_truth_andor_1): Likewise.
1910 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
1911 Update for new type of m_mode.
1912 (get_best_mode): As above.
1914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1915 Alan Hayward <alan.hayward@arm.com>
1916 David Sherwood <david.sherwood@arm.com>
1918 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
1919 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
1920 (store_bit_field): Check is_a <scalar_int_mode> before calling
1921 strict_volatile_bitfield_p.
1922 (extract_bit_field): Likewise.
1924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1925 Alan Hayward <alan.hayward@arm.com>
1926 David Sherwood <david.sherwood@arm.com>
1928 * target.def (cstore_mode): Return a scalar_int_mode.
1929 * doc/tm.texi: Regenerate.
1930 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
1931 * targhooks.h (default_cstore_mode): Likewise.
1932 * targhooks.c (default_cstore_mode): Likewise, using a forced
1934 * expmed.c (emit_cstore): Expect the target of the cstore to be
1937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1938 Alan Hayward <alan.hayward@arm.com>
1939 David Sherwood <david.sherwood@arm.com>
1941 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
1943 (niter_desc): Likewise mode.
1944 (iv_analyze): Add a mode parameter.
1946 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
1947 and change its type to scalar_int_mode.
1948 * loop-iv.c: Update commentary at head of file.
1949 (iv_constant): Pass the mode paraeter before the rtx it describes
1950 and change its type to scalar_int_mode. Remove VOIDmode handling.
1951 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
1952 (iv_extend): Likewise.
1953 (shorten_into_mode): Likewise.
1954 (iv_add): Use scalar_int_mode.
1955 (iv_mult): Likewise.
1956 (iv_shift): Likewise.
1957 (canonicalize_iv_subregs): Likewise.
1958 (get_biv_step_1): Pass the outer_mode parameter before the rtx
1959 it describes and change its mode to scalar_int_mode. Also change
1960 the type of the returned inner_mode to scalar_int_mode.
1961 (get_biv_step): Likewise, turning outer_mode from a pointer
1962 into a direct parameter. Update call to get_biv_step_1.
1963 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
1964 iv_constant and get_biv_step.
1965 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
1966 and change its type to scalar_int_mode. Don't initialise iv->mode
1967 to VOIDmode and remove later checks for its still being VOIDmode.
1968 Update calls to iv_analyze_op and iv_analyze_expr. Check
1969 is_a <scalar_int_mode> when changing the mode under consideration.
1970 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
1971 Update call to iv_analyze_expr.
1972 (iv_analyze_op): Add a mode parameter. Reject subregs whose
1973 inner register is not also a scalar_int_mode. Update call to
1975 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
1976 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
1977 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
1978 separate mode class checks. Update calls to iv_analyze. Remove
1979 fix-up of VOIDmodes after iv_analyze_biv.
1980 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
1981 don't have a scalar_int_mode. Update call to biv_p.
1983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1984 Alan Hayward <alan.hayward@arm.com>
1985 David Sherwood <david.sherwood@arm.com>
1987 * cfgexpand.c (convert_debug_memory_address): Use
1988 as_a <scalar_int_mode>.
1989 * combine.c (expand_compound_operation): Likewise.
1990 (make_extraction): Likewise.
1991 (change_zero_ext): Likewise.
1992 (simplify_comparison): Likewise.
1993 * cse.c (cse_insn): Likewise.
1994 * dwarf2out.c (minmax_loc_descriptor): Likewise.
1995 (mem_loc_descriptor): Likewise.
1996 (loc_descriptor): Likewise.
1997 * expmed.c (init_expmed_one_mode): Likewise.
1998 (synth_mult): Likewise.
1999 (emit_store_flag_1): Likewise.
2000 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2001 of a comparison with size.
2002 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2003 (reduce_to_bit_field_precision): Likewise.
2004 * function.c (expand_function_end): Likewise.
2005 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2006 * loop-doloop.c (doloop_modify): Likewise.
2007 * optabs.c (expand_binop): Likewise.
2008 (expand_unop): Likewise.
2009 (expand_copysign_absneg): Likewise.
2010 (prepare_cmp_insn): Likewise.
2011 (maybe_legitimize_operand): Likewise.
2012 * recog.c (const_scalar_int_operand): Likewise.
2013 * rtlanal.c (get_address_mode): Likewise.
2014 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2015 (simplify_cond_clz_ctz): Likewise.
2016 * tree-nested.c (get_nl_goto_field): Likewise.
2017 * tree.c (build_vector_type_for_mode): Likewise.
2018 * var-tracking.c (use_narrower_mode): Likewise.
2020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2021 Alan Hayward <alan.hayward@arm.com>
2022 David Sherwood <david.sherwood@arm.com>
2024 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2025 * builtins.c (expand_builtin_signbit): Use it.
2026 * cfgexpand.c (expand_debug_expr): Likewise.
2027 * dojump.c (do_jump): Likewise.
2028 (do_compare_and_jump): Likewise.
2029 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2030 * expmed.c (make_tree): Likewise.
2031 * expr.c (expand_expr_real_2): Likewise.
2032 (expand_expr_real_1): Likewise.
2033 (try_casesi): Likewise.
2034 * fold-const-call.c (fold_const_call_ss): Likewise.
2035 * fold-const.c (unextend): Likewise.
2036 (extract_muldiv_1): Likewise.
2037 (fold_single_bit_test): Likewise.
2038 (native_encode_int): Likewise.
2039 (native_encode_string): Likewise.
2040 (native_interpret_int): Likewise.
2041 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2042 * internal-fn.c (expand_addsub_overflow): Likewise.
2043 (expand_neg_overflow): Likewise.
2044 (expand_mul_overflow): Likewise.
2045 (expand_arith_overflow): Likewise.
2046 * match.pd: Likewise.
2047 * stor-layout.c (layout_type): Likewise.
2048 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2049 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2050 * tree-ssanames.c (get_range_info): Likewise.
2051 * tree-switch-conversion.c (array_value_type) Likewise.
2052 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2053 (vect_recog_divmod_pattern): Likewise.
2054 (vect_recog_mixed_size_cond_pattern): Likewise.
2055 * tree-vrp.c (extract_range_basic): Likewise.
2056 (simplify_float_conversion_using_ranges): Likewise.
2057 * tree.c (int_fits_type_p): Likewise.
2058 * ubsan.c (instrument_bool_enum_load): Likewise.
2059 * varasm.c (mergeable_string_section): Likewise.
2060 (narrowing_initializer_constant_valid_p): Likewise.
2061 (output_constant): Likewise.
2063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2064 Alan Hayward <alan.hayward@arm.com>
2065 David Sherwood <david.sherwood@arm.com>
2067 * machmode.h (NARROWEST_INT_MODE): New macro.
2068 * expr.c (alignment_for_piecewise_move): Use it instead of
2069 GET_CLASS_NARROWEST_MODE (MODE_INT).
2070 (push_block): Likewise.
2071 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2073 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2076 Alan Hayward <alan.hayward@arm.com>
2077 David Sherwood <david.sherwood@arm.com>
2079 * postreload.c (move2add_valid_value_p): Change the type of the
2080 mode parameter to scalar_int_mode.
2081 (move2add_use_add2_insn): Add a mode parameter and use it instead
2083 (move2add_use_add3_insn): Likewise.
2084 (reload_cse_move2add): Update accordingly.
2086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2087 Alan Hayward <alan.hayward@arm.com>
2088 David Sherwood <david.sherwood@arm.com>
2090 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2092 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2093 * optabs.c (expand_unop): Likewise.
2095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2096 Alan Hayward <alan.hayward@arm.com>
2097 David Sherwood <david.sherwood@arm.com>
2099 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2100 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2101 (popcount_loc_descriptor): Likewise.
2102 (bswap_loc_descriptor): Likewise.
2103 (rotate_loc_descriptor): Likewise.
2104 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2105 calling the functions above.
2107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2108 Alan Hayward <alan.hayward@arm.com>
2109 David Sherwood <david.sherwood@arm.com>
2111 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2113 (try_combine): Likewise.
2114 (simplify_if_then_else): Likewise.
2115 * cse.c (cse_insn): Likewise.
2116 * dwarf2out.c (mem_loc_descriptor): Likewise.
2117 * emit-rtl.c (gen_lowpart_common): Likewise.
2118 * simplify-rtx.c (simplify_truncation): Likewise.
2119 (simplify_binary_operation_1): Likewise.
2120 (simplify_const_relational_operation): Likewise.
2121 (simplify_ternary_operation): Likewise.
2122 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2125 Alan Hayward <alan.hayward@arm.com>
2126 David Sherwood <david.sherwood@arm.com>
2128 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2129 * reload.c (push_reload): Likewise.
2130 (find_reloads): Likewise.
2132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2133 Alan Hayward <alan.hayward@arm.com>
2134 David Sherwood <david.sherwood@arm.com>
2136 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2137 (make_compound_operation_int): Likewise.
2138 (change_zero_ext): Likewise.
2139 * expr.c (convert_move): Likewise.
2140 (convert_modes): Likewise.
2141 * fwprop.c (forward_propagate_subreg): Likewise.
2142 * loop-iv.c (get_biv_step_1): Likewise.
2143 * optabs.c (widen_operand): Likewise.
2144 * postreload.c (move2add_valid_value_p): Likewise.
2145 * recog.c (simplify_while_replacing): Likewise.
2146 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2147 (simplify_binary_operation_1): Likewise. Remove redundant
2148 mode equality check.
2150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2151 Alan Hayward <alan.hayward@arm.com>
2152 David Sherwood <david.sherwood@arm.com>
2154 * combine.c (combine_simplify_rtx): Add checks for
2155 is_a <scalar_int_mode>.
2156 (simplify_if_then_else): Likewise.
2157 (make_field_assignment): Likewise.
2158 (simplify_comparison): Likewise.
2159 * ifcvt.c (noce_try_bitop): Likewise.
2160 * loop-invariant.c (canonicalize_address_mult): Likewise.
2161 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2164 Alan Hayward <alan.hayward@arm.com>
2165 David Sherwood <david.sherwood@arm.com>
2167 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2168 is_a <scalar_int_mode> instead of != BLKmode.
2170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2171 Alan Hayward <alan.hayward@arm.com>
2172 David Sherwood <david.sherwood@arm.com>
2174 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2175 instead of != VOIDmode.
2176 * combine.c (if_then_else_cond): Likewise.
2177 (change_zero_ext): Likewise.
2178 * dwarf2out.c (mem_loc_descriptor): Likewise.
2179 (loc_descriptor): Likewise.
2180 * rtlanal.c (canonicalize_condition): Likewise.
2181 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2184 Alan Hayward <alan.hayward@arm.com>
2185 David Sherwood <david.sherwood@arm.com>
2187 * simplify-rtx.c (simplify_binary_operation_1): Use
2188 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2191 Alan Hayward <alan.hayward@arm.com>
2192 David Sherwood <david.sherwood@arm.com>
2194 * wide-int.h (int_traits<unsigned char>) New class.
2195 (int_traits<unsigned short>) Likewise.
2196 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2197 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2199 * combine.c (set_nonzero_bits_and_sign_copies): Use
2200 is_a <scalar_int_mode>.
2201 (find_split_point): Likewise.
2202 (combine_simplify_rtx): Likewise.
2203 (simplify_logical): Likewise.
2204 (expand_compound_operation): Likewise.
2205 (expand_field_assignment): Likewise.
2206 (make_compound_operation): Likewise.
2207 (extended_count): Likewise.
2208 (change_zero_ext): Likewise.
2209 (simplify_comparison): Likewise.
2210 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2211 (ucompare_loc_descriptor): Likewise.
2212 (minmax_loc_descriptor): Likewise.
2213 (mem_loc_descriptor): Likewise.
2214 (loc_descriptor): Likewise.
2215 * expmed.c (init_expmed_one_mode): Likewise.
2216 * lra-constraints.c (lra_constraint_offset): Likewise.
2217 * optabs.c (prepare_libcall_arg): Likewise.
2218 * postreload.c (move2add_note_store): Likewise.
2219 * reload.c (operands_match_p): Likewise.
2220 * rtl.h (load_extend_op): Likewise.
2221 * rtlhooks.c (gen_lowpart_general): Likewise.
2222 * simplify-rtx.c (simplify_truncation): Likewise.
2223 (simplify_unary_operation_1): Likewise.
2224 (simplify_binary_operation_1): Likewise.
2225 (simplify_const_binary_operation): Likewise.
2226 (simplify_const_relational_operation): Likewise.
2227 (simplify_subreg): Likewise.
2228 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2229 * var-tracking.c (adjust_mems): Likewise.
2230 (prepare_call_arguments): Likewise.
2232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2233 Alan Hayward <alan.hayward@arm.com>
2234 David Sherwood <david.sherwood@arm.com>
2236 * machmode.h (is_int_mode): New fuction.
2237 * combine.c (find_split_point): Use it.
2238 (combine_simplify_rtx): Likewise.
2239 (simplify_if_then_else): Likewise.
2240 (simplify_set): Likewise.
2241 (simplify_shift_const_1): Likewise.
2242 (simplify_comparison): Likewise.
2243 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2244 * cse.c (notreg_cost): Likewise.
2245 (cse_insn): Likewise.
2246 * cselib.c (cselib_lookup_1): Likewise.
2247 * dojump.c (do_jump_1): Likewise.
2248 (do_compare_rtx_and_jump): Likewise.
2249 * dse.c (get_call_args): Likewise.
2250 * dwarf2out.c (rtl_for_decl_init): Likewise.
2251 (native_encode_initializer): Likewise.
2252 * expmed.c (emit_store_flag_1): Likewise.
2253 (emit_store_flag): Likewise.
2254 * expr.c (convert_modes): Likewise.
2255 (store_field): Likewise.
2256 (expand_expr_real_1): Likewise.
2257 * fold-const.c (fold_read_from_constant_string): Likewise.
2258 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2259 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2260 * optabs.c (expand_binop): Likewise.
2261 (expand_unop): Likewise.
2262 (expand_abs_nojump): Likewise.
2263 (expand_one_cmpl_abs_nojump): Likewise.
2264 * simplify-rtx.c (mode_signbit_p): Likewise.
2265 (val_signbit_p): Likewise.
2266 (val_signbit_known_set_p): Likewise.
2267 (val_signbit_known_clear_p): Likewise.
2268 (simplify_relational_operation_1): Likewise.
2269 * tree.c (vector_type_mode): Likewise.
2271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * machmode.h (smallest_mode_for_size): Fix formatting.
2276 (smallest_int_mode_for_size): New function.
2277 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2278 instead of smallest_mode_for_size.
2279 * combine.c (make_extraction): Likewise.
2280 * config/arc/arc.c (arc_expand_movmem): Likewise.
2281 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2282 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2283 * config/s390/s390.c (s390_expand_insv): Likewise.
2284 * config/sparc/sparc.c (assign_int_registers): Likewise.
2285 * config/spu/spu.c (spu_function_value): Likewise.
2286 (spu_function_arg): Likewise.
2287 * coverage.c (get_gcov_type): Likewise.
2288 (get_gcov_unsigned_t): Likewise.
2289 * dse.c (find_shift_sequence): Likewise.
2290 * expmed.c (store_bit_field_1): Likewise.
2291 * expr.c (convert_move): Likewise.
2292 (store_field): Likewise.
2293 * internal-fn.c (expand_arith_overflow): Likewise.
2294 * optabs-query.c (get_best_extraction_insn): Likewise.
2295 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2296 * stor-layout.c (layout_type): Likewise.
2297 (initialize_sizetypes): Likewise.
2298 * targhooks.c (default_get_mask_mode): Likewise.
2299 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2302 Alan Hayward <alan.hayward@arm.com>
2303 David Sherwood <david.sherwood@arm.com>
2305 * machmode.h (opt_mode::else_blk): New function.
2306 (int_mode_for_mode): Declare.
2307 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2308 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2310 * cfgexpand.c (expand_debug_expr): Likewise.
2311 * combine.c (gen_lowpart_or_truncate): Likewise.
2312 (gen_lowpart_for_combine): Likewise.
2313 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2314 * config/avr/avr.c (avr_to_int_mode): Likewise.
2315 (avr_out_plus_1): Likewise.
2316 (avr_out_plus): Likewise.
2317 (avr_out_round): Likewise.
2318 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2319 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2320 (s390_expand_vcond): Likewise.
2321 * config/spu/spu.c (spu_split_immediate): Likewise.
2322 (spu_expand_mov): Likewise.
2323 * dse.c (get_stored_val): Likewise.
2324 * expmed.c (store_bit_field_1): Likewise.
2325 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2327 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2328 (extract_low_bits): Likewise.
2329 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2330 handling rather than repeating the check.
2331 (emit_group_store): Likewise.
2332 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2333 * optabs.c (expand_absneg_bit): Likewise.
2334 (expand_copysign_absneg): Likewise.
2335 (expand_copysign_bit): Likewise.
2336 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2337 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2338 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2339 * var-tracking.c (prepare_call_arguments): Likewise.
2340 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2341 int_mode_for_mode instead of mode_for_size.
2342 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2345 Alan Hayward <alan.hayward@arm.com>
2346 David Sherwood <david.sherwood@arm.com>
2348 * machmode.h (int_mode_for_size): New function.
2349 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2350 instead of mode_for_size.
2351 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2353 * combine.c (expand_field_assignment): Use int_mode_for_size
2354 instead of mode_for_size.
2355 (make_extraction): Likewise.
2356 (simplify_shift_const_1): Likewise.
2357 (simplify_comparison): Likewise.
2358 * dojump.c (do_jump): Likewise.
2359 * dwarf2out.c (mem_loc_descriptor): Likewise.
2360 * emit-rtl.c (init_derived_machine_modes): Likewise.
2361 * expmed.c (flip_storage_order): Likewise.
2362 (convert_extracted_bit_field): Likewise.
2363 * expr.c (copy_blkmode_from_reg): Likewise.
2364 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2365 * internal-fn.c (expand_mul_overflow): Likewise.
2366 * lower-subreg.c (simple_move): Likewise.
2367 * optabs-libfuncs.c (init_optabs): Likewise.
2368 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2369 * tree.c (vector_type_mode): Likewise.
2370 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2371 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2372 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2373 * tree-vect-stmts.c (vectorizable_load): Likewise.
2374 (vectorizable_store): Likewise.
2376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2377 Alan Hayward <alan.hayward@arm.com>
2378 David Sherwood <david.sherwood@arm.com>
2380 * coretypes.h (pod_mode): New type.
2381 (scalar_int_mode_pod): New typedef.
2382 * machmode.h (pod_mode): New class.
2383 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2384 * genmodes.c (emit_mode_int_n): Update accordingly.
2385 * lower-subreg.h (target_lower_subreg): Change type to
2386 scalar_int_mode_pod.
2387 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2388 scalar_int_mode_pod.
2390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2391 Alan Hayward <alan.hayward@arm.com>
2392 David Sherwood <david.sherwood@arm.com>
2394 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2395 machine_mode to scalar_int_mode.
2396 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2397 (rs6000_option_override_internal): Remove cast to int.
2398 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2399 machine_mode to scalar_int_mode.
2400 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2401 (rs6000_option_override_internal): Remove cast to int.
2402 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2403 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2405 * config/s390/s390.c (s390_expand_builtin): Likewise.
2406 * coretypes.h (scalar_int_mode): New type.
2407 (opt_scalar_int_mode): New typedef.
2408 * machmode.h (scalar_int_mode): New class.
2409 (scalar_int_mode::includes_p): New function.
2410 (byte_mode): Change type to scalar_int_mode.
2411 (word_mode): Likewise.
2412 (ptr_mode): Likewise.
2413 * emit-rtl.c (byte_mode): Likewise.
2414 (word_mode): Likewise.
2415 (ptr_mode): Likewise.
2416 (init_derived_machine_modes): Update accordingly.
2417 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2418 and MODE_PARTIAL_INT.
2419 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2420 opt_scalar_int_mode.
2422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2423 Alan Hayward <alan.hayward@arm.com>
2424 David Sherwood <david.sherwood@arm.com>
2426 * target.def (libgcc_floating_mode_supported_p): Take a
2428 * doc/tm.texi: Regenerate.
2429 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2431 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2432 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2440 * doc/tm.texi: Regenerate.
2441 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2442 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2443 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2444 * targhooks.h (default_floatn_mode): Likewise.
2445 * targhooks.c (default_floatn_mode): Likewise.
2446 * tree.c (build_common_tree_nodes): Update accordingly.
2448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2449 Alan Hayward <alan.hayward@arm.com>
2450 David Sherwood <david.sherwood@arm.com>
2452 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2453 (mode_iterator::iterate_p): Likewise.
2454 (mode_iterator::get_wider): Likewise.
2455 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2458 Alan Hayward <alan.hayward@arm.com>
2459 David Sherwood <david.sherwood@arm.com>
2461 * coretypes.h (opt_scalar_float_mode): New typedef.
2462 * machmode.h (float_mode_for_size): New function.
2463 * emit-rtl.c (double_mode): Delete.
2464 (init_emit_once): Use float_mode_for_size.
2465 * stor-layout.c (layout_type): Likewise.
2466 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
2473 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2474 * config/arm/arm.md (consttable_4): Likewise.
2475 (consttable_8): Likewise.
2476 (consttable_16): Likewise.
2477 * config/mips/mips.md (consttable_float): Likewise.
2478 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2479 * varasm.c (assemble_real): Take a scalar_float_mode.
2480 (output_constant_pool_2): Update accordingly.
2481 (output_constant): Likewise.
2483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2484 Alan Hayward <alan.hayward@arm.com>
2485 David Sherwood <david.sherwood@arm.com>
2487 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2488 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2489 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2490 (native_encode_real): Likewise.
2491 (native_interpret_real): Likewise.
2492 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2493 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2500 <scalar_float_mode>. Simplify.
2501 (gen_extend_conv_libfunc): Likewise.
2503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2504 Alan Hayward <alan.hayward@arm.com>
2505 David Sherwood <david.sherwood@arm.com>
2507 * coretypes.h (scalar_float_mode): New type.
2508 * machmode.h (mode_traits::from_int): Use machine_mode if
2509 USE_ENUM_MODES is defined.
2510 (is_a): New function.
2512 (dyn_cast): Likewise.
2513 (scalar_float_mode): New class.
2514 (scalar_float_mode::includes_p): New function.
2515 (is_float_mode): Likewise.
2516 * gdbhooks.py (MachineModePrinter): New class.
2517 (build_pretty_printer): Use it for scalar_float_mode.
2518 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2519 (format_helper::format_helper): Turn into a template.
2520 * genmodes.c (get_mode_class): New function.
2521 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2522 or machine_mode if none.
2523 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2524 as_a <scalar_float_mode>.
2525 * dwarf2out.c (mem_loc_descriptor): Likewise.
2526 (insert_float): Likewise.
2527 (add_const_value_attribute): Likewise.
2528 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2529 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2530 (expand_unop): Update accordingly.
2531 (expand_abs_nojump): Likewise.
2532 (expand_copysign_absneg): Take a scalar_float_mode.
2533 (expand_copysign_bit): Likewise.
2534 (expand_copysign): Update accordingly.
2536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2537 Alan Hayward <alan.hayward@arm.com>
2538 David Sherwood <david.sherwood@arm.com>
2540 * coretypes.h (opt_mode): New class.
2541 * machmode.h (opt_mode): Likewise.
2542 (opt_mode::else_void): New function.
2543 (opt_mode::require): Likewise.
2544 (opt_mode::exists): Likewise.
2545 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2546 (GET_MODE_2XWIDER_MODE): Likewise.
2547 (mode_iterator::get_wider): Update accordingly.
2548 (mode_iterator::get_2xwider): Likewise.
2549 (mode_iterator::get_known_wider): Likewise, turning into a template.
2550 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2551 forcing a wider mode to exist.
2552 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2553 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2554 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2555 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2556 * lower-subreg.c (init_lower_subreg): Likewise.
2557 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2558 on the final iteration.
2559 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
2560 a wider mode exists before asking for a move pattern.
2561 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
2562 forcing a wider mode to exist.
2563 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
2564 returning false if no such mode exists.
2565 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
2566 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
2567 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
2568 Avoid checking for a MODE_INT if we already know the mode is not a
2570 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
2571 forcing a wider mode to exist.
2572 (expmed_mult_highpart_optab): Likewise.
2573 (expmed_mult_highpart): Likewise.
2574 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
2576 * lto-streamer-in.c (lto_input_mode_table): Likewise.
2577 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
2578 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2579 * internal-fn.c (expand_mul_overflow): Update use of
2580 GET_MODE_2XWIDER_MODE.
2581 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2582 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
2583 GET_MODE_WIDER_MODE.
2584 (convert_plusminus_to_widen): Likewise.
2585 * tree-switch-conversion.c (array_value_type): Likewise.
2586 * var-tracking.c (emit_note_insn_var_location): Likewise.
2587 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2588 Return false inside rather than outside the loop if no wider mode
2590 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
2591 and GET_MODE_2XWIDER_MODE
2592 (can_compare_p): Use else_void.
2593 * gdbhooks.py (OptMachineModePrinter): New class.
2594 (build_pretty_printer): Use it for opt_mode.
2596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2597 Alan Hayward <alan.hayward@arm.com>
2598 David Sherwood <david.sherwood@arm.com>
2600 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
2601 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
2603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2604 Alan Hayward <alan.hayward@arm.com>
2605 David Sherwood <david.sherwood@arm.com>
2607 * machmode.h (mode_traits): New structure.
2608 (get_narrowest_mode): New function.
2609 (mode_iterator::start): Likewise.
2610 (mode_iterator::iterate_p): Likewise.
2611 (mode_iterator::get_wider): Likewise.
2612 (mode_iterator::get_known_wider): Likewise.
2613 (mode_iterator::get_2xwider): Likewise.
2614 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
2615 (FOR_EACH_MODE): Likewise.
2616 (FOR_EACH_MODE_FROM): Likewise.
2617 (FOR_EACH_MODE_UNTIL): Likewise.
2618 (FOR_EACH_WIDER_MODE): Likewise.
2619 (FOR_EACH_2XWIDER_MODE): Likewise.
2620 * builtins.c (expand_builtin_strlen): Use new mode iterators.
2621 * combine.c (simplify_comparison): Likewise
2622 * config/i386/i386.c (type_natural_mode): Likewise.
2623 * cse.c (cse_insn): Likewise.
2624 * dse.c (find_shift_sequence): Likewise.
2625 * emit-rtl.c (init_derived_machine_modes): Likewise.
2626 (init_emit_once): Likewise.
2627 * explow.c (hard_function_value): Likewise.
2628 * expmed.c (extract_fixed_bit_field_1): Likewise.
2629 (extract_bit_field_1): Likewise.
2630 (expand_divmod): Likewise.
2631 (emit_store_flag_1): Likewise.
2632 * expr.c (init_expr_target): Likewise.
2633 (convert_move): Likewise.
2634 (alignment_for_piecewise_move): Likewise.
2635 (widest_int_mode_for_size): Likewise.
2636 (emit_block_move_via_movmem): Likewise.
2637 (copy_blkmode_to_reg): Likewise.
2638 (set_storage_via_setmem): Likewise.
2639 (compress_float_constant): Likewise.
2640 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2641 * optabs-query.c (get_best_extraction_insn): Likewise.
2642 * optabs.c (expand_binop): Likewise.
2643 (expand_twoval_unop): Likewise.
2644 (expand_twoval_binop): Likewise.
2645 (widen_leading): Likewise.
2646 (widen_bswap): Likewise.
2647 (expand_parity): Likewise.
2648 (expand_unop): Likewise.
2649 (prepare_cmp_insn): Likewise.
2650 (prepare_float_lib_cmp): Likewise.
2651 (expand_float): Likewise.
2652 (expand_fix): Likewise.
2653 (expand_sfix_optab): Likewise.
2654 * postreload.c (move2add_use_add2_insn): Likewise.
2655 * reg-stack.c (reg_to_stack): Likewise.
2656 * reginfo.c (choose_hard_reg_mode): Likewise.
2657 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2658 * stor-layout.c (mode_for_size): Likewise.
2659 (smallest_mode_for_size): Likewise.
2660 (mode_for_vector): Likewise.
2661 (finish_bitfield_representative): Likewise.
2662 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
2663 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
2664 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2665 * var-tracking.c (prepare_call_arguments): Likewise.
2667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2668 Alan Hayward <alan.hayward@arm.com>
2669 David Sherwood <david.sherwood@arm.com>
2671 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
2672 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
2673 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
2674 * machmode.h (mode_size): Move earlier in file.
2675 (mode_precision): Likewise.
2676 (mode_inner): Likewise.
2677 (mode_nunits): Likewise.
2678 (mode_unit_size): Likewise.
2679 (unit_unit_precision): Likewise.
2680 (mode_wider): Likewise.
2681 (mode_2xwider): Likewise.
2682 (machine_mode): New class.
2683 (mode_to_bytes): New function.
2684 (mode_to_bits): Likewise.
2685 (mode_to_precision): Likewise.
2686 (mode_to_inner): Likewise.
2687 (mode_to_unit_size): Likewise.
2688 (mode_to_unit_precision): Likewise.
2689 (mode_to_nunits): Likewise.
2690 (GET_MODE_SIZE): Use mode_to_bytes.
2691 (GET_MODE_BITSIZE): Use mode_to_bits.
2692 (GET_MODE_PRECISION): Use mode_to_precision.
2693 (GET_MODE_INNER): Use mode_to_inner.
2694 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
2695 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
2696 (GET_MODE_NUNITS): Use mode_to_nunits.
2697 * system.h (ALWAYS_INLINE): New macro.
2698 * config/powerpcspe/powerpcspe-c.c
2699 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
2700 int for arg1_mode and arg2_mode.
2702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2703 Alan Hayward <alan.hayward@arm.com>
2704 David Sherwood <david.sherwood@arm.com>
2706 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2707 Prefix mode names with E_ in case statements.
2708 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2709 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
2710 (aarch64_split_simd_move): Likewise.
2711 (aarch64_gen_storewb_pair): Likewise.
2712 (aarch64_gen_loadwb_pair): Likewise.
2713 (aarch64_gen_store_pair): Likewise.
2714 (aarch64_gen_load_pair): Likewise.
2715 (aarch64_get_condition_code_1): Likewise.
2716 (aarch64_constant_pool_reload_icode): Likewise.
2717 (get_rsqrte_type): Likewise.
2718 (get_rsqrts_type): Likewise.
2719 (get_recpe_type): Likewise.
2720 (get_recps_type): Likewise.
2721 (aarch64_gimplify_va_arg_expr): Likewise.
2722 (aarch64_simd_container_mode): Likewise.
2723 (aarch64_emit_load_exclusive): Likewise.
2724 (aarch64_emit_store_exclusive): Likewise.
2725 (aarch64_expand_compare_and_swap): Likewise.
2726 (aarch64_gen_atomic_cas): Likewise.
2727 (aarch64_emit_bic): Likewise.
2728 (aarch64_emit_atomic_swap): Likewise.
2729 (aarch64_emit_atomic_load_op): Likewise.
2730 (aarch64_evpc_trn): Likewise.
2731 (aarch64_evpc_uzp): Likewise.
2732 (aarch64_evpc_zip): Likewise.
2733 (aarch64_evpc_ext): Likewise.
2734 (aarch64_evpc_rev): Likewise.
2735 (aarch64_evpc_dup): Likewise.
2736 (aarch64_gen_ccmp_first): Likewise.
2737 (aarch64_gen_ccmp_next): Likewise.
2738 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2739 (alpha_emit_xfloating_libcall): Likewise.
2740 (emit_insxl): Likewise.
2741 (alpha_arg_type): Likewise.
2742 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2743 (arc_preferred_simd_mode): Likewise.
2744 (arc_secondary_reload): Likewise.
2745 (get_arc_condition_code): Likewise.
2746 (arc_print_operand): Likewise.
2747 (arc_legitimate_constant_p): Likewise.
2748 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2749 * config/arc/arc.md (casesi_load): Likewise.
2750 (casesi_compact_jump): Likewise.
2751 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2752 (cc_use_register): Likewise.
2753 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2754 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2755 (arm_init_iwmmxt_builtins): Likewise.
2756 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2757 (neon_expand_vector_init): Likewise.
2758 (arm_attr_length_move_neon): Likewise.
2759 (maybe_get_arm_condition_code): Likewise.
2760 (arm_emit_vector_const): Likewise.
2761 (arm_preferred_simd_mode): Likewise.
2762 (arm_output_iwmmxt_tinsr): Likewise.
2763 (thumb1_output_casesi): Likewise.
2764 (thumb2_output_casesi): Likewise.
2765 (arm_emit_load_exclusive): Likewise.
2766 (arm_emit_store_exclusive): Likewise.
2767 (arm_expand_compare_and_swap): Likewise.
2768 (arm_evpc_neon_vuzp): Likewise.
2769 (arm_evpc_neon_vzip): Likewise.
2770 (arm_evpc_neon_vrev): Likewise.
2771 (arm_evpc_neon_vtrn): Likewise.
2772 (arm_evpc_neon_vext): Likewise.
2773 (arm_validize_comparison): Likewise.
2774 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
2775 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
2776 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
2777 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
2778 (c6x_preferred_simd_mode): Likewise.
2779 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
2780 (epiphany_rtx_costs): Likewise.
2781 * config/epiphany/predicates.md (proper_comparison_operator):
2783 * config/frv/frv.c (condexec_memory_operand): Likewise.
2784 (frv_emit_move): Likewise.
2785 (output_move_single): Likewise.
2786 (output_condmove_single): Likewise.
2787 (frv_hard_regno_mode_ok): Likewise.
2788 (frv_matching_accg_mode): Likewise.
2789 * config/h8300/h8300.c (split_adds_subs): Likewise.
2790 (h8300_rtx_costs): Likewise.
2791 (h8300_print_operand): Likewise.
2792 (compute_mov_length): Likewise.
2793 (output_logical_op): Likewise.
2794 (compute_logical_op_length): Likewise.
2795 (compute_logical_op_cc): Likewise.
2796 (h8300_shift_needs_scratch_p): Likewise.
2797 (output_a_shift): Likewise.
2798 (compute_a_shift_length): Likewise.
2799 (compute_a_shift_cc): Likewise.
2800 (expand_a_rotate): Likewise.
2801 (output_a_rotate): Likewise.
2802 * config/i386/i386.c (classify_argument): Likewise.
2803 (function_arg_advance_32): Likewise.
2804 (function_arg_32): Likewise.
2805 (function_arg_64): Likewise.
2806 (function_value_64): Likewise.
2807 (ix86_gimplify_va_arg): Likewise.
2808 (ix86_legitimate_constant_p): Likewise.
2809 (put_condition_code): Likewise.
2810 (split_double_mode): Likewise.
2811 (ix86_avx256_split_vector_move_misalign): Likewise.
2812 (ix86_expand_vector_logical_operator): Likewise.
2813 (ix86_split_idivmod): Likewise.
2814 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
2815 (ix86_build_const_vector): Likewise.
2816 (ix86_build_signbit_mask): Likewise.
2817 (ix86_match_ccmode): Likewise.
2818 (ix86_cc_modes_compatible): Likewise.
2819 (ix86_expand_branch): Likewise.
2820 (ix86_expand_sse_cmp): Likewise.
2821 (ix86_expand_sse_movcc): Likewise.
2822 (ix86_expand_int_sse_cmp): Likewise.
2823 (ix86_expand_vec_perm_vpermi2): Likewise.
2824 (ix86_expand_vec_perm): Likewise.
2825 (ix86_expand_sse_unpack): Likewise.
2826 (ix86_expand_int_addcc): Likewise.
2827 (ix86_split_to_parts): Likewise.
2828 (ix86_vectorize_builtin_gather): Likewise.
2829 (ix86_vectorize_builtin_scatter): Likewise.
2830 (avx_vpermilp_parallel): Likewise.
2831 (inline_memory_move_cost): Likewise.
2832 (ix86_tieable_integer_mode_p): Likewise.
2833 (x86_maybe_negate_const_int): Likewise.
2834 (ix86_expand_vector_init_duplicate): Likewise.
2835 (ix86_expand_vector_init_one_nonzero): Likewise.
2836 (ix86_expand_vector_init_one_var): Likewise.
2837 (ix86_expand_vector_init_concat): Likewise.
2838 (ix86_expand_vector_init_interleave): Likewise.
2839 (ix86_expand_vector_init_general): Likewise.
2840 (ix86_expand_vector_set): Likewise.
2841 (ix86_expand_vector_extract): Likewise.
2842 (emit_reduc_half): Likewise.
2843 (ix86_emit_i387_round): Likewise.
2844 (ix86_mangle_type): Likewise.
2845 (ix86_expand_round_sse4): Likewise.
2846 (expand_vec_perm_blend): Likewise.
2847 (canonicalize_vector_int_perm): Likewise.
2848 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
2849 (expand_vec_perm_1): Likewise.
2850 (expand_vec_perm_interleave3): Likewise.
2851 (expand_vec_perm_even_odd_pack): Likewise.
2852 (expand_vec_perm_even_odd_1): Likewise.
2853 (expand_vec_perm_broadcast_1): Likewise.
2854 (ix86_vectorize_vec_perm_const_ok): Likewise.
2855 (ix86_expand_vecop_qihi): Likewise.
2856 (ix86_expand_mul_widen_hilo): Likewise.
2857 (ix86_expand_sse2_abs): Likewise.
2858 (ix86_expand_pextr): Likewise.
2859 (ix86_expand_pinsr): Likewise.
2860 (ix86_preferred_simd_mode): Likewise.
2861 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
2862 * config/i386/sse.md (*andnot<mode>3): Likewise.
2863 (<mask_codefor><code><mode>3<mask_name>): Likewise.
2864 (*<code><mode>3): Likewise.
2865 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
2866 (ia64_expand_atomic_op): Likewise.
2867 (ia64_arg_type): Likewise.
2868 (ia64_mode_to_int): Likewise.
2869 (ia64_scalar_mode_supported_p): Likewise.
2870 (ia64_vector_mode_supported_p): Likewise.
2871 (expand_vec_perm_broadcast): Likewise.
2872 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
2873 (iq2000_function_arg_advance): Likewise.
2874 (iq2000_function_arg): Likewise.
2875 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
2876 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
2877 (m68k_libcall_value): Likewise.
2878 (m68k_function_value): Likewise.
2879 (sched_attr_op_type): Likewise.
2880 * config/mcore/mcore.c (mcore_output_move): Likewise.
2881 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
2883 (microblaze_function_arg): Likewise.
2884 * config/mips/mips.c (mips16_build_call_stub): Likewise.
2885 (mips_print_operand): Likewise.
2886 (mips_mode_ok_for_mov_fmt_p): Likewise.
2887 (mips_vector_mode_supported_p): Likewise.
2888 (mips_preferred_simd_mode): Likewise.
2889 (mips_expand_vpc_loongson_even_odd): Likewise.
2890 (mips_expand_vec_unpack): Likewise.
2891 (mips_expand_vi_broadcast): Likewise.
2892 (mips_expand_vector_init): Likewise.
2893 (mips_expand_vec_reduc): Likewise.
2894 (mips_expand_msa_cmp): Likewise.
2895 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
2896 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
2897 (cc_flags_for_mode): Likewise.
2898 * config/msp430/msp430.c (msp430_print_operand): Likewise.
2899 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
2900 (nds32_output_casesi_pc_relative): Likewise.
2901 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2902 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
2903 (nvptx_gen_unpack): Likewise.
2904 (nvptx_gen_pack): Likewise.
2905 (nvptx_gen_shuffle): Likewise.
2906 (nvptx_gen_wcast): Likewise.
2907 (nvptx_preferred_simd_mode): Likewise.
2908 * config/pa/pa.c (pa_secondary_reload): Likewise.
2909 * config/pa/predicates.md (base14_operand): Likewise.
2910 * config/powerpcspe/powerpcspe-c.c
2911 (altivec_resolve_overloaded_builtin): Likewise.
2912 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
2914 (rs6000_preferred_simd_mode): Likewise.
2915 (output_vec_const_move): Likewise.
2916 (rs6000_expand_vector_extract): Likewise.
2917 (rs6000_split_vec_extract_var): Likewise.
2918 (reg_offset_addressing_ok_p): Likewise.
2919 (rs6000_legitimate_offset_address_p): Likewise.
2920 (rs6000_legitimize_address): Likewise.
2921 (rs6000_emit_set_const): Likewise.
2922 (rs6000_const_vec): Likewise.
2923 (rs6000_emit_move): Likewise.
2924 (spe_build_register_parallel): Likewise.
2925 (rs6000_darwin64_record_arg_recurse): Likewise.
2926 (swap_selector_for_mode): Likewise.
2927 (spe_init_builtins): Likewise.
2928 (paired_init_builtins): Likewise.
2929 (altivec_init_builtins): Likewise.
2930 (do_load_for_compare): Likewise.
2931 (rs6000_generate_compare): Likewise.
2932 (rs6000_expand_float128_convert): Likewise.
2933 (emit_load_locked): Likewise.
2934 (emit_store_conditional): Likewise.
2935 (rs6000_output_function_epilogue): Likewise.
2936 (rs6000_handle_altivec_attribute): Likewise.
2937 (rs6000_function_value): Likewise.
2938 (emit_fusion_gpr_load): Likewise.
2939 (emit_fusion_p9_load): Likewise.
2940 (emit_fusion_p9_store): Likewise.
2941 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
2942 (fusion_gpr_mem_load): Likewise.
2943 (fusion_addis_mem_combo_load): Likewise.
2944 (fusion_addis_mem_combo_store): Likewise.
2945 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
2946 (fusion_gpr_mem_load): Likewise.
2947 (fusion_addis_mem_combo_load): Likewise.
2948 (fusion_addis_mem_combo_store): Likewise.
2949 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2951 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
2952 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
2953 (rs6000_preferred_simd_mode): Likewise.
2954 (output_vec_const_move): Likewise.
2955 (rs6000_expand_vector_extract): Likewise.
2956 (rs6000_split_vec_extract_var): Likewise.
2957 (reg_offset_addressing_ok_p): Likewise.
2958 (rs6000_legitimate_offset_address_p): Likewise.
2959 (rs6000_legitimize_address): Likewise.
2960 (rs6000_emit_set_const): Likewise.
2961 (rs6000_const_vec): Likewise.
2962 (rs6000_emit_move): Likewise.
2963 (rs6000_darwin64_record_arg_recurse): Likewise.
2964 (swap_selector_for_mode): Likewise.
2965 (paired_init_builtins): Likewise.
2966 (altivec_init_builtins): Likewise.
2967 (rs6000_expand_float128_convert): Likewise.
2968 (emit_load_locked): Likewise.
2969 (emit_store_conditional): Likewise.
2970 (rs6000_output_function_epilogue): Likewise.
2971 (rs6000_handle_altivec_attribute): Likewise.
2972 (rs6000_function_value): Likewise.
2973 (emit_fusion_gpr_load): Likewise.
2974 (emit_fusion_p9_load): Likewise.
2975 (emit_fusion_p9_store): Likewise.
2976 * config/rx/rx.c (rx_gen_move_template): Likewise.
2977 (flags_from_mode): Likewise.
2978 * config/s390/predicates.md (s390_alc_comparison): Likewise.
2979 (s390_slb_comparison): Likewise.
2980 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
2981 (s390_vector_mode_supported_p): Likewise.
2982 (s390_cc_modes_compatible): Likewise.
2983 (s390_match_ccmode_set): Likewise.
2984 (s390_canonicalize_comparison): Likewise.
2985 (s390_emit_compare_and_swap): Likewise.
2986 (s390_branch_condition_mask): Likewise.
2987 (s390_rtx_costs): Likewise.
2988 (s390_secondary_reload): Likewise.
2989 (__SECONDARY_RELOAD_CASE): Likewise.
2990 (s390_expand_cs): Likewise.
2991 (s390_preferred_simd_mode): Likewise.
2992 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
2993 * config/sh/sh.c (sh_print_operand): Likewise.
2994 (dump_table): Likewise.
2995 (sh_secondary_reload): Likewise.
2996 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2997 * config/sh/sh.md (casesi_worker_1): Likewise.
2998 (casesi_worker_2): Likewise.
2999 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3000 (fcc_comparison_operator): Likewise.
3001 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3002 (emit_soft_tfmode_cvt): Likewise.
3003 (sparc_preferred_simd_mode): Likewise.
3004 (output_cbranch): Likewise.
3005 (sparc_print_operand): Likewise.
3006 (sparc_expand_vec_perm_bmask): Likewise.
3007 (vector_init_bshuffle): Likewise.
3008 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3009 (spu_vector_mode_supported_p): Likewise.
3010 (spu_expand_insv): Likewise.
3011 (spu_emit_branch_or_set): Likewise.
3012 (spu_handle_vector_attribute): Likewise.
3013 (spu_builtin_splats): Likewise.
3014 (spu_builtin_extract): Likewise.
3015 (spu_builtin_promote): Likewise.
3016 (spu_expand_sign_extend): Likewise.
3017 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3018 (tilegx_simd_int): Likewise.
3019 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3020 (tilepro_simd_int): Likewise.
3021 * config/v850/v850.c (const_double_split): Likewise.
3022 (v850_print_operand): Likewise.
3023 (ep_memory_offset): Likewise.
3024 * config/vax/vax.c (vax_rtx_costs): Likewise.
3025 (vax_output_int_move): Likewise.
3026 (vax_output_int_add): Likewise.
3027 (vax_output_int_subtract): Likewise.
3028 * config/visium/predicates.md (visium_branch_operator): Likewise.
3029 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3030 (visium_print_operand_address): Likewise.
3031 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3032 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3033 (xtensa_expand_conditional_branch): Likewise.
3034 (xtensa_copy_incoming_a7): Likewise.
3035 (xtensa_output_literal): Likewise.
3036 * dfp.c (decimal_real_maxval): Likewise.
3037 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3040 Alan Hayward <alan.hayward@arm.com>
3041 David Sherwood <david.sherwood@arm.com>
3043 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3044 (mode_nunits_inline): Likewise.
3045 (mode_inner_inline): Likewise.
3046 (mode_unit_size_inline): Likewise.
3047 (mode_unit_precision_inline): Likewise.
3048 (emit_insn_modes_h): Likewise. Also emit a #define of the
3050 (emit_mode_wider): Add an E_ prefix to mode names.
3051 (emit_mode_complex): Likewise.
3052 (emit_mode_inner): Likewise.
3053 (emit_mode_adjustments): Likewise.
3054 (emit_mode_int_n): Likewise.
3055 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3056 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3057 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3058 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3059 (CRC32_BUILTIN, ENTRY): Likewise.
3060 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3061 (aarch64_pop_regs): Likewise.
3062 (aarch64_process_components): Likewise.
3063 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3064 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3065 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3066 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3067 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3068 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3070 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3071 (pcmpistr): Likewise.
3072 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3073 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3074 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3076 * config/rl78/rl78.c (mduc_regs): Likewise.
3077 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3078 (htm_expand_builtin): Likewise.
3079 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3080 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3081 * config/xtensa/xtensa.c (print_operand): Likewise.
3082 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3083 (NUM_MODE_VECTOR_INT): Likewise.
3084 * genoutput.c (null_operand): Likewise.
3085 (output_operand_data): Likewise.
3086 * genrecog.c (print_parameter_value): Likewise.
3087 * lra.c (debug_operand_data): Likewise.
3089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3091 * dwarf2out.c (scompare_loc_descriptor_wide)
3092 (scompare_loc_descriptor_narrow): New functions, split out from...
3093 (scompare_loc_descriptor): ...here.
3094 * expmed.c (emit_store_flag_int): New function, split out from...
3095 (emit_store_flag): ...here.
3097 2017-08-30 Richard Biener <rguenther@suse.de>
3099 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3100 (dwarf2out_early_finish): Move setting of AT_pubnames from
3101 early debug output to early finish.
3103 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3105 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3106 and -mdata-region to the assembler.
3108 2017-08-30 Richard Biener <rguenther@suse.de>
3110 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3112 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3113 (dwarf2out_early_global_decl): Only generate a DIE for the
3114 abstract origin if it doesn't already exist or is a declaration DIE.
3115 (resolve_addr): Do not add the linkage name twice when
3116 generating a stub DIE for the DW_TAG_GNU_call_site target.
3118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3120 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3121 Use machine_mode rather than int for arg1_mode.
3123 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3126 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3127 that the second argument of the built-in functions to unpack
3128 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3129 switch statement instead a lot of if statements.
3130 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3131 Allow 64-bit values to be in Altivec registers as well as
3132 traditional floating point registers.
3133 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3135 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3137 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3138 MAX_REGS_PER_ADDRESS == 1.
3140 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3142 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3143 * config/i386/i386.c (ix86_option_override_internal): Simplify
3144 setting of opts->x_flag_entry.
3146 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3147 Jakub Jelinek <jakub@redhat.com>
3148 Richard Biener <rguenther@suse.de>
3150 PR tree-optimization/81503
3151 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3152 folded constant fits in the target type; reorder tests for clarity.
3154 2017-08-29 Martin Liska <mliska@suse.cz>
3156 * passes.def: Include pass_lower_switch.
3157 * stmt.c (dump_case_nodes): Remove and move to
3158 tree-switch-conversion.
3159 (case_values_threshold): Likewise.
3160 (expand_switch_as_decision_tree_p): Likewise.
3161 (emit_case_decision_tree): Likewise.
3162 (expand_case): Likewise.
3163 (balance_case_nodes): Likewise.
3164 (node_has_low_bound): Likewise.
3165 (node_has_high_bound): Likewise.
3166 (node_is_bounded): Likewise.
3167 (emit_case_nodes): Likewise.
3168 (struct simple_case_node): New struct.
3169 (add_case_node): Remove.
3170 (emit_case_dispatch_table): Use vector instead of case_list.
3171 (reset_out_edges_aux): Remove.
3172 (compute_cases_per_edge): Likewise.
3173 (expand_case): Build list of simple_case_node.
3174 (expand_sjlj_dispatch_table): Use it.
3175 * tree-switch-conversion.c (struct case_node): Moved from
3176 stmt.c and adjusted.
3177 (emit_case_nodes): Likewise.
3178 (node_has_low_bound): Likewise.
3179 (node_has_high_bound): Likewise.
3180 (node_is_bounded): Likewise.
3181 (case_values_threshold): Likewise.
3182 (reset_out_edges_aux): Likewise.
3183 (compute_cases_per_edge): Likewise.
3184 (add_case_node): Likewise.
3185 (dump_case_nodes): Likewise.
3186 (balance_case_nodes): Likewise.
3187 (expand_switch_as_decision_tree_p): Likewise.
3188 (emit_jump): Likewise.
3189 (emit_case_decision_tree): Likewise.
3190 (try_switch_expansion): Likewise.
3191 (do_jump_if_equal): Likewise.
3192 (emit_cmp_and_jump_insns): Likewise.
3193 (fix_phi_operands_for_edge): New function.
3194 (record_phi_operand_mapping): Likewise.
3195 (class pass_lower_switch): New pass.
3196 (pass_lower_switch::execute): New function.
3197 (make_pass_lower_switch): Likewise.
3198 (conditional_probability):
3199 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3200 * tree-pass.h: Add make_pass_lower_switch.
3202 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3205 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3208 2017-08-29 Richard Biener <rguenther@suse.de>
3210 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3211 we do not add a DW_AT_inline attribute twice.
3212 (gen_subprogram_die): Remove code setting DW_AT_inline on
3213 DECL_ABSTRACT_P nodes.
3215 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3217 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3218 calls to internal functions.
3219 (gimplify_modify_expr): Likewise.
3220 * tree-call-cdce.c (use_internal_fn): Likewise.
3221 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3222 (convert_to_divmod): Set the nothrow flag.
3223 * tree-if-conv.c (predicate_mem_writes): Likewise.
3224 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3225 (vectorizable_call): Likewise.
3226 (vectorizable_store): Likewise.
3227 (vectorizable_load): Likewise.
3228 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3229 (vect_recog_mask_conversion_pattern): Likewise.
3231 2017-08-29 Martin Liska <mliska@suse.cz>
3234 * gcc.c (driver_handle_option): Add new argument.
3235 * opts-common.c (handle_option): Pass
3236 target_option_override_hook.
3237 * opts-global.c (lang_handle_option): Add new option.
3238 (set_default_handlers): Add new argument.
3239 (decode_options): Likewise.
3240 * opts.c (target_handle_option): Likewise.
3241 (common_handle_option): Call target_option_override_hook.
3242 * opts.h (struct cl_option_handler_func): Add hook for
3243 target option override.
3244 (struct cl_option_handlers): Likewise.
3245 (set_default_handlers): Add new argument.
3246 (decode_options): Likewise.
3247 (common_handle_option): Likewise.
3248 (target_handle_option): Likewise.
3249 * toplev.c (toplev::main): Pass targetm.target_option.override
3252 2017-08-29 Richard Biener <rguenther@suse.de>
3253 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3255 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3256 life to the active subtree.
3258 2017-08-28 Jeff Law <law@redhat.com>
3260 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3261 derive_equivalences.
3262 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3264 (edge_info::derive_equivalences): New private member function
3266 * tree-ssa-dom.c (class edge_info): Changed from a struct
3267 to a class. Add ctor/dtor, methods and data members.
3268 (edge_info::edge_info): Renamed from allocate_edge_info.
3269 Initialize additional members.
3270 (edge_info::~edge_info): New.
3271 (free_dom_edge_info): Delete the edge info.
3272 (record_edge_info): Use new class & associated member functions.
3273 Tighten forms for testing for edge equivalences.
3274 (record_temporary_equivalences): Iterate over the simple
3275 equivalences rather than assuming there's only one per edge.
3276 (cprop_into_successor_phis): Iterate over the simple
3277 equivalences rather than assuming there's only one per edge.
3278 (optimize_stmt): Use operand_equal_p rather than pointer
3279 equality for mini-DSE code.
3281 2017-08-28 Nathan Sidwell <nathan@acm.org>
3283 * gcc.c (execute): Fold SIGPIPE handling into switch
3284 statement. Adjust internal error message.
3286 2017-08-28 Richard Biener <rguenther@suse.de>
3289 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3290 Do nothing for removed DIEs.
3292 2017-08-28 Richard Biener <rguenther@suse.de>
3294 PR tree-optimization/81977
3295 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3298 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3301 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3302 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3303 using targetm.gen_mem_thread_fence.
3305 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3308 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3309 predicate to register_operand. Reorder operands.
3310 (*btr<mode>): Ditto.
3311 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3312 (*btr<mode>_mask): Ditto.
3314 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3316 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3317 * config/rs6000/xmmintrin.h: New file.
3318 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3320 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3323 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3324 parameter and_insn and return it.
3325 (recombine_lvx_pattern): Insert a copy to ensure availability of
3326 the base register of the copied masking operation at the point of
3327 the instruction replacement.
3328 (recombine_stvx_pattern): Likewise.
3330 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3332 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3333 undocumented switches.
3334 (-mpower9-dform-vector): Likewise.
3335 (-mpower9-dform): Likewise.
3336 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3337 comments to delete references to -mpower9-dform* switches.
3338 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3339 Delete reference to -mpower9-dform* switches, test for
3340 -mpower9-vector instead.
3341 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3342 (OTHER_P9_VECTOR_MASKS): Likewise.
3343 (POWERPC_MASKS): Likewise.
3344 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3345 tests against -mpower9-dform* to -mpower9-vector. Delete code
3346 that checked for -mpower9-dform* consistancy with other options.
3347 Add test for -mpower9-misc to enable other power9 switches.
3348 (rs6000_init_hard_regno_mode_ok): Likewise.
3349 (rs6000_option_override_internal): Likewise.
3350 (rs6000_emit_prologue): Likewise.
3351 (rs6000_emit_epilogue): Likewise.
3352 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3353 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3354 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3356 (emit_fusion_p9_store): Likewise.
3357 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3358 resetting these macros if the assembler does not support ISA 3.0
3360 (TARGET_P9_DFORM_VECTOR): Likewise.
3361 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3362 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3364 2017-08-25 Alan Modra <amodra@gmail.com>
3366 PR rtl-optimization/81747
3367 * cse.c (cse_extended_basic_block): Don't attempt to record
3368 equivalences for degenerate conditional branches that branch
3369 to their fall-through.
3371 2017-08-24 Martin Sebor <msebor@redhat.com>
3374 * gimple-fold.c (size_must_be_zero_p): New function.
3375 (gimple_fold_builtin_memory_op): Call it.
3377 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3379 * config/rs6000/mm_malloc.h: New file.
3381 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3383 PR tree-optimization/81913
3384 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3385 analysis when either IVs in condition can wrap.
3387 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3389 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3390 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3392 2017-08-24 Richard Biener <rguenther@suse.de>
3395 * targhooks.c (default_target_can_inline_p): Properly
3396 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3397 is present and always compare.
3398 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3399 infer -mfpmath=sse from TARGET_SSE_P.
3400 (ix86_can_inline_p): Properly use target_option_default_node when
3401 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3403 2017-08-24 Richard Biener <rguenther@suse.de>
3406 * dwarf2out.c (output_die): Handle flag_generate_offload like
3408 (output_comp_unit): Likewise.
3409 (gen_array_type_die): Likewise.
3410 (dwarf2out_early_finish): Likewise.
3411 (note_variable_value_in_expr): Likewise.
3412 (dwarf2out_finish): Likewise. Adjust assert.
3413 * cgraphunit.c (symbol_table::compile): Move setting of
3414 flag_generate_offload earlier ...
3415 (symbol_table::finalize_compilation_unit): ... here, before
3416 early debug finalization.
3418 2017-08-24 Richard Biener <rguenther@suse.de>
3420 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3421 and ipa-fnsummary.h.
3422 (ix86_can_inline_p): When ix86_fpmath flags do not match
3423 check whether the callee uses FP math at all.
3425 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3428 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3429 instead of TYPE_PRECISION.
3431 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3432 Alan Hayward <alan.hayward@arm.com>
3433 David Sherwood <david.sherwood@arm.com>
3435 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3437 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3439 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3441 (ix86_frame::stack_realign_allocate): New field.
3442 (struct machine_frame_state): Modify comments.
3443 (machine_frame_state::sp_realigned_fp_end): New field.
3444 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3446 (sp_valid_at): Add assertion to assure no attempt to access invalid
3447 offset of a realigned stack.
3448 (fp_valid_at): Likewise.
3449 (choose_baseaddr): Modify comments.
3450 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3451 ix86_expand_prologue.
3452 (ix86_expand_prologue): Modify stack realignment and allocation.
3453 (ix86_expand_epilogue): Modify comments.
3454 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3455 avx2_runtime, avx512f, and avx512f_runtime.
3457 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3459 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3460 (mstackrealign): Do not init to -1.
3461 * config/i386/i386.c (ix86_option_override_internal):
3462 Check opts_set, not opts when setting default value of
3463 opts->x_ix86_force_align_arg_pointer.
3465 2017-08-23 Richard Biener <rguenther@suse.de>
3467 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3468 lang_hooks.decl_printable_name.
3469 * print-rtl-function.c (print_rtx_function): Likewise.
3470 * tree-pretty-print.c (dump_function_header): Likewise.
3472 2017-08-23 Richard Biener <rguenther@suse.de>
3475 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3476 -g0 at compile-time.
3478 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3481 * doc/sourcebuild.texi (Other hardware attributes):
3484 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3487 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3488 Fix single-use check.
3490 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3492 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3494 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3496 * doc/install.texi: Modify to add more details on running selected
3499 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3501 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3502 is combined with -mabi=ms.
3503 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3506 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3508 PR tree-optimization/81488
3509 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3510 and cached_basis fields.
3511 (MAX_SPREAD): New constant.
3512 (alloc_cand_and_find_basis): Initialize new fields.
3513 (clear_visited): New function.
3514 (create_phi_basis_1): Rename from create_phi_basis, set visited
3515 and cached_basis fields.
3516 (create_phi_basis): New wrapper function.
3517 (phi_add_costs_1): Rename from phi_add_costs, add spread
3518 parameter, set visited field, short-circuit when limits reached.
3519 (phi_add_costs): New wrapper function.
3520 (record_phi_increments_1): Rename from record_phi_increments, set
3522 (record_phi_increments): New wrapper function.
3523 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3524 (phi_incr_cost): New wrapper function.
3525 (all_phi_incrs_profitable_1): Rename from
3526 all_phi_incrs_profitable, set visited field.
3527 (all_phi_incrs_profitable): New wrapper function.
3529 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3530 Alan Hayward <alan.hayward@arm.com>
3531 David Sherwood <david.sherwood@arm.com>
3533 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3534 that takes the outer and inner modes.
3535 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3536 comparison as the canonical test for a paradoxical subreg.
3537 * combine.c (simplify_set): Use paradoxical_subreg_p.
3538 (make_extraction): Likewise.
3539 (force_to_mode): Likewise.
3540 (rtx_equal_for_field_assignment_p): Likewise.
3541 (gen_lowpart_for_combine): Likewise.
3542 (simplify_comparison): Likewise.
3543 * cse.c (equiv_constant): Likewise.
3544 * expmed.c (store_bit_field_1): Likewise.
3545 * final.c (alter_subreg): Likewise.
3546 * fwprop.c (propagate_rtx): Likewise.
3547 (forward_propagate_subreg): Likewise.
3548 * ira-conflicts.c (ira_build_conflicts): Likewise.
3549 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3550 * lra-constraints.c (curr_insn_transform): Likewise.
3551 (split_reg): Likewise.
3552 * lra-eliminations.c (move_plus_up): Likewise.
3553 (lra_eliminate_regs_1): Likewise.
3554 * recog.c (general_operand): Likewise.
3555 * ree.c (combine_reaching_defs): Likewise.
3556 * reload.c (push_reload): Likewise.
3557 (find_reloads): Likewise.
3558 * reload1.c (elimination_effects): Likewise.
3559 (compute_reload_subreg_offset): Likewise.
3560 (choose_reload_regs): Likewise.
3561 * rtlanal.c (subreg_lsb_1): Likewise.
3562 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3563 (simplify_subreg): Likewise.
3564 * var-tracking.c (track_loc_p): Likewise.
3565 * emit-rtl.c (byte_lowpart_offset): Likewise.
3566 (paradoxical_subreg_p): Delete out-of-line definition.
3568 2017-08-22 Jeff Law <law@redhat.com>
3570 PR tree-optimization/81741
3571 PR tree-optimization/71947
3572 * tree-ssa-dom.c: Include tree-inline.h.
3573 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
3574 equivalences if one is more expensive to compute than the other.
3575 * tree-ssa-scopedtables.h (class const_or_copies): Make
3576 record_const_or_copy_raw method private.
3577 (class avail_exprs_stack): New method simplify_binary_operation.
3578 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
3579 avail_exprs_stack::simplify_binary_operation as needed.
3580 (avail_exprs_stack::simplify_binary_operation): New function.
3582 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
3584 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
3585 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
3586 (DOT_SYMBOLS): Likewise.
3587 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
3588 (RELOCATABLE_NEEDS_FIXUP): Likewise.
3589 (RS6000_ABI_NAME): Likewise.
3590 (TARGET_CMODEL): Likewise.
3591 (TOC_SECTION_ASM_OP): Likewise.
3592 (SET_CMODEL): New macro.
3593 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
3595 2017-08-22 Richard Biener <rguenther@suse.de>
3597 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
3598 to allow for free-lang-data replacements similar to verify_type_variant.
3600 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3601 Alan Hayward <alan.hayward@arm.com>
3602 David Sherwood <david.sherwood@arm.com>
3604 * config/aarch64/aarch64.md (casesi): Use DImode rather than
3605 VOIDmode for the LABEL_REF.
3607 2017-08-22 Richard Biener <rguenther@suse.de>
3609 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
3611 2017-08-22 Richard Biener <rguenther@suse.de>
3613 * common.opt (feliminate-dwarf2-dups): Ignore.
3614 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
3615 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
3616 same_die_p_wrap, compute_section_prefix,
3617 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
3618 (comdat_symbol_id, comdat_symbol_number): Likewise.
3619 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
3621 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
3622 (output_die): Mark unreachable path unreachable.
3623 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
3624 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
3625 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
3626 (dwarf2out_early_finish): Likewise.
3628 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
3630 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
3632 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
3635 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
3636 not VAR_P. Filter attribute warnings with OPT_Wattributes.
3637 (avr_attribute_table) <io, io_low, address>: Initialize
3638 .decl_required with true.
3640 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
3642 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
3643 undocumented debugging options.
3644 (-mvsx-scalar-double): Likewise.
3645 (-mallow-df-permute): Likewise.
3646 (-mvectorize-builtins): Likewise.
3647 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3648 (rs6000_builtin_vectorized_function): Likewise.
3649 (rs6000_builtin_md_vectorized_function): Likewise.
3650 (rs6000_opt_vars): Likewise.
3652 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
3655 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
3656 (*btrq_imm): Rename from *btrq.
3657 (*btcq_imm): Rename from *btcq.
3658 (btsc): New code attribute.
3659 (*<btsc><mode>): New insn pattern.
3660 (*btr<mode>): Ditto.
3661 (*<btsc><mode>_mask): New insn_and_split pattern.
3662 (*btr<mode>_mask): Ditto.
3664 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3665 Alan Hayward <alan.hayward@arm.com>
3666 David Sherwood <david.sherwood@arm.com>
3668 * function.c (pad_below): Simplify padding calculation.
3670 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3671 Alan Hayward <alan.hayward@arm.com>
3672 David Sherwood <david.sherwood@arm.com>
3674 * target.def (function_prologue): Remove frame size argument.
3675 (function_epilogue): Likewise.
3676 * doc/tm.texi: Regenerate.
3677 * final.c (final_start_function): Update call to function_prologue.
3678 (final_end_function): Update call to function_epilogue.
3679 (default_function_pro_epilogue): Remove frame size argument.
3680 * output.h (default_function_pro_epilogue): Likewise.
3681 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
3682 (arm_output_function_prologue): Likewise.
3683 * config/frv/frv.c (frv_function_prologue): Likewise.
3684 (frv_function_epilogue): Likewise.
3685 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
3686 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
3687 (ia64_output_function_epilogue): Likewise.
3688 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
3689 (m32r_output_function_epilogue): Likewise.
3690 * config/microblaze/microblaze.c (microblaze_function_prologue)
3691 (microblaze_function_epilogue): Likewise.
3692 * config/mips/mips.c (mips_output_function_prologue): Likewise.
3693 (mips_output_function_epilogue): Likewise.
3694 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
3695 (mmix_target_asm_function_epilogue): Likewise.
3696 * config/msp430/msp430.c (msp430_start_function): Likewise.
3697 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
3698 (nds32_asm_function_epilogue): Likewise.
3699 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
3700 * config/pa/pa.c (pa_output_function_prologue): Likewise.
3701 (pa_output_function_epilogue): Likewise.
3702 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
3703 (rs6000_output_function_epilogue): Likewise.
3704 * config/rl78/rl78.c (rl78_start_function): Likewise.
3705 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
3706 (rs6000_output_function_epilogue): Likewise.
3707 * config/rx/rx.c (rx_output_function_prologue): Likewise.
3708 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
3709 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
3710 (sparc_asm_function_epilogue): Likewise.
3712 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3714 * tree.h (type_has_mode_precision_p): New function.
3715 * convert.c (convert_to_integer_1): Use it.
3716 * expr.c (expand_expr_real_2): Likewise.
3717 (expand_expr_real_1): Likewise.
3718 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
3719 * match.pd: Likewise.
3720 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
3721 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
3722 * tree-tailcall.c (process_assignment): Likewise.
3723 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3724 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
3725 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
3726 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3727 (vectorizable_assignment): Likewise.
3728 (vectorizable_shift): Likewise.
3729 (vectorizable_operation): Likewise.
3730 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3732 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3734 * match.pd: Add pow (C, x) simplification.
3736 2017-08-21 Richard Biener <rguenther@suse.de>
3738 PR tree-optimization/81900
3739 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3740 for blocks with abnormal predecessors.
3741 (compute_antic): Do not set visited flag prematurely.
3743 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3746 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3748 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3750 * stor-layout.h (vector_type_mode): Move to...
3751 * tree.h (vector_type_mode): ...here.
3752 * stor-layout.c (vector_type_mode): Move to...
3753 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3755 2017-08-21 Richard Biener <rguenther@suse.de>
3757 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3758 register_external_die hooks.
3759 (debug_false_tree_charstarstar_uhwistar): Declare.
3760 (debug_nothing_tree_charstar_uhwi): Likewise.
3761 * debug.c (do_nothing_debug_hooks): Adjust.
3762 (debug_false_tree_charstarstar_uhwistar): New do nothing.
3763 (debug_nothing_tree_charstar_uhwi): Likewise.
3764 * dbxout.c (dbx_debug_hooks): Adjust.
3765 (xcoff_debug_hooks): Likewise.
3766 * sdbout.c (sdb_debug_hooks): Likewise.
3767 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3768 * dwarf2out.c (macinfo_label_base): New global.
3769 (dwarf2out_register_external_die): New function for the
3770 register_external_die hook.
3771 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
3772 (dwarf2_debug_hooks): Use them.
3773 (dwarf2_lineno_debug_hooks): Adjust.
3774 (struct die_struct): Add with_offset flag.
3775 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
3776 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
3777 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
3778 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
3779 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
3780 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
3781 defining section names for the early LTO debug variants.
3782 (reset_indirect_string): New helper.
3783 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
3784 (print_dw_val): Add support for offsetted symbol references.
3785 (get_ultimate_context): Split out from is_cxx.
3786 (is_cxx): Use get_ultimate_context.
3787 (is_fortran): Add decl overload.
3788 (compute_comp_unit_symbol): Split out worker from
3789 compute_section_prefix.
3790 (compute_section_prefix): Call compute_comp_unit_symbol and
3791 set comdat_type_p here.
3792 (output_die): Skip DIE symbol output for the LTO added one.
3793 Handle DIE symbol references with offset.
3794 (output_comp_unit): Guard section name mangling properly.
3795 For LTO debug sections emit a symbol at the section beginning
3796 which we use to refer to its DIEs.
3797 (add_abstract_origin_attribute): For DIEs registered via
3798 dwarf2out_register_external_die directly refer to the early
3799 DIE rather than indirectly through the shadow one we created.
3800 Remove obsolete call to dwarf2out_abstract_function for
3801 non-function/block origins.
3802 (gen_array_type_die): When generating early LTO debug do
3803 not emit DW_AT_string_length.
3804 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
3805 late when in LTO. As suggested place a gcc_unreachable for
3806 the DECL_ABSTRACT_P case.
3807 (gen_subprogram_die): Avoid another specification DIE
3808 for early built declarations/definitions for the late LTO case.
3809 (gen_variable_die): Add type references for late duplicated VLA dies
3811 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
3812 we have the abstract instance already.
3813 (process_scope_var): Adjust decl DIE contexts in LTO which
3814 first puts them in limbo.
3815 (gen_decl_die): Do not generate type DIEs late apart from
3816 types for VLAs or for decls we do not yet have a DIE. Do not
3817 call dwarf2out_abstract_function late.
3818 (dwarf2out_early_global_decl): Make sure to create DIEs
3819 for abstract instances of a decl first.
3820 (dwarf2out_late_global_decl): Adjust comment.
3821 (output_macinfo_op): With multiple macro sections use
3822 macinfo_label_base to distinguish labels.
3823 (output_macinfo): Likewise. Update macinfo_label_base.
3824 Pass in the line info label.
3825 (note_variable_value_in_expr): When generating LTO resolve
3826 all variable values here by generating DIEs as needed.
3827 (init_sections_and_labels): Add early LTO debug flag parameter
3828 and generate different sections and names if set. Add generation
3829 counter for the labels so we can have multiple of them.
3830 (reset_dies): Helper to allow DIEs to be output multiple times.
3831 (dwarf2out_finish): When outputting DIEs to the fat part of an
3832 LTO object first reset DIEs.
3833 (dwarf2out_early_finish): Output early DIEs when generating LTO.
3834 (modified_type_die): Check for decl_ultimate_origin being self
3836 (gen_type_die_with_usage): Likewise.
3837 (gen_typedef_die): Allow decl_ultimate_origin being self.
3838 (set_decl_abstract_flags): Remove.
3839 (set_block_abstract_flags): Likewise.
3840 (dwarf2out_abstract_function): Treat the early generated DIEs
3841 as the abstract copy and only add DW_AT_inline and
3842 DW_AT_artificial here and call set_decl_origin_self.
3843 If the DIE has an abstract origin don't do anything.
3844 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
3845 if we have none yet (Go fails to build one, PR78628).
3846 (variably_modified_type_p): Prevent endless recursion for Ada
3847 cyclic pointer types.
3848 * lto-streamer-in.c: Include debug.h.
3849 (dref_queue): New global.
3850 (lto_read_tree_1): Stream in DIE references.
3851 (lto_input_tree): Register DIE references.
3852 (input_function): Stream DECL_DEBUG_ARGS.
3853 * lto-streamer-out.c: Include debug.h.
3854 (lto_write_tree_1): Output DIE references.
3855 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
3856 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
3857 (output_function): Stream DECL_DEBUG_ARGS.
3858 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
3859 Stream DECL_ABSTRACT_ORIGIN.
3860 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
3861 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
3862 DECL_CONTEXT for file-scope decls.
3863 * lto-streamer.h (struct dref_entry): Declare.
3864 (dref_queue): Likewise.
3865 * cfgexpand.c (pass_expand::execute): Do not call the
3866 outlining_inline_function hook here.
3867 * lto-wrapper.c (debug_obj): New global.
3868 (tool_cleanup): Unlink it if required.
3869 (debug_objcopy): New function.
3870 (run_gcc): Handle early debug sections in the IL files by
3871 extracting them to separate files, partially linkin them and
3872 feeding the result back as result to the linker.
3873 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
3874 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
3875 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
3876 sections into a separate segment.
3877 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
3879 (darwin_asm_dwarf_section): Likewise.
3880 (darwin_asm_output_dwarf_offset): Likewise.
3881 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
3883 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3884 Alan Hayward <alan.hayward@arm.com>
3885 David Sherwood <david.sherwood@arm.com>
3887 * read-md.h (md_reader::record_potential_iterator_use): Replace
3888 pointer argument with an rtx and an index.
3889 * read-rtl.c (iterator_group::apply_iterator): Likewise.
3890 (apply_mode_iterator): Likewise.
3891 (apply_code_iterator): Likewise.
3892 (apply_int_iterator): Likewise.
3893 (apply_subst_iterator): Likewise.
3894 (record_iterator_use): Likewise.
3895 (record_attribute_use): Likewise.
3896 (md_reader::record_potential_iterator_use): Likewise. Update calls
3897 to record_iterator_use and apply_iterator.
3898 (iterator_use): Replace ptr with x and index.
3899 (attribute_use): Likewise.
3900 (apply_attribute_uses): Update calls to apply_iterator.
3901 (apply_iterators): Likewise. Update initialization of iterator_use.
3902 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
3903 and record_potential_iterator_use.
3904 (rtx_reader::read_rtx_operand): Likewise.
3906 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3907 Alan Hayward <alan.hayward@arm.com>
3908 David Sherwood <david.sherwood@arm.com>
3910 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
3913 2017-08-21 Richard Biener <rguenther@suse.de>
3916 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
3917 at struct end conservatively when comparing common bases.
3919 2017-08-21 Richard Biener <rguenther@suse.de>
3921 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
3922 (mem_ref_in_stmt): Remove.
3923 (determine_max_movement): Use ref index to get at the reference.
3924 (invariantness_dom_walker::before_dom_children): Deal with
3925 lim data already initialized.
3926 (gather_mem_refs_stmt): Initialize lim data and record ref index.
3928 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3930 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
3931 (TARGET_ISA_ROUND): Ditto.
3932 (TARGET_ROUND): Ditto.
3933 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
3934 * config/i386/i386.md: Ditto.
3935 * config/i386/sse.md: Ditto.
3936 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
3937 with OPTION_MASK_ISA_SSE4_1.
3939 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3942 * doc/extend.texi (x86 Built-in Functions): Correct the name of
3943 __builtin_ia32_lzcnt_u16.
3945 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3948 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
3949 (rs6000_set_current_function): Rewrite function to use it.
3951 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
3954 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
3955 and TYPE_WARN_IF_NOT_ALIGN.
3956 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
3957 (handle_warn_if_not_align): New.
3958 (place_union_field): Call handle_warn_if_not_align.
3959 (place_field): Call handle_warn_if_not_align.
3960 Copy TYPE_WARN_IF_NOT_ALIGN.
3961 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
3962 (layout_type): Likewise.
3963 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
3965 (tree_decl_common): Add warn_if_not_align.
3966 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
3967 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
3968 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
3969 (DECL_WARN_IF_NOT_ALIGN): Likewise.
3970 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
3971 * doc/extend.texi: Document warn_if_not_aligned attribute.
3972 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
3974 2017-08-17 Martin Liska <mliska@suse.cz>
3977 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
3978 (get_data_dependence): Use it as pointer type.
3979 (distribute_loop): Likewise.
3981 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3983 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
3984 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
3985 (p8_vmrgow_<mode>_direct): New define_insn.
3986 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
3987 handle endianness for vmrgew and vmrgow permute patterns.
3989 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3991 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
3992 * config/rs6000/rs6000-cpus.def: Remove comment.
3993 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
3994 (POWERPC_MASKS): Likewise.
3995 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
3996 use of TARGET_VSX_TIMODE.
3997 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
3998 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
3999 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4000 (rs6000_option_override_internal): Remove dead code.
4001 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4002 (rs6000_legitimize_reload_address): Likewise.
4003 (rs6000_legitimate_address_p): Likewise.
4004 (rs6000_opt_masks): Delete "vsx-timode".
4005 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4006 from function comment.
4007 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4008 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4009 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4011 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4012 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4013 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4016 (define_peephole2): Likewise.
4018 2017-08-17 Martin Sebor <msebor@redhat.com>
4021 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4022 past the end of an array.
4023 (test_pp_format): Add test cases.
4025 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4027 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4028 missing ECF_NOTHROW flags.
4030 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4033 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4034 operands residing in integer registers.
4035 (*vsx_le_perm_load_<mode>): Likewise.
4036 (*vsx_le_perm_store_<mode>): Likewise.
4037 (define_peephole2): Add peepholes to optimize the above.
4039 2017-08-17 Marek Polacek <polacek@redhat.com>
4042 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4043 to mimic what shorten_compare did. Change the return type to bool.
4044 (fold_cond_expr_with_comparison): Update call to
4045 operand_equal_for_comparison_p.
4046 (fold_ternary_loc): Likewise.
4048 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4050 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4052 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4054 2017-08-17 Richard Biener <rguenther@suse.de>
4056 * tree-ssa-structalias.c (solve_graph): When propagating
4057 to successors update the graphs succ edges and avoid duplicate work.
4059 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4062 * config/i386/i386.c (ix86_option_override_internal): Save target
4063 specific options after ix86_stack_protector_guard_reg was changed.
4065 2017-08-17 Richard Biener <rguenther@suse.de>
4067 PR tree-optimization/81827
4068 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4069 (new_var_info): Initialize it conservatively.
4070 (get_call_vi): Mark register vars.
4071 (new_scalar_tmp_constraint_exp): Likewise.
4072 (handle_rhs_call): Likewise.
4073 (handle_const_call): Likewise.
4074 (create_function_info_for): Likewise.
4075 (solve_constraints): Sort varinfos to separate register from
4076 non-register vars to pack points-to solution bitmaps during
4079 2017-08-17 Marek Polacek <polacek@redhat.com>
4081 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4083 2017-08-17 Richard Biener <rguenther@suse.de>
4085 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4086 to true when overflow is undefined and we saturated the result.
4088 2017-08-17 Alan Modra <amodra@gmail.com>
4091 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4092 Don't use store multiple if only one reg needs saving.
4093 (interesting_frame_related_regno): New function.
4094 (rs6000_frame_related): Don't emit frame info for regs that
4096 (rs6000_emit_epilogue): Likewise.
4098 2017-08-16 Nathan Sidwell <nathan@acm.org>
4100 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4101 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4102 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4103 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4104 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4105 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4106 (verify_type): Adjust for TYPE_BINFO move.
4107 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4108 process TYPE_BINFO directly.
4109 (hash_tree): Likewise.
4110 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4112 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4115 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4117 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4119 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4122 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4123 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4124 Add ix86_binary_operator_ok to insn constraint.
4125 (*iordi_1_bts): Ditto.
4126 (*xordi_1_btc): Ditto.
4127 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4128 Update corresponding peephole2 pattern.
4132 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4134 PR tree-optimization/81832
4135 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4136 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4138 2017-08-16 Marek Polacek <polacek@redhat.com>
4141 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4142 regarding size_zero_node.
4144 2017-08-16 Martin Liska <mliska@suse.cz>
4147 * config.gcc: Respect previously set extra_objs in case
4150 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4152 PR tree-optimization/81835
4153 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4154 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4155 not depend on the phi.
4157 2017-08-16 Alan Modra <amodra@gmail.com>
4159 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4162 2017-08-16 Alan Modra <amodra@gmail.com>
4164 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4165 (save_reg_p): ..into this. Update all callers.
4166 (first_reg_to_save): Simplify.
4168 2017-08-16 Alan Modra <amodra@gmail.com>
4170 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4173 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4177 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4178 constant count if that count is less than 32.
4180 2017-08-15 Nathan Sidwell <nathan@acm.org>
4182 * gcc.c (execute): Emit friendlier message if inferior is killed
4183 by an external cause.
4185 2017-08-15 Richard Biener <rguenther@suse.de>
4187 PR tree-optimization/81790
4188 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4189 CONSTRUCTORs from simplifying and VN.
4191 2017-08-14 Martin Sebor <msebor@redhat.com>
4193 * builtin-attrs.def: Add comments.
4195 2017-08-14 Martin Sebor <msebor@redhat.com>
4198 * doc/extend.texi (attribute nonstring): Document new attribute.
4200 2017-08-14 Martin Sebor <msebor@redhat.com>
4203 * tree-diagnostic.c (default_tree_printer): Handle %G.
4204 * gimple-pretty-print.h (percent_G_format): Declare new function.
4205 * gimple-pretty-print.c (percent_G_format): Define.
4206 * tree-pretty-print.c (percent_K_format): Add argument.
4208 2017-08-14 Martin Sebor <msebor@redhat.com>
4210 PR translation/79998
4211 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4212 Remove a stray space.
4214 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4217 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4218 (*iordi_1_bts): Ditto.
4219 (*xordi_1_btc): Ditto.
4221 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4224 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4225 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4227 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4228 quoted strings, and make more translator-friendly.
4229 (darwin_rs6000_override_options): Likewise.
4230 (rs6000_option_override_internal): Likewise.
4231 (rs6000_return_in_memory): Fix overlong line.
4232 (init_cmulative_args): Use quoted strings, and make more
4233 translator-friendly.
4234 (rs6000_pass_by_reference): Fix overlong line.
4235 (def_builtin): Use quoted strings.
4236 (altivec_expand_predicate_builtin): Use quoted strings, and make
4237 more translator-friendly.
4238 (htm_expand_builtin): Use quoted strings.
4239 (cpu_expand_builtin): Use quoted strings, and make more
4240 translator-friendly.
4241 (altivec_expand_builtin): Likewise.
4242 (paired_expand_predicate_builtin): Likewise.
4243 (rs6000_invalid_builtin): Likewise.
4244 (builtin_function_type): Use quoted strings.
4245 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4246 more translator-friendly.
4247 (rs6000_trampoline_init): Likewise.
4248 (rs6000_handle_altivec_attribute): Likewise.
4249 (rs6000_inner_target_options): Use quoted strings.
4250 (rs6000_disable_incompatible_switches): Likewise.
4251 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4252 strings, and make more translator-friendly.
4253 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4255 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4257 PR tree-optimization/81799
4258 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4259 cond_expr to simple gimple operand.
4261 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4264 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4266 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4270 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4271 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4272 TARGET_GASISR_PROLOGUES.
4273 * config/avr/avr.c (avr_option_override): Same.
4274 (avr_pass_pre_proep::execute): Same.
4276 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4279 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4280 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4282 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4284 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4285 define_insn_and_split. Split to a memory load from 0 in
4286 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4287 using PTR mode iterator.
4288 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4289 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4290 (*add_tp_<mode>): Redefine as define_insn_and_split.
4291 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4292 address space. Merge with *add_tp_x32 using PTR mode iterator.
4293 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4294 Split to an add with a memory load from 0 in
4295 DEFAULT_TLS_SEG_REG address space.
4297 2017-08-12 Andrew Pinski <apinski@cavium.com>
4299 * config/aarch64/aarch64-option-extensions.def (rdma):
4300 Fix feature string to what Linux prints out in /proc/cpuinfo.
4302 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4305 * dwarf2out.c (modified_type_die): For C typedef types that have
4306 an ultimate origin, process the ultimate origin instead of the
4308 (gen_typedef_die): Assert that input DECLs have no ultimate
4310 (gen_type_die_with_usage): For typedef variants that have an
4311 ultimate origin, just call gen_decl_die on the original DECL.
4312 (process_scope_var): Avoid creating DIEs for local typedefs and
4313 concrete static variables.
4315 2017-08-12 Alan Modra <amodra@gmail.com>
4319 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4320 match gnu-user.h startfile.
4321 (ENDFILE_LINUX_SPEC): Similarly.
4323 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4326 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4328 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4330 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4331 * config/aarch64/aarch64.md (mov<mode>): Change.
4332 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4333 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4334 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4336 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4338 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4339 for storage order barriers.
4341 2017-08-11 Martin Liska <mliska@suse.cz>
4344 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4345 variables of void type.
4347 2017-08-11 Martin Liska <mliska@suse.cz>
4349 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4350 TARGET_SUPPORTS_ALIASES.
4351 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4352 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4353 (optimize_weakref): Likewise.
4354 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4355 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4356 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4358 2017-08-11 Martin Liska <mliska@suse.cz>
4361 * config/i386/i386.c (make_resolver_func): Do complete
4362 refactoring of the function.
4364 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4367 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4368 * config/i386/i386.c (ix86_stack_protect_guard): Use
4369 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4370 * doc/invoke.texi (x86 Options): Document
4371 -mstack-protector-guard-symbol= option.
4373 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4375 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4376 * config/i386/i386.c (ix86_split_stack_guard): New function.
4377 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4378 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4379 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4380 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4381 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4382 (split_stack_space_check): Call ix86_split_stack_guard.
4384 2017-08-10 Martin Sebor <msebor@redhat.com>
4386 * print-tree.c (print_node): Print location using the established
4388 Replace spaces with colons.
4389 (debug_raw, debug): Ditto.
4391 2017-08-10 Martin Sebor <msebor@redhat.com>
4394 * pretty-print.c (pp_format): Correct the handling of %s precision.
4396 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4399 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4401 (ix86_finalize_stack_frame_flags): This. Also clear
4402 frame_pointer_needed if -fno-omit-frame-pointer is used without
4404 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4405 with ix86_finalize_stack_frame_flags.
4406 (ix86_expand_epilogue): Likewise.
4407 (ix86_expand_split_stack_prologue): Likewise.
4408 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4410 2017-08-10 Martin Liska <mliska@suse.cz>
4413 * c-attribs.c (handle_target_attribute):
4414 Report warning for an empty string argument of target attribute.
4416 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4419 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4421 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4422 or DECL_NONLOCAL labels.
4423 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4424 or DECL_NONLOCAL labels here.
4426 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4428 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4429 to indicate when early gimple folding has been disabled.
4430 (rs6000_gimple_fold_builtin): Add debug content.
4431 (rs6000_invalid_builtin): Fix whitespace.
4432 (rs6000_expand_builtin): Fix whitespace.
4433 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4435 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4438 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4439 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4441 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4443 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4444 (qdf24xx): Likewise.
4445 * config/aarch64/aarch64.md: Include falkor.md.
4446 * config/aarch64/falkor.md: New.
4448 2017-08-09 Marek Polacek <polacek@redhat.com>
4451 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4452 * diagnostic.c (emit_diagnostic): Add a comment.
4453 (emit_diagnostic_valist): New function.
4455 2017-08-09 Marek Polacek <polacek@redhat.com>
4458 * input.c (make_location): New overload.
4459 * input.h (make_location): Declare.
4461 2017-08-08 Alan Modra <amodra@gmail.com>
4462 H.J. Lu <hongjiu.lu@intel.com>
4465 * gcc.c (NO_PIE_SPEC): Delete.
4466 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4468 (LINK_PIE_SPEC): ..to here.
4469 (LINK_COMMAND_SPEC): Support -no-pie.
4470 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4471 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4473 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4474 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4475 (ENDFILE_CRTEND_SPEC): Similarly.
4477 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4480 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4481 (mstack-protector-guard-offset=): Ditto.
4482 * config/i386/i386.c (ix86_option_override): Handle
4483 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4485 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4486 ix86_stack_protect_guard_offset variables.
4487 (TARGET_STACK_PROTECT_GUARD): Always define.
4488 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4489 and -mstack-protector-guard-offset= options.
4491 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4493 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4494 boundary case for the last candidate.
4496 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4498 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4499 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4501 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4504 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4505 * config/aarch64/aarch64-builtins.c
4506 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4507 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4508 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4510 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4511 Andrew Pinski <pinskia@gmail.com>
4514 * internal-fn.def (XORSIGN): New.
4515 * optabs.def (xorsign_optab): New.
4516 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4517 (convert_expand_mult_copysign): New.
4518 (pass_optimize_widening_mul::execute): Call
4519 convert_expand_mult_copysign.
4521 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4523 PR tree-optimization/81354
4524 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4525 Insert on edges rather than explicitly creating landing pads.
4526 (analyze_candidates_and_replace): Commit edge inserts.
4528 2017-08-08 Richard Biener <rguenther@suse.de>
4531 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4532 (expand_simple_operations): Also look through ADDR_EXPRs with
4533 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4535 2017-08-08 Richard Biener <rguenther@suse.de>
4537 PR tree-optimization/81723
4538 * tree-vect-slp.c (struct bst_traits): New hash traits.
4539 (bst_fail): New global.
4540 (vect_build_slp_tree_2): New worker, split out from ...
4541 (vect_build_slp_tree): ... this now wrapping it with using
4542 bst_fail set to cache SLP tree build fails. Properly handle
4544 (vect_analyze_slp_instance): Allocate and free bst_fail.
4546 2017-08-08 Martin Liska <mliska@suse.cz>
4549 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4550 LABEL_DECLs that can be from a different function.
4552 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4554 PR tree-optimization/81744
4555 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4556 loop's number of iterations.
4558 2017-08-08 Martin Liska <mliska@suse.cz>
4560 * asan.c: Include header files.
4561 * attribs.c (build_decl_attribute_variant): New function moved
4563 (build_type_attribute_qual_variant): Likewise.
4564 (cmp_attrib_identifiers): Likewise.
4565 (simple_cst_list_equal): Likewise.
4566 (omp_declare_simd_clauses_equal): Likewise.
4567 (attribute_value_equal): Likewise.
4568 (comp_type_attributes): Likewise.
4569 (build_type_attribute_variant): Likewise.
4570 (lookup_ident_attribute): Likewise.
4571 (remove_attribute): Likewise.
4572 (merge_attributes): Likewise.
4573 (merge_type_attributes): Likewise.
4574 (merge_decl_attributes): Likewise.
4575 (merge_dllimport_decl_attributes): Likewise.
4576 (handle_dll_attribute): Likewise.
4577 (attribute_list_equal): Likewise.
4578 (attribute_list_contained): Likewise.
4579 * attribs.h (lookup_attribute): New function moved from tree.[ch].
4580 (lookup_attribute_by_prefix): Likewise.
4581 * bb-reorder.c: Include header files.
4582 * builtins.c: Likewise.
4583 * calls.c: Likewise.
4584 * cfgexpand.c: Likewise.
4585 * cgraph.c: Likewise.
4586 * cgraphunit.c: Likewise.
4587 * convert.c: Likewise.
4588 * dwarf2out.c: Likewise.
4589 * final.c: Likewise.
4590 * fold-const.c: Likewise.
4591 * function.c: Likewise.
4592 * gimple-expr.c: Likewise.
4593 * gimple-fold.c: Likewise.
4594 * gimple-pretty-print.c: Likewise.
4595 * gimple.c: Likewise.
4596 * gimplify.c: Likewise.
4597 * hsa-common.c: Likewise.
4598 * hsa-gen.c: Likewise.
4599 * internal-fn.c: Likewise.
4600 * ipa-chkp.c: Likewise.
4601 * ipa-cp.c: Likewise.
4602 * ipa-devirt.c: Likewise.
4603 * ipa-fnsummary.c: Likewise.
4604 * ipa-inline.c: Likewise.
4605 * ipa-visibility.c: Likewise.
4607 * lto-cgraph.c: Likewise.
4608 * omp-expand.c: Likewise.
4609 * omp-general.c: Likewise.
4610 * omp-low.c: Likewise.
4611 * omp-offload.c: Likewise.
4612 * omp-simd-clone.c: Likewise.
4613 * opts-global.c: Likewise.
4614 * passes.c: Likewise.
4615 * predict.c: Likewise.
4616 * sancov.c: Likewise.
4617 * sanopt.c: Likewise.
4618 * symtab.c: Likewise.
4619 * toplev.c: Likewise.
4620 * trans-mem.c: Likewise.
4621 * tree-chkp.c: Likewise.
4622 * tree-eh.c: Likewise.
4623 * tree-into-ssa.c: Likewise.
4624 * tree-object-size.c: Likewise.
4625 * tree-parloops.c: Likewise.
4626 * tree-profile.c: Likewise.
4627 * tree-ssa-ccp.c: Likewise.
4628 * tree-ssa-live.c: Likewise.
4629 * tree-ssa-loop.c: Likewise.
4630 * tree-ssa-sccvn.c: Likewise.
4631 * tree-ssa-structalias.c: Likewise.
4632 * tree-ssa.c: Likewise.
4633 * tree-streamer-in.c: Likewise.
4634 * tree-vectorizer.c: Likewise.
4635 * tree-vrp.c: Likewise.
4637 * ubsan.c: Likewise.
4638 * varasm.c: Likewise.
4639 * varpool.c: Likewise.
4640 * tree.c: Remove functions moved to attribs.[ch].
4642 * config/aarch64/aarch64.c: Add attrs.h header file.
4643 * config/alpha/alpha.c: Likewise.
4644 * config/arc/arc.c: Likewise.
4645 * config/arm/arm.c: Likewise.
4646 * config/avr/avr.c: Likewise.
4647 * config/bfin/bfin.c: Likewise.
4648 * config/c6x/c6x.c: Likewise.
4649 * config/cr16/cr16.c: Likewise.
4650 * config/cris/cris.c: Likewise.
4651 * config/darwin.c: Likewise.
4652 * config/epiphany/epiphany.c: Likewise.
4653 * config/fr30/fr30.c: Likewise.
4654 * config/frv/frv.c: Likewise.
4655 * config/ft32/ft32.c: Likewise.
4656 * config/h8300/h8300.c: Likewise.
4657 * config/i386/winnt.c: Likewise.
4658 * config/ia64/ia64.c: Likewise.
4659 * config/iq2000/iq2000.c: Likewise.
4660 * config/lm32/lm32.c: Likewise.
4661 * config/m32c/m32c.c: Likewise.
4662 * config/m32r/m32r.c: Likewise.
4663 * config/m68k/m68k.c: Likewise.
4664 * config/mcore/mcore.c: Likewise.
4665 * config/microblaze/microblaze.c: Likewise.
4666 * config/mips/mips.c: Likewise.
4667 * config/mmix/mmix.c: Likewise.
4668 * config/mn10300/mn10300.c: Likewise.
4669 * config/moxie/moxie.c: Likewise.
4670 * config/msp430/msp430.c: Likewise.
4671 * config/nds32/nds32-isr.c: Likewise.
4672 * config/nds32/nds32.c: Likewise.
4673 * config/nios2/nios2.c: Likewise.
4674 * config/nvptx/nvptx.c: Likewise.
4675 * config/pa/pa.c: Likewise.
4676 * config/pdp11/pdp11.c: Likewise.
4677 * config/powerpcspe/powerpcspe.c: Likewise.
4678 * config/riscv/riscv.c: Likewise.
4679 * config/rl78/rl78.c: Likewise.
4680 * config/rx/rx.c: Likewise.
4681 * config/s390/s390.c: Likewise.
4682 * config/sh/sh.c: Likewise.
4683 * config/sol2.c: Likewise.
4684 * config/sparc/sparc.c: Likewise.
4685 * config/spu/spu.c: Likewise.
4686 * config/stormy16/stormy16.c: Likewise.
4687 * config/tilegx/tilegx.c: Likewise.
4688 * config/tilepro/tilepro.c: Likewise.
4689 * config/v850/v850.c: Likewise.
4690 * config/vax/vax.c: Likewise.
4691 * config/visium/visium.c: Likewise.
4692 * config/xtensa/xtensa.c: Likewise.
4694 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4697 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
4698 constraints since the -mupper-regs-* switches have been
4700 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
4701 into a vector from a double word element that was extracted from
4702 another vector, and eliminate extra XXPERMDI instructions.
4703 (vsx_concat_<mode>_2): Likewise.
4704 (vsx_concat_<mode>_3): Likewise.
4705 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
4706 concat to allow optimizing inserts from previous extracts.
4708 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
4710 * config/i386/i386.c (ix86_stack_protect_guard): Generate
4711 memory reference to a SSP offset in TLS address space.
4712 (ix86_print_operand) <case '@'>: Remove.
4713 (ix86_print_operand_punct_valid_p): Remove '@' code.
4714 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
4716 (stack_tls_protect_set_<mode>): Remove.
4717 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
4718 (stack_tls_protect_test_<mode>): Remove.
4719 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
4721 2017-08-07 Olivier Hainque <hainque@adacore.com>
4724 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
4726 2017-08-07 Douglas Rupp <rupp@adacore.com>
4728 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4729 variable was referenced as multidir in command.
4731 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4734 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4737 2017-08-07 Martin Liska <mliska@suse.cz>
4739 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4740 * config/rl78/rl78.c: Add include of attribs.h.
4741 * config/sh/sh.c: Likewise.
4742 * config/v850/v850.c: Likewise.
4744 2017-08-07 Tom de Vries <tom@codesourcery.com>
4747 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4749 2017-08-07 Martin Liska <mliska@suse.cz>
4751 * config/mips/mips.c: Include attribs.h.
4753 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4756 * doc/invoke.texi: Document change in behvaior for -Ofast for
4759 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4761 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
4763 (aarch64_pop_regs): Likewise.
4764 (aarch64_gen_load_pair): Likewise.
4765 (aarch64_save_callee_saves): Likewise.
4766 (aarch64_restore_callee_saves): Likewise.
4768 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4770 * config/i386/i386.c: Revert the last change.
4772 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4775 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4777 (ix86_finalize_stack_frame_flags): This. Also clear
4778 frame_pointer_needed if -fno-omit-frame-pointer is used without
4780 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4781 with ix86_finalize_stack_frame_flags.
4782 (ix86_expand_epilogue): Likewise.
4783 (ix86_expand_split_stack_prologue): Likewise.
4785 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4788 * config/i386/i386.c (get_builtin_code_for_version): Set priority
4789 to P_AES for Westmere.
4791 2017-08-07 Jonathan Yong <10walls@gmail.com>
4793 * config/i386/mingw.opt (fset-stack-executable): Removed.
4794 * config/i386/cygming.opt (fset-stack-executable): Moved
4796 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
4798 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
4800 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
4802 2017-08-07 Marek Polacek <polacek@redhat.com>
4805 * fold-const.c (fold_indirect_ref_1): Check type_domain.
4807 2017-08-07 Martin Liska <mliska@suse.cz>
4809 * attribs.h (canonicalize_attr_name): New function.
4810 (cmp_attribs): Move from c-format.c and adjusted.
4811 (is_attribute_p): Moved from tree.h.
4812 * tree-inline.c: Add new includes.
4813 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
4814 (private_is_attribute_p): Remove.
4815 (private_lookup_attribute): Likewise.
4816 (private_lookup_attribute_by_prefix): Simplify.
4817 (remove_attribute): Use is_attribute_p.
4818 * tree.h: Remove removed declarations.
4820 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4823 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
4824 instead of computing it in the function. Formatting fix.
4825 (expand_case): Don't rely on default_edge being the first edge,
4826 clear it if removing it, pass default_edge to
4827 emit_case_dispatch_table.
4828 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
4831 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
4833 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
4834 insn in the function, emit NOP after the insn.
4836 2017-08-06 Tom de Vries <tom@codesourcery.com>
4838 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
4841 2017-08-06 Tom de Vries <tom@codesourcery.com>
4843 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
4846 2017-08-04 Yury Gribov <tetra2005@gmail.com>
4848 PR tree-optimization/57371
4849 * match.pd: New pattern.
4851 2017-08-04 Marek Polacek <polacek@redhat.com>
4854 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
4855 perform the computation in offset_int.
4857 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4859 PR tree-optimization/81136
4860 * tree-vectorizer.h: Include tree-hash-traits.h.
4861 (vec_base_alignments): New typedef.
4862 (vec_info): Add a base_alignments field.
4863 (vect_record_base_alignments): Declare.
4864 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
4866 (DR_IS_CONDITIONAL_IN_STMT): New macro.
4867 (create_data_ref): Add an is_conditional_in_stmt argument.
4868 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
4869 the is_conditional_in_stmt field.
4870 (data_ref_loc): Add an is_conditional_in_stmt field.
4871 (get_references_in_stmt): Set the is_conditional_in_stmt field.
4872 (find_data_references_in_stmt): Update call to create_data_ref.
4873 (graphite_find_data_references_in_stmt): Likewise.
4874 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
4875 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
4876 (vect_record_base_alignment): New function.
4877 (vect_record_base_alignments): Likewise.
4878 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
4879 for nested statements even if we fail to compute a misalignment.
4880 Use pooled base alignments for unconditional references.
4881 (vect_find_same_alignment_drs): Compare base addresses instead
4883 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
4884 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
4886 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4888 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
4889 Add an explicit name for the enum. Use auto_vec for slp_instances
4891 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
4893 (_bb_vec_info): Add a constructor and destructor.
4894 (vinfo_for_stmt): Return NULL for uids of -1 as well.
4895 (destroy_loop_vec_info): Delete.
4896 (vect_destroy_datarefs): Likewise.
4897 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
4898 (vec_info::vec_info): New function.
4899 (vec_info::~vec_info): Likewise.
4900 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
4901 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
4902 destroy_loop_vec_info.
4903 * tree-vect-loop.c (new_loop_vec_info): Replace with...
4904 (_loop_vec_info::_loop_vec_info): ...this.
4905 (destroy_loop_vec_info): Replace with...
4906 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
4907 the stmt_vec_infos. Leave handling of vec_info information to its
4908 destructor. Remove explicit vector releases.
4909 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
4910 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
4911 * tree-vect-slp.c (new_bb_vec_info): Replace with...
4912 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
4913 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
4914 (destroy_bb_vec_info): Replace with...
4915 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
4916 information to its destructor.
4917 (vect_slp_analyze_bb_1): Use new and delete instead of
4918 new_bb_vec_info and destroy_bb_vec_info.
4919 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
4922 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4924 * tree-data-ref.h (subscript): Add access_fn field.
4925 (data_dependence_relation): Add could_be_independent_p.
4926 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
4927 (same_access_functions): Move to tree-data-ref.c.
4928 * tree-data-ref.c (ref_contains_union_access_p): New function.
4929 (access_fn_component_p): Likewise.
4930 (access_fn_components_comparable_p): Likewise.
4931 (dr_analyze_indices): Add a reference to access_fn_component_p.
4932 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
4934 (constant_access_functions): Likewise.
4935 (add_other_self_distances): Likewise.
4936 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
4937 (initialize_data_dependence_relation): Use XCNEW and remove
4938 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
4939 of access functions that have the same type. Allow the
4940 subsequence to end with different bases in some circumstances.
4941 Record the chosen access functions in SUB_ACCESS_FN.
4942 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
4943 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
4944 (subscript_dependence_tester_1): Likewise dra and drb.
4945 (build_classic_dist_vector): Update calls accordingly.
4946 (subscript_dependence_tester): Likewise.
4947 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
4948 DDR_COULD_BE_INDEPENDENT_P.
4949 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
4950 comp_alias_ddrs instead of may_alias_ddrs.
4951 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
4953 (vect_analyze_data_ref_dependence): Use it if
4954 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
4955 distance vectors if that fails.
4956 (dependence_distance_ge_vf): New function.
4957 (vect_prune_runtime_alias_test_list): Use it. Don't clear
4958 LOOP_VINFO_MAY_ALIAS_DDRS.
4960 2017-08-04 Richard Biener <rguenther@suse.de>
4963 * fold-const.c (fold_binary_loc): Properly restrict
4964 minus_var0 && minus_var1 case when associating undefined overflow
4967 2017-08-04 Tom de Vries <tom@codesourcery.com>
4969 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
4971 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4973 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4974 Don't start diagnostic messages with a capital letter.
4975 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4977 (rs6000_invalid_builtin): Likewise.
4978 (rs6000_trampoline_init): Likewise.
4980 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4983 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
4984 after setting changeable df flags.
4986 2017-08-03 Richard Biener <rguenther@suse.de>
4988 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
4989 up if the use is in USE - X.
4991 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
4993 * toplev.c (dumpfile.h): New include.
4994 (internal_error_reentered): New static function. Use it...
4995 (internal_error_function): ...here to handle reentered internal_error.
4997 2017-08-03 Richard Biener <rguenther@suse.de>
5000 * fold-const.c (split_tree): Add minus_var and minus_con
5001 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5002 here but always use minus_*.
5003 (associate_trees): Assert we never associate with MINUS_EXPR
5004 and NULL first operand. Do not recurse for PLUS_EXPR operands
5005 when associating as MINUS_EXPR either.
5006 (fold_binary_loc): Track minus_var and minus_con.
5008 2017-08-03 Tom de Vries <tom@codesourcery.com>
5011 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5012 ACCEL_COMPILER, apply finish_options on
5013 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5015 2017-08-03 Tom de Vries <tom@codesourcery.com>
5018 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5019 function_entry_patch_area_size > 0.
5021 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5024 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5025 instead of 10??LU, perform unit multiplication in wide_int,
5026 don't change alloc_object_size_limit if the limit is larger
5029 PR tree-optimization/81655
5030 PR tree-optimization/81588
5031 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5032 the case when ranges[i].low and high are 1 for unsigned type with
5036 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5037 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5039 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5041 * tree-vrp.h: Add include guard.
5043 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5046 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5047 (ud2): New insn pattern.
5048 * config/i386/i386.c (ix86_expand_epilogue):
5049 For naked functions, generate ud2 instead of trap insn.
5051 2017-08-02 Marek Polacek <polacek@redhat.com>
5054 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5056 2017-08-02 Tom de Vries <tom@codesourcery.com>
5057 Cesar Philippidis <cesar@codesourcery.com>
5059 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5060 Add missing edge probabilities.
5062 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5064 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5067 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5070 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5071 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5072 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5073 prologue_seq sequences - if any.
5075 2017-08-02 Richard Biener <rguenther@suse.de>
5077 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5078 via vectors if supported, integer extracts via punning if supported
5079 or otherwise vector extracts.
5081 2017-08-02 Richard Biener <rguenther@suse.de>
5083 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5085 (bitmap_insert_into_set): ... this.
5087 2017-08-02 Richard Biener <rguenther@suse.de>
5089 PR tree-optimization/81633
5091 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5093 PR tree-optimization/71752
5094 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5096 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5098 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5099 (machine_function::call_ms2sysv_pad_out): Remove field.
5100 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5101 (ix86_compute_frame_layout): Likewise.
5103 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5106 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5107 attribute with interrupt attribute.
5109 2017-08-01 Andrew Pinski <apinski@cavium.com>
5111 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5112 BIT_INSERT_EXPR's operand 1
5113 to see if the types precision matches.
5115 2017-08-01 Martin Liska <mliska@suse.cz>
5118 * builtins.c (expand_builtin_memcpy_args): Remove.
5119 (expand_builtin_memcpy): Call newly added function
5120 expand_builtin_memory_copy_args.
5121 (expand_builtin_memcpy_with_bounds): Likewise.
5122 (expand_builtin_mempcpy): Remove last argument.
5123 (expand_builtin_mempcpy_with_bounds): Likewise.
5124 (expand_builtin_memory_copy_args): New function created from
5125 expand_builtin_mempcpy_args with small modifications.
5126 (expand_builtin_mempcpy_args): Remove.
5127 (expand_builtin_stpcpy): Remove unused argument.
5128 (expand_builtin): Likewise.
5129 (expand_builtin_with_bounds): Likewise.
5131 2017-08-01 Martin Liska <mliska@suse.cz>
5134 Make mempcpy more optimal (PR middle-end/70140).
5136 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5139 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5140 __builtin_vec_cmpne verify both arguments are compatible vectors
5141 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5142 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5143 move computation of aligned to after checking the argument types.
5147 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5148 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5151 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5152 Doug Rupp <rupp@adacore.com>
5153 Olivier Hainque <hainque@adacore.com>
5155 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5156 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5158 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5159 for TARGET_OS_CPP_BUILTIN.
5160 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5161 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5163 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5164 passing required abi options to the assembler for EABI configurations.
5165 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5166 of .text.hot and .text.unlikely sections for kernel modules when
5167 using ARM style exceptions.
5168 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5169 options. Add EXTRA_CC1_SPEC.
5170 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5172 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5174 (ARM_TARGET2_DWARF_FORMAT): Define.
5175 * config/arm/t-vxworks: Adjust multilib control to removal of the
5176 Diab command line options.
5178 2017-08-01 Martin Liska <mliska@suse.cz>
5180 PR gcov-profile/81561
5181 * gcov.c (unblock): Make unblocking safe as we need to preserve
5182 index correspondence of blocks and block_lists.
5184 2017-08-01 Richard Biener <rguenther@suse.de>
5186 PR tree-optimization/81181
5187 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5188 (compute_antic): ... end of iteration here.
5190 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5192 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5193 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5194 (ftree-slp-vectorize): Likewise.
5195 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5196 can no longer be set independent of flag_tree_loop_vectorize.
5197 * omp-general.c (emp_max_vf): Likewise.
5198 * opts.c (enable_fdo_optimizations): Remove references to
5199 flag_tree_vectorize, these are now implicit.
5200 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5201 and leave it for the options machinery.
5203 2017-08-01 Martin Liska <mliska@suse.cz>
5206 * builtins.c (expand_builtin_memcpy_args): Remove.
5207 (expand_builtin_memcpy): Call newly added function
5208 expand_builtin_memory_copy_args.
5209 (expand_builtin_memcpy_with_bounds): Likewise.
5210 (expand_builtin_mempcpy): Remove last argument.
5211 (expand_builtin_mempcpy_with_bounds): Likewise.
5212 (expand_builtin_memory_copy_args): New function created from
5213 expand_builtin_mempcpy_args with small modifications.
5214 (expand_builtin_mempcpy_args): Remove.
5215 (expand_builtin_stpcpy): Remove unused argument.
5216 (expand_builtin): Likewise.
5217 (expand_builtin_with_bounds): Likewise.
5219 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5222 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5223 print "ds:" only for immediates in generic address space.
5225 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5228 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5229 (ix86_function_ok_for_sibcall): Return false for naked functions.
5231 2017-08-01 Richard Biener <rguenther@suse.de>
5233 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5234 (compute_antic): Seed worklist with exit block predecessors.
5235 * cfganal.c (dfs_find_deadend): For a cycle return the source
5236 of the edge closing it.
5238 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5240 * config/aarch64/aarch64.c
5241 (aarch64_can_const_movi_rtx_p): Move 0 check.
5243 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5245 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5246 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5248 * match.pd: Ditto in address comparison pattern.
5250 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5252 PR tree-optimization/81627
5253 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5254 closed ssa form for store-store chain.
5256 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5258 PR tree-optimization/81620
5259 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5260 for store-store chain.
5262 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5264 PR tree-optimization/81588
5265 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5266 ranges[i].in_p, invert comparison code ccode. For >/>=,
5267 swap rhs1 and rhs2 and comparison code unconditionally,
5268 for </<= don't do that. Don't swap rhs1/rhs2 again if
5269 ranges[i].in_p, instead invert comparison code ccode if
5270 opcode or oe->rank is BIT_IOR_EXPR.
5273 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5274 a direct optab to conversion optab.
5275 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5276 with GET_MODE_INNER as last argument instead of optab_handler.
5277 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5278 vector extraction if possible and optab is available.
5279 * expr.c (store_constructor): Use convert_optab_handler instead
5280 of optab_handler. Use vector initialization from smaller
5281 vectors if possible and optab is available.
5282 * tree-vect-stmts.c (vectorizable_load): Likewise.
5283 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5285 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5286 of vec_init from half-sized vectors with the same element mode.
5287 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5288 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5289 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5290 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5291 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5292 after mode in gen_vec_extract* calls.
5293 (vec_extract<mode>): Renamed to ...
5294 (vec_extract<mode><ssescalarmodelower>): ... this.
5295 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5296 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5297 element mode after mode in gen_vec_init* calls.
5298 (VEC_INIT_HALF_MODE): New mode iterator.
5299 (vec_init<mode>): Renamed to ...
5300 (vec_init<mode><ssescalarmodelower>): ... this.
5301 (vec_init<mode><ssehalfvecmodelower>): New expander.
5302 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5303 (vec_extractv2sfsf): ... this.
5304 (vec_initv2sf): Renamed to ...
5305 (vec_initv2sfsf): ... this.
5306 (vec_extractv2si): Renamed to ...
5307 (vec_extractv2sisi): ... this.
5308 (vec_initv2si): Renamed to ...
5309 (vec_initv2sisi): ... this.
5310 (vec_extractv4hi): Renamed to ...
5311 (vec_extractv4hihi): ... this.
5312 (vec_initv4hi): Renamed to ...
5313 (vec_initv4hihi): ... this.
5314 (vec_extractv8qi): Renamed to ...
5315 (vec_extractv8qiqi): ... this.
5316 (vec_initv8qi): Renamed to ...
5317 (vec_initv8qiqi): ... this.
5318 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5319 (vec_init<mode>): Renamed to ...
5320 (vec_init<mode><VEC_base_l>): ... this.
5321 (vec_extract<mode>): Renamed to ...
5322 (vec_extract<mode><VEC_base_l>): ... this.
5323 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5324 (vec_initv2sfsf): ... this.
5325 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5326 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5327 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5328 element mode after mode in gen_vec_init* calls.
5329 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5330 (vec_init<mode><Vel>): ... this.
5331 (vec_extract<mode>): Renamed to ...
5332 (vec_extract<mode><Vel>): ... this.
5333 * config/aarch64/iterators.md (Vel): New mode attribute.
5334 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5335 Add element mode after mode in gen_vec_extract* calls.
5336 * config/s390/vector.md (non_vec_l): New mode attribute.
5337 (vec_extract<mode>): Renamed to ...
5338 (vec_extract<mode><non_vec_l>): ... this.
5339 (vec_init<mode>): Renamed to ...
5340 (vec_init<mode><non_vec_l>): ... this.
5341 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5342 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5344 * config/arm/iterators.md (V_elem_l): New mode attribute.
5345 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5346 (vec_extract<mode><V_elem_l>): ... this.
5347 (vec_extractv2di): Renamed to ...
5348 (vec_extractv2didi): ... this.
5349 (vec_init<mode>): Renamed to ...
5350 (vec_init<mode><V_elem_l>): ... this.
5351 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5352 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5353 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5354 Add element mode after gen_vec_extract* calls.
5355 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5356 (vec_init<mode><unitmode>): ... this.
5357 (vec_extract<mode>): Renamed to ...
5358 (vec_extract<mode><unitmode>): ... this.
5359 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5360 (vec_init<mode><unitmode>): ... this.
5361 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5362 (vec_initv2sfsf): ... this.
5363 (vec_extractv2sf): Renamed to ...
5364 (vec_extractv2sfsf): ... this.
5365 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5366 Add element mode after gen_vec_extract* calls.
5367 * config/mips/mips.md (unitmode): New mode iterator.
5368 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5369 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5370 * config/spu/spu.md (inner_l): New mode attribute.
5371 (vec_init<mode>): Renamed to ...
5372 (vec_init<mode><inner_l>): ... this.
5373 (vec_extract<mode>): Renamed to ...
5374 (vec_extract<mode><inner_l>): ... this.
5375 * config/sparc/sparc.md (veltmode): New mode iterator.
5376 (vec_init<VMALL:mode>): Renamed to ...
5377 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5378 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5379 (vec_initv2sisi): ... this.
5380 (vec_initv2sf): Renamed to ...
5381 (vec_initv2sfsf): ... this.
5382 (vec_extractv2sf): Renamed to ...
5383 (vec_extractv2sfsf): ... this.
5384 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5385 (vec_init<mode>): Renamed to ...
5386 (vec_init<mode><VEC_base_l>): ... this.
5387 (vec_extract<mode>): Renamed to ...
5388 (vec_extract<mode><VEC_base_l>): ... this.
5389 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5390 (vec_initv2sfsf): ... this.
5391 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5392 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5393 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5394 gen_vec_init* calls.
5396 2017-08-01 Richard Biener <rguenther@suse.de>
5398 PR tree-optimization/81297
5399 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5400 TREE_OVERFLOW from INTEGER_CSTs.
5402 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5404 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5406 2017-07-31 Carl Love <cel@us.ibm.com>
5408 * config/rs6000/rs6000-c: Add support for built-in functions
5409 vector signed char vec_xl_be (signed long long, signed char *);
5410 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5411 vector signed int vec_xl_be (signed long long, signed int *);
5412 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5413 vector signed long long vec_xl_be (signed long long, signed long long *);
5414 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5415 vector signed short vec_xl_be (signed long long, signed short *);
5416 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5417 vector double vec_xl_be (signed long long, double *);
5418 vector float vec_xl_be (signed long long, float *);
5419 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5420 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5421 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5423 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5424 (altivec_expand_builtin): Add switch statement to call
5425 altivec_expand_xl_be for each builtin.
5426 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5427 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5428 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5429 __builtin_vsx_le_be_v16qi.
5430 * doc/extend.texi: Update the built-in documentation file for the
5431 new built-in functions.
5433 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5436 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5438 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5440 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5442 * config.gcc: Add z14.
5443 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5444 CPU model numbers for z13s and z14.
5445 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5447 * config/s390/s390-opts.h (enum processor_type): Rename
5448 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5449 * config/s390/s390.c (processor_table): Add field for CPU name to
5450 be passed to Binutils.
5451 (s390_asm_output_machine_for_arch): Use the new field in
5452 processor_table for Binutils.
5453 (s390_expand_builtin): Replace arch12 with z14.
5454 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5455 (s390_get_sched_attrmask): Likewise.
5456 (s390_get_unit_mask): Likewise.
5457 * config/s390/s390.opt: Add z14 to processor_type enum.
5459 2017-07-31 Martin Jambor <mjambor@suse.cz>
5462 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5463 regardless of optimization level.
5465 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5466 Martin Liska <mliska@suse.cz>
5468 * predict.def: Remove old comment and adjust probability.
5469 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5472 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5475 * config/i386/i386.c (ix86_function_naked): New function.
5476 (ix86_can_use_return_insn_p): Return false for naked functions.
5477 (ix86_expand_prologue): Skip prologue for naked functions.
5478 (ix86_expand_epilogue): Skip epilogue for naked functions
5479 and emit trap instruction.
5480 (ix86_warn_func_return): New function.
5481 (ix86_attribute_table): Add "naked" attribute specification.
5482 (TARGET_WARN_FUNC_RETURN): Define.
5483 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5485 2017-07-31 Martin Liska <mliska@suse.cz>
5487 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5488 (dump_gimple_bb_header): Always dump BB info.
5489 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5491 2017-07-31 Martin Liska <mliska@suse.cz>
5494 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5495 also with current_function_decl non-null equality.
5497 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5500 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5501 change type to the element type, instead add eltype variable and
5502 use it where we are interested in the element type.
5504 PR tree-optimization/81603
5505 * ipa-polymorphic-call.c
5506 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5507 offset arithmetic in offset_int, bail out if the resulting bit offset
5508 doesn't fit into shwi.
5510 2017-07-31 Martin Liska <mliska@suse.cz>
5512 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5513 (gimplify_save_expr): Fix comment.
5515 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5518 * config/i386/i386.c (ix86_function_arg): Update arguments for
5520 (ix86_compute_frame_layout): Set the initial stack offset to
5521 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5522 INCOMING_FRAME_SP_OFFSET.
5523 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5524 stack before exception handler returns.
5525 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5526 the 'ERROR_CODE' for exception handler.
5528 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5530 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5531 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5532 (ASM_OUTPUT_REG_POP): Ditto.
5533 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5534 instead of asm_fprintf to output pure string.
5536 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5538 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5539 to imported_module_or_decl hook.
5540 (debug_nothing_tree_tree_tree_bool): Remove.
5541 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5542 * debug.c (do_nothing_debug_hooks): Use
5543 debug_nothing_tree_tree_tree_bool_bool instead of
5544 debug_nothing_tree_tree_tree_bool.
5545 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5546 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5547 * sdbout.c (sdb_debug_hooks): Likewise.
5548 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5549 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5551 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5552 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5553 attribute, don't add anything.
5555 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5557 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5558 (fold_build2_stat_loc): Likewise.
5559 (fold_build3_stat_loc): Likewise.
5560 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
5561 (fold_build1_loc): Remove macro.
5562 (fold_build2_loc): Likewise.
5563 (fold_build3_loc): Likewise.
5565 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5567 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
5568 (gimple_build_debug_bind_source_stat): Likewise.
5569 * gimple.h (gimple_build_debug_bind): Remove macro.
5570 (gimple_build_debug_bind_source): Likewise.
5572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5574 * bitmap.c (bitmap_alloc): Adjust.
5575 (bitmap_gc_alloc): Likewise.
5576 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
5578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5580 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
5581 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
5582 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
5583 (bitmap_gc_alloc_stat): Likewise.
5584 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
5586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5588 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
5589 * rtl.h (shallow_copy_rtx): Remove macro.
5591 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5593 * emit-rtl.c (gen_raw_REG): Adjust.
5594 * gengenrtl.c (gendef): Likewise.
5595 * rtl.c (rtx_alloc_stat): Remove _stat from name.
5596 * rtl.h (rtx_alloc): Remove macro.
5598 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5600 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
5601 (build_tree_list_stat): Likewise.
5602 * tree.h (build_tree_list): Remove macro.
5603 (build_tree_list_vec): Likewise.
5605 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5607 * tree.c (make_vector_stat): Remove _stat from name.
5608 (build_vector_stat): Likewise.
5609 * tree.h (make_vector_stat): Remove macro.
5610 (build_vector_stat): Likewise.
5612 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5614 * tree.h (build_var_debug_value): Remove prototype.
5616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5618 * tree.c (tree_cons_stat): Remove _stat from name.
5619 * tree.h (tree_cons): Remove macro.
5621 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5623 * tree.c (build_vl_exp_stat): Remove _stat from name.
5624 * tree.h (build_vl_exp): Remove macro.
5626 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5628 * tree.c (build_decl_stat): Remove _stat from name.
5629 * tree.h (build_decl): Remove macro.
5631 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5633 * gimple.c (gimple_build_with_ops_stat): Adjust.
5634 (gimple_alloc_stat): Remove _stat from name.
5635 * gimple.h (gimple_alloc): Remove macro.
5637 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5639 * tree.c (make_tree_vec_stat): Remove _stat from name.
5640 (grow_tree_vec_stat): Likewise.
5641 * tree.h (make_tree_vec_stat): Adjust prototype.
5642 (grow_tree_vec_stat): Likewise.
5643 (make_tree_vec): Remove macro.
5644 (grow_tree_vec): Likewise.
5646 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5648 * fold-const.c (fold_build1_stat_loc): Adjust.
5649 (fold_build2_stat_loc): Likewise.
5650 (fold_build3_stat_loc): Likewise.
5651 * tree.c (build0_stat): Remove _stat from name.
5652 (build1_stat): Likewise.
5653 (build2_stat): Likewise.
5654 (build3_stat): Likewise.
5655 (build4_stat): Likewise.
5656 (build5_stat): Likewise.
5657 * tree.h (build1_loc): Remove macro, and rename _stat function
5659 (build2_loc): Likewise.
5660 (build3_loc): Likewise.
5661 (build4_loc): Likewise.
5662 (build5_loc): Likewise.
5664 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5666 * tree.c (make_int_cst_stat): Remove _stat from name.
5667 * tree.h (make_int_cst_stat): Adjust prototype.
5668 (make_int_cst): Remove macro.
5670 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5672 * tree.c (make_tre_binfo_stat): Remove _stat from name.
5673 * tree.h (make_tree_binfo_stat): Adjust prototype.
5674 (make_tree_binfo): Remove.
5676 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5678 * tree.c (copy_node_stat): Rename to copy_node.
5679 (build_distinct_type_copy): Adjust.
5680 * tree.h (copy_node_stat): Adjust prototype.
5681 (copy_node): Remove macro.
5683 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5685 * tree.c (make_node_stat): rename to make_node.
5686 (build_tree_list_stat): Adjust.
5687 (build0_stat): Likewise.
5688 (build2_stat): Likewise.
5689 (build3_stat): Likewise.
5690 (build4_stat): Likewise.
5691 (build5_stat): Likewise.
5692 (build_decl_stat): Likewise.
5693 * tree.h (make_node_stat): Adjust prototype.
5694 (make_node): remove macro.
5696 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
5698 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
5699 (PPC_FEATURE2_SCV): Likewise.
5700 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
5702 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5704 * config/aarch64/aarch64.c
5705 (aarch64_internal_mov_immediate): Add new special pattern.
5706 * config/aarch64/aarch64.md (*movdi_aarch64):
5707 Add reg/32bit const mov case.
5709 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5710 Richard Sandiford <richard.sandiford@linaro.org>
5712 * config/aarch64/aarch64.md (mov<mode>): Generalize.
5713 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5714 Add integer and movi cases.
5715 (movi-split-hf-df-sf split, fp16): New.
5716 (enabled): Added TARGET_FP_F16INST.
5717 * config/aarch64/iterators.md (GPF_HF): New.
5718 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
5720 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5722 * config/aarch64/aarch64.c
5723 (aarch64_simd_container_mode): Add prototype.
5724 (aarch64_expand_mov_immediate): Add HI support.
5725 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
5726 (aarch64_can_const_movi_rtx_p): New.
5727 (aarch64_preferred_reload_class):
5728 Remove restrictions of using FP registers for certain SIMD operations.
5729 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5730 (aarch64_valid_floating_const): Add integer move validation.
5731 (aarch64_simd_imm_scalar_p): Remove.
5732 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5733 (aarch64_legitimate_constant_p): Expand list of supported cases.
5734 * config/aarch64/aarch64-protos.h
5735 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5736 (aarch64_reinterpret_float_as_int): New.
5737 (aarch64_simd_imm_scalar_p): Remove.
5738 * config/aarch64/constraints.md (Uvi): New.
5739 (Dd): Split into Ds and new Dd.
5740 * config/aarch64/aarch64.md (*movsi_aarch64):
5742 (*movdi_aarch64): Add SIMD mov case.
5744 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5746 * tree-predcom.c: (struct chain): Handle store-store chain in which
5747 stores for elimination only store loop invariant values.
5748 (execute_pred_commoning_chain): Ditto.
5749 (prepare_initializers_chain_store_elim): Ditto.
5750 (prepare_finalizers): Ditto.
5751 (is_inv_store_elimination_chain): New function.
5752 (initialize_root_vars_store_elim_1): New function.
5754 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5756 * tree-predcom.c: Revise general description of the pass.
5757 (enum chain_type): New enum type for store elimination.
5758 (struct chain): New field supporting store elimination.
5759 (struct component): Ditto.
5760 (dump_chain): Dump store-stores chain.
5761 (release_chain): Release resources.
5762 (split_data_refs_to_components): Compute and create component
5763 contains only stores for elimination.
5764 (get_chain_last_ref_at): New function.
5765 (make_invariant_chain): Initialization.
5766 (make_rooted_chain): Specify chain type in parameter and record it.
5767 (add_looparound_copies): Skip for store-stores chain.
5768 (determine_roots_comp): Compute type of chain and pass it to
5770 (initialize_root_vars_store_elim_2): New function.
5771 (finalize_eliminated_stores): New function.
5772 (remove_stmt): Handle store for elimination.
5773 (execute_pred_commoning_chain): Execute predictive commoning on
5775 (determine_unroll_factor): Skip unroll for store-stores chain.
5776 (prepare_initializers_chain_store_elim): New function.
5777 (prepare_initializers_chain): Hanlde store-store chain.
5778 (prepare_finalizers_chain, prepare_finalizers): New function.
5779 (tree_predictive_commoning_loop): Return integer value indicating
5780 if loop is unrolled or lcssa form is corrupted.
5781 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
5783 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5785 * tree-predcom.c (initialize_root): Delete.
5786 (execute_pred_commoning_chain): Initialize root vars and replace
5787 reference of non-combined chain directly, rather than call above
5790 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5792 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
5793 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
5795 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5797 * tree-predcom.c (struct chain): New field init_seq.
5798 (release_chain): Release init_seq.
5799 (prepare_initializers_chain): Record intialization stmts in above
5801 (insert_init_seqs): New function.
5802 (tree_predictive_commoning_loop): Call insert_init_seqs.
5804 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5806 * tree-predcom.c (determine_roots_comp): Skip trivial components.
5808 2017-07-28 Richard Biener <rguenther@suse.de>
5810 * match.pd: Remove superfluous :c.
5811 * genmatch.c (simplify::id): Add member.
5812 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
5814 (current_id): New global.
5815 (dt_node::parent): Move from ...
5816 (dt_operand::parent): ... here. Add for_id member.
5817 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
5818 (decision_tree::find_node): Relax order requirement when
5819 merging DT_TRUE nodes to ones inbetween the current simplify
5820 and the one we try to merge with. Add diagnostic whenever
5821 we need to enforce pattern order by not merging.
5822 (decision_tree::insert): Set current_id.
5823 (decision_tree::print_node): Dump parent node and for_id.
5824 (parser::last_id): Add member.
5825 (parser::push_simplify): Assign unique id.
5826 (parser::parser): Initialize last_id.
5828 2017-07-28 Martin Liska <mliska@suse.cz>
5831 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
5832 gimple_build_debug_bind.
5834 2017-07-28 Richard Biener <rguenther@suse.de>
5836 PR tree-optimization/81502
5837 * match.pd: Add pattern combining BIT_INSERT_EXPR with
5839 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
5841 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
5842 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
5843 for BIT_FIELD_REF args.
5844 * fold-const.c (make_bit_field_ref): Likewise.
5845 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5847 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5850 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
5851 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
5852 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
5853 Or it into SANITIZER_UNDEFINED.
5854 * ubsan.c: Include gimple-fold.h and varasm.h.
5855 (ubsan_expand_ptr_ifn): New function.
5856 (instrument_pointer_overflow): New function.
5857 (maybe_instrument_pointer_overflow): New function.
5858 (instrument_object_size): Formatting fix.
5859 (pass_ubsan::execute): Call instrument_pointer_overflow
5860 and maybe_instrument_pointer_overflow.
5861 * internal-fn.c (expand_UBSAN_PTR): New function.
5862 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
5863 * sanitizer.def (__ubsan_handle_pointer_overflow,
5864 __ubsan_handle_pointer_overflow_abort): New builtins.
5865 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
5866 * internal-fn.def (UBSAN_PTR): New internal function.
5867 * opts.c (sanitizer_opts): Add pointer-overflow.
5868 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
5869 * fold-const.c (build_range_check): Compute pointer range check in
5870 integral type if pointer arithmetics would be needed. Formatting
5873 2017-07-28 Martin Liska <mliska@suse.cz>
5876 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
5877 parameters that are of a variable-length.
5879 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
5881 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
5883 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
5884 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
5885 (CRT_CALL_STATIC_FUNCTION): Likewise.
5886 (ASM_DEFAULT_SPEC): New define.
5887 (ASM_SPEC32): Likewise.
5888 (ASM_SPEC64): Likewise.
5889 (ASM_SPEC_COMMON): Likewise.
5890 (ASM_SPEC): Likewise.
5891 (INVALID_64BIT): Likewise.
5892 (LINK_OS_DEFAULT_SPEC): Likewise.
5893 (LINK_OS_SPEC32): Likewise.
5894 (LINK_OS_SPEC64): Likewise.
5895 (POWERPC_LINUX): Likewise.
5896 (PTRDIFF_TYPE): Likewise.
5897 (RESTORE_FP_PREFIX): Likewise.
5898 (RESTORE_FP_SUFFIX): Likewise.
5899 (SAVE_FP_PREFIX): Likewise.
5900 (SAVE_FP_SUFFIX): Likewise.
5901 (SIZE_TYPE): Likewise.
5902 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
5903 (TARGET_64BIT): Likewise.
5904 (TARGET_64BIT): Likewise.
5905 (TARGET_AIX): Likewise.
5906 (WCHAR_TYPE_SIZE): Likewise.
5907 (WCHAR_TYPE): Undefine.
5908 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
5909 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
5910 (CPP_OS_RTEMS_SPEC): Delete.
5911 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
5912 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
5914 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
5916 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5918 PR tree-optimization/81578
5919 * tree-parloops.c (build_new_reduction): Bail out if
5920 reduction_code isn't one of the standard OpenMP reductions.
5921 Move the details printing after that decision.
5923 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5925 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
5926 related to reload_in_progress.
5927 (splat_input_operand): Likewise.
5928 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
5930 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
5932 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
5933 (TARGET_INSTANTIATE_DECLS): Likewise.
5934 (legitimate_indexed_address_p): Delete reload_in_progress code.
5935 (rs6000_debug_legitimate_address_p): Likewise.
5936 (rs6000_eliminate_indexed_memrefs): Likewise.
5937 (rs6000_emit_le_vsx_store): Likewise.
5938 (rs6000_emit_move_si_sf_subreg): Likewise.
5939 (rs6000_emit_move): Likewise.
5940 (register_to_reg_type): Likewise.
5941 (rs6000_pre_atomic_barrier): Likewise.
5942 (rs6000_machopic_legitimize_pic_address): Likewise.
5943 (rs6000_allocate_stack_temp): Likewise.
5944 (rs6000_address_for_fpconvert): Likewise.
5945 (rs6000_address_for_altivec): Likewise.
5946 (rs6000_secondary_memory_needed_rtx): Delete function.
5947 (rs6000_check_sdmode): Likewise.
5948 (rs6000_alloc_sdmode_stack_slot): Likewise.
5949 (rs6000_instantiate_decls): Likewise.
5950 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
5951 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
5952 Delete reload_in_progress.
5953 (*vec_reload_and_plus_<mptrsize>): Likewise.
5954 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
5955 (vsx_div_v2di): Likewise.
5956 (vsx_udiv_v2di): Likewise.
5958 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5960 * config/rs6000/rs6000.opt (mlra): Replace with stub.
5961 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
5962 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
5963 (rs6000_debug_reg_global): Delete print of LRA status.
5964 (rs6000_option_override_internal): Delete dead LRA related code.
5965 (rs6000_lra_p): Delete function.
5966 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
5968 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
5970 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
5971 * config/riscv/rtems.h: New file.
5973 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5974 Sudakshina Das <sudi.das@arm.com>
5976 * config/aarch64/aarch64.md
5977 (define_split for and<mode>3nr_compare): Move
5978 non aarch64_logical_operand to a register.
5979 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
5980 register immediate operand to a register.
5981 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
5983 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5986 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
5988 2017-07-27 Richard Biener <rguenther@suse.de>
5990 PR tree-optimization/81573
5991 PR tree-optimization/81494
5992 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
5993 multi defuse cycle case.
5995 2017-07-27 Richard Biener <rguenther@suse.de>
5997 PR tree-optimization/81571
5998 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6001 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6003 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6004 earlier and only if MASK_FPU is set. Adjust formatting.
6006 2017-07-27 Martin Liska <mliska@suse.cz>
6008 * opt-functions.awk: Add validation of value of Init.
6009 * optc-gen.awk: Pass new argument.
6011 2017-07-27 Martin Liska <mliska@suse.cz>
6013 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6014 Fix wrong condition.
6016 2017-07-27 Martin Liska <mliska@suse.cz>
6018 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6019 BBs and edges seen by autoFDO.
6021 2017-07-27 Richard Biener <rguenther@suse.de>
6023 PR tree-optimization/81502
6024 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6025 with incompatible but same sized type.
6026 (execute_update_addresses_taken): Likewise.
6028 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6030 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6031 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6033 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6036 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6037 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6038 Change s_operand to memory_operand.
6040 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6042 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6043 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6044 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6045 Emit instructions rather than returning an expression. Handle TFmode
6046 and KFmode by casting to TImode.
6047 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6048 (rs6000_emit_le_vsx_store): Likewise.
6049 * config/rs6000/vsx.md (VSX_TI): New iterator.
6050 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6051 (*vsx_le_undo_permute_<mode>): Likewise.
6052 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6053 emit the split sequence.
6054 (*vsx_le_perm_store_<mode>): Likewise.
6056 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6058 PR tree-optimization/81555
6059 PR tree-optimization/81556
6060 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6061 if true, force CHANGED for the recursive invocation.
6062 (reassociate_bb): Remember original length of ops array, pass
6063 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6065 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6066 attributes for noipa attribute. For naked attribute use
6067 lookup_attribute first before lookup_attribute_spec.
6068 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6070 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6071 for functions with noipa attribute.
6072 (cgraph_externally_visible_p): Return true for functions with noipa
6074 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6075 for functions with noipa attribute.
6076 * doc/extend.texi: Document noipa function attribute.
6077 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6078 also for functions with noipa attribute.
6079 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6081 2017-07-26 Andrew Pinski <apinski@cavium.com>
6083 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6084 vec_unalign_load_cost and vec_unalign_store_cost.
6086 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6088 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6089 -mvsx-small-integer option.
6090 (ISA_3_0_MASKS_IEEE): Likewise.
6091 (OTHER_VSX_VECTOR_MASKS): Likewise.
6092 (POWERPC_MASKS): Likewise.
6093 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6094 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6095 code, only testing for DImode being allowed in non-VSX floating
6097 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6098 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6100 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6101 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6102 TARGET_P8_VECTOR test.
6103 (rs6000_secondary_reload_simple_move): Likewise.
6104 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6105 since TARGET_P9_VECTOR was already tested.
6106 (rs6000_opt_masks): Remove -mvsx-small-integer.
6107 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6108 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6110 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6111 test for TARGET_VEXTRACTUB was used, and that uses
6113 (p9 extract splitter): Likewise.
6114 (vsx_extract_<mode>_di_p9): Likewise.
6115 (vsx_extract_<mode>_store_p9): Likewise.
6116 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6117 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6118 the elimination of TARGET_VSX_SMALL_INTEGER.
6119 (vsx_extract_<mode>_p8): Likewise.
6120 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6121 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6122 (vsx_set_<mode>_p9): Likewise.
6123 (vsx_set_v4sf_p9): Likewise.
6124 (vsx_set_v4sf_p9_zero): Likewise.
6125 (vsx_insert_extract_v4sf_p9): Likewise.
6126 (vsx_insert_extract_v4sf_p9_2): Likewise.
6127 * config/rs6000/rs6000.md (sign extend splitter): Change
6128 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6129 (floatsi<mode>2_lfiwax_mem): Likewise.
6130 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6131 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6132 since a test for TARGET_P9_VECTOR was used.
6133 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6134 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6135 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6136 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6137 TARGET_P8_VECTOR test.
6138 (fix_trunc<mode>si2_stfiwx): Likewise.
6139 (fix_trunc<mode>si2_internal): Likewise.
6140 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6141 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6143 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6144 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6145 TARGET_P8_VECTOR test.
6146 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6147 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6148 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6150 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6151 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6152 since a test for TARGET_P9_VECTOR was used.
6153 (splitter for loading small constants): Likewise.
6155 2017-07-26 Andrew Pinski <apinski@cavium.com>
6157 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6160 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6163 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6164 (fp_valid_at): Likewise.
6166 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6168 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6169 (qdf24xx_addrcost_table): Likewise.
6170 (cortexa57_tunings): Update to use generic_branch_cost.
6171 (cortexa72_tunings): Likewise.
6172 (cortexa73_tunings): Likewise.
6173 (qdf24xx_tunings): Likewise.
6175 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6177 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6178 (thunderx2t99_branch_cost): Likewise.
6179 (cortexa35_tunings): Update to use generic_branch_cost.
6180 (cortexa53_tunings): Likewise.
6181 (cortexa57_tunings): Likewise.
6182 (cortexa72_tunings): Likewise.
6183 (cortexa73_tunings): Likewise.
6184 (thunderx2t99_tunings): Likewise.
6186 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6188 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6189 (sparc_option_override): Honour MASK_FSMULD.
6190 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6191 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6192 * config/sparc/sparc.opt (mfsmuld): New option.
6193 * doc/invoke.texi (mfsmuld): Document option.
6195 2017-07-26 Marek Polacek <polacek@redhat.com>
6198 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6200 2017-07-26 Richard Biener <rguenther@suse.de>
6202 * gimple-match-head.c (do_valueize): Return OP if valueize
6204 (get_def): New helper to get at the def stmt of a SSA name
6206 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6207 do_valueize to get at the def stmt.
6208 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6210 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6213 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6215 2017-07-26 Martin Liska <mliska@suse.cz>
6218 * function.c (expand_function_start): Make expansion of
6219 nonlocal_goto_save_area after parm_birth_insn.
6221 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6223 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6224 from all CPU target flags enable members.
6226 2017-07-26 Richard Biener <rguenther@suse.de>
6228 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6229 (decision_tree::gen): Make 'type' const.
6230 (write_predicate): Likewise.
6232 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6234 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6235 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6236 (rs6000_option_override_internal): Likewise.
6237 (rs6000_expand_vector_set): Likewise.
6238 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6239 (TARGET_UPPER_REGS_SF): Likewise.
6240 (TARGET_UPPER_REGS_DI): Likewise.
6241 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6242 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6243 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6244 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6245 (Splitters for DI constants in Altivec registers): Likewise.
6246 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6247 (vsx_set_v4sf_p9): Likewise.
6248 (vsx_set_v4sf_p9_zero): Likewise.
6249 (vsx_insert_extract_v4sf_p9): Likewise.
6250 (vsx_insert_extract_v4sf_p9_2): Likewise.
6252 2017-07-25 Carl Love <cel@us.ibm.com>
6254 * doc/extend.texi: Update the built-in documentation file for the
6255 existing built-in functions
6256 vector signed char vec_cnttz (vector signed char);
6257 vector unsigned char vec_cnttz (vector unsigned char);
6258 vector signed short vec_cnttz (vector signed short);
6259 vector unsigned short vec_cnttz (vector unsigned short);
6260 vector signed int vec_cnttz (vector signed int);
6261 vector unsigned int vec_cnttz (vector unsigned int);
6262 vector signed long long vec_cnttz (vector signed long long);
6263 vector unsigned long long vec_cnttz (vector unsigned long long);
6265 2017-07-25 Andrew Pinski <apinski@cavium.com>
6267 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6268 accesses where the use is for the first operand of a BIT_INSERT.
6270 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6273 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6274 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6276 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6278 * config/i386/gstabs.h: Delete.
6279 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6281 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6283 * config/i386/i386.c (ix86_decompose_address): Do not check for
6284 register RTX when looking at index_reg or base_reg.
6285 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6287 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6289 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6290 to update EH info here.
6292 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6294 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6296 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6298 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6300 2017-07-25 Torsten Duwe <duwe@suse.de>
6302 * common.opt: Introduce -fpatchable-function-entry
6303 command line option, and its variables function_entry_patch_area_size
6304 and function_entry_patch_area_start.
6305 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6306 including a two-value parser.
6307 * target.def (print_patchable_function_entry): New target hook.
6308 * targhooks.h (default_print_patchable_function_entry): New function.
6309 * targhooks.c (default_print_patchable_function_entry): Likewise.
6310 * toplev.c (process_options): Switch off IPA-RA if
6311 patchable function entries are being generated.
6312 * varasm.c (assemble_start_function): Look at the
6313 patchable-function-entry command line switch and current
6314 function attributes and maybe generate NOP instructions by
6315 calling the print_patchable_function_entry hook.
6316 * doc/extend.texi: Document patchable_function_entry attribute.
6317 * doc/invoke.texi: Document -fpatchable_function_entry
6318 command line option.
6319 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6321 * doc/tm.texi: Re-generate.
6323 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6326 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6327 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6329 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6331 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6333 2017-07-25 Richard Biener <rguenther@suse.de>
6335 PR tree-optimization/81455
6336 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6337 not walk in cycles when looking for guards.
6339 2017-07-25 Richard Biener <rguenther@suse.de>
6341 PR tree-optimization/81529
6342 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6343 when optimizing backedge uses.
6345 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6347 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6349 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6350 to dl_section_ref. On AIX, append an expression to subtract
6351 the size of the section length to dl_section_ref.
6353 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6355 * configure.ac: If any of the config.* scripts fail, exit 1.
6356 * configure: Regenerate.
6358 2017-07-25 Richard Biener <rguenther@suse.de>
6361 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6362 of immediate uses, be more verbose on errors.
6364 2017-07-25 Richard Biener <rguenther@suse.de>
6366 PR tree-optimization/81510
6367 * tree-vect-loop.c (vect_is_simple_reduction): When the
6368 reduction stmt is not inside the loop bail out.
6370 2017-07-25 Richard Biener <rguenther@suse.de>
6372 PR tree-optimization/81303
6373 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6374 profitability check against LOOP_VINFO_NITERSM1.
6376 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6378 * domwalk.c (cmp_bb_postorder): Simplify.
6379 (sort_bbs_postorder): New function. Use it...
6380 (dom_walker::walk): ...here to optimize common cases.
6382 2017-07-25 Martin Liska <mliska@suse.cz>
6385 * ipa-visibility.c (function_and_variable_visibility): Make the
6386 redirection just on target that supports aliasing.
6387 Fix GNU coding style.
6389 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6392 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6394 * config/arm/rtems.h: Likewise.
6395 * config/bfin/rtems.h: Likewise.
6396 * config/i386/rtemself.h: Likewise.
6397 * config/lm32/rtems.h: Likewise.
6398 * config/m32c/rtems.h: Likewise.
6399 * config/m68k/rtemself.h: Likewise.
6400 * config/microblaze/rtems.h: Likewise.
6401 * config/mips/rtems.h: Likewise.
6402 * config/moxie/rtems.h: Likewise.
6403 * config/nios2/rtems.h: Likewise.
6404 * config/powerpcspe/rtems.h: Likewise.
6405 * config/rs6000/rtems.h: Likewise.
6406 * config/rtems.h: Likewise.
6407 * config/sh/rtems.h: Likewise.
6408 * config/sh/rtemself.h: Likewise.
6409 * config/sparc/rtemself.h: Likewise.
6411 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6414 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6415 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6416 * tree-ssa-structalias.c (alias_get_name): Same.
6418 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6421 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6422 instructions if no du chain is found.
6424 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6426 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6428 2017-07-25 Richard Biener <rguenther@suse.de>
6431 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6434 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6436 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6438 (ISA_2_7_MASKS_SERVER): Likewise.
6439 (ISA_3_0_MASKS_IEEE): Likewise.
6440 (OTHER_P8_VECTOR_MASKS): Likewise.
6441 (OTHER_VSX_VECTOR_MASKS): Likewise.
6442 (POWERPC_MASKS): Likewise.
6443 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6444 duplicate list of options.
6445 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6446 explicit -mupper-regs options.
6447 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6448 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6449 alias for -mupper-regs-df.
6450 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6451 (rs6000_init_hard_regno_mode_ok): Likewise.
6452 (rs6000_option_override_internal): Likewise.
6453 (rs6000_opt_masks): Likewise.
6454 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6455 options in terms of whether -mvsx or -mpower8-vector was used.
6456 (TARGET_UPPER_REGS_DI): Likewise.
6457 (TARGET_UPPER_REGS_SF): Likewise.
6458 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6459 -mupper-regs-* options.
6461 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6463 * passes.c (emergency_dump_function): Print some empty lines and a
6464 header before the RTL dump.
6466 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6468 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6470 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6473 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6474 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6476 2017-07-24 Carl Love <cel@us.ibm.com>
6478 * config/rs6000/rs6000-c.c: Add support for built-in functions
6479 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6480 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6481 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6482 vec_extract_fp_from_shortl): Add defines for the two builtins.
6483 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6484 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6486 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6487 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6488 * doc/extend.texi: Update the built-in documentation file for the
6489 new built-in function.
6491 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6494 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6496 * doc/generic.texi: Likewise.
6497 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6498 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6500 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6502 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6503 (aarch64_mls_elt_merge<mode>): Likewise.
6505 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6507 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6508 having __cxa_atexit.
6510 2017-07-23 Michael Collison <michael.collison@arm.com>
6512 * config/arm/arm.c (arm_option_override): Deprecate
6513 use of -mstructure-size-boundary.
6514 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6515 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6517 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6520 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6521 Reduce cost estimate for direct moves.
6523 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6526 * config/i386/i386.c (ix86_option_override_internal): Disable
6527 BMI, BMI2 and TBM instructions for -m16.
6529 2017-07-21 Carl Love <cel@us.ibm.com>
6531 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6532 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6533 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6534 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6535 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6536 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6537 VMULOSW): New enum "unspec" values.
6538 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6539 altivec_vmulosw): New patterns.
6540 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6541 VMULOSW): Add definitions.
6543 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6545 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6546 (qdf24xx): Likewise.
6547 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6548 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6549 (AARCH64_FL_V8_1): Renumber.
6550 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6551 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6552 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6553 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6554 rdma to feature modifiers list.
6556 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6559 * ipa-visibility (function_and_variable_visibility): Convert
6560 recursive PLT call to direct call if appropriate.
6562 2017-07-21 Andrew Pinski <apinski@cavium.com>
6564 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
6565 operand 1 to see if the types precision matches.
6566 * fold-const.c (operand_equal_p): Likewise.
6568 2017-07-21 Richard Biener <rguenther@suse.de>
6570 PR tree-optimization/81303
6571 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
6572 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
6573 (vect_peeling_hash_get_lowest_cost): Adjust.
6574 (vect_enhance_data_refs_alignment): Likewise. Use
6575 vect_get_peeling_costs_all_drs to compute the penalty for no
6576 peeling to match up costs.
6578 2017-07-21 Richard Biener <rguenther@suse.de>
6580 PR tree-optimization/81500
6581 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
6582 we didn't identify a reduction path.
6584 2017-07-21 Tom de Vries <tom@codesourcery.com>
6585 Cesar Philippidis <cesar@codesourcery.com>
6587 PR gcov-profile/81442
6588 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
6591 2017-07-21 Tom de Vries <tom@codesourcery.com>
6594 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
6596 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
6597 nvptx_override_options_after_change.
6599 2017-07-21 Ulrich Drepper <drepper@redhat.com>
6601 * dwarf2out.c (output_file_names): Avoid double testing for
6604 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
6606 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
6608 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6610 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
6612 (try_crossjump_to_edge): Do not punt on partitioned functions.
6614 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6616 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6617 Put all BBs reachable only via paths crossing cold region to cold
6619 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
6621 2016-07-21 Richard Biener <rguenther@suse.de>
6623 PR tree-optimization/81303
6624 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
6625 into account prologue and epilogue iterations when raising
6626 min_profitable_iters to sth at least covering one vector iteration.
6628 2017-07-21 Tamar Christina <tamar.christina@arm.com>
6630 * config/arm/arm.c (arm_test_cpu_arch_dat):
6633 2017-07-20 Nathan Sidwell <nathan@acm.org>
6635 Remove TYPE_METHODS.
6636 * tree.h (TYPE_METHODS): Delete.
6637 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
6638 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
6639 (dbxout_type_methods): Scan TYPE_FIELDS.
6640 (dbxout_type): Don't check TYPE_METHODS here.
6641 * function.c (use_register_for_decl): Always ignore register for
6642 class types when not optimizing.
6643 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
6644 * tree.c (free_lang_data_in_type): Stitch out member functions and
6645 templates from TYPE_FIELDS.
6646 (build_distinct_type_copy, verify_type_variant,
6647 verify_type): Member fns are on TYPE_FIELDS.
6648 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
6649 * tree-pretty-print.c (dump_generic_node): Likewise.
6651 2017-07-20 Jakub Jelinek <jakub@redhat.com>
6654 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
6655 V2TImode and V4TImode.
6656 (ix86_expand_vector_extract): Likewise.
6657 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
6658 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
6659 (ssescalarmode): Handle V4TImode and V2TImode.
6660 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
6661 (*vec_extractv2ti, *vec_extractv4ti): New insns.
6662 (VEXTRACTI128_MODE): New mode iterator.
6663 (splitter for *vec_extractv?ti first element): New.
6664 (VEC_INIT_MODE): New mode iterator.
6665 (vec_init<mode>): Consolidate 3 expanders into one using
6666 VEC_INIT_MODE mode iterator.
6668 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6670 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
6671 non_spilled_static_chain_regno_p.
6673 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6675 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
6677 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
6679 * bb-reorder.c (connect_traces): Allow copying of blocks within
6682 2017-07-20 Richard Biener <rguenther@suse.de>
6684 * gimple.h (gimple_phi_result): Add gphi * overload.
6685 (gimple_phi_result_ptr): Likewise.
6686 (gimple_phi_arg): Likewise. Adjust index assert to only
6687 allow actual argument accesses rather than all slots available
6689 (gimple_phi_arg_def): Add gphi * overload.
6690 * tree-phinodes.c (make_phi_node): Initialize only actual
6692 (resize_phi_node): Clear memory not covered by old node,
6693 do not initialize excess argument slots.
6694 (reserve_phi_args_for_new_edge): Initialize new argument slot
6697 2017-07-20 Bin Cheng <bin.cheng@arm.com>
6699 PR tree-optimization/81388
6701 2016-07-21 Bin Cheng <bin.cheng@arm.com>
6703 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
6704 by removing computation of may_be_zero.
6706 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6707 Tom de Vries <tom@codesourcery.com>
6710 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
6711 when gimple level profile disagrees with what RTL expander did.
6713 2017-07-20 Richard Biener <rguenther@suse.de>
6715 PR tree-optimization/61171
6716 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
6717 (vect_analyze_stmt): Add slp instance parameter.
6718 (vectorizable_reduction): Likewise.
6719 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
6720 (vect_is_simple_reduction): Deal with chains not detected
6721 as SLP reduction chain, specifically not properly associated
6722 chains containing a mix of plus/minus.
6723 (get_reduction_op): Remove.
6724 (get_initial_defs_for_reduction): Simplify, pass in whether
6725 this is a reduction chain, pass in the SLP node for the PHIs.
6726 (vect_create_epilog_for_reduction): Get the SLP instance as
6728 (vectorizable_reduction): Get the SLP instance as arg.
6729 During analysis remember the SLP node with the PHIs in the
6730 instance. Simplify getting at the vectorized reduction PHIs.
6731 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6732 through SLP instance.
6733 (vect_slp_analyze_operations): Likewise.
6734 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6735 (vect_transform_stmt): Likewise.
6737 2017-07-20 Tom de Vries <tom@codesourcery.com>
6739 PR tree-optimization/81489
6740 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6741 read of phi arg location to before loop that modifies phi.
6743 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6745 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6748 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6751 * except.c (execute): Fix ordering issue.
6753 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6755 PR rtl-optimization/81423
6756 * combine.c (make_compound_operation_int): Don't try to optimize
6757 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6759 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6761 PR rtl-optimization/81423
6762 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
6763 with a constant that is -1 in the truncated to mode.
6765 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6767 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
6768 (determine_unlikely_bbs): ... here.
6769 * predict.h (propagate_unlikely_bbs_forward): Declare.
6770 * cfgexpand.c (pass_expand::execute): Use it.
6771 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
6773 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
6774 propagate_unlikely_bbs_forward.
6776 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6779 * except.c (maybe_add_nop_after_section_switch): New function.
6782 2017-07-19 Tom de Vries <tom@codesourcery.com>
6784 * gimple.h (gimple_phi_set_arg): Make assert more strict.
6786 2017-07-19 Tom de Vries <tom@codesourcery.com>
6788 * gimple.h (gimple_phi_arg): Make assert more strict.
6790 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
6792 * config.gcc (powerpc*-*-*): Add mmintrin.h.
6793 * config/rs6000/mmintrin.h: New file.
6794 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
6796 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6798 PR tree-optimization/81346
6799 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
6801 2017-07-19 Tom de Vries <tom@codesourcery.com>
6803 * config/nvptx/nvptx.md (VECIM): Add V2DI.
6805 2017-07-19 Tom de Vries <tom@codesourcery.com>
6807 * config/nvptx/nvptx-modes.def: Add V2DImode.
6808 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
6809 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
6810 (nvptx_output_mov_insn): Handle lack of mov.b128.
6811 (nvptx_print_operand): Handle 'H' and 'L' codes.
6812 (nvptx_vector_mode_supported): Allow V2DImode.
6813 (nvptx_preferred_simd_mode): New function.
6814 (nvptx_data_alignment): New function.
6815 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
6816 nvptx_preferred_simd_mode.
6817 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
6819 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
6821 2017-07-19 Tom de Vries <tom@codesourcery.com>
6823 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
6824 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
6825 (nvptx_vector_mode_supported): New function. Allow V2SImode.
6826 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
6827 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
6828 (mov<VECIM>_insn): New define_insn.
6829 (define_expand "mov<VECIM>): New define_expand.
6831 2017-07-19 Tom de Vries <tom@codesourcery.com>
6833 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
6835 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6837 PR tree-optimization/81346
6838 * fold-const.h (fold_div_compare, range_check_type): Declare.
6839 * fold-const.c (range_check_type): New function.
6840 (build_range_check): Use range_check_type.
6841 (fold_div_compare): No longer static, rewritten into
6842 a match.pd helper function.
6843 (fold_comparison): Don't call fold_div_compare here.
6844 * match.pd (X / C1 op C2): New optimization using fold_div_compare
6847 2017-07-19 Nathan Sidwell <nathan@acm.org>
6849 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
6850 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
6851 * tree.c (find_decls_types_r, verify_type): Use
6852 TYPE_{MIN,MAX}_VALUE_RAW.
6853 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
6854 (hash_tree): Likewise.
6855 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6857 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6860 2017-07-18 Tom de Vries <tom@codesourcery.com>
6863 * omp-expand.c (expand_omp_for_static_chunk): Handle
6864 equal-argument loop exit phi.
6866 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
6869 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
6870 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
6871 operand 2 predicate.
6872 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
6873 operand 2 predicate.
6874 (ror,rol -> rorx splitters): Use const_int_operand as
6875 operand 2 predicate.
6877 2017-06-18 Richard Biener <rguenther@suse.de>
6879 PR tree-optimization/81410
6880 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
6881 the gap in the ! slp_perm SLP case after each group.
6883 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6886 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
6889 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6892 * predict.c (set_even_probabilities): Cleanup; do not affect
6893 probabilities that are already known.
6894 (combine_predictions_for_bb): Call even when count is set.
6896 2017-07-18 Nathan Sidwell <nathan@acm.org>
6898 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
6901 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6904 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
6905 optimization for loop niter analysis.
6907 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
6910 * config/avr/avr.c (avr_optimize_casesi): Don't use
6911 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
6913 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6915 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
6916 body_cost_vec from _vect_peel_extended_info.
6917 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
6918 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
6921 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6923 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
6925 2017-07-18 Richard Biener <rguenther@suse.de>
6927 PR tree-optimization/80620
6928 PR tree-optimization/81403
6929 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
6930 info when re-using a VN table entry.
6932 2017-07-18 Richard Biener <rguenther@suse.de>
6934 PR tree-optimization/81418
6935 * tree-vect-loop.c (vectorizable_reduction): Properly compute
6936 vectype_in. Verify that with lane-reducing reduction operations
6937 we have a single def-use cycle.
6939 2017-07-17 Carl Love <cel@us.ibm.com>
6941 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
6943 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6944 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6945 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6946 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6947 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6948 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6949 VMULOSW): New enum "unspec" values.
6950 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
6951 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
6952 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6953 altivec_vmulosw): New patterns.
6954 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6955 VMULOSW): Add definitions.
6956 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
6958 * config/alpha/alpha.c: Include predict.h.
6960 2017-07-17 Yury Gribov <tetra2005@gmail.com>
6962 * tree-vrp.c (compare_assert_loc): Fix comparison function
6963 to return predictable results.
6965 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6967 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
6970 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
6971 * doc/invoke.texi (mexpand-adddi): Update text.
6973 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6975 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
6976 that also clobbers the CC register. The old expand code is moved
6978 (*arc_clzsi2): ... here.
6979 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
6980 the CC register. The old expand code is moved to ...
6981 (arc_ctzsi2): ... here.
6983 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6985 * config/arc/arc.opt (mindexed-loads): Use initial value
6986 TARGET_INDEXED_LOADS_DEFAULT.
6987 (mauto-modify-reg): Use initial value
6988 TARGET_AUTO_MODIFY_REG_DEFAULT.
6989 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6990 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6991 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6992 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6994 2017-07-17 Martin Liska <mliska@suse.cz>
6997 * opts.c (finish_options): Do not allow -fgnu-tm
6998 w/ -fsanitize={kernel-,}address. Say sorry.
7000 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7003 * tree-loop-distribution.c (classify_partition): Only assert on
7004 numer of iterations.
7005 (merge_dep_scc_partitions): Delete prameter. Update function call.
7006 (distribute_loop): Remove code handling loop with unknown niters.
7007 (pass_loop_distribution::execute): Skip loop with unknown niters.
7009 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7012 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7013 function sort_partitions_by_post_order.
7015 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7017 PR tree-optimization/81374
7018 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7019 the max index of basic blocks, rather than number of basic blocks.
7021 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7023 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7025 (arc_legitimate_pic_operand_p): Likewise.
7026 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7028 (arc_needs_pcl_p): Likewise.
7029 (arc_legitimate_pc_offset_p): Likewise.
7030 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7031 function is also used in constrains.md.
7032 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7033 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7034 PLUS. Only return true/false in known cases, otherwise assert.
7035 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7036 is already called in arc_legitimate_constant_p.
7037 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7039 (LEGITIMATE_PIC_OPERAND_P): Use
7040 arc_raw_symbolic_reference_mentioned_p function.
7041 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7046 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7047 Andrew Burgess <andrew.burgess@embecosm.com>
7049 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7050 (arc_return_address_register): New function.
7051 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7052 (arc_handle_fndecl_attribute): Add naked attribute.
7053 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7054 (TARGET_WARN_FUNC_RETURN): Likewise.
7055 (arc_allocate_stack_slots_for_args): New function.
7056 (arc_warn_func_return): Likewise.
7057 (machine_function): Change type fn_type.
7058 (arc_compute_function_type): Consider new naked function type,
7059 change function return type.
7060 (arc_must_save_register): Adapt to handle new
7061 arc_compute_function_type's return type.
7062 (arc_expand_prologue): Likewise.
7063 (arc_expand_epilogue): Likewise.
7064 (arc_return_address_regs): Delete.
7065 (arc_return_address_register): New function.
7066 (arc_epilogue_uses): Use above function.
7067 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7068 (arc_function_type): Change encoding, add naked type.
7069 (ARC_INTERRUPT_P): Change to handle the new encoding.
7070 (ARC_FAST_INTERRUPT_P): Likewise.
7071 (ARC_NORMAL_P): Define.
7072 (ARC_NAKED_P): Likewise.
7073 (arc_compute_function_type): Delete prototype.
7074 * config/arc/arc.md (in_ret_delay_slot): Use
7075 arc_return_address_register function.
7076 (simple_return): Likewise.
7077 (p_return_i): Likewise.
7079 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7081 PR tree-optimization/81428
7082 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7083 can't be built for those types.
7085 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7087 Remove stuff dead since r239246.
7089 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7090 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7091 (avr_inform_devices): Remove dead stuff.
7093 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7095 * config/arm/arm_neon.h: Fix softp typo.
7097 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7099 PR tree-optimization/81365
7100 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7101 aggregate moves onto bb predecessor edges, make sure there are no
7102 loads that could alias the lhs in between the start of bb and the
7105 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7108 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7109 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7110 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7112 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7114 PR tree-optimization/81396
7115 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7116 (init_symbolic_number): Initialize it to 1.
7117 (perform_symbolic_merge): Add n_ops from both operands into the new
7119 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7120 without base_addr as useless if they need more than one operation.
7121 (bswap_replace): Handle !bswap case for NULL base_addr.
7123 2017-07-17 Tom de Vries <tom@codesourcery.com>
7126 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7129 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7131 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7132 conditional builtin define __FIX_LEON3FT_B2BST.
7134 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7136 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7137 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7140 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7142 PR rtl-optimization/81424
7143 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7144 to remove potential trapping from operands if -fnon-call-exceptions.
7146 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7148 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7149 profile_proability for scalling.
7150 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7152 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7154 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7156 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7158 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7159 fixpoint arithmetics.
7161 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7163 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7164 fixpoint arithmetics.
7166 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7168 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7169 fixpoint arithmetics.
7171 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7173 * profile-count.h (profile_probability::from_reg_br_prob_note,
7174 profile_probability::to_reg_br_prob_note): New functions.
7175 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7176 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7177 * predict.c (probability_reliable_p): Update.
7178 (edge_probability_reliable_p): Update.
7179 (br_prob_note_reliable_p): Update.
7180 (invert_br_probabilities): Update.
7181 (add_reg_br_prob_note): New function.
7182 (combine_predictions_for_insn): Update.
7183 * asan.c (asan_clear_shadow): Update.
7184 * cfgbuild.c (compute_outgoing_frequencies): Update.
7185 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7186 (update_br_prob_note): Update.
7187 (rtl_verify_edges): Update.
7188 (purge_dead_edges): Update.
7189 (fixup_reorder_chain): Update.
7190 * emit-rtl.c (try_split): Update.
7191 * ifcvt.c (cond_exec_process_insns): Update.
7192 (cond_exec_process_if_block): Update.
7193 (dead_or_predicable): Update.
7194 * internal-fn.c (expand_addsub_overflow): Update.
7195 (expand_neg_overflow): Update.
7196 (expand_mul_overflow): Update.
7197 * loop-doloop.c (doloop_modify): Update.
7198 * loop-unroll.c (compare_and_jump_seq): Update.
7199 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7200 * predict.h: Update.
7201 * reorg.c (mostly_true_jump): Update.
7203 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7204 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7205 * config/arc/arc.c: (emit_unlikely_jump): Update.
7206 * config/arm/arm.c: (emit_unlikely_jump): Update.
7207 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7208 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7209 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7210 (ix86_print_operand): Update.
7211 (ix86_split_fp_branch): Update.
7212 (predict_jump): Update.
7213 * config/ia64/ia64.c (ia64_print_operand): Update.
7214 * config/mmix/mmix.c (mmix_print_operand): Update.
7215 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7216 (rs6000_expand_split_stack_prologue): Update.
7217 * config/rs6000/rs6000.c: Update.
7218 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7219 (s390_expand_vec_movstr): Update.
7220 (s390_expand_cs_tdsi): Update.
7221 (s390_expand_split_stack_prologue): Update.
7222 * config/sh/sh.c (sh_print_operand): Update.
7223 (expand_cbranchsi4): Update.
7224 (expand_cbranchdi4): Update.
7225 * config/sparc/sparc.c (output_v9branch): Update.
7226 * config/spu/spu.c (get_branch_target): Update.
7227 (ea_load_store_inline): Update.
7228 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7229 * config/tilepro/tilepro.c: Update.
7231 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7233 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7234 (gimplify_save_expr): Likewise.
7236 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7238 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7240 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7242 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7244 * timevar.def (TV_IPA_FNSUMMARY): Define.
7246 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7248 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7249 to back store errata sensitive sequence from being generated.
7250 (sqrtdf2_fix): Likewise.
7252 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7254 * tree-ssa-threadupdate.c (compute_path_counts,
7255 update_joiner_offpath_counts): Use profile_probability.
7257 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7260 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7262 * config/arm/arm-c.c (arm_cpu_builtins): Define
7263 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7265 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7268 array entries to represent __ieee128 versions of the
7269 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7270 scalar_extract_sig, and scalar_insert_exp built-in functions.
7271 (altivec_resolve_overloaded_builtin): Add special case handling
7272 for the __builtin_scalar_insert_exp function, as represented by
7273 the P9V_BUILTIN_VEC_VSIEDP constant.
7274 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7275 exponent support for __ieee128 argument.
7276 (VSESQP): Add scalar extract signature support for __ieee128
7278 (VSTDCNQP): Add scalar test negative support for __ieee128
7280 (VSIEQP): Add scalar insert exponent support for __int128 argument
7281 with __ieee128 result.
7282 (VSIEQPF): Add scalar insert exponent support for __ieee128
7283 argument with __ieee128 result.
7284 (VSTDCQP): Add scalar test data class support for __ieee128
7286 (VSTDCNQP): Add overload support for scalar test negative with
7288 (VSTDCQP): Add overload support for scalar test data class
7290 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7292 (UNSPEC_VSX_SIEXPQP): New constant.
7293 (xsxexpqp): New insn for VSX scalar extract exponent quad
7295 (xsxsigqp): New insn for VSX scalar extract significand quad
7297 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7298 precision with floating point argument.
7299 (xststdcqp): New expand for VSX scalar test data class quad
7301 (xststdcnegqp): New expand for VSX scalar test negative quad
7303 (xststdcqp): New insn to match expansions for VSX scalar test data
7304 class quad precision and VSX scalar test negative quad precision.
7305 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7306 special case operand checking to enforce that second operand of
7307 VSX scalar test data class with quad precision argument is a 7-bit
7309 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7310 prototypes and descriptions of __ieee128 versions of
7311 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7312 scalar_test_data_class, and scalar_test_neg built-in functions.
7314 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7316 PR tree-optimization/81162
7317 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7318 replace a negate with an add.
7320 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7322 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7324 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7326 * config/arm/arm-c.c (arm_cpu_builtins): Define
7327 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7329 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7331 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7332 (armv8-r): Set ARM Cortex-R52 as default CPU.
7333 * config/arm/arm-tables.opt: Regenerate.
7334 * config/arm/arm-tune.md: Regenerate.
7335 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7337 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7338 extension for -mcpu=cortex-r52.
7340 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7342 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7343 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7344 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7345 (fp-armv8): Define it as FP_ARMv8 only.
7346 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7347 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7349 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7350 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7351 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7352 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7353 than TARGET_FPU_ARMV8.
7354 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7355 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7356 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7358 * config/arm/neon.md (neon_vrint): Likewise.
7359 (neon_vcvt): Likewise.
7360 (neon_<fmaxmin_op><mode>): Likewise.
7361 (<fmaxmin><mode>3): Likewise.
7362 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7363 * config/arm/predicates.md (arm_cond_move_operator): Check against
7364 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7366 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7368 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7371 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7373 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7374 loop in comment with memset.
7376 2017-07-14 Martin Liska <mliska@suse.cz>
7378 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7379 * dwarf2out.c (is_java): Remove the function.
7380 (output_pubname): Remove usage of the function.
7381 (lower_bound_default): Remove usage of DW_LANG_Java.
7382 (gen_compile_unit_die): Likewise.
7383 * gcc.c: Remove compiler defaults for .java and .zip files.
7384 * gimple-expr.c (remove_suffix): Change as there's no longer
7385 extension than 4-letter one.
7386 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7387 (gimplify_save_expr): Likewise.
7388 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7389 as it's possible even for other languages than Java.
7390 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7391 * lto-opts.c (lto_write_options): Remove reference to Java.
7392 * opts.c (strip_off_ending): Update file extension handling.
7393 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7394 * tree-eh.c (lower_resx): Likewise.
7395 * tree.c (free_lang_data_in_type): Remove dead code.
7396 (find_decls_types_r): Likewise.
7397 (build_common_builtin_nodes): Remove Java from a comment.
7398 (verify_type): Remove dead code.
7399 * varasm.c (assemble_external): Remove Java from a comment.
7401 2017-07-14 Martin Liska <mliska@suse.cz>
7403 * opts.c (finish_options): Add quotes.
7404 (common_handle_option): Likewise.
7406 2017-07-14 Martin Liska <mliska@suse.cz>
7408 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7409 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7411 * dwarf2out.c (lower_bound_default): Do not handle
7413 (gen_compile_unit_die): Likewise.
7414 * gcc.c: Remove default extension binding for GNU Pascal.
7415 * stmt.c: Remove Pascal language from a comment.
7416 * xcoffout.c: Likewise.
7418 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7421 * diagnostic-show-locus.c (fixit_cmp): New function.
7422 (layout::layout): Sort m_fixit_hints.
7423 (column_range::column_range): Assert that the values are valid.
7424 (struct char_span): New struct.
7425 (correction::overwrite): New method.
7426 (struct source_line): New struct.
7427 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7428 calls in terms of classes source_line and char_span, and
7429 correction::overwrite.
7430 (selftest::test_overlapped_fixit_printing_2): New function.
7431 (selftest::diagnostic_show_locus_c_tests): Call it.
7433 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7435 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7436 early if there is no lhs.
7438 2017-07-13 Martin Liska <mliska@suse.cz>
7440 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7441 (gen_reference_type_die): Likewise.
7442 * stor-layout.c: Remove Pascal-related comment.
7444 2017-07-13 Martin Liska <mliska@suse.cz>
7446 * opts.c (finish_options): Add quotes to error messages.
7447 (parse_sanitizer_options): Likewise.
7449 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7451 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7453 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7455 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7457 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7459 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7461 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7463 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7466 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7467 provides the hardware capability bits, define the macro
7468 __BUILTIN_CPU_SUPPORTS__.
7469 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7470 if GLIBC does not provide the hardware capability bits. Add a
7471 gcc_unreachable call if the built-in cpu function is neither
7472 __builtin_cpu_is nor __builtin_cpu_supports.
7473 (rs6000_get_function_versions_dispatcher): Change the warning
7474 that an old GLIBC is used which does not export the capability
7475 bits to be an error.
7476 * doc/extend.texi (target_clones attribute): Document the
7477 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7478 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7479 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7480 the macros defined by GCC if the newer GLIBC is available.
7482 2017-07-12 Jeff Law <law@redhat.com>
7484 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7485 remaining includes slightly.
7486 * config/riscv/riscv-builtins.c: Include profile-count.h.
7488 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7491 * config/avr/avr.c (avr_set_current_function): In diagnostic
7492 messages: Quote keywords and (parts of) identifiers.
7493 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7496 2017-07-12 Carl Love <cel@us.ibm.com>
7498 * config/rs6000/rs6000-c.c: Add support for built-in functions
7499 vector bool char vec_revb (vector bool char);
7500 vector bool short vec_revb (vector short char);
7501 vector bool int vec_revb (vector bool int);
7502 vector bool long long vec_revb (vector bool long long);
7503 * doc/extend.texi: Update the built-in documentation file for the
7504 new built-in functions.
7506 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7508 * config/s390/s390.md: Remove movcc splitter.
7510 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7512 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7513 load/store on condition.
7515 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7518 * config/avr/avr.c (avr_encode_section_info)
7519 [progmem && !TREE_READONLY]: Error if progmem object needs
7522 2017-07-11 Michael Collison <michael.collison@arm.com>
7524 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7527 2017-07-11 Carl Love <cel@us.ibm.com>
7529 * config/rs6000/rs6000-c.c: Add support for builtins
7530 vector unsigned int vec_parity_lsbb (vector signed int);
7531 vector unsigned int vec_parity_lsbb (vector unsigned int);
7532 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7533 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7534 vector unsigned long long vec_parity_lsbb (vector signed long long);
7535 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7536 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7537 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7538 * doc/extend.texi: Update the built-in documentation file for the
7539 new built-in functions.
7541 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7543 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7544 (layout::m_primary_loc): New field.
7545 (layout::layout): Initialize new field. Move location filtering
7546 logic from here to...
7547 (layout::maybe_add_location_range): ...this new method. Add
7548 support for filtering to just the lines already specified by other
7550 (layout::will_show_line_p): New method.
7551 (gcc_rich_location::add_location_if_nearby): New method.
7552 (selftest::test_add_location_if_nearby): New test function.
7553 (selftest::diagnostic_show_locus_c_tests): Call it.
7554 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7557 2017-07-11 Tom de Vries <tom@codesourcery.com>
7559 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
7560 (bb_first_real_insn): New function.
7561 (nvptx_single): Add extra initialization of broadcasted condition
7564 2017-07-11 Nathan Sidwell <nathan@acm.org>
7566 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
7568 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
7570 * doc/extend.texi (AVR Function Attributes): Remove weblink to
7571 Binutils doc as TEXI will mess them up.
7572 * doc/invoke.texi (AVR Options): Same here.
7574 2017-07-11 Daniel Cederman <cederman@gaisler.com>
7576 * config/sparc/sparc.opt (mfix-ut700): New option.
7577 (mfix-gr712rc): Likewise.
7578 (sparc_fix_b2bst): New variable.
7579 * doc/invoke.texi (SPARC options): Document them.
7580 (ARM options): Fix warnings.
7581 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
7582 instructions to prevent sequences that can trigger the store-store
7583 errata for certain LEON3FT processors.
7584 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
7585 (sparc_option_override): Set sparc_fix_b2bst appropriately.
7586 * config/sparc/sparc.md (fix_b2bst): New attribute.
7587 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
7589 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
7592 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
7594 (*rsqrtsf2_sse): Ditto.
7596 (div<mode>3): Macroize insn from divdf3 and divsf3
7597 using MODEF mode iterator.
7599 2017-07-10 Martin Sebor <msebor@redhat.com>
7601 PR tree-optimization/80397
7602 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
7603 instead of testing for equality to INTEGER_TYPE.
7605 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
7607 * config.gcc: Remove uclibc from arc target spec.
7609 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
7611 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
7613 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7616 * lto-wrapper.c (remove_option): New function.
7617 (merge_and_complain): Merge PIC/PIE options more realistically.
7619 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7621 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
7625 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
7626 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
7627 * config.in: Regenerate.
7628 * configure: Regenerate.
7629 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
7630 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
7631 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
7632 (TARGET_GASISR_PROLOGUES): ...target mask.
7633 * common/config/avr/avr-common.c
7634 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
7635 Set -mgas-isr-prologues.
7636 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
7637 INSERT_PASS_BEFORE for it.
7638 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
7639 * config/avr/avr.c (avr_option_override)
7640 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
7641 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
7642 (avr_attribute_table) <no_gccisr>: Add new function attribute.
7643 (avr_set_current_function) <is_no_gccisr>: Init machine field.
7644 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
7646 (make_avr_pass_pre_proep): New function.
7647 (emit_push_sfr) <treg>: Add argument to function and use it
7649 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
7650 and set machine->gasisr.yes.
7651 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
7652 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
7653 __gcc_isr.n_pushed to .L__stack_usage.
7654 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
7655 (avr_asm_final_postscan_insn): ...this new static function.
7656 * config/avr/avr.h (machine_function)
7657 <is_no_gccisr, use_L__stack_usage>: New fields.
7658 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
7659 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
7660 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
7661 (gasisr, *gasisr): New expander and insn.
7662 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
7663 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
7664 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
7666 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
7668 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
7671 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7673 Move jump-tables out of .text again.
7676 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
7677 (ASM_OUTPUT_ADDR_VEC): New function.
7678 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
7679 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
7680 INSN_ADDRESSes as asm comment.
7681 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
7682 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
7683 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
7684 * config/avr/avr.md (*tablejump): Adjust comment.
7685 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
7686 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
7688 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
7689 (avr_output_addr_vec): New proto.
7690 (avr_log_t) <insn_addresses>: New field.
7692 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
7695 * config/i386/i386.c (ix86_function_arg_advance): Set
7696 outgoing_args_on_stack to true if there are outgoing arguments
7698 (ix86_function_arg): Likewise.
7699 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
7700 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
7701 * config/i386/i386.h (machine_function): Add
7702 outgoing_args_on_stack.
7704 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
7706 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7708 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
7710 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
7712 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
7713 (REAL_H): Remove $(MACHMODE_H).
7714 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
7716 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
7717 $(MACHMODE_H) and double-int.h.
7718 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
7720 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
7723 2017-07-07 Andrew Pinski <apinski@cavium.com>
7725 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
7726 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7728 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7730 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7731 Add warning if GCC was not configured to link against a GLIBC that
7732 exports the hardware capability bits.
7733 (make_resolver_func): Make resolver function private and not a
7734 COMDAT function. Create the name with clone_function_name instead
7735 of make_unique_name.
7738 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7739 correct operand in doing the split.
7741 2017-07-07 Carl Love <cel@us.ibm.com>
7743 * config/rs6000/rs6000-c: Add support for built-in function
7744 vector unsigned short vec_pack_to_short_fp32 (vector float,
7746 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7747 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7748 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7749 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7750 (convert_4f32_8i16): Add define_expand.
7751 * doc/extend.texi: Update the built-in documentation file for the
7752 new built-in function.
7754 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7756 * config/sparc/m8.md: New file.
7757 * config/sparc/sparc.md: Include m8.md.
7759 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7761 * config/sparc/sparc.opt: New option -mvis4b.
7762 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
7763 (sparc_option_override): Handle VIS4B.
7764 (enum sparc_builtins): Define
7765 SPARC_BUILTIN_DICTUNPACK{8,16,32},
7766 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
7767 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
7768 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
7769 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
7770 (check_constant_argument): New function.
7771 (sparc_vis_init_builtins): Define builtins
7772 __builtin_vis_dictunpack{8,16,32},
7773 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
7774 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
7775 __builtin_vis_fpcmpde{8,16,32}shl and
7776 __builtin_vis_fpcmpur{8,16,32}shl.
7777 (sparc_expand_builtin): Check that the constant operands to
7778 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
7779 constant and in range.
7780 * config/sparc/sparc-c.c (sparc_target_macros): Handle
7782 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
7783 (SPARC_IMM5_P): Likewise.
7784 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
7785 (enabled): Handle vis4b.
7786 (UNSPEC_DICTUNPACK): New unspec.
7787 (UNSPEC_FPCMPSHL): Likewise.
7788 (UNSPEC_FPUCMPSHL): Likewise.
7789 (UNSPEC_FPCMPDESHL): Likewise.
7790 (UNSPEC_FPCMPURSHL): Likewise.
7791 (cpu_feature): New CPU feature `vis4b'.
7792 (dictunpack{8,16,32}): New insns.
7793 (FPCSMODE): New mode iterator.
7794 (fpcscond): New code iterator.
7795 (fpcsucond): Likewise.
7796 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
7797 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
7798 (fpcmpde{8,16,32}{si,di}shl): Likewise.
7799 (fpcmpur{8,16,32}{si,di}shl): Likewise.
7800 * config/sparc/constraints.md: Define constraints `q' for unsigned
7801 2-bit integer constants and `t' for unsigned 5-bit integer
7803 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
7805 (imm5_operand_dictunpack16): Likewise.
7806 (imm5_operand_dictunpack32): Likewise.
7807 (imm2_operand): Likewise.
7808 * doc/invoke.texi (SPARC Options): Document -mvis4b.
7809 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
7810 ditunpack* and fpcmp*shl builtins.
7812 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7814 * config.gcc: Handle m8 in --with-{cpu,tune} options.
7815 * config.in: Add HAVE_AS_SPARC6 define.
7816 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
7818 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
7820 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
7821 (CPP_CPU_SPEC): Handle m8.
7822 (ASM_CPU_SPEC): Likewise.
7823 * config/sparc/sparc-opts.h (enum processor_type): Add
7825 * config/sparc/sparc.c (m8_costs): New struct.
7826 (sparc_option_override): Handle TARGET_CPU_m8.
7827 (sparc32_initialize_trampoline): Likewise.
7828 (sparc64_initialize_trampoline): Likewise.
7829 (sparc_issue_rate): Likewise.
7830 (sparc_register_move_cost): Likewise.
7831 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
7832 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
7833 (ASM_CPU64_DEFAULT_SPEC): Likewise.
7834 (CPP_CPU_SPEC): Handle M8.
7835 (ASM_CPU_SPEC): Likewise.
7836 (AS_M8_FLAG): Define.
7837 * config/sparc/sparc.md: Add m8 to the cpu attribute.
7838 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
7839 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
7841 * configure: Regenerate.
7842 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
7845 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7847 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
7849 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
7850 ("*movdi_insn_sp32"): Do not set v3pipe.
7851 ("*movsi_insn"): Likewise.
7852 ("*movdi_insn_sp64"): Likewise.
7853 ("*movsf_insn"): Likewise.
7854 ("*movdf_insn_sp32"): Likewise.
7855 ("*movdf_insn_sp64"): Likewise.
7856 ("*zero_extendsidi2_insn_sp64"): Likewise.
7857 ("*sign_extendsidi2_insn"): Likewise.
7858 ("*mov<VM32:mode>_insn"): Likewise.
7859 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7860 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7861 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7862 ("<vlop:code><VL:mode>3"): Likewise.
7863 ("*not_<vlop:code><VL:mode>3"): Likewise.
7864 ("*nand<VL:mode>_vis"): Likewise.
7865 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
7866 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
7867 ("one_cmpl<VL:mode>2"): Likewise.
7868 ("faligndata<VM64:mode>_vis"): Likewise.
7869 ("alignaddrsi_vis"): Likewise.
7870 ("alignaddrdi_vis"): Likweise.
7871 ("alignaddrlsi_vis"): Likewise.
7872 ("alignaddrldi_vis"): Likewise.
7873 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7874 ("bmaskdi_vis"): Likewise.
7875 ("bmasksi_vis"): Likewise.
7876 ("bshuffle<VM64:mode>_vis"): Likewise.
7877 ("cmask8<P:mode>_vis"): Likewise.
7878 ("cmask16<P:mode>_vis"): Likewise.
7879 ("cmask32<P:mode>_vis"): Likewise.
7880 ("pdistn<P:mode>_vis"): Likewise.
7881 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7883 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7885 * config/sparc/sparc.md ("subtype"): New insn attribute.
7886 ("*wrgsr_sp64"): Set insn subtype.
7887 ("*rdgsr_sp64"): Likewise.
7888 ("alignaddrsi_vis"): Likewise.
7889 ("alignaddrdi_vis"): Likewise.
7890 ("alignaddrlsi_vis"): Likewise.
7891 ("alignaddrldi_vis"): Likewise.
7892 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7893 ("fexpand_vis"): Likewise.
7894 ("fpmerge_vis"): Likewise.
7895 ("faligndata<VM64:mode>_vis"): Likewise.
7896 ("bshuffle<VM64:mode>_vis"): Likewise.
7897 ("cmask8<P:mode>_vis"): Likewise.
7898 ("cmask16<P:mode>_vis"): Likewise.
7899 ("cmask32<P:mode>_vis"): Likewise.
7900 ("fchksm16_vis"): Likewise.
7901 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
7902 ("fmean16_vis"): Likewise.
7903 ("fp<plusminus_insn>64_vis"): Likewise.
7904 ("<plusminus_insn>v8qi3"): Likewise.
7905 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7906 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
7907 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
7908 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
7909 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
7910 ("*movqi_insn"): Likewise.
7911 ("*movhi_insn"): Likewise.
7912 ("*movsi_insn"): Likewise.
7913 ("movsi_pic_gotdata_op"): Likewise.
7914 ("*movdi_insn_sp32"): Likewise.
7915 ("*movdi_insn_sp64"): Likewise.
7916 ("movdi_pic_gotdata_op"): Likewise.
7917 ("*movsf_insn"): Likewise.
7918 ("*movdf_insn_sp32"): Likewise.
7919 ("*movdf_insn_sp64"): Likewise.
7920 ("*zero_extendhisi2_insn"): Likewise.
7921 ("*zero_extendqihi2_insn"): Likewise.
7922 ("*zero_extendqisi2_insn"): Likewise.
7923 ("*zero_extendqidi2_insn"): Likewise.
7924 ("*zero_extendhidi2_insn"): Likewise.
7925 ("*zero_extendsidi2_insn_sp64"): Likewise.
7926 ("ldfsr"): Likewise.
7927 ("prefetch_64"): Likewise.
7928 ("prefetch_32"): Likewise.
7929 ("tie_ld32"): Likewise.
7930 ("tie_ld64"): Likewise.
7931 ("*tldo_ldub_sp32"): Likewise.
7932 ("*tldo_ldub1_sp32"): Likewise.
7933 ("*tldo_ldub2_sp32"): Likewise.
7934 ("*tldo_ldub_sp64"): Likewise.
7935 ("*tldo_ldub1_sp64"): Likewise.
7936 ("*tldo_ldub2_sp64"): Likewise.
7937 ("*tldo_ldub3_sp64"): Likewise.
7938 ("*tldo_lduh_sp32"): Likewise.
7939 ("*tldo_lduh1_sp32"): Likewise.
7940 ("*tldo_lduh_sp64"): Likewise.
7941 ("*tldo_lduh1_sp64"): Likewise.
7942 ("*tldo_lduh2_sp64"): Likewise.
7943 ("*tldo_lduw_sp32"): Likewise.
7944 ("*tldo_lduw_sp64"): Likewise.
7945 ("*tldo_lduw1_sp64"): Likewise.
7946 ("*tldo_ldx_sp64"): Likewise.
7947 ("*mov<VM32:mode>_insn"): Likewise.
7948 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7949 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7951 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7953 * config/sparc/sparc.md ("type"): New insn type viscmp.
7954 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
7956 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
7957 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
7958 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7959 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
7961 ("n7_vis_logical_11cycle"): Likewise.
7962 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
7963 * config/sparc/niagara2.md ("niag3_vis": Likewise.
7964 * config/sparc/niagara.md ("niag_vis"): Likewise.
7965 * config/sparc/ultra3.md ("us3_fga"): Likewise.
7966 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
7968 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7970 * config/sparc/sparc.md: New instruction type `bmask'.
7971 (bmaskdi_vis): Use the `bmask' type.
7972 (bmasksi_vis): Likewise.
7973 * config/sparc/ultra3.md (us3_array): Likewise.
7974 * config/sparc/niagara7.md (n7_array): Likewise.
7975 * config/sparc/niagara4.md (n4_array): Likewise.
7976 * config/sparc/niagara2.md (niag2_vis): Likewise.
7977 (niag3_vis): Likewise.
7978 * config/sparc/niagara.md (niag_vis): Likewise.
7980 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7982 * ipa-comdats.c: Remove optimize check from gate.
7983 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
7984 for functions not optimized.
7985 (ipa_fn_summary_read): Skip optimize check.
7986 (ipa_fn_summary_write): Likewise.
7987 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
7989 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
7991 (can_inline_edge_p): Check flag_pcc_struct_return for match.
7992 (check_callers): Give up on caller which is not optimized.
7993 (inline_small_functions): Likewise.
7994 (ipa_inline): Do not give up when not optimizing.
7995 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
7996 away unoptimizes cdtors.
7997 (whole_program_function_and_variable_visibility): Do
7998 ipa_discover_readonly_nonaddressable_vars in LTO mode.
7999 * ipa.c (process_references): Do not check optimize.
8000 (symbol_table::remove_unreachable_nodes): Update optimize check.
8001 (set_writeonly_bit): Update optimize check.
8002 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8003 (pass_ipa_single_use::gate): Remove.
8005 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8007 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8008 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8009 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8010 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8011 permute_load, permute_store, adjust_extract, adjust_splat,
8012 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8013 replace_swap_with_copy, dump_swap_insn_table,
8014 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8015 recombine_lvx_pattern, recombine_stvx_pattern,
8016 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8017 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8018 to file rs6000-p8swap.c.
8019 * config/rs6000/rs6000-p8swap.c: New file.
8020 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8021 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8022 and rs6000*-*-* targets.
8024 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8026 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8028 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8030 * lto-wrapper.c (merge_and_complain): Do not merge
8031 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8032 fsigned_zeros, ftrapping_math, fwrapv.
8033 (append_compiler_options): Do not track these options.
8034 (append_linker_options): Likewie
8036 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8038 * cgraphunit.c (cgraph_node::finalize_function): When
8039 !flag_toplevel_reorde set no_reorder flag.
8040 (varpool_node::finalize_decl): Likewise.
8041 (symbol_table::compile): Drop no toplevel reorder path.
8043 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8045 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8046 edges; zero probability is not better than uninitialized.
8048 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8050 * asan.h (asan_sanitize_allocas_p): Declare.
8051 * asan.c (asan_sanitize_allocas_p): New function.
8052 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8053 (handle_builtin_alloca): Likewise.
8054 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8055 if !asan_sanitize_allocas_p.
8056 * params.def (asan-instrument-allocas): Add new option.
8057 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8058 * opts.c (common_handle_option): Disable allocas sanitization for
8061 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8063 * asan.c: Include gimple-fold.h.
8064 (get_last_alloca_addr): New function.
8065 (handle_builtin_stackrestore): Likewise.
8066 (handle_builtin_alloca): Likewise.
8067 (asan_emit_allocas_unpoison): Likewise.
8068 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8069 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8070 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8071 (instrument_builtin_call): Pass gimple iterator to
8072 get_mem_refs_of_builtin_call.
8073 (last_alloca_addr): New global.
8074 * asan.h (asan_emit_allocas_unpoison): Declare.
8075 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8076 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8077 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8078 if function calls alloca.
8079 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8080 * gimple-fold.h (replace_call_with_value): Declare.
8081 * internal-fn.c: Include asan.h.
8082 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8083 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8085 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8087 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8088 (C_SELFTEST_FLAGS): New.
8089 (CPP_SELFTEST_FLAGS): New.
8090 (SELFTEST_DEPS): New, from deps of s-selftest.
8091 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8092 (CPP_SELFTEST_DEPS): New.
8093 (selftest): Add dependency on s-selftest-c++.
8094 (s-selftest): Rename to...
8095 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8096 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8097 than SELFTEST_FLAGS.
8098 (selftest-gdb): Rename to...
8099 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8101 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8102 (selftest-valgrind): Rename to...
8103 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8105 (selftest-valgrind): Reintroduce as an alias for
8106 selftest-c-valgrind.
8107 (s-selftest-c++): New.
8108 (selftest-c++-gdb): New.
8109 (selftest-c++-valgrind): New.
8111 2017-07-06 Olivier Hainque <hainque@adacore.com>
8113 * gcc.c (process_command): When deciding if undefined variables
8114 should be ignored when processing specs, accept "gcc -v" as well.
8116 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8118 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8119 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8121 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8123 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8124 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8125 * config/arm/arm-tables.opt: Regenerate.
8126 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8128 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8130 2017-07-06 Carl Love <cel@us.ibm.com>
8132 * ChangeLog: Clean up from mid air collision
8134 2017-07-06 Carl Love <cel@us.ibm.com>
8136 * config/rs6000/rs6000-c.c: Add support for built-in functions
8137 vector signed int vec_subc (vector signed int, vector signed int);
8138 vector signed __int128 vec_subc (vector signed __int128,
8139 vector signed __int128);
8140 vector unsigned __int128 vec_subc (vector unsigned __int128,
8141 vector unsigned __int128);
8142 vector signed int vec_sube (vector signed int, vector signed int,
8144 vector unsigned int vec_sube (vector unsigned int,
8145 vector unsigned int,
8146 vector unsigned int);
8147 vector signed __int128 vec_sube (vector signed __int128,
8148 vector signed __int128,
8149 vector signed__int128);
8150 vector unsigned __int128 vec_sube (vector unsigned __int128,
8151 vector unsigned __int128,
8152 vector unsigned __int128);
8153 vector signed int vec_subec (vector signed int, vector signed int,
8155 vector unsigned int vec_subec (vector unsigned int,
8156 vector unsigned int,
8157 vector unsigned int);
8158 vector signed __int128 vec_subec (vector signed __int128,
8159 vector signed __int128,
8160 vector signed__int128);
8161 vector unsigned __int128 vec_subec (vector unsigned __int128,
8162 vector unsigned __int128,
8163 vector unsigned __int128);
8164 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8165 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8166 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8167 BU_ALTIVEC_OVERLOAD_X definitions.
8168 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8169 * doc/extend.texi: Update the built-in documentation file for the new
8172 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8175 * diagnostic-show-locus.c (layout::layout): Use start and finish
8176 spelling location for the start and finish of each range.
8177 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8178 Add unused aspect param.
8179 * input.c (expand_location_1): Add "aspect" param, and use it
8180 to access the correct part of the location.
8181 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8183 (expand_location_to_spelling_point): Likewise.
8184 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8185 param, and pass it to expand_location_1.
8187 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8189 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8190 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8191 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8192 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8193 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8194 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8195 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8196 _mm_maskz_getmant_ss): New intrinsics.
8197 (__builtin_ia32_getexpss128_mask): Changed to ...
8198 __builtin_ia32_getexpss128_round ... this.
8199 (__builtin_ia32_getexpsd128_mask): Changed to ...
8200 __builtin_ia32_getexpsd128_round ... this.
8201 * config/i386/i386-builtin-types.def
8202 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8203 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8204 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8205 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8206 __builtin_ia32_getmantss_mask_round): New builtins.
8207 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8208 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8209 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8210 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8211 * config/i386/sse.md
8212 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8213 avx512f_sgetexp<mode><mask_scalar_name>
8214 <round_saeonly_scalar_name> ... this.
8215 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8216 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8217 vgetexp<ssescalarmodesuffix>
8218 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8219 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8220 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8221 avx512f_vgetmant<mode><mask_scalar_name>
8222 <round_saeonly_scalar_name> ... this.
8223 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8224 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8225 vgetmant<ssescalarmodesuffix>
8226 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8227 %0<mask_scalar_operand4>, %1, %2
8228 <round_saeonly_scalar_mask_op4>, %3} ... this.
8229 * config/i386/subst.md (mask_scalar_operand4,
8230 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8231 round_saeonly_scalar_nimm_predicate): New subst attributes.
8233 2017-07-06 Julia Koval <julia.koval@intel.com>
8235 * config/i386/i386.c (ix86_erase_embedded_rounding):
8236 Remove code for old rounding pattern.
8238 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8240 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8242 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8244 * doc/sourcebuild.texi (Test Directives, Variants of
8245 dg-require-support): Add documentation for dg-require-stack-check.
8247 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8249 * config/i386/subst.md (mask_scalar, round_scalar,
8250 round_saeonly_scalar): New meta-templates.
8251 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8252 round_scalar_mask_operand3, round_scalar_mask_op3,
8253 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8254 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8255 round_saeonly_scalar_constraint,
8256 round_saeonly_scalar_prefix): New subst attribute.
8257 * config/i386/sse.md
8258 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8259 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8260 <round_scalar_name> ... this.
8261 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8262 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8263 <round_scalar_name> ... this.
8264 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8265 <sse>_vm<code><mode>3<mask_scalar_name>
8266 <round_saeonly_scalar_name> ... this.
8267 (v<plusminus_mnemonic><ssescalarmodesuffix>
8268 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8269 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8270 v<plusminus_mnemonic><ssescalarmodesuffix>
8271 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8272 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8273 (v<multdiv_mnemonic><ssescalarmodesuffix>
8274 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8275 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8276 v<multdiv_mnemonic><ssescalarmodesuffix>
8277 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8278 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8279 (v<maxmin_float><ssescalarmodesuffix>
8280 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8281 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8282 v<maxmin_float><ssescalarmodesuffix>
8283 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8284 %0<mask_scalar_operand3>, %1, %<iptr>2
8285 <round_saeonly_scalar_mask_op3>} ... this.
8287 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8289 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8290 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8292 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8293 Alan Hayward <alan.hayward@arm.com>
8294 David Sherwood <david.sherwood@arm.com>
8296 * combine.c (simplify_if_then_else): Remove "enum" before
8298 * compare-elim.c (can_eliminate_compare): Likewise.
8299 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8301 (aarch64_lookup_simd_builtin_type): Likewise.
8302 (aarch64_simd_builtin_type): Likewise.
8303 (aarch64_init_simd_builtin_types): Likewise.
8304 (aarch64_simd_expand_args): Likewise.
8305 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8307 (aarch64_reverse_mask): Likewise.
8308 (aarch64_simd_emit_reg_reg_move): Likewise.
8309 (aarch64_gen_adjusted_ldpstp): Likewise.
8310 (aarch64_ccmp_mode_to_code): Likewise.
8311 (aarch64_operands_ok_for_ldpstp): Likewise.
8312 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8313 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8315 (aarch64_min_divisions_for_recip_mul): Likewise.
8316 (aarch64_reassociation_width): Likewise.
8317 (aarch64_get_condition_code_1): Likewise.
8318 (aarch64_simd_emit_reg_reg_move): Likewise.
8319 (aarch64_simd_attr_length_rglist): Likewise.
8320 (aarch64_reverse_mask): Likewise.
8321 (aarch64_operands_ok_for_ldpstp): Likewise.
8322 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8323 (aarch64_gen_adjusted_ldpstp): Likewise.
8324 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8326 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8327 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8328 (arm_lookup_simd_builtin_type): Likewise.
8329 (arm_simd_builtin_type): Likewise.
8330 (arm_init_simd_builtin_types): Likewise.
8331 (arm_expand_builtin_args): Likewise.
8332 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8333 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8334 (ft32_setup_incoming_varargs): Likewise.
8335 (ft32_function_arg): Likewise.
8336 (ft32_function_arg_advance): Likewise.
8337 (ft32_pass_by_reference): Likewise.
8338 (ft32_arg_partial_bytes): Likewise.
8339 (ft32_valid_pointer_mode): Likewise.
8340 (ft32_addr_space_pointer_mode): Likewise.
8341 (ft32_addr_space_legitimate_address_p): Likewise.
8342 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8344 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8345 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8346 (iamcu_alignment): Likewise.
8347 (canonicalize_vector_int_perm): Likewise.
8348 (ix86_noce_conversion_profitable_p): Likewise.
8349 (ix86_mpx_bound_mode): Likewise.
8350 (ix86_operands_ok_for_move_multiple): Likewise.
8351 * config/microblaze/microblaze-protos.h
8352 (microblaze_expand_conditional_branch_reg): Likewise.
8353 * config/microblaze/microblaze.c
8354 (microblaze_expand_conditional_branch_reg): Likewise.
8355 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8357 (rs6000_reassociation_width): Likewise.
8358 (rs6000_invalid_binary_op): Likewise.
8359 (fusion_p9_p): Likewise.
8360 (emit_fusion_p9_load): Likewise.
8361 (emit_fusion_p9_store): Likewise.
8362 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8364 (riscv_hard_regno_mode_ok_p): Likewise.
8365 (riscv_address_insns): Likewise.
8366 (riscv_split_symbol): Likewise.
8367 (riscv_legitimize_move): Likewise.
8368 (riscv_function_value): Likewise.
8369 (riscv_hard_regno_nregs): Likewise.
8370 (riscv_expand_builtin): Likewise.
8371 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8372 (riscv_build_integer): Likewise.
8373 (riscv_split_integer): Likewise.
8374 (riscv_legitimate_constant_p): Likewise.
8375 (riscv_cannot_force_const_mem): Likewise.
8376 (riscv_regno_mode_ok_for_base_p): Likewise.
8377 (riscv_valid_base_register_p): Likewise.
8378 (riscv_valid_offset_p): Likewise.
8379 (riscv_valid_lo_sum_p): Likewise.
8380 (riscv_classify_address): Likewise.
8381 (riscv_legitimate_address_p): Likewise.
8382 (riscv_address_insns): Likewise.
8383 (riscv_load_store_insns): Likewise.
8384 (riscv_force_binary): Likewise.
8385 (riscv_split_symbol): Likewise.
8386 (riscv_force_address): Likewise.
8387 (riscv_legitimize_address): Likewise.
8388 (riscv_move_integer): Likewise.
8389 (riscv_legitimize_const_move): Likewise.
8390 (riscv_legitimize_move): Likewise.
8391 (riscv_address_cost): Likewise.
8392 (riscv_subword): Likewise.
8393 (riscv_output_move): Likewise.
8394 (riscv_canonicalize_int_order_test): Likewise.
8395 (riscv_emit_int_order_test): Likewise.
8396 (riscv_function_arg_boundary): Likewise.
8397 (riscv_pass_mode_in_fpr_p): Likewise.
8398 (riscv_pass_fpr_single): Likewise.
8399 (riscv_pass_fpr_pair): Likewise.
8400 (riscv_get_arg_info): Likewise.
8401 (riscv_function_arg): Likewise.
8402 (riscv_function_arg_advance): Likewise.
8403 (riscv_arg_partial_bytes): Likewise.
8404 (riscv_function_value): Likewise.
8405 (riscv_pass_by_reference): Likewise.
8406 (riscv_setup_incoming_varargs): Likewise.
8407 (riscv_print_operand): Likewise.
8408 (riscv_elf_select_rtx_section): Likewise.
8409 (riscv_save_restore_reg): Likewise.
8410 (riscv_for_each_saved_reg): Likewise.
8411 (riscv_register_move_cost): Likewise.
8412 (riscv_hard_regno_mode_ok_p): Likewise.
8413 (riscv_hard_regno_nregs): Likewise.
8414 (riscv_class_max_nregs): Likewise.
8415 (riscv_memory_move_cost): Likewise.
8416 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8417 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8418 (rl78_addr_space_address_mode): Likewise.
8419 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8421 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8422 (rs6000_reassociation_width): Likewise.
8423 (rs6000_invalid_binary_op): Likewise.
8424 (fusion_p9_p): Likewise.
8425 (emit_fusion_p9_load): Likewise.
8426 (emit_fusion_p9_store): Likewise.
8427 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8428 (ok_for_simple_move_operands): Likewise.
8429 (ok_for_simple_move_strict_operands): Likewise.
8430 (ok_for_simple_arith_logic_operands): Likewise.
8431 (visium_legitimize_reload_address): Likewise.
8432 (visium_select_cc_mode): Likewise.
8433 (output_cbranch): Likewise.
8434 (visium_split_double_move): Likewise.
8435 (visium_expand_copysign): Likewise.
8436 (visium_expand_int_cstore): Likewise.
8437 (visium_expand_fp_cstore): Likewise.
8438 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8439 (visium_function_arg): Likewise.
8440 (visium_function_arg_advance): Likewise.
8441 (visium_libcall_value): Likewise.
8442 (visium_setup_incoming_varargs): Likewise.
8443 (visium_legitimate_constant_p): Likewise.
8444 (visium_legitimate_address_p): Likewise.
8445 (visium_legitimize_address): Likewise.
8446 (visium_secondary_reload): Likewise.
8447 (visium_register_move_cost): Likewise.
8448 (visium_memory_move_cost): Likewise.
8449 (prepare_move_operands): Likewise.
8450 (ok_for_simple_move_operands): Likewise.
8451 (ok_for_simple_move_strict_operands): Likewise.
8452 (ok_for_simple_arith_logic_operands): Likewise.
8453 (visium_function_value_1): Likewise.
8454 (rtx_ok_for_offset_p): Likewise.
8455 (visium_legitimize_reload_address): Likewise.
8456 (visium_split_double_move): Likewise.
8457 (visium_expand_copysign): Likewise.
8458 (visium_expand_int_cstore): Likewise.
8459 (visium_expand_fp_cstore): Likewise.
8460 (visium_split_cstore): Likewise.
8461 (visium_select_cc_mode): Likewise.
8462 (visium_split_cbranch): Likewise.
8463 (output_cbranch): Likewise.
8464 (visium_print_operand_address): Likewise.
8465 * expmed.c (flip_storage_order): Likewise.
8466 * expmed.h (emit_cstore): Likewise.
8467 (flip_storage_order): Likewise.
8468 * genrecog.c (validate_pattern): Likewise.
8469 * hsa-gen.c (gen_hsa_addr): Likewise.
8470 * internal-fn.c (expand_arith_overflow): Likewise.
8471 * ira-color.c (allocno_copy_cost_saving): Likewise.
8472 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8473 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8474 (process_invariant_for_inheritance): Likewise.
8475 * lra-eliminations.c (move_plus_up): Likewise.
8476 * omp-low.c (lower_oacc_reductions): Likewise.
8477 * simplify-rtx.c (simplify_subreg): Likewise.
8478 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8479 (TARGET_CHKP_BOUND_MODE): Likewise..
8480 * targhooks.c (default_chkp_bound_mode): Likewise.
8481 (default_setup_incoming_vararg_bounds): Likewise.
8482 * targhooks.h (default_chkp_bound_mode): Likewise.
8483 (default_setup_incoming_vararg_bounds): Likewise.
8484 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8485 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8486 (have_whole_vector_shift): Likewise.
8487 * tree-vect-stmts.c (vectorizable_load): Likewise.
8488 * doc/tm.texi: Regenerate.
8490 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8492 Graceful degrade if Binutils PR21472 is not available.
8495 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8496 .rodata in flash test fails.
8497 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8498 * confgure: Regenerate.
8499 * config.in: Regenerate.
8500 * config/avr/avr.c (avr_asm_named_section)
8501 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8502 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8503 (avr_asm_init_sections): Same.
8505 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8507 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8508 (fma<VH:mode>4_intrinsic): Likewise.
8509 (*fmsub<VCVTF:mode>4): Likewise.
8510 (*fmsub<VH:mode>4_intrinsic): Likewise.
8512 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8515 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8516 Don't depend on "optimize > 0".
8517 (out_movhi_r_mr, out_movqi_mr_r): Same.
8518 (out_movhi_mr_r, out_movqi_r_mr): Same.
8519 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8520 io_address_operand on "optimize > 0".
8522 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8524 * tree-loop-distribution.c: Add general explanantion on the pass.
8525 (generate_loops_for_partition): Mark distributed loop.
8526 (pg_add_dependence_edges): New parameter. Handle alias data
8527 dependence specially and record it in the parameter if asked.
8528 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8529 (init_partition_graph_vertices, add_partition_graph_edge): New.
8530 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8531 (free_partition_graph_vdata, build_partition_graph): New.
8532 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8533 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8534 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8535 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8536 (version_for_distribution_p, finalize_partitions): New.
8537 (distribute_loop): Handle alias data dependence specially. Factor
8538 out loop fusion code as functions and call these functions.
8540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8542 * tree-loop-distribution.c (classify_partition): New parameter and
8543 better handle reduction statement.
8544 (rdg_build_partitions): Revise comment.
8545 (distribute_loop): Compute statements in all partitions and pass it
8546 to classify_partition.
8548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8550 * tree-loop-distribution.c (enum partition_type): New.
8551 (struct partition): New field type.
8552 (partition_merge_into): Add parameter. Update partition type.
8553 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8554 (build_rdg_partition_for_vertex): Compute partition type.
8555 (rdg_build_partitions): Dump partition type.
8556 (distribute_loop): Update calls to partition_merge_into.
8558 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8560 * tree-loop-distribution.c (struct ddr_hasher): New.
8561 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
8563 (classify_partition): Call get_data_dependence.
8564 (pg_add_dependence_edges): Ditto.
8565 (distribute_loop): Release data dependence hash table.
8567 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8569 * tree-loop-distribution.c (ref_base_address): Delete.
8570 (similar_memory_accesses): Rename ...
8571 (share_memory_accesses): ... to this. Check if partitions access
8572 the same memory reference.
8573 (distribute_loop): Call share_memory_accesses.
8575 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8577 * tree-loop-distribution.c (struct partition): New field recording
8579 (partition_alloc, partition_free): Init and release data refs.
8580 (partition_merge_into): Merge data refs.
8581 (build_rdg_partition_for_vertex): Collect data refs for partition.
8582 (pg_add_dependence_edges): Change parameters from vector to bitmap.
8584 (distribute_loop): Remve data refs from vertice data of partition
8587 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8589 * tree-loop-distribution.c (params.h): Include header file.
8590 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
8591 (datarefs_vec): New global var.
8592 (create_rdg_vertices): Use datarefs_vec directly.
8593 (free_rdg): Don't free data references.
8594 (build_rdg): Update use. Don't free data references.
8595 (distribute_loop): Compute global variable for data references.
8596 Bail out if there are too many data references.
8598 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8600 * tree-loop-distribution.c (loop_nest): New global var.
8601 (build_rdg): Use loop directly, rather than loop nest.
8602 (pg_add_dependence_edges): Remove loop nest parameter. Use global
8604 (distribute_loop): Compute global variable loop nest. Update use.
8606 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8608 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
8609 (partition_merge_into): New parameter. Dump reason for fusion.
8610 (distribute_loop): Update use of partition_merge_into.
8612 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8614 * tree-loop-distribution.c (bb_top_order_index): New.
8615 (bb_top_order_index_size, bb_top_order_cmp): New.
8616 (stmts_from_loop): Use topological order.
8617 (pass_loop_distribution::execute): Compute and release topological
8618 order for basic blocks.
8620 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8622 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
8625 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8627 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
8628 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
8629 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
8630 * internal-fn.def (LOOP_DIST_ALIAS): New.
8631 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
8632 (fold_loop_internal_call): ... this.
8633 (vect_loop_dist_alias_call): New function.
8634 (set_uid_loop_bbs): Call fold_loop_internal_call.
8635 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
8638 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8641 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
8642 Require dead FLAGS_REG at the beginning of a peephole.
8644 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8647 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
8648 arguments in the call to __builtin_ia32_sbb_u32.
8649 (_subborrow_u64): Swap _X and _Y arguments in the call to
8650 __builtin_ia32_sbb_u64.
8652 2017-07-04 Jakub Jelinek <jakub@redhat.com>
8655 * tree-vrp.c (compare_assert_loc): Turn into a function template
8656 with stable template parameter. Only test if a->e is NULL,
8657 !a->e == !b->e has been verified already. Use e == NULL or
8658 e != NULL instead of e or ! e tests. If stable is true, don't use
8659 iterative_hash_expr, on the other side allow a or b or both NULL
8660 and sort the NULLs last.
8661 (process_assert_insertions): Sort using compare_assert_loc<false>
8662 instead of compare_assert_loc, later sort using
8663 compare_assert_loc<true> before calling process_assert_insertions_for
8664 in a loop. Use break instead of continue once seen NULL pointer.
8666 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8668 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8669 Cortex-R7 and Cortex-R8 processors.
8671 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8673 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
8674 uninitialized while src is not.
8676 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
8678 * common/config/arm/arm-common.c: Adjust include path for
8680 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
8681 (arm-cpu.h): Create in build directory. Adjust dependency rules.
8682 (arm-cpu-data.h): Likewise.
8683 (arm-cpu-cdata.h): Likewise.
8684 * config/arm/arm-cpu.h: Delete.
8685 * config/arm/arm-cpu-cdata.h: Delete.
8686 * config/arm/arm-cpu-data.h: Delete.
8688 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
8690 * config/arm/arm-cpus.in (cortex-a55): New.
8691 (cortex-a75): Likewise.
8692 (cortex-a75.cortex-a55): Likewise.
8693 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
8695 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
8696 * config/arm/arm-cpu-cdata.h: Regenerate.
8697 * config/arm/arm-cpu-data.h: Regenerate.
8698 * config/arm/arm-cpu.h: Regenerate.
8699 * config/arm/arm-tables.opt: Regenerate.
8700 * config/arm/arm-tune.md: Regenerate.
8702 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8704 * haifa-sched.c (sched_create_recovery_edges): Update profile.
8706 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8708 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
8711 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
8713 PR tree-optimization/81292
8714 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
8715 full_string_p, also call adjust_related_strinfos if the adjustment
8716 is simple, otherwise invalidate related strinfos.
8718 2017-07-04 Martin Liska <mliska@suse.cz>
8721 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
8722 newly created variable as DECL_IGNORED_P.
8724 2017-07-04 Martin Liska <mliska@suse.cz>
8727 * ipa-inline.c (inline_small_functions):
8728 Use xstrdup_for_dump.
8730 2017-07-04 Tom de Vries <tom@codesourcery.com>
8732 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8734 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8737 * config/darwin.c (darwin_function_switched_text_sections):
8740 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8742 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8744 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8746 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8748 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8750 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8751 min_profitable_iters, and th as inclusive lower bounds.
8752 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8753 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8754 for min_profitable_iters and min_profitable_estimate.
8755 (vect_transform_loop): Treat th as an inclusive lower bound.
8756 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8758 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8761 * config/darwin.c (darwin_function_switched_text_sections):
8762 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
8763 in two pieces, and suppress the use of buf.
8765 2017-07-03 Nathan Sidwell <nathan@acm.org>
8767 * hash-table.h (hash_table_mod1): Fix indentation.
8769 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8772 * predict.c (force_edge_cold): Be more careful about propagation
8774 * profile-count.h (profile_probability::guessed,
8775 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
8777 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
8779 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8781 * doc/invoke.texi (rcpc architecture extension): Document it.
8783 2017-07-03 Richard Biener <rguenther@suse.de>
8785 PR tree-optimization/60510
8786 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
8787 the scalar reduction PHI and use it.
8788 (vectorizable_reduction): Properly guard the single_defuse_cycle
8789 path for non-SLP reduction chains where we cannot use it.
8790 Rework reduc_def/index and vector type deduction. Rework
8791 vector operand gathering during reduction op code-gen.
8792 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
8793 chains dissolve the chain and leave it to non-SLP reduction
8796 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8798 * tree-data-ref.h (dr_alignment): Declare.
8799 * tree-data-ref.c (dr_alignment): New function.
8800 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
8801 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
8803 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
8805 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8807 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
8808 and base_misalignment fields.
8809 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
8810 * tree-data-ref.c: Include builtins.h.
8811 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
8812 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
8813 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
8814 * tree-vect-data-refs.c: Include tree-cfg.h.
8815 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
8816 fields instead of calculating an alignment here.
8817 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
8818 innermost_loop_behavior fields.
8820 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8822 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
8824 (DR_STEP_ALIGNMENT): New macro.
8825 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
8826 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
8827 (create_data_ref): Print it.
8828 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
8829 to tell whether the step preserves vector (mis)alignment.
8830 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8831 Move the check for an integer step and generalise to all INTEGER_CST.
8832 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
8833 Print the outer step alignment.
8835 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8837 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
8838 with offset_alignment.
8839 (DR_ALIGNED_TO): Delete.
8840 (DR_OFFSET_ALIGNMENT): New macro.
8841 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
8842 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
8843 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
8844 (create_data_ref): Likewise.
8845 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8846 (vect_analyze_data_refs): Likewise.
8847 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
8848 creating dummy innermost behavior.
8850 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8852 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
8853 with a "innermost_loop_behavior *" and refeence tree.
8854 * tree-data-ref.c (dr_analyze_innermost): Likewise.
8855 (create_data_ref): Update call accordingly.
8856 * tree-predcom.c (find_looparound_phi): Likewise.
8858 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8860 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
8861 fields with dr_wrt_vec_loop.
8862 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
8863 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
8864 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
8865 (vect_dr_behavior): New function.
8866 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8867 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
8868 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
8869 track whether the step preserves the misalignment.
8870 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8871 Use vect_dr_behavior.
8872 (vect_setup_realignment): Update call accordingly.
8873 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
8874 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
8875 call to vect_create_addr_base_for_vector_ref.
8876 (vect_create_cond_for_align_checks): Likewise.
8877 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
8878 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
8879 (vect_recog_mask_conversion_pattern): Likewise.
8880 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
8881 (new_stmt_vec_info): Remove redundant zeroing.
8883 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8885 * common/config/arm/arm-common.c (arm_be8_option): New function.
8886 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
8887 (ISA_ARMv6): Add isa_bit_be8.
8888 * config/arm/arm.h (arm_be8_option): Add prototype.
8889 (BE8_SPEC_FUNCTION): New define.
8890 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
8891 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
8892 (mlittle-endian): Similarly.
8893 (mbe8, mbe32): New options.
8894 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
8895 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
8897 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8899 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
8901 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8903 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
8904 (cleanup_tree_cfg_bb): Use it.
8905 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
8907 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
8909 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8912 * loop-doloop.c (add_test): Update profile.
8914 2017-07-03 Martin Liska <mliska@suse.cz>
8917 * sanopt.c (rewrite_usage_of_param): New function.
8918 (sanitize_rewrite_addressable_params): Likewise.
8919 (pass_sanopt::execute): Call rewrite_usage_of_param.
8921 2017-07-03 Richard Biener <rguenther@suse.de>
8923 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
8924 back to using VIEW_CONVERT_EXPR.
8926 2017-07-03 Martin Liska <mliska@suse.cz>
8929 * doc/extend.texi: Document when a resolver function is
8930 generated for target_clones.
8932 2017-07-03 Martin Liska <mliska@suse.cz>
8934 * asan.c (asan_emit_stack_protection): Unpoison just red zones
8935 and shadow memory of auto variables which are subject of
8936 use-after-scope sanitization.
8937 (asan_expand_mark_ifn): Add do set only when is_poison.
8939 2016-07-03 Richard Biener <rguenther@suse.de>
8941 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
8943 (vect_force_simple_reduction): Record reduction def -> phi mapping.
8944 (vectorizable_reduction): Perform reduction PHI creation when
8945 visiting a reduction PHI and adjust and simplify code generation
8946 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
8947 (vect_transform_loop): Visit reduction PHIs.
8948 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
8949 defs into the SLP tree.
8950 (vect_build_slp_tree): Reduction defs terminate the recursion.
8951 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
8953 (vect_get_vec_defs_for_stmt_copy): Export.
8954 (vect_get_vec_defs): Likewise.
8955 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
8957 (vect_get_vec_defs_for_stmt_copy): Declare.
8958 (vect_get_vec_defs): Likewise.
8960 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8962 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
8963 parameter with a "loop" parameter and use it instead of the
8964 loop containing DR_STMT. Don't check simple_iv when doing
8965 BB analysis. Describe the two analysis modes in the comment.
8967 2017-07-03 Tom de Vries <tom@codesourcery.com>
8969 PR tree-optimization/69468
8970 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
8971 (find_same_succ_bb): Handle ignore_edge_flags.
8973 2017-07-03 Tom de Vries <tom@codesourcery.com>
8975 PR tree-optimization/81192
8976 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
8978 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
8980 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
8982 2017-07-03 Tom de Vries <tom@codesourcery.com>
8984 PR tree-optimization/81192
8985 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
8986 BB_SAME_SUCC (bb) == NULL.
8988 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8990 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
8993 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8995 * dumpfile.c: Include profile-count.h
8996 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
8998 (insert_cond_bb): Update profile.
8999 * tree-cfg.h (insert_cond_bb): Update prototype.
9000 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9001 * tree-dump.c: Do not include tree-cfg.
9003 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9005 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9007 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9009 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9012 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9014 * tree-complex.c (expand_complex_div_wide): update profile.
9016 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9017 Alan Hayward <alan.hayward@arm.com>
9018 David Sherwood <david.sherwood@arm.com>
9020 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9021 (CORETYPES_H): New define.
9022 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9023 (insn-modes-inline.h, s-modes-inline-h): New rules.
9024 (generated_files): Add insn-modes-inline.h.
9025 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9026 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9027 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9028 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9029 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9030 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9031 (build/gencodes.o, build/genconditions.o): Likewise.
9032 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9033 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9034 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9035 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9036 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9037 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9038 * coretypes.h: Include everything up to real.h for generators.
9039 Include insn-modes.h first. Include wide-int-print.h after
9040 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9041 * machmode.h: Don't include insn-modes.h here.
9042 * function-tests.c: Remove includes of signop.h, machmode.h,
9043 double-int.h and wide-int.h.
9045 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9047 * optc-save-gen.awk: Likewise.
9048 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9049 * godump.c: Remove include of wide-int-print.h.
9050 * pretty-print.h: Likewise.
9051 * wide-int-print.cc: Likewise.
9052 * wide-int.cc: Likewise.
9053 * hash-map-tests.c: Remove include of signop.h.
9054 * hash-set-tests.c: Likewise.
9055 * rtl-tests.c: Likewise.
9056 * mkconfig.sh: Remove include of machmode.h.
9057 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9059 (emit_insn_modes_inline_h): ...this new function. Emit the code
9060 into an insn-modes-inline.h header file, adding appropriate
9061 include guards and end comments.
9062 (emit_insn_modes_c_header): Remove include of machmode.h.
9063 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9065 (main): Handle -i flag and call emit_insn_modes_inline_h when
9068 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9070 * tree-ssa-strlen.c (strinfo): Rename the length field to
9071 nonzero_chars. Add a full_string_p field.
9072 (compare_nonzero_chars, zero_length_string_p): New functions.
9073 (get_addr_stridx): Add an offset_out parameter.
9074 Use compare_nonzero_chars.
9075 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9076 (new_strinfo): Update after above changes to strinfo.
9077 (set_endptr_and_length): Set full_string_p.
9078 (get_string_length): Update after above changes to strinfo.
9079 (unshare_strinfo): Update call to new_strinfo.
9080 (maybe_invalidate): Likewise.
9081 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9082 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9083 as a uhwi instead of an shwi. Update after above changes to
9084 strinfo and new_strinfo.
9085 (zero_length_string): Assert that chainsi contains full strings.
9086 Use zero_length_string_p. Update call to new_strinfo.
9087 (adjust_related_strinfos): Update after above changes to strinfo.
9088 Copy full_string_p from origsi.
9089 (adjust_last_stmt): Use zero_length_string_p.
9090 (handle_builtin_strlen): Update after above changes to strinfo and
9091 new_strinfo. Install the lhs as the string length if the previous
9092 entry didn't describe a full string.
9093 (handle_builtin_strchr): Update after above changes to strinfo
9095 (handle_builtin_strcpy): Likewise.
9096 (handle_builtin_strcat): Likewise.
9097 (handle_builtin_malloc): Likewise.
9098 (handle_pointer_plus): Likewise.
9099 (handle_builtin_memcpy): Likewise. Track nonzero characters
9100 that aren't necessarily followed by a nul terminator.
9101 (handle_char_store): Likewise.
9103 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9105 PR tree-optimization/80769
9106 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9107 for malloc and calloc. Document the new invariant that all related
9108 strinfos have delayed lengths or none do.
9109 (verify_related_strinfos): Move earlier in file.
9110 (set_endptr_and_length): New function, split out from...
9111 (get_string_length): ...here. Also set the lengths of related
9113 (zero_length_string): Assert that chainsi has known (rather than
9115 (adjust_related_strinfos): Likewise.
9117 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9119 PR tree-optimization/81136
9120 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9121 assert that two references with the same misalignment have the same
9122 compile-time misalignment if those compile-time misalignments
9125 2017-07-01 Andi Kleen <ak@linux.intel.com>
9127 * print-tree.c (print_node): Print all attributes.
9129 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9131 * cfg.c (scale_bbs_frequencies): New function.
9132 * cfg.h (scale_bbs_frequencies): Declare it.
9133 * cfgloopanal.c (single_likely_exit): Cleanup.
9134 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9136 (scale_loop_profile): Likewise.
9137 (loop_version): Likewise.
9138 (create_empty_loop_on_edge): Update.
9139 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9140 scale_loop_frequencies, scale_loop_profile, loopify,
9141 loop_version): Update prototypes.
9142 * modulo-sched.c (sms_schedule): Update.
9143 * predict.c (unlikely_executed_edge_p): Also check probability.
9144 (probably_never_executed_edge_p): Fix typo.
9145 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9146 * tree-parloops.c (gen_parallel_loop): Update.
9147 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9148 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9149 * tree-ssa-loop-split.c (split_loop): Update.
9150 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9151 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9152 (vect_loop_versioning): Update.
9153 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9155 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9157 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9159 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9161 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9162 to keep profile consistent.
9164 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9166 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9167 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9168 * profile-count.h (max_safe_multiplier): Make unsigned.
9169 (profile_count::guessed_zero): New.
9171 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9173 * bb-reorder.c (fix_up_crossing_landing_pad,
9174 fix_crossing_conditional_branches): Use make_single_succ_edge
9175 to keep profile consistent.
9177 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9179 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9182 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9185 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9186 the right scopes, make sure cond_jump isn't preserved between multiple
9187 iterations. Search for fallthru edge whenever there are 3+ edges and
9188 use find_fallthru_edge for it.
9190 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9192 Patch by Alexander Monakov <amonakov@ispras.ru>
9193 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9194 probabilities consistently.
9196 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9198 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9199 emit_cmp_and_jump_insns.
9201 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9204 * tree-inline.c (expand_call_inline): Combine profile statuses.
9206 2017-06-30 Andrew Pinski <apinski@cavium.com>
9208 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9209 fold_stmt returned true.
9211 2017-06-30 Nathan Sidwell <nathan@acm.org>
9213 * ggc.h (empty_string): Delete.
9214 * cfgexpand.c (expand_asm_stmt): Use plain "".
9215 * optabs.c (expand_asm_memory_barrier): Likewise.
9216 * stringpool.c (empty_string): Delete.
9217 (digit_vector, digit_string): Delete.
9218 (ggc_alloc_string): Use plain "", don't optimize single digit
9219 strings. Use ggc_alloc_atomic.
9221 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9223 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9224 comparison set and one other set, use the cost of the non-comparison
9227 2017-06-30 Nathan Sidwell <nathan@acm.org>
9229 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9232 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9234 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9235 loops. Remove now unneeded calls to gimple_switch_set_label() that
9236 just set removed labels to NULL_TREE.
9238 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9240 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9241 (set_range_info): ...here. Only call set_range_info_raw if domain
9243 (set_nonzero_bits): Call set_range_info_raw.
9244 * tree-ssanames.h (set_range_info_raw): New.
9246 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9249 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9250 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9251 of nonimmediate_operand and <store_mask_constraint> instead of m
9252 for the input operand. For V8FI iterator, always split if input
9253 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9254 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9255 <store_mask_predicate> instead of register_operand and
9256 <store_mask_constraint> instead of v for the input operand. Make
9257 sure both operands aren't MEMs for if not <mask_applied>.
9259 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9261 * lto-wrapper.c (copy_file) Close both file descriptors before
9264 2017-06-30 Martin Liska <mliska@suse.cz>
9267 * multiple_target.c (create_dispatcher_calls): Make ifunc
9268 also for function that don't have calls or are not referenced.
9270 2017-06-30 Richard Biener <rguenther@suse.de>
9272 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9273 analyze the first scalar stmt. Move vector type computation
9274 for the BB case here from ...
9275 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9276 live operation processing in the SLP case properly.
9278 2017-06-30 Richard Biener <rguenther@suse.de>
9280 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9282 2017-06-30 Martin Liska <mliska@suse.cz>
9285 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9286 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9288 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9290 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9293 2017-06-30 Richard Biener <rguenther@suse.de>
9295 PR tree-optimization/81249
9296 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9297 condition reduction result to original scalar type.
9299 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9301 * profile-count.h (enum profile_quality): Fix typos and whitespace
9304 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9306 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9307 type for branch probabilities.
9309 2017-06-29 Julian Brown <julian@codesourcery.com>
9310 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9312 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9313 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9314 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9315 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9317 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9319 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9320 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9321 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9322 CC usage from generic code to here.
9323 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9324 CC usage into the target macros.
9326 2017-06-29 Maya Rashish <coypu@sdf.org>
9328 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9331 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9333 * arm/arm-builtins.c: Include profile-count.h
9334 * except.c (sjlj_emit_function_enter): Use
9335 profile_probability::unlikely.
9337 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9339 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9340 and tocrel_offset be pointer args rather than implicitly using
9342 (legitimate_constant_pool_address_p, rs6000_emit_move,
9343 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9344 tocrel_offset and use in toc_relative_expr_p call.
9345 (print_operand, print_operand_address): Use static tocrel_base_oac
9346 and tocrel_offset_oac.
9347 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9350 2017-06-29 Maya Rashish <coypu@sdf.org>
9352 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9354 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9356 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9357 objects, take into account only the alignment of 'op0' and 'mode1' if
9360 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9362 * ccmp.c (ccmp_tree_comparison_p): New function.
9363 (ccmp_candidate_p): Update to use above function.
9364 (get_compare_parts): New function.
9365 (expand_ccmp_next): Update to use new functions.
9366 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9368 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9369 take mode as argument.
9370 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9371 * expr.c (expand_expr_real_1): Pass mode as argument.
9373 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9375 * combine.c (combine_instructions): Print insns to dump_file, together
9378 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9380 * asan.c (asan_emit_stack_protection): Update.
9381 (create_cond_insert_point): Update.
9382 * auto-profile.c (afdo_propagate_circuit): Update.
9383 * basic-block.h (struct edge_def): Turn probability to
9384 profile_probability.
9385 (EDGE_FREQUENCY): Update.
9386 * bb-reorder.c (find_traces_1_round): Update.
9387 (better_edge_p): Update.
9388 (sanitize_hot_paths): Update.
9389 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9390 (make_single_succ_edge): Update.
9391 (check_bb_profile): Update.
9392 (dump_edge_info): Update.
9393 (update_bb_profile_for_threading): Update.
9394 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9395 probabilitycount to 0.
9396 * cfgbuild.c (compute_outgoing_frequencies): Update.
9397 * cfgcleanup.c (try_forward_edges): Update.
9398 (outgoing_edges_match): Update.
9399 (try_crossjump_to_edge): Update.
9400 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9401 (expand_gimple_tailcall): Update.
9402 (construct_init_block): Use make_single_succ_edge.
9403 (construct_exit_block): Use make_single_succ_edge.
9404 * cfghooks.c (verify_flow_info): Update.
9405 (redirect_edge_succ_nodup): Update.
9406 (split_edge): Update.
9407 (account_profile_record): Update.
9408 * cfgloopanal.c (single_likely_exit): Update.
9409 * cfgloopmanip.c (scale_loop_profile): Update.
9410 (set_zero_probability): Remove.
9411 (duplicate_loop_to_header_edge): Update.
9412 * cfgloopmanip.h (loop_version): Update prototype.
9413 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9414 (force_nonfallthru_and_redirect): Update.
9415 (update_br_prob_note): Update.
9416 (rtl_verify_edges): Update.
9417 (purge_dead_edges): Update.
9418 (rtl_lv_add_condition_to_bb): Update.
9419 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9420 * cgraphunit.c (init_lowered_empty_function): Update.
9421 (cgraph_node::expand_thunk): Update.
9422 * cilk-common.c: Include profile-count.h
9423 * dojump.c (inv): Remove.
9424 (jumpifnot): Update.
9425 (jumpifnot_1): Update.
9426 (do_jump_1): Update.
9428 (do_jump_by_parts_greater_rtx): Update.
9429 (do_compare_rtx_and_jump): Update.
9430 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9431 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9432 * dwarf2cfi.c: Include profile-count.h
9433 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9434 (sjlj_emit_dispatch_table): Likewise.
9435 * explow.c: Include profile-count.h
9436 * expmed.c (emit_store_flag_force): Update.
9437 (do_cmp_and_jump): Update.
9438 * expr.c (compare_by_pieces_d::generate): Update.
9439 (compare_by_pieces_d::finish_mode): Update.
9440 (emit_block_move_via_loop): Update.
9441 (store_expr_with_bounds): Update.
9442 (store_constructor): Update.
9443 (expand_expr_real_2): Update.
9444 (expand_expr_real_1): Update.
9445 * expr.h (try_casesi, try_tablejump): Update prototypes.
9446 * gimple-pretty-print.c (dump_probability): Update.
9447 (dump_profile): New.
9448 (dump_gimple_label): Update.
9449 (dump_gimple_bb_header): Update.
9450 * graph.c (draw_cfg_node_succ_edges): Update.
9451 * hsa-gen.c (convert_switch_statements): Update.
9452 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9453 (find_if_case_1): Update.
9454 (find_if_case_2): Update.
9455 * internal-fn.c (expand_arith_overflow_result_store): Update.
9456 (expand_addsub_overflow): Update.
9457 (expand_neg_overflow): Update.
9458 (expand_mul_overflow): Update.
9459 (expand_vector_ubsan_overflow): Update.
9460 * ipa-cp.c (good_cloning_opportunity_p): Update.
9461 * ipa-split.c (split_function): Use make_single_succ_edge.
9462 * ipa-utils.c (ipa_merge_profiles): Update.
9463 * loop-doloop.c (add_test): Update.
9464 (doloop_modify): Update.
9465 * loop-unroll.c (compare_and_jump_seq): Update.
9466 (unroll_loop_runtime_iterations): Update.
9467 * lra-constraints.c (lra_inheritance): Update.
9468 * lto-streamer-in.c (input_cfg): Update.
9469 * lto-streamer-out.c (output_cfg): Update.
9470 * mcf.c (adjust_cfg_counts): Update.
9471 * modulo-sched.c (sms_schedule): Update.
9472 * omp-expand.c (expand_omp_for_init_counts): Update.
9473 (extract_omp_for_update_vars): Update.
9474 (expand_omp_ordered_sink): Update.
9475 (expand_omp_for_ordered_loops): Update.
9476 (expand_omp_for_generic): Update.
9477 (expand_omp_for_static_nochunk): Update.
9478 (expand_omp_for_static_chunk): Update.
9479 (expand_cilk_for): Update.
9480 (expand_omp_simd): Update.
9481 (expand_omp_taskloop_for_outer): Update.
9482 (expand_omp_taskloop_for_inner): Update.
9483 * omp-simd-clone.c (simd_clone_adjust): Update.
9484 * optabs.c (expand_doubleword_shift): Update.
9485 (expand_abs): Update.
9486 (emit_cmp_and_jump_insn_1): Update.
9487 (expand_compare_and_swap_loop): Update.
9488 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9489 * predict.c (predictable_edge_p): Update.
9490 (edge_probability_reliable_p): Update.
9491 (set_even_probabilities): Update.
9492 (combine_predictions_for_insn): Update.
9493 (combine_predictions_for_bb): Update.
9494 (propagate_freq): Update.
9495 (estimate_bb_frequencies): Update.
9496 (force_edge_cold): Update.
9497 * profile-count.c (profile_count::dump): Add missing space into dump.
9498 (profile_count::debug): Add newline.
9499 (profile_count::differs_from_p): Explicitly convert to unsigned.
9500 (profile_count::stream_in): Update.
9501 (profile_probability::dump): New member function.
9502 (profile_probability::debug): New member function.
9503 (profile_probability::differs_from_p): New member function.
9504 (profile_probability::differs_lot_from_p): New member function.
9505 (profile_probability::stream_in): New member function.
9506 (profile_probability::stream_out): New member function.
9507 * profile-count.h (profile_count_quality): Rename to ...
9508 (profile_quality): ... this one.
9509 (profile_probability): New.
9510 (profile_count): Update.
9511 * profile.c (compute_branch_probabilities): Update.
9512 * recog.c (peep2_attempt): Update.
9513 * sched-ebb.c (schedule_ebbs): Update.
9514 * sched-rgn.c (find_single_block_region): Update.
9515 (compute_dom_prob_ps): Update.
9516 (schedule_region): Update.
9517 * sel-sched-ir.c (compute_succs_info): Update.
9518 * stmt.c (struct case_node): Update.
9519 (do_jump_if_equal): Update.
9520 (get_outgoing_edge_probs): Update.
9521 (conditional_probability): Update.
9522 (emit_case_dispatch_table): Update.
9523 (expand_case): Update.
9524 (expand_sjlj_dispatch_table): Update.
9525 (emit_case_nodes): Update.
9526 * targhooks.c: Update.
9527 * tracer.c (better_p): Update.
9528 (find_best_successor): Update.
9529 * trans-mem.c (expand_transaction): Update.
9530 * tree-call-cdce.c: Update.
9531 * tree-cfg.c (gimple_split_edge): Upate.
9532 (move_sese_region_to_fn): Upate.
9533 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9534 * tree-eh.c (lower_resx): Upate.
9535 (cleanup_empty_eh_move_lp): Upate.
9536 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9537 * tree-inline.c (copy_edges_for_bb): Update.
9538 (copy_cfg_body): Update.
9539 * tree-parloops.c (gen_parallel_loop): Update.
9540 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9541 (gimple_gen_time_profiler): Update.
9542 * tree-ssa-dce.c (remove_dead_stmt): Update.
9543 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9544 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9545 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9546 (unloop_loops): Update.
9547 (try_peel_loop): Update.
9548 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9549 * tree-ssa-loop-split.c (connect_loops): Update.
9550 (split_loop): Update.
9551 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9552 (hoist_guard): Update.
9553 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9554 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9555 (value_replacement): Update.
9556 * tree-ssa-reassoc.c (branch_fixup): Update.
9557 * tree-ssa-tail-merge.c (replace_block_by): Update.
9558 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9559 (create_edge_and_update_destination_phis): Update.
9560 (compute_path_counts): Update.
9561 (recompute_probabilities): Update.
9562 (update_joiner_offpath_counts): Update.
9563 (freqs_to_counts_path): Update.
9564 (duplicate_thread_path): Update.
9565 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9566 (struct switch_conv_info): Update.
9567 (gen_inbound_check): Update.
9568 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9569 (vect_do_peeling): Update.
9570 (vect_loop_versioning): Update.
9571 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9572 (optimize_mask_stores): Update.
9573 * ubsan.c (ubsan_expand_null_ifn): Update.
9574 * value-prof.c (gimple_divmod_fixed_value): Update.
9575 (gimple_divmod_fixed_value_transform): Update.
9576 (gimple_mod_pow2): Update.
9577 (gimple_mod_pow2_value_transform): Update.
9578 (gimple_mod_subtract): Update.
9579 (gimple_mod_subtract_transform): Update.
9580 (gimple_ic): Update.
9581 (gimple_stringop_fixed_value): Update.
9582 (gimple_stringops_transform): Update.
9583 * value-prof.h: Update.
9585 2017-06-29 Carl Love <cel@us.ibm.com>
9587 * config/rs6000/rs6000-c.c: Add support for built-in functions
9588 vector signed int vec_signed (vector float);
9589 vector signed long long vec_signed (vector double);
9590 vector signed int vec_signed2 (vector double, vector double);
9591 vector signed int vec_signede (vector double);
9592 vector signed int vec_signedo (vector double);
9593 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
9594 instruction generator.
9595 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9596 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
9597 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
9599 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
9600 vunsignede_v2df): Add define_expands.
9601 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
9602 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
9603 VEC_UNSIGNEDO): Add definitions.
9604 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9605 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
9606 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
9607 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
9608 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
9609 * config/rs6000/altivec.h (vec_signed, vec_signed2,
9610 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
9611 vec_unsignede, vec_unsignedo): Add builtin defines.
9612 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
9614 * doc/extend.texi: Update the built-in documentation file for the
9615 new built-in functions.
9617 2017-06-29 Richard Biener <rguenther@suse.de>
9619 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
9620 reduction chains to LOOP_VINFO_REDUCTIONS.
9621 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
9622 SLP reductions after processing reduction chains.
9624 2017-06-29 Nathan Sidwell <nathan@acm.org>
9626 * builtins.c (fold_builtin_FUNCTION): Use
9627 lang_hooks.decl_printable_name.
9629 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
9632 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
9633 with only one label.
9634 * stmt.c (expand_case): Assert NCASES is greater than one.
9636 2017-06-29 Richard Biener <rguenther@suse.de>
9638 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
9640 (group_case_labels): Likewise.
9641 (find_taken_edge): Push sanity checking on val to workers...
9642 (find_taken_edge_cond_expr): ... here
9643 (find_taken_edge_switch_expr): ... and here, handle cases
9644 with just a default label.
9645 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
9646 (group_case_labels): Likewise.
9647 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
9648 group_case_labels does anything cleanup the CFG again.
9650 2017-06-29 Bin Cheng <bin.cheng@arm.com>
9652 PR tree-optimization/81196
9653 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
9654 exit condition comparing two IVs.
9656 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
9658 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
9659 profile to the dummy entry at the end of the list of architectures.
9660 * config/arm/arm-cpu-cdata.h: Regenerated.
9662 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9663 Michael Collison <michael.collison@arm.com>
9666 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
9668 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
9669 (*aarch64_reg_<mode>3_minus_mask): New pattern.
9670 (*aarch64_<optab>_reg_di3_mask2): New pattern.
9671 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
9672 of shift when the shift amount is masked with constant equal to
9673 the size of the mode.
9674 * config/aarch64/predicates.md (subreg_lowpart_operator): New
9677 2017-06-29 Martin Liska <mliska@suse.cz>
9679 * config/i386/i386.opt: Change range from [1,5] to [0,5].
9681 2017-06-29 Yury Gribov <tetra2005@gmail.com>
9684 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
9687 (edge_growth_cache_entry::edge_growth_cache_entry): New
9689 (reset_edge_growth_cache): Update to use constructor.
9691 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9693 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
9694 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9695 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9697 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
9699 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
9700 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
9702 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
9704 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
9705 (*-linux-uclibc*): Add t-uclibc tmake_file.
9706 * config/t-musl: New.
9707 * config/t-uclibc: New.
9709 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
9711 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
9713 (gen_comm_data): Emit architectural setting of arch_prof.
9714 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
9716 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
9717 (armv8-m.base, armv8-m.main): Likewise.
9718 * arm-protos.h (arm_build_target): Add profile field.
9719 (arch_option): Likewise.
9720 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
9722 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
9723 arm_active_target.profile.
9725 2017-06-28 Richard Biener <rguenther@suse.de>
9728 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9729 TYPE_OVERFLOW_WRAPS.
9730 * match.pd (negate_expr_p): Likewise.
9731 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9732 fold_build2, not fold_binary.
9734 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9736 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9737 Convert memory address to Pmode.
9738 (aarch64_print_operand): Assert MEM operands are always Pmode.
9740 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9743 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9744 Remove redundant if.
9745 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9746 * config/arm/aarch-common-protos.h
9747 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9748 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9750 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9753 * multiple_target.c (create_dispatcher_calls): Set the default
9754 clone to be static, not public.
9756 2017-06-28 Richard Biener <rguenther@suse.de>
9758 * tree-vect-loop.c (vectorizable_reduction): Move special
9759 cond reduction IV var creation ...
9760 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9761 parameter. Use STMT_VINFO_VECTYPE.
9762 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
9765 2017-06-28 Martin Liska <mliska@suse.cz>
9768 * ipa-visibility.c (non_local_p): Handle visibility.
9770 2017-06-28 Martin Liska <mliska@suse.cz>
9773 * common.opt: Add IntegerRange to various options.
9774 * opt-functions.awk (integer_range_info): New function.
9775 * optc-gen.awk: Add integer_range_info to cl_options struct.
9776 * opts-common.c (decode_cmdline_option): Handle
9777 CL_ERR_INT_RANGE_ARG.
9778 (cmdline_handle_error): Likewise.
9779 * opts.c (print_filtered_help): Show valid interval in
9780 when --help is provided.
9781 * opts.h (struct cl_option): Add range_min and range_max fields.
9782 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
9784 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
9786 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
9787 (x * C EQ/NE y * C): New transformation.
9789 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
9791 * genmultilib (combination_space): Accept '+' in option names.
9793 2017-06-28 Martin Liska <mliska@suse.cz>
9796 * asan.c (instrument_derefs): Bail out inner references
9797 that are hard register variables.
9799 2017-06-28 Jakub Jelinek <jakub@redhat.com>
9802 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
9803 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
9805 2017-06-28 Richard Biener <rguenther@suse.de>
9807 * tree-vectorizer.h (vect_get_vec_defs): Remove.
9808 (vect_get_slp_defs): Adjust.
9809 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
9811 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
9813 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9814 get_initial_defs_for_reduction instead of vect_get_vec_defs.
9815 (vectorizable_reduction): Adjust.
9816 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
9818 (vect_get_slp_defs): Likewise.
9819 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
9820 (vectorizable_bswap): Adjust.
9821 (vectorizable_call): Likewise.
9822 (vectorizable_conversion): Likewise.
9823 (vectorizable_assignment): Likewise.
9824 (vectorizable_shift): Likewise.
9825 (vectorizable_operation): Likewise.
9826 (vectorizable_store): Likewise.
9827 (vectorizable_condition): Likewise.
9828 (vectorizable_comparison): Likewise.
9830 2017-06-28 Michael Collison <michael.collison@arm.com>
9833 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
9835 (arm_gen_movmemqi): Removed unused variable 'i'.
9836 Convert 'for' loop into 'while' loop.
9837 (arm_expand_prologue): Remove last unnecessary set of insn.
9838 (thumb_pop): Remove unused variable 'pushed_words'.
9839 (thumb_exit): Remove last unnecessary set of regs_to_pop.
9841 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9843 * config/s390/predicates.md: Use s390_rel_address_ok_p.
9844 * config/s390/s390-protos.h: Add prototype of
9845 s390_rel_address_ok_p.
9846 * config/s390/s390.c (s390_got_symbol): New function.
9847 (s390_rel_address_ok_p): New function.
9848 (legitimize_pic_address): Use s390_rel_address_ok_p.
9849 (s390_load_got): Use s390_got_symbol.
9850 (s390_option_override): Issue error if
9851 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
9852 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
9854 * config/s390/s390.opt: New option mpic-data-is-text-relative.
9856 2017-06-27 Andrew Pinski <apinski@cavium.com>
9858 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
9859 (X * copysign (1.0, X)): New pattern.
9860 (X * copysign (1.0, -X)): New pattern.
9861 (copysign (-1.0, CST)): New pattern.
9863 2017-06-27 Joseph Myers <joseph@codesourcery.com>
9865 * genmultilib (combination_space): Remove variable.
9866 Validate reuse rules against regular expression for any sequence
9867 of multilib options in any order.
9869 2017-06-27 Michael Collison <michael.collison@arm.com>
9871 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
9872 call aarch64_split_simd_combine.
9873 * (aarch64_combine_internal<mode>): Delete pattern.
9874 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
9875 Allow register and subreg operands.
9877 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9879 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
9880 specific need, just fallback on defaults.
9881 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
9883 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9884 Olivier Hainque <hainque@adacore.com>
9886 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
9888 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
9889 targets. Pick a default if no particular attempt applied.
9890 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
9893 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9895 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
9896 (x86_64-wrs-vxworks7): Likewise.
9898 2017-06-27 Marek Polacek <polacek@redhat.com>
9901 * ubsan.c (instrument_null): Check get_base_address's result for null.
9903 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9905 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
9907 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9909 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
9910 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
9911 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
9912 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
9914 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
9915 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
9916 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
9917 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
9918 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
9919 BUILT_IN_FEUPDATEENV): New builtins.
9920 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
9921 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
9922 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
9923 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
9925 (builtin_structptr_types): Adjust size.
9926 * tree.c (builtin_structptr_types): Add four entries.
9928 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9929 Olivier Hainque <hainque@adacore.com>
9931 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
9932 (TLS_SYM): New local macro, forcing reference to __tls__ on
9933 link command lines for VxWorks 7 RTPs, triggering initialization
9935 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
9936 OS features TLS support, true for RTPs on VxWorks 7.
9937 * config/vxworks.c (vxworks_override_options): Setup emutls
9940 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9942 * predict.c (test_prediction_value_range): Use -1U instead of -1
9943 to avoid narrowing conversion warning.
9944 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
9945 to avoid narrowing conversion warning.
9946 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
9948 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
9950 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9952 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
9953 64bit configurations.
9954 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
9955 (SIZE_TYPE): Likewise.
9956 * config/vxworks.c (vxworks_emutls_var_fields): Use
9957 long_unsigned_type_node instead of unsigned_type_node as the offset
9958 field type, which is "pointer" mode in emutls.c.
9960 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9963 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
9966 * gimple-fold.c (replace_call_with_call_and_fold): Handle
9967 gimple_vuse copying separately from gimple_vdef copying.
9969 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9971 * value-prof.c (free_hist): Remove call to memset and the enclosing if
9974 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9975 Olivier Hainque <hainque@adacore.com>
9977 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
9978 for all vxworks7 targets.
9979 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
9980 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
9981 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
9982 variations for VX6/VX7 and 32/64bits later on in ...
9983 (VXWORKS_LIB_SPEC): Leverage new macros.
9984 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
9985 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
9987 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9989 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
9990 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
9992 2017-06-26 Carl Love <cel@us.ibm.com>
9994 * config/rs6000/rs6000-c.c: Add support for built-in functions
9995 vector bool char vec_reve (vector bool char);
9996 vector signed char vec_reve (vector signed char);
9997 vector unsigned char vec_reve (vector unsigned char);
9998 vector bool int vec_reve (vector bool int);
9999 vector signed int vec_reve (vector signed int);
10000 vector unsigned int vec_reve (vector unsigned int);
10001 vector bool long long vec_reve (vector bool long long);
10002 vector signed long long vec_reve (vector signed long long);
10003 vector unsigned long long vec_reve (vector unsigned long long);
10004 vector bool short vec_reve (vector bool short);
10005 vector signed short vec_reve (vector signed short);
10006 vector double vec_reve (vector double);
10007 vector float vec_reve (vector float);
10008 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10009 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10010 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10011 (altivec_vreve): New pattern.
10012 * config/rs6000/altivec.h (vec_reve): New define.
10013 * doc/extend.texi (vec_rev): Update the built-in documentation file
10014 for the new built-in functions.
10016 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10018 PR tree-optimization/71815
10019 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10021 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10023 (slsr_process_phi): Likewise.
10024 (replace_uncond_cands_and_profitable_phis): Don't replace a
10025 multiply candidate with a stride of 1 (copy or cast).
10026 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10028 (lowest_cost_path): Likewise.
10029 (total_savings): Likewise.
10031 2017-06-26 Richard Biener <rguenther@suse.de>
10034 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10035 Use def_builtin_pure for all gather builtins.
10037 2017-06-26 Richard Biener <rguenther@suse.de>
10039 PR tree-optimization/81203
10040 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10041 non-dominating BBs.
10043 2017-06-26 Marek Polacek <polacek@redhat.com>
10046 * doc/invoke.texi: Document -Wmultistatement-macros.
10048 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10050 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10051 arm_neon_ok_no_float_abi effective target.
10053 2017-06-26 Richard Biener <rguenther@suse.de>
10055 PR tree-optimization/80928
10056 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10057 (copy_bbs): Set BB_DUPLICATED flag early.
10058 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10060 (execute_on_shrinking_pred): Likewise.
10061 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10062 BB_DUPLICATED blocks.
10063 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10064 iterate over all PHIs considering removal of *gsi.
10066 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10068 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10071 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10073 * config/rs6000/rs6000-string.c: (expand_block_clear,
10074 do_load_for_compare, select_block_compare_mode,
10075 compute_current_alignment, expand_block_compare,
10076 expand_strncmp_align_check, expand_strn_compare,
10077 expand_block_move, rs6000_output_load_multiple)
10078 Move functions related to string/block move/compare
10079 to a separate file.
10080 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10081 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10082 for this function which is now used in two files.
10083 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10084 * config.gcc: Add rs6000-string.o to extra_objs for
10085 targets powerpc*-*-* and rs6000*-*-*.
10087 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10090 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10091 32-bit, since indexed is not valid for DImode.
10092 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10093 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10094 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10095 (define_peephole2 for Altivec d-form store): Likewise.
10098 * multiple_target.c (create_dispatcher_calls): Only create the
10099 dispatcher call if the function is the default clone of a
10100 versioned function.
10102 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10104 PR middle-end/80902
10105 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10106 a call, force the call to not be a tail call.
10108 2017-06-23 Jeff Law <law@redhat.com>
10110 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10113 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10115 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10116 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10117 handling for early expansion of vector shifts (sl,sr,sra,rl).
10118 (builtin_function_type): Add vector shift right instructions
10119 to the unsigned argument list.
10121 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10123 rtl-optimizatoin/79286
10124 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10125 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10126 trap. PIC register plus a const unspec without offset can never trap.
10128 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10130 * tree.h (builtin_structptr_type): New type.
10131 (builtin_structptr_types): Declare new array.
10132 * tree.c (builtin_structptr_types): New array.
10133 (free_lang_data, build_common_tree_nodes): Use it.
10135 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10138 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10141 2017-06-22 Matt Turner <mattst88@gmail.com>
10143 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10144 Lake models to skylake case. Assume skylake for unknown
10145 models with clflushopt.
10147 2017-06-22 Jeff Law <law@redhat.com>
10149 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10150 frame sizes that do not satisfy aarch64_uimm12_shift.
10152 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10154 * profile-count.h (apply_probability,
10155 apply_scale, probability_in): Fix checks for zero.
10157 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10159 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10160 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10162 2016-06-22 Richard Biener <rguenther@suse.de>
10164 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10165 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10166 REDUC_MAX_EXPR support.
10167 (vectorizable_reduction): Likewise.
10168 (vect_create_epilog_for_reduction): Likewise.
10170 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10172 * match.pd (A / (1 << B) -> A >> B): New.
10173 * generic-match-head.c: Include optabs-tree.h.
10174 * gimple-match-head.c: Likewise.
10175 * optabs-tree.h (target_supports_op_p): New.
10176 * optabs-tree.c (target_supports_op_p): New.
10178 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10180 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10181 $gcc_cv_ld --help output.
10182 (gcc_cv_ld_demangle): Likewise.
10183 (gcc_cv_ld_eh_frame_hdr): Likewise.
10184 (gcc_cv_ld_pie): Likewise.
10185 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10186 (gcc_cv_ld_buildid): Likewise.
10187 (gcc_cv_ld_sysroot): Likewise.
10188 (ld_bndplt_support): Likewise.
10189 (ld_pushpopstate_support): Likewise.
10190 * configure: Regenerate.
10191 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10193 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10196 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10197 operands indexes to avoid gap between operands and match_dups.
10199 2017-06-21 Andrew Pinski <apinski@cavium.com>
10201 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10202 Increment Arith_shift and Arith_shift_reg by 1.
10203 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10205 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10206 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10207 (aarch64_strip_extend): Add new argument and test for it.
10208 (aarch64_cheap_mult_shift_p): New function.
10209 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10210 add a cost if it is true.
10211 Update calls to aarch64_strip_extend.
10212 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10214 2017-06-21 Andrew Pinski <apinski@cavium.com>
10216 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10218 (thunderxt88): Likewise.
10219 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10220 (thunderx_prefetch_tune): New variable.
10221 (thunderx2t99_prefetch_tune): Update for the correct values.
10222 (thunderxt88_tunings): New variable.
10223 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10224 generic_prefetch_tune.
10225 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10227 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10229 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10230 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10231 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10232 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10233 (aarch64_atomic_cas<mode>, GPI): Likewise.
10235 2017-06-21 Martin Liska <mliska@suse.cz>
10237 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10238 statements on cold and hot labels.
10239 * predict.c (tree_estimate_probability_bb): Remove the
10240 prediction from this place.
10242 2017-06-21 Martin Liska <mliska@suse.cz>
10244 PR tree-optimization/79489
10245 * gimplify.c (maybe_add_early_return_predict_stmt): New
10247 (gimplify_return_expr): Call the function.
10248 * predict.c (tree_estimate_probability_bb): Remove handling
10250 * predict.def: Update comment about early return predictor.
10251 * gimple-predict.h (is_gimple_predict): New function.
10252 * predict.def: Change default value of early return to 66.
10253 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10255 * passes.def: Put pass_strip_predict_hints to the beginning of
10258 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10260 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10261 FUNCTION_DECL declarations.
10262 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10264 (dwaf2out_decl): Likewise.
10265 * godump.c (go_early_global_decl): Skip call to the real debug hook
10266 for FUNCTION_DECL declarations.
10267 * passes.c (rest_of_decl_compilation): Skip call to the
10268 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10269 -fdump-go-spec is passed.
10271 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10273 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10274 Reorder for compactness.
10275 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10276 (def_builtin_pure, def_builtin_pure2): New functions.
10277 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10279 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10281 * match.pd (nop_convert): New predicate.
10282 ((A +- CST1) +- CST2): Allow some NOP conversions.
10284 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10287 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10288 with ctors/dtors if GOVD_SHARED is set.
10290 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10292 * config/aarch64/aarch64.md (movti_aarch64):
10293 Emit mov rather than orr.
10294 (movtf_aarch64): Likewise.
10295 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10296 Emit mov rather than orr.
10298 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10300 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10301 Swap alternatives, make integer dup more expensive.
10303 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10305 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10306 Return true for non-tls symbols.
10308 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10310 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10311 (cortex-a75): Likewise.
10312 (cortex-a75.cortex-a55): Likewise.
10313 * config/aarch64/aarch64-tune.md: Regenerate.
10314 * doc/invoke.texi (-mtune): Document new values for -mtune.
10316 2017-06-21 Tom de Vries <tom@codesourcery.com>
10318 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10319 stack_size feature.
10320 (Effective-Target Keywords, Other attributes): Suggest using
10321 dg-add-options stack_size feature to get stack limit in stack_size
10322 effective target documentation.
10324 2017-06-21 Julian Brown <julian@codesourcery.com>
10325 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10327 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10328 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10329 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10331 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10332 attribute type list for neon_multiply.
10333 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10334 attribute type list for neon_multiply.
10335 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10336 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10337 attribute type list for neon_multiply.
10338 * config/arm/types.md (crypto_pmull): Add.
10339 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10340 attribute type list.
10342 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10344 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10347 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10349 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10350 to make sure not to dereference a NULL cost_classes_ptr pointer.
10352 2017-06-20 Carl Love <cel@us.ibm.com>
10354 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10355 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10356 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10357 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10358 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10359 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10360 VMULOSW): New enum "unspec" values.
10361 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10362 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10363 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10364 altivec_vmulosw): New patterns.
10365 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10366 VMULOSW): Add definitions.
10368 2017-06-20 Julia Koval <julia.koval@intel.com>
10370 * config/i386/i386.c: Fix rounding expand for new pattern.
10371 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10373 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10375 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10376 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10378 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10380 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10383 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10385 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10386 architecture, then by implementer ID.
10387 * config/aarch64/aarch64-tune.md: Regenerate.
10389 2017-06-20 Richard Biener <rguenther@suse.de>
10391 PR middle-end/81097
10392 * fold-const.c (split_tree): Fold to type before negating.
10394 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10396 * diagnostic-show-locus.c
10397 (selftest::test_fixit_deletion_affecting_newline): New function.
10398 (selftest::diagnostic_show_locus_c_tests): Call it.
10400 2017-06-20 Andreas Schwab <schwab@suse.de>
10403 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10406 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10408 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10409 __ARM_FEATURE_COPROC according to support.
10411 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10413 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10414 Rewritten to avoid overflow for > 32-bit pointers.
10417 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10418 by removing enum keyword.
10419 (ubsan_type_descriptor): Likewise. Formatting fix.
10422 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10423 splitter): Require TARGET_SSE2 in the condition.
10425 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10428 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10429 for doing vector set of SFmode on ISA 3.0.
10430 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10431 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10433 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10434 SFmode value into a V4SF variable that was extracted from another
10435 V4SF variable without converting the element to double precision
10436 and back to single precision vector format.
10437 (vsx_insert_extract_v4sf_p9_2): Likewise.
10439 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10441 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10442 in UWHI to avoid undefined overflow.
10445 * ubsan.h (enum ubsan_encode_value_phase): New.
10446 (ubsan_encode_value): Change second argument to
10447 enum ubsan_encode_value_phase with default value of
10448 UBSAN_ENCODE_VALUE_GENERIC.
10449 * ubsan.c (ubsan_encode_value): Change second argument to
10450 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10451 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10452 create_tmp_var_raw instead of create_tmp_var and use a
10454 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10455 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10456 ubsan_encode_value callers.
10459 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10460 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10461 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10463 2017-06-19 Richard Biener <rguenther@suse.de>
10465 PR middle-end/81118
10466 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10467 estimates if we changed anything.
10469 2017-06-19 Richard Biener <rguenther@suse.de>
10471 PR tree-optimization/80887
10472 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10473 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10474 simplified lookups, then reset mprts_hook.
10475 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10477 (try_to_simplify): Likewise.
10479 2017-06-19 Martin Liska <mliska@suse.cz>
10482 * gimplify.c (gimplify_switch_expr):
10483 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10485 2017-06-19 Martin Liska <mliska@suse.cz>
10487 * doc/install.texi: Document that PGO runs in 4 stages.
10489 2017-06-19 Martin Liska <mliska@suse.cz>
10492 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10493 to dispatcher function name.
10494 * multiple_target.c (replace_function_decl): New function.
10495 (create_dispatcher_calls): Redirect both edges and references.
10497 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10499 * profile-count.c (profile_count::dump): Dump quality.
10500 (profile_count::differs_from_p): Update for unsigned val.
10501 * profile-count.h (profile_count_quality): New enum.
10502 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10504 2017-06-19 Richard Biener <rguenther@suse.de>
10506 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10507 struct function as arg.
10508 (estimate_numbers_of_iterations): Export overload with loop arg.
10509 (free_numbers_of_iterations_estimates_loop): Use an overload of
10510 free_numbers_of_iterations_estimates instead.
10511 * tree-cfg.c (remove_bb): Adjust.
10512 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10513 * tree-parloops.c (gen_parallel_loop): Likewise.
10514 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10516 (tree_unroll_loops_completely): Likewise.
10517 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10518 Use an overload instead and export.
10519 (estimated_loop_iterations): Adjust.
10520 (max_loop_iterations): Likewise.
10521 (likely_max_loop_iterations): Likewise.
10522 (estimate_numbers_of_iterations): Take struct function as arg
10524 (loop_exits_before_overflow): Adjust.
10525 (free_numbers_of_iterations_estimates_loop): Use an overload.
10526 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10527 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10529 2017-06-19 Richard Biener <rguenther@suse.de>
10532 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10533 RANGE_EXPR conservatively.
10535 2017-06-16 Carl Love <cel@us.ibm.com>
10537 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10538 definitions for vec_float, vec_float2, vec_floato,
10539 vec_floate built-ins.
10540 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10541 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10543 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10544 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10545 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10546 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10547 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10548 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10549 vec_floato): Add builtin defines.
10550 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10551 Update the built-in documentation file for the new built-in
10554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10556 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10557 (mthumb): Mark as the negative of -marm.
10559 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10561 * doc/invoke.texi (ARM Options, -mcpu): Document supported
10563 (ARM Options, -mtune): Document that this accepts the same
10564 extension options as -mcpu.
10565 (ARM Options, -mfpu): Document addition of -mfpu=auto.
10567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10569 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
10570 permitted extensions.
10572 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10574 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
10575 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
10576 (armv8-m.main): Add option +nodsp.
10577 * config/arm/arm-cpu-cdata.h: Regenerated.
10579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10581 * config/arm/t-fuchsia: New file.
10582 * config.gcc (arm*-*-fuchsia*): Use it.
10584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10586 * config/arm/t-symbian: Rewrite for new option infrastructure.
10588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10590 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
10591 (MULTILIB_REQUIRED): Likewise.
10593 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10595 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
10596 (MULTILIB_RESUE): Likewise.
10597 (MULTILIB_MATCHES): Likewise.
10598 (MULTLIB_REQUIRED): Likewise.
10600 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10602 * config/arm/t-rtems: Rewrite for new option framework.
10604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10606 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
10607 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
10608 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
10609 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
10610 * config/arm/t-multilib: ... here.
10611 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
10612 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
10613 armv7-a and armv8*-a when A-profile libraries have not been built.
10614 * config/arm/t-rmprofile: Rewrite.
10616 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10618 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
10619 with a backslash. Remove the backslash after substituting unescaped
10621 * doc/fragments.texi (MULTILIB_REUSE): Document it.
10623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10625 * config.gcc: (arm*-*-*): When building a-profile libraries, force
10626 the driver to pass through the default setting of -mfloat-abi.
10627 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
10629 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
10630 (all_feat_combs): New rule.
10631 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
10633 * config/arm/t-aprofile: Rewrite.
10635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10637 * config/arm/arm.h (FPUTYPE_AUTO): Define.
10638 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
10639 fpu is not specified by the user/command-line.
10640 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
10641 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
10642 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
10643 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
10644 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
10645 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
10647 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10649 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
10650 * config/arm/t-arm-elf: Rewritten.
10652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10654 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
10655 have some floating-point instructions.
10656 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
10657 (TARGET_MAYBE_HARD_FLOAT): New macro.
10658 * config/arm/arm-builtins.c (arm_init_builtins): Use
10659 TARGET_MAYBE_HARD_FLOAT.
10660 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
10662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10664 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
10665 (configargs.h): Include it.
10666 (arm_print_hint_for_fpu_option): New function.
10667 (arm_parse_fpu_option): New function.
10668 (candidate_extension): New class.
10669 (arm_canon_for_multilib): New function.
10670 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
10671 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
10672 (ARCH_CANONICAL_SPECS): New macro.
10673 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
10675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10677 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
10678 are set after handling multilib fragments. Set target_cpu_default2
10681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10683 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
10685 (arm*-*-*): Set target_cpu_default2 to a quoted string.
10686 * config/arm/parsecpu.awk (check_cpu): Validate any extension
10688 (check_arch): Likewise.
10689 * config/arm/arm.c (arm_configure_build_target): Handle
10690 TARGET_CPU_DEFAULT being a string constant. Scan any feature
10691 options in the default.
10693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10695 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
10696 when an option is an alias of another.
10697 * config/arm/parsecpu.awk (optalias): New parser token.
10698 (gen_comm_data): Mark non-alias options as such. Emit entries
10699 for extension aliases.
10700 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
10701 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
10702 (armv6kz, armv6zk, armv6t2): Likewise.
10703 (armv7): Make vfpv3-d16 an alias.
10704 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
10706 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
10707 Sort in canonical order.
10708 (armv8-a): Sort in canonical order.
10709 (armv8.1-a, armv8.2-a): Likewise.
10710 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
10712 (cortex-a9): Sort in canonical order.
10713 * config/arm/arm.c (selftests.h): Include it.
10714 (arm_test_cpu_arch_data): New function.
10715 (arm_run_self_tests): New function.
10716 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
10717 (targetm): Move declaration to the end of the file.
10718 * arm-cpu-cdata.h: Regenerated.
10720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10722 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
10723 call to target_mode_check describing the type of option passed.
10724 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
10725 (arm_target_thumb_only): Use arm_parse_arch_option_name or
10726 arm_parse_cpu_option_name to match parameters against list of
10728 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10729 arm_arch_core_flags data structure.
10730 * config/arm/arm-cpu_cdata.h: Regenerated.
10732 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10734 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10736 (arm_print_hint_for_cpu_option): Likewise.
10737 (arm_print_hint_for_arch_option): Likewise.
10738 (arm_parse_cpu_option_name): Likewise.
10739 (arm_parse_arch_option_name): Likewise.
10740 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10741 of entries in the all_fpus list.
10742 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10743 (arm_parse_cpu_option_name): Declare.
10744 (arm_parse_arch_option_name): Declare.
10745 (arm_parse_option_features): Declare.
10746 (arm_intialize_isa): Declare.
10747 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10749 (gen_comm_data): ... here. Make definitions non-static.
10750 * config/arm/arm-cpu-data.h: Regenerated.
10751 * config/arm/arm-cpu-cdata.h: Regenerated.
10753 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10755 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10756 (cpu_arch_extension): New structure.
10757 (cpu_arch_option, arch_option, cpu_option): New structures.
10758 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10759 architecture types.
10760 (gen_data): Generate new format data tables.
10761 * config/arm/arm.c (cpu_tune): New structure.
10762 (cpu_option, processors): Delete.
10763 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
10764 (arm_print_hint_for_cpu_option): ... this and ...
10765 (arm_print_hint_for_arch_option): ... this.
10766 (arm_parse_arch_cpu_name): Delete. Replace with ...
10767 (arm_parse_cpu_option_name): ... this and ...
10768 (arm_parse_arch_option_name): ... this.
10769 (arm_unrecognized_feature): Change type of target parameter to
10771 (arm_parse_arch_cpu_features): Delete. Replace with ...
10772 (arm_parse_option_features): ... this.
10773 (arm_configure_build_target): Rework to use new configuration data
10775 (arm_print_tune_info): Rework for new configuration data tables.
10776 * config/arm/arm-cpu-data.h: Regenerated.
10777 * config/arm/arm-cpu.h: Regenerated.
10779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10781 * Makefile.in (OBJS): Move sbitmap.o from here ...
10782 (OBJS-libcommon): ... to here.
10784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10786 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
10787 (ISA_ALL_CRYPTO): New macro.
10788 (ISA_ALL_SIMD): New macro
10789 (ISA_ALL_FP): New macro.
10790 * config/arm/arm.c (fpu_bitlist): Update initializer.
10791 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
10793 (arm9e): Add fpu. Add option for nofp
10794 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
10795 (arm926ej-s, arm1026ej-s): Likewise.
10796 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
10797 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
10798 neon-fp16, neon-vfpv4, nofp and nosimd.
10799 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
10800 (cortex-a8): Add fpu. Add option for nofp.
10801 (cortex-a9): Add fpu. Add options for nosimd and nofp.
10802 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
10803 (cortex-r4f): Add fpu.
10804 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
10805 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
10807 (cortex-r8): Likewise.
10808 (cortex-m4): Add fpu. Add option for nofp.
10809 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
10810 (cortex-a17.cortex-a7): Likewise.
10811 (cortex-a32): Add fpu. Add options for crypto and nofp.
10812 (cortex-a35, cortex-a53): Likewise.
10813 (cortex-a57): Add fpu. Add option for crypto.
10814 (cortex-a72, cortex-a73): Likewise.
10815 (exynos-m1): Likewise.
10816 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
10817 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
10818 (cortex-m33): Add fpu. Add option for nofp.
10819 * config/arm/arm-cpu-cdata.h: Regenerated
10820 * config/arm/arm-cpu-data.h: Regenerated.
10822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10824 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
10825 (armv5te, armv5tej): Likewise.
10826 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
10827 (armv7): Add options fp and vfpv3-d16.
10828 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
10829 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
10831 (armv7ve): Likewise.
10832 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
10833 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
10834 (armv8-a): Add nocrypto option.
10835 (armv8.1-a, armv8.2-a): Likewise.
10836 (armv8-m.main): add options fp, fp.dp and nofp.
10838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10840 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
10842 (armv8-a+crc): Delete.
10843 (armv8.1-a): Add options simd, crypto and nofp.
10844 (armv8.2-a): Add options fp16, simd, crypto and nofp.
10845 (armv8.2-a+fp16): Delete.
10846 (armv8-m.main): Add option dsp.
10847 (armv8-m.main+dsp): Delete.
10848 (cortex-a8): Add fpu. Add nofp option.
10849 (cortex-a9): Add fpu. Add nofp and nosimd options.
10850 * config/arm/parsecpu.awk (gen_data): Generate option tables and
10851 link to main cpu and architecture data structures.
10852 (gen_comm_data): Only put isa attributes from the main architecture
10854 (option): New statement for architecture and CPU entries.
10855 * arm.c (struct cpu_option): New structure.
10856 (struct processors): Add entry for options.
10857 (arm_unrecognized_feature): New function.
10858 (arm_parse_arch_cpu_name): Ignore any characters after the first
10860 (arm_parse_arch_cpu_feature): New function.
10861 (arm_configure_build_target): Separate out any CPU and architecture
10862 features and parse separately. Don't error out if -mfpu=auto is
10863 used with only an architecture string.
10864 (arm_print_asm_arch_directives): New function.
10865 (arm_file_start): Call it.
10866 * config/arm/arm-cpu-cdata.h: Regenerated.
10867 * config/arm/arm-cpu-data.h: Likewise.
10868 * config/arm/arm-tables.opt: Likewise.
10870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10872 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
10873 assembler when it is not -mfpu=auto.
10875 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10877 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
10878 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
10879 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
10880 (ASM_CPU_SPEC): Rewrite.
10881 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
10882 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
10883 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
10884 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
10885 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
10886 copied string is NUL-terminated. Also strip any characters prefixed
10888 (arm_rewrite_selected_arch): New function.
10889 (arm_rewrite_march): New function.
10891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10893 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
10894 (x_arm_cpu_string, x_arm_tune_string): Likewise.
10895 (march, mcpu, mtune): Convert to string-based options.
10896 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
10897 (arm_parse_arch_cpu_name): New function.
10898 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
10899 identify selected architecture or CPU.
10900 (arm_option_save): New function.
10901 (TARGET_OPTION_SAVE): Redefine.
10902 (arm_option_restore): Restore string options.
10903 (arm_option_print): Print string options.
10905 2017-06-16 Martin Sebor <msebor@redhat.com>
10907 PR tree-optimization/80933
10908 PR tree-optimization/80934
10909 * builtins.c (fold_builtin_3): Do not handle bcmp here.
10910 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
10911 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
10912 (gimple_fold_builtin): Call them.
10914 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10916 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
10917 as unlikely; update profile.
10919 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10921 * predict.c (force_edge_cold): Handle declaring edges impossible
10924 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10926 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
10928 (try_unroll_loop_completely): Fix reporting.
10930 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10932 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
10934 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
10937 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
10938 if given a non-constant argument for an intrinsic which requires a
10941 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10943 * profile.c (compare_freqs): New function.
10944 (branch_prob): Sort edge list.
10945 (find_spanning_tree): Assume that the list is priority sorted.
10947 2017-06-16 Richard Biener <rguenther@suse.de>
10949 PR tree-optimization/81090
10950 * passes.def (pass_record_bounds): Remove.
10951 * tree-pass.h (make_pass_record_bounds): Likewise.
10952 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
10953 make_pass_record_bounds): Likewise.
10954 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
10955 not free niter estimates at the beginning but at the end.
10956 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
10958 2017-06-16 Richard Biener <rguenther@suse.de>
10960 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
10961 initializer to workaround ICE in host GCC 4.8.
10963 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10965 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
10967 (clone_inlined_nodes): Update.
10969 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10971 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
10972 prefetch settings, and enable prefetching by default at -O3.
10974 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10976 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10977 Set flag_prefetch_loop_arrays according to tuning data.
10979 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10981 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
10982 New tune structure.
10983 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
10984 [Unrelated to main purpose of the patch] Place the pointer field last
10985 to enable type checking errors when tune structure are wrongly merged.
10986 * config/aarch64/aarch64.c (generic_prefetch_tune,)
10987 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
10988 (thunderx2t99_prefetch_tune): New tune constants.
10989 (tune_params *_tunings): Update all tunings (no functional change).
10990 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
10991 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
10992 from tunings structures.
10994 2017-06-16 Jakub Jelinek <jakub@redhat.com>
10997 * ubsan.c (instrument_null): Add T argument, use it instead
10998 of computing it based on IS_LHS.
10999 (instrument_object_size): Likewise.
11000 (pass_ubsan::execute): Adjust instrument_null and
11001 instrument_object_size callers to pass gimple_get_lhs or
11002 gimple_assign_rhs1 result to it. Use instrument_null instead of
11003 calling get_base_address and instrument_mem_ref. Handle
11004 aggregate call arguments for object-size sanitization.
11006 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11008 PR tree-optimization/81089
11009 * tree-vrp.c (is_masked_range_test): Validate operands of
11012 2017-06-15 Martin Sebor <msebor@redhat.com>
11015 * dumpfile.c (dump_register): Avoid calling memset to initialize
11016 a class with a default ctor.
11017 * gcc.c (struct compiler): Remove const qualification.
11018 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11019 * hash-table.h: Ditto.
11020 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11022 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11023 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11025 * params.h (struct param_info): Make struct members non-const.
11026 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11027 with default initialization.
11028 * vec.h (vec_copy_construct, vec_default_construct): New helper
11030 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11031 with vec_copy_construct.
11032 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11033 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11034 * doc/invoke.texi (-Wclass-memaccess): Document.
11036 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11038 * emit-rtl.h (is_leaf): Update comment about local
11039 register allocator.
11041 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11044 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11045 for a variable to have a section before checking if the section has a
11047 Set section to.persistent if persistent attribute is set.
11048 Warn if .persistent attribute is used on an automatic variable.
11050 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11052 PR rtl-optimization/80474
11053 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11055 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11057 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11060 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11062 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11063 (casesi): Emit bounds checking as RTL.
11064 (casesi_internal_mips16_<mode>): Remove bounds checking.
11066 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11068 * config/xtensa/xtensa.c (xtensa_option_override): Append
11069 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11070 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11071 xtensa_doloop_hooks): Define unconditionally.
11072 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11074 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11075 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11076 for it in xtensa_option_override.
11077 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11078 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11080 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11082 * doc/cppopts.texi: Document '-' special value to -MF.
11084 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11086 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11087 (cortex_a53_fconst): Likewise.
11088 (cortex_a53_fpmul): Likewise.
11089 (cortex_a53_f_load_64): Likewise.
11090 (cortex_a53_f_load_many): Likewise.
11091 (cortex_a53_advsimd_alu): Likewise.
11092 (cortex_a53_advsimd_alu_q): Likewise.
11093 (cortex_a53_advsimd_mul): Likewise.
11094 (cortex_a53_advsimd_mul_q): Likewise.
11095 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11096 Add missing fmul, r2f_cvt and fconst cases.
11098 2017-06-14 Richard Biener <rguenther@suse.de>
11100 PR middle-end/81088
11101 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11103 (fold_binary_loc): When associating do not treat pre-existing
11104 TREE_OVERFLOW on literal constants as a reason to allow
11105 TREE_OVERFLOW on associated literal constants.
11107 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11109 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11110 (MASK_FEATURES): New macro.
11111 * config/sparc/sparc.c (sparc_option_override): Remove the special
11112 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11114 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11116 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11117 a division of 0 if non-call exceptions are enabled.
11119 2017-06-14 Andrew Pinski <apinski@cavium.com>
11120 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11123 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11124 Improve vector initialization code gen for only variable case.
11126 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11128 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11130 2017-06-14 Richard Biener <rguenther@suse.de>
11132 PR tree-optimization/81083
11133 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11136 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11138 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11139 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11140 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11141 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11142 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11143 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11145 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11147 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11148 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11150 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11152 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11154 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11156 * config/rs6000/t-rtems: Don't handle SPE.
11158 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11160 * config/rs6000/t-linux: Don't handle SPE.
11162 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11164 * config/rs6000/eabispe.h: Delete file.
11166 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11168 * config/rs6000/t-spe: Delete file.
11170 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11172 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11173 (rs6000_legitimate_offset_address_p): Return false for anything in
11174 V2SImode or V2SFmode.
11176 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11178 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11179 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11180 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11182 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11183 (rs6000_legitimate_offset_address_p): Ditto.
11184 (rs6000_emit_move): Ditto.
11185 (rs6000_init_builtins): Remove V4HI_type_node.
11187 2017-06-13 Martin Liska <mliska@suse.cz>
11190 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11191 (gate_asan): Likewise.
11192 * asan.h (asan_no_sanitize_address_p): Remove the function.
11193 (sanitize_flags_p): New function.
11194 * builtins.def: Fix coding style.
11195 * common.opt: Use renamed enum value.
11196 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11197 * doc/extend.texi: Document no_sanitize attribute.
11198 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11199 to SANITIZE_UNDEFINED_NONDEFAULT.
11200 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11201 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11202 Use sanitize_flags_p.
11203 * gimplify.c (gimplify_function_tree): Likewise.
11204 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11205 * opts.c (parse_no_sanitize_attribute): New function.
11206 (common_handle_option): Use renamed enum value.
11207 * opts.h (parse_no_sanitize_attribute): Declare.
11208 * tree.c (sanitize_flags_p): New function.
11209 * tree.h: Declared here.
11210 * tsan.c: Use sanitize_flags_p.
11211 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11212 (instrument_mem_ref): Likewise.
11213 (instrument_bool_enum_load): Likewise.
11214 (do_ubsan_in_current_function): Remove the function.
11215 (pass_ubsan::execute): Use sanitize_flags_p.
11216 * ubsan.h: Remove do_ubsan_in_current_function
11217 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11218 (dump_function_to_file): Use it here.
11220 2017-06-13 Martin Jambor <mjambor@suse.cz>
11222 PR tree-optimization/80803
11223 PR tree-optimization/81063
11224 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11225 (propagate_subaccesses_across_link): Enqueue subtree whenever
11226 necessary instead of relying on the caller.
11228 2017-06-13 Martin Jambor <mjambor@suse.cz>
11230 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11231 that have a first_link.
11232 (sort_and_splice_var_accesses): Do not check first_link before
11234 (subtree_mark_written_and_enqueue): Likewise.
11235 (propagate_all_subaccesses): Likewise and do not stop at first
11236 parent with a first_link.
11238 2017-06-13 Martin Jambor <mjambor@suse.cz>
11240 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11243 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11245 * match.pd: New pattern.
11247 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11249 * tree-vrp.c (is_masked_range_test): New function.
11250 (register_edge_assert_for): Determine ranges for
11253 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11255 PR tree-optimization/67328
11256 * fold-const.c (maskable_range_p): New function.
11257 (build_range_check): Generate bittests if possible.
11259 2017-06-13 Martin Liska <mliska@suse.cz>
11261 * gimple-pretty-print.c (dump_probability): Add new argument.
11262 (dump_edge_probability): Dump both probability and count.
11263 (dump_gimple_label): Likewise.
11264 (dump_gimple_bb_header): Likewise.
11266 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11269 * config/avr/avr-devices.c: Fix indentation.
11270 * config/avr/gen-avr-mmcu-specs.c: Dito.
11272 2017-06-13 Richard Biener <rguenther@suse.de>
11274 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11275 instead get vector type from stmt_info.
11276 (vectorizable_reduction): Adjust. Remove dead code.
11278 2017-06-13 Richard Biener <rguenther@suse.de>
11280 PR middle-end/81065
11281 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11282 case of C * (x * C2 + C3).
11283 (fold_addr_of_array_ref_difference): Properly fold index difference.
11285 2017-06-12 David S. Miller <davem@davemloft.net>
11288 * config/sparc/sparc.md (return expander): Emit frame blockage if
11289 function uses alloca.
11291 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11293 * combine.c (make_field_assignment): Check len rather than the mode
11294 precision when calling force_to_mode.
11296 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11298 Support multilibs and devices that see flash in RAM address range.
11301 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11302 (avr_mcu_t) <flash_pm_offset>: New field.
11303 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11304 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11305 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11306 (AVR_TINY_PM_OFFSET): Remove macro.
11307 * config/avr/avr.opt (-mshort-calls): New option.
11308 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11309 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11310 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11311 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11312 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11313 instead of avr_arch->have_jmp_call.
11314 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11315 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11316 avr_arch->flash_pm_offset to define.
11317 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11318 new field flash_pm_offset. Add entry for avrxmega3.
11319 (avr_texinfo): Add entry for avrxmega3.
11320 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11321 attiny212, attiny214,
11322 attiny412, attiny414, attiny416, attiny417,
11323 attiny814, attiny816, attiny817,
11324 attiny1614, attiny1616, attiny1617,
11325 attiny3214, attiny3216, attiny3217.
11326 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11327 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11328 (avr_print_operand_address) [AVR_TINY]: Same.
11329 (avr_asm_init_sections) <readonly_data_section>: Only patch
11330 callback if avr_arch->flash_pm_offset = 0.
11331 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11332 for rodata if avr_arch->flash_pm_offset != 0.
11333 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11334 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11335 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11336 (m_options): Append opt_rcall.
11337 (m_dirnames): Append dir_rcall.
11338 * config/avr/t-multilib: Regenerate.
11340 * configure.ac [target=avr]: Check whether avrxmega3 default
11341 linker description file works as needed.
11342 * configure: Regenerate.
11343 * doc/avr-mmcu.texi: Regenerate.
11344 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11345 <__AVR_ARCH__>: Document avrxmega3 and 103.
11346 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11347 <__AVR_SHORT_CALLS__>: Document it.
11348 <__AVR_PM_BASE_ADDRESS__>: Document it.
11349 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11350 (AVR Variable Attributes) <progmem>: Document this is
11351 not needed for avrxmega3.
11352 (AVR Named Address Spaces) <__flash>: Dito.
11354 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11356 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11358 2017-06-12 Doug Rupp <rupp@adacore.com>
11360 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11361 Append vxworks-stdint.h to the tm_file list.
11362 * config/vxworks-stdint.h: New file.
11364 2017-06-12 Martin Liska <mliska@suse.cz>
11366 PR tree-optimization/81041
11367 * tree-profile.c (gimple_gen_ic_func_profiler):
11368 Create an extra BB in profile-generate
11369 (gimple_gen_time_profiler): Likewise.
11371 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11373 PR tree-optimization/81003
11374 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11375 (update_range_test): Use it instead of force_gimple_operand_gsi.
11377 2017-06-12 Richard Biener <rguenther@suse.de>
11379 PR tree-optimization/81053
11380 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11381 with backedge value not defined in loop. Simplify def stmt
11384 2017-06-11 Tom de Vries <tom@codesourcery.com>
11387 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11389 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11390 nvptx_cannot_force_const_mem.
11392 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11394 * opts.c (finish_options): Move test for flag_split_stack after
11395 it has been initialized.
11397 2017-06-11 Jason Merrill <jason@redhat.com>
11399 * tree.h (id_equal): New.
11400 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11401 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11402 instead of strcmp of IDENTIFIER_POINTER.
11404 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11406 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11407 (mark_all_inlined_calls_cdtor): Fix formating.
11408 (inline_transform): Rescale profile before inlining.
11410 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11412 * cgraph.h (cgraph_edge::clone): Update prototype.
11413 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11414 (cgraph_node::create_clone): Update.
11415 (cgraph_node::create_version_clone): Update.
11416 * tree-inline.c (copy_bb): Update.
11417 (expand_call_inline): Update.
11419 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11421 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11422 factored out from ...
11423 (rs6000_emit_prologue): ... here.
11425 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11427 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11428 factored out from ...
11429 (rs6000_emit_prologue): ... here.
11431 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11433 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11435 (handle_missing_profiles): Fix computation of tp_first_run.
11436 (counts_to_freqs): Do not touch freqs when count is 0.
11438 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11440 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11443 2017-06-10 Tom de Vries <tom@codesourcery.com>
11445 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11446 attributes): Document signal effective target.
11448 2017-06-10 Tom de Vries <tom@codesourcery.com>
11450 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11451 Document effective target stack_size.
11453 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11455 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11456 to the edit_context if they can be auto-applied.
11458 2017-06-9 Ian Lance Taylor <iant@golang.org>
11460 * opts.c (finish_options): If -fsplit-stack, disable implicit
11461 -forder-blocks-and-partition.
11462 * doc/invoke.texi (Optimize Options): Document that when using
11463 -fsplit-stack -forder-blocks-and-partition is not implicitly
11466 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11468 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11469 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11470 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11471 * builtins.def (abort, trap, unreachable): Declare cold.
11472 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11473 * tree-core.h (ECF_COLD): New.
11474 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11475 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11477 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11479 * predict.c (unlikely_executed_stmt_p): Cleanup.
11481 2017-06-09 Richard Biener <rguenther@suse.de>
11483 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11484 model if the ref is always written to.
11486 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11488 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11490 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11492 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11495 2017-06-09 Tom de Vries <tom@codesourcery.com>
11498 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11499 "target cannot support label values" when encountering LABEL_REF.
11501 2017-06-09 Martin Liska <mliska@suse.cz>
11503 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11504 (gimple_gen_ic_func_profiler): Emit direct comparison
11505 of __gcov_indirect_call_callee with NULL.
11506 (gimple_gen_time_profiler): Change probability from
11507 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11509 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11511 * profile.c (edge_gcov_counts): Turn to pointer.
11512 (compute_branch_probabilities, compute_branch_probabilities): Update.
11513 (branch_prob): Do not clear edge_gcov_count.
11514 * profile.h (edge_gcov_counts): Turn to pointer.
11515 (edge_gcov_count): Update.
11517 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11519 * gimple.h (gimple_check_failed): Mark cold.
11521 2017-06-09 Richard Biener <rguenther@suse.de>
11523 PR tree-optimization/66623
11524 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11525 refactor check_reduction into two parts, properly computing
11526 whether we have to check reduction validity for outer loop
11529 2017-06-09 Richard Biener <rguenther@suse.de>
11531 PR tree-optimization/79483
11532 * graphite-scop-detection.c (order): New global.
11533 (get_order): Compute bb to order mapping that satisfies code
11534 generation constraints.
11535 (cmp_pbbs): New helper.
11536 (build_scops): Start domwalk at entry block, sort generated
11539 2017-06-09 Richard Biener <rguenther@suse.de>
11541 PR middle-end/81007
11542 * ipa-polymorphic-call.c
11543 (ipa_polymorphic_call_context::restrict_to_inner_class):
11544 Skip FIELD_DECLs with error_mark_node type.
11545 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11548 2017-06-09 Martin Liska <mliska@suse.cz>
11550 * predict.c (struct branch_predictor): New struct.
11551 (test_prediction_value_range): New test.
11552 (predict_c_tests): New function.
11553 * selftest-run-tests.c (selftest::run_tests): Run the function.
11554 * selftest.h: Declare new tests.
11556 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11559 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
11560 gen_add3_insn did not fail.
11561 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
11562 r0, construct that number in a temporary reg and add that reg to r0.
11563 If asked to put the result in r0 as well, fail.
11565 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
11567 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11568 for early expansion of vec_eqv.
11570 2017-06-08 Jakub Jelinek <jakub@redhat.com>
11572 PR middle-end/81005
11573 * ubsan.c (instrument_null): Avoid pointless code temporary.
11574 (pass_ubsan::execute): Instrument aggregate arguments of calls.
11576 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
11580 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
11583 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
11586 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11588 * predict.c (unlikely_executed_edge_p): Move ahead.
11589 (probably_never_executed_edge_p): Use it.
11591 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11593 PR middle-end/79988
11594 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
11595 gimple_call_builtin_p call.
11597 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11599 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
11600 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
11601 rtl_check_failed_type2, rtl_check_failed_code1,
11602 rtl_check_failed_code2, rtl_check_failed_code_mode,
11603 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
11604 rtvec_check_failed_bounds, rtl_check_failed_flag,
11605 _fatal_insn_not_found, _fatal_insn): Likewise.
11606 * tree.h (tree_contains_struct_check_failed,
11607 tree_check_failed, tree_not_check_failed,
11608 tree_class_check_failed, tree_range_check_failed,
11609 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
11610 tree_vec_elt_check_failed, phi_node_elt_check_failed,
11611 tree_operand_check_failed, omp_clause_check_failed,
11612 omp_clause_operand_check_failed, omp_clause_range_check_failed):
11615 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11617 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
11618 flag_branch_probabilities.
11619 * ipa-inline.c (edge_badness): Likewise.
11620 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11621 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
11622 * predict.c (maybe_hot_frequency_p): Likewise.
11623 (probably_never_executed): Likewise.
11624 * sched-ebb.c (schedule_ebbs): Likewise.
11625 * sched-rgn.c (find_single_block_region): Likewise.
11626 * tracer.c (tail_duplicate): Likewise.
11628 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11630 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
11631 longer requires x_flag_profile_use.
11633 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11635 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
11636 instead of flag_reorder_blocks_and_partition.
11637 * dbxout.c (dbxout_function_end): Likewise.
11638 * dwarf2out.c (gen_subprogram_die): Likewise.
11639 * haifa-sched.c (sched_create_recovery_edges): Likewise.
11640 * hw-doloop.c (reorg_loops): Likewise.
11641 * varasm.c (assemble_start_function,
11642 assemble_end_function): Likewise.
11643 (decide_function_section): Do not check for
11644 flag_reorder_blocks_and_partition.
11646 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11648 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11650 (chkp_get_hard_register_fake_addr_expr): Ditto.
11651 (chkp_build_addr_expr): Add check for hard reg case.
11652 (chkp_parse_array_and_component_ref): Ditto.
11653 (chkp_find_bounds_1): Ditto.
11654 (chkp_process_stmt): Don't generate bounds store for
11657 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11659 * predict.c (maybe_hot_bb_p): Do not check profile status.
11660 (maybe_hot_edge_p): Likewise.
11661 (probably_never_executed): Check for zero counts even if profile
11663 (unlikely_executed_edge_p): New function.
11664 (unlikely_executed_stmt_p): New function.
11665 (unlikely_executed_bb_p): New function.
11666 (set_even_probabilities): Use unlikely predicates.
11667 (combine_predictions_for_bb): Likewise.
11668 (predict_paths_for_bb): Likewise.
11669 (predict_paths_leading_to_edge): Likewise.
11670 (determine_unlikely_bbs): New function.
11671 (estimate_bb_frequencies): Use it.
11672 (compute_function_frequency): Use zero counts even if profile is
11674 * profile-count.h: Fix typo.
11676 2017-08-08 Julia Koval <julia.koval@intel.com>
11678 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
11679 _mm512_mask_cvtsepi16_storeu_epi8,
11680 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
11681 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
11682 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
11683 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
11684 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
11685 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
11686 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
11687 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
11688 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
11689 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
11690 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
11691 __builtin_ia32_pmovuswb256mem_mask,
11692 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
11693 __builtin_ia32_pmovwb512mem_mask): New builtins.
11695 2017-08-08 Julia Koval <julia.koval@intel.com>
11697 PR target/73350,80862
11698 * config/i386/subst.md (round): Fix round pattern.
11699 * config/i386/i386.c (ix86_erase_embedded_rounding):
11700 Fix erasing rounding for the fixed pattern.
11702 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11704 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
11706 2017-06-08 Martin Liska <mliska@suse.cz>
11708 PR gcov-profile/80911
11709 * gcov.c (block_info::block_info): New constructor.
11711 2017-06-07 Carl Love <cel@us.ibm.com>
11713 * config/rs6000/rs6000-c: The return type of the following
11714 built-in functions was implemented as int not long long. Fix sign
11715 of return value for the unsigned version of vec_mulo and vec_mule.
11716 vector unsigned long long vec_bperm (vector unsigned long long,
11717 vector unsigned char)
11718 vector signed long long vec_mule (vector signed int,
11720 vector unsigned long long vec_mule (vector unsigned int,
11721 vector unsigned int)
11722 vector signed long long vec_mulo (vector signed int,
11724 vector unsigned long long vec_mulo (vector unsigned int,
11725 vector unsigned int)
11726 * doc/extend.texi: Fix the documentation for the built-in
11729 2017-06-07 Carl Love <cel@us.ibm.com>
11732 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11735 2017-06-07 Carl Love <cel@us.ibm.com>
11737 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11738 support, Generate doublehv for signed int/float for BE case only.
11740 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11742 * doc/invoke.texi (mcx16): Rewrite.
11744 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11746 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11747 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11748 *mov<mode>_softfloat, and an anonymous splitter): Use
11749 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11751 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11753 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11755 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11756 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11757 (rs6000_debug_reg_global): Adjust.
11758 (rs6000_init_hard_regno_mode_ok): Adjust.
11759 (rs6000_dbx_register_number): Adjust.
11760 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11761 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11762 Remove SPE_ACC and SPEFSCR.
11763 (REG_ALLOC_ORDER): Ditto.
11764 (FRAME_POINTER_REGNUM): Change to 111.
11765 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
11766 (REG_CLASS_NAMES): Ditto.
11767 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
11768 (REGISTER_NAMES): Ditto.
11769 (ADDITIONAL_REG_NAMES): Ditto.
11770 (rs6000_reg_names): Ditto.
11771 * config/rs6000/rs6000.md: Renumber some register number
11774 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11776 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
11778 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11779 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
11781 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
11782 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11783 Delete the SPE high registers.
11784 (REG_ALLOC_ORDER): Ditto.
11785 (enum reg_class): Remove SPE_HIGH_REGS.
11786 (REG_CLASS_NAMES): Ditto.
11787 (REG_CLASS_CONTENTS): Delete the SPE high registers.
11788 (REGISTER_NAMES): Ditto.
11789 (rs6000_reg_names): Ditto.
11790 * doc/tm.texi.in: Remove SPE as example.
11791 * doc/tm.texi: Regenerate.
11793 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11795 * config/rs6000/8540.md (ppc8540_brinc): Delete.
11796 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
11797 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
11798 * config/rs6000/rs6000.md (type): Remove "brinc".
11800 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11802 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
11803 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
11804 * config/rs6000/linuxspe.h: Delete file.
11805 * config/rs6000/rs6000.md: Don't include spe.md.
11806 * config/rs6000/spe.h: Delete file.
11807 * config/rs6000/spe.md: Delete file.
11808 * config/rs6000/t-rs6000: Remove spe.md.
11810 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11812 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
11813 (reg_or_none500mem_operand): Delete.
11814 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
11815 instead of reg_or_none500mem_operand.
11817 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11819 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11820 handling of SPE flags.
11821 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
11823 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11825 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11827 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
11828 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
11829 paired_divv2sf3): Similar.
11830 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
11831 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
11832 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
11834 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
11835 Rename the paired_* instruction patterns.
11836 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
11838 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
11839 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
11840 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
11841 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
11842 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
11843 PAIRED_VECTOR_MODE.
11844 (struct machine_function): Delete field spe_insn_chain_scanned_p.
11845 (spe_func_has_64bit_regs_p): Delete.
11846 (spe_expand_predicate_builtin): Delete.
11847 (spe_expand_evsel_builtin): Delete.
11848 (TARGET_DWARF_REGISTER_SPAN): Do not define.
11849 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
11850 (invalid_e500_subreg): Delete.
11851 (rs6000_legitimize_address): Always force_reg op2 as well, for
11852 paired single memory accesses.
11853 (rs6000_member_type_forces_blk): Delete.
11854 (rs6000_spe_function_arg): Delete.
11855 (rs6000_expand_unop_builtin): Delete SPE handling.
11856 (rs6000_expand_binop_builtin): Ditto.
11857 (spe_expand_stv_builtin): Delete.
11858 (bdesc_2arg_spe): Delete.
11859 (spe_expand_builtin): Delete.
11860 (spe_expand_predicate_builtin): Delete.
11861 (spe_expand_evsel_builtin): Delete.
11862 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
11863 (spe_init_builtins): Delete.
11864 (spe_func_has_64bit_regs_p): Delete.
11865 (savres_routine_name): Delete "info" parameter. Adjust callers.
11866 (rs6000_emit_stack_reset): Ditto.
11867 (rs6000_dwarf_register_span): Delete.
11868 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
11869 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
11870 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
11872 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
11874 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
11875 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
11876 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
11877 mulv2sf3, divv2sf3): Delete expanders.
11879 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11881 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
11883 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11885 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
11886 * config/rs6000/rs6000.c: Ditto.
11888 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11890 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
11891 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
11892 comparison_operator.
11894 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11896 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
11897 * config/rs6000/rs6000.opt: Ditto.
11898 * config/rs6000/t-rtems: Ditto.
11900 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11902 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
11903 TARGET_E500_SINGLE by 0, simplify.
11904 * config/rs6000/rs6000.c: Ditto.
11905 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
11906 (spe_build_register_parallel): Delete.
11907 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
11908 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
11909 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
11910 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
11911 (E500_CONVERT): Delete.
11912 * config/rs6000/spe.md: Remove many patterns and all define_constants.
11914 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11916 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
11917 * config/rs6000/dfp.md: Ditto.
11918 (negdd2, *negdd2_fpr): Merge.
11919 (absdd2, *absdd2_fpr): Merge.
11920 (negtd2, *negtd2_fpr): Merge.
11921 (abstd2, *abstd2_fpr): Merge.
11922 * config/rs6000/e500.h: Delete file.
11923 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
11924 TARGET_FPRS by 1 and simplify.
11925 * config/rs6000/rs6000-c.c: Ditto.
11926 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
11927 TARGET_DF_SPE by 0.
11928 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
11930 * config/rs6000/rs6000.md: Ditto.
11931 (floatdidf2, *floatdidf2_fpr): Merge.
11932 (move_from_CR_gt_bit): Delete.
11933 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
11934 (E500_CR_IOR_COMPARE): Delete.
11935 (All patterns that require !TARGET_FPRS): Delete.
11936 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
11938 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11940 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
11942 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11944 * graphds.c (add_edge): Intitialize edge's attached data.
11945 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
11946 pointer parameter. Call pointed function on each edge during
11947 graph traversing. Skip traversing the edge when the function
11949 (graphds_dfs, graphds_scc): Ditto.
11950 (for_each_edge): New parameter. Pass the new parameter to callback
11952 * graphds.h (skip_edge_callback): New function pointer type.
11953 (graphds_dfs, graphds_scc): New function pointer parameter.
11954 (graphds_edge_callback, for_each_edge): New parameter.
11956 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11958 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
11959 out code checking if runtime alias check is possible to below ...
11960 Call the new function.
11961 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
11962 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
11964 2017-06-07 Marek Polacek <polacek@redhat.com>
11967 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
11968 TYPE_OVERFLOW_WRAPS check.
11970 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11972 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
11973 if versioning is required.
11974 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
11975 peeling with the check for versioning.
11977 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11979 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
11980 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
11981 Set true to new parameter if new ssa variable is defined.
11982 (vect_gen_vector_loop_niters): Refactor. Set range information
11983 for the new vector loop bound variable.
11984 (vect_do_peeling): Ditto.
11986 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11988 * tree-affine.c (ssa.h): Include header file.
11989 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
11990 has wrapping overflow behavior.
11992 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11994 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
11996 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11998 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
11999 (tree_to_aff_combination): ... here.
12001 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12003 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12004 reg_pressure model function.
12005 (ivopts_global_cost_for_size): Delete.
12006 (determine_set_costs, iv_ca_recount_cost): Call new model function
12007 ivopts_estimate_reg_pressure.
12009 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12011 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12012 expensive than udiv. Remove floating point cases from mod.
12014 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12016 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12017 Increase idiv cost.
12019 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12021 * config/aarch64/aarch64.md
12022 (copysignsf3): Fix mask generation.
12024 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12026 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12028 (class dump_manager): Add register_dumps method.
12029 * dumpfile.c: Include langhooks.h.
12030 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12031 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12032 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12033 (dump_manager::dump_register): Start with 512 entries instead of 32.
12034 (dump_manager::register_dumps): New method.
12035 * toplev.c (general_init): Instead of invoking register_dumps
12036 langhook, invoke register_dumps method on the dump manager.
12037 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12040 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12042 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12043 Say that # requires an associated define_split to exist, and that
12044 the define_split must be suitable for use after register allocation.
12046 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12048 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12049 (compute_outgoing_frequencies): Also initialize zero counts.
12050 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12051 around loops; preserve more of profile when nothing changes.
12053 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12055 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12057 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12058 * config/arm/arm-cpu-cdata.h: Regenerate.
12059 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12060 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12061 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12062 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12063 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12065 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12067 * config/arm/t-rmprofile: Likewise.
12068 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12070 2017-06-06 David S. Miller <davem@davemloft.net>
12073 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12074 blockage if function uses alloca.
12076 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12078 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12079 New "uid" fields to hold pretty-print IDs of group and ref.
12080 Memory references are now identified as <group_id>:<ref_id>
12081 instead of using [random] addresses.
12082 (dump_mem_details): Simplify, no functional change.
12083 (dump_mem_ref): Simplify and make output more concise.
12084 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12085 (find_or_create_group): Initialize group uid.
12086 (record_ref): Initialize ref uid. Improve debug output.
12087 (prune_group_by_reuse, should_issue_prefetch_p,)
12088 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12089 (mark_nontemporal_store, determine_loop_nest_reuse):
12090 Improve debug output.
12092 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12094 * dbgcnt.def (prefetch): New debug counter.
12095 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12096 (schedule_prefetches): Stop issueing prefetches if debug counter
12099 2017-06-06 Tom de Vries <tom@codesourcery.com>
12101 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12102 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12104 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12106 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12107 Use aarch64_reg_or_zero predicate for operand 4.
12108 (aarch64_compare_and_swap<mode> define_insn_and_split):
12109 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12110 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12112 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12114 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12115 (arm_compute_save_core_reg_mask): This.
12116 (thumb1_compute_save_reg_mask): Rename into ...
12117 (thumb1_compute_save_core_reg_mask): This.
12118 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12119 (arm_compute_frame_layout): Likewise.
12121 2017-06-06 Richard Biener <rguenther@suse.de>
12123 PR tree-optimization/80974
12124 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12125 keep or clear leaders SSA info.
12127 2017-06-06 Tom de Vries <tom@codesourcery.com>
12129 * config/nvptx/nvptx.c (split_mode_p): New function.
12130 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12132 2017-06-06 Tom de Vries <tom@codesourcery.com>
12134 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12136 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12139 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12142 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12144 * shrink-wrap.c (handle_simple_exit): Update profile.
12145 (try_shrink_wrapping): Upate profile.
12147 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12149 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12150 (tree_guess_outgoing_edge_probabilities): New.
12151 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12152 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12154 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12156 * ipa-split.c (split_function): Initialize return bb profile.
12158 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12160 * profile.c (compute_branch_probabilities): Also initialize
12161 EXIT_BLOCK profile.
12163 2017-06-06 Richard Biener <rguenther@suse.de>
12165 PR tree-optimization/80928
12166 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12167 (vect_analyze_loop_operations): Properly guard analysis for
12169 (vect_transform_loop): Likewise.
12170 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12171 (vect_model_induction_cost): Do not cost for pure SLP.
12172 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12173 of induction in inner loop vectorization.
12174 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12175 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12176 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12178 (vect_analyze_slp_cost_1): Cost induction.
12179 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12180 (vect_get_slp_vect_defs): Likewise.
12181 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12182 (vect_transform_stmt): Handle SLP reductions.
12183 * tree-vectorizer.h (vectorizable_induction): Adjust.
12185 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12187 * config/rs6000/rs6000.c (make_resolver_func): Update
12188 init_lowered_empty_function call.
12190 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12192 * doc/invoke.texi: Document the -fprofile-abs-path option.
12193 * common.opt (fprofile-abs-path): New option.
12194 * gcov-io.h (gcov_write_filename): Declare.
12195 * gcov-io.c (gcov_write_filename): New function.
12196 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12197 * profile.c (output_location): Likewise.
12199 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12201 * shring-wrap.c: Revert accidental commit.
12203 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12205 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12207 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12209 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12211 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12212 profile in callgraph edge.
12213 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12214 (apply_scale): Likewise.
12215 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12216 Also scale profile when inlining function with zero profile.
12217 (initialize_cfun): Update exit block profile even when it is zero.
12218 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12219 when profile is read.
12221 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12223 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12224 (CLONE_*): New constants to define the processors we can generate
12225 code for with the target_clone attribute.
12226 (rs6000_clone_map): New array to identify which clone processors
12227 the current program is running on.
12228 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12229 target_clone attribute.
12230 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12231 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12232 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12233 (cpu_expand_builtin): Add support for target_clone attribute.
12234 (rs6000_valid_attribute_p): Allow "default" attribute.
12235 (get_decl_name): New debug function to simplify printing the
12236 current function name in debugging statements.
12237 (rs6000_clone_priority): New functions to support the target_clone
12238 attribute, and be able to generate code to switch between ISA 2.05
12239 through ISA 3.0 (power6 through power9).
12240 (rs6000_compare_version_priority): Likewise.
12241 (rs6000_get_function_versions_dispatcher): Likewise.
12242 (make_resolver_func): Likewise.
12243 (add_condition_to_bb): Likewise.
12244 (dispatch_function_versions): Likewise.
12245 (rs6000_generate_version_dispatcher_body): Likewise.
12246 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12247 (fusion_gpr_load_p): Fix a spacing issue.
12248 * doc/extend.texi (Common Function Attributes): Document that the
12249 PowerPC supports the target_clone attribute.
12251 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12253 * config/arm/arm.h: explain F symbol found in description of ARM
12254 register allocation in its legend.
12256 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12258 * config/mips/frame-header-opt.c: Include profile-count.h.
12259 * config/riscv/riscv.c: Include profile-count.h
12261 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12263 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12265 (sm_set_flag_if_changed): Add bbs field.
12266 (execute_sm_if_changed_flag_set): Pass BBS.
12267 (execute_sm): Update.
12269 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12271 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12274 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12276 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12277 (peephole2): New peephole2 to emit the above.
12278 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12280 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12282 * config/aarch64/aarch64.c (define_peephole2 above
12283 *sub_<shift>_<mode>): New peephole.
12285 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12287 * config/i386/i386.c (make_resolver_func): Update.
12288 * Makefile.in: Add profile-count.h and profile-count.o
12289 * auto-profile.c (afdo_indirect_call): Update to new API.
12290 (afdo_set_bb_count): Update.
12291 (afdo_propagate_edge): Update.
12292 (afdo_propagate_circuit): Update.
12293 (afdo_calculate_branch_prob): Update.
12294 (afdo_annotate_cfg): Update.
12295 * basic-block.h: Include profile-count.h
12296 (struct edge_def): Turn count to profile_count.
12297 (struct basic_block_def): Likewie.
12298 (REG_BR_PROB_BASE): Move to profile-count.h
12299 (RDIV): Move to profile-count.h
12300 * bb-reorder.c (max_entry_count): Turn to profile_count.
12301 (find_traces): Update.
12302 (rotate_loop):Update.
12303 (connect_traces):Update.
12304 (sanitize_hot_paths):Update.
12305 * bt-load.c (migrate_btr_defs): Update.
12306 * cfg.c (RDIV): Remove.
12307 (init_flow): Use alloc_block.
12308 (alloc_block): Uninitialize count.
12309 (unchecked_make_edge): Uninitialize count.
12310 (check_bb_profile): Update.
12311 (dump_edge_info): Update.
12312 (dump_bb_info): Update.
12313 (update_bb_profile_for_threading): Update.
12314 (scale_bbs_frequencies_int): Update.
12315 (scale_bbs_frequencies_gcov_type): Update.
12316 (scale_bbs_frequencies_profile_count): New.
12317 * cfg.h (update_bb_profile_for_threading): Update.
12318 (scale_bbs_frequencies_profile_count): Declare.
12319 * cfgbuild.c (compute_outgoing_frequencies): Update.
12320 (find_many_sub_basic_blocks): Update.
12321 * cfgcleanup.c (try_forward_edges): Update.
12322 (try_crossjump_to_edge): Update.
12323 * cfgexpand.c (expand_gimple_tailcall): Update.
12324 (construct_exit_block): Update.
12325 * cfghooks.c (verify_flow_info): Update.
12326 (dump_bb_for_graph): Update.
12327 (split_edge): Update.
12328 (make_forwarder_block): Update.
12329 (duplicate_block): Update.
12330 (account_profile_record): Update.
12331 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12332 (get_estimated_loop_iterations): Update.
12333 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12334 (single_likely_exit): Update.
12335 * cfgloopmanip.c (scale_loop_profile): Update.
12337 (set_zero_probability): Update.
12338 (lv_adjust_loop_entry_edge): Update.
12339 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12340 (purge_dead_edges): Update.
12341 (rtl_account_profile_record): Update.
12342 * cgraph.c (cgraph_node::create): Uninitialize count.
12343 (symbol_table::create_edge): Uninitialize count.
12344 (cgraph_update_edges_for_call_stmt_node): Update.
12345 (cgraph_edge::dump_edge_flags): Update.
12346 (cgraph_node::dump): Update.
12347 (cgraph_edge::maybe_hot_p): Update.
12348 * cgraph.h: Include profile-count.h
12349 (create_clone), create_edge, create_indirect_edge): Update.
12350 (cgraph_node): Turn count to profile_count.
12351 (cgraph_edge0: Likewise.
12352 (make_speculative, clone): Update.
12353 (create_edge): Update.
12354 (init_lowered_empty_function): Update.
12355 * cgraphclones.c (cgraph_edge::clone): Update.
12356 (duplicate_thunk_for_node): Update.
12357 (cgraph_node::create_clone): Update.
12358 * cgraphunit.c (cgraph_node::analyze): Update.
12359 (cgraph_node::expand_thunk): Update.
12360 * final.c (dump_basic_block_info): Update.
12361 * gimple-streamer-in.c (input_bb): Update.
12362 * gimple-streamer-out.c (output_bb): Update.
12363 * graphite.c (print_global_statistics): Update.
12364 (print_graphite_scop_statistics): Update.
12365 * hsa-brig.c: Include basic-block.h.
12366 * hsa-dump.c: Include basic-block.h.
12367 * hsa-gen.c (T sum_slice): Update.
12368 (convert_switch_statements):Update.
12369 * hsa-regalloc.c: Include basic-block.h.
12370 * ipa-chkp.c (chkp_produce_thunks): Update.
12371 * ipa-cp.c (struct caller_statistics): Update.
12372 (init_caller_stats): Update.
12373 (gather_caller_stats): Update.
12374 (ipcp_cloning_candidate_p): Update.
12375 (good_cloning_opportunity_p): Update.
12376 (get_info_about_necessary_edges): Update.
12377 (dump_profile_updates): Update.
12378 (update_profiling_info): Update.
12379 (update_specialized_profile): Update.
12380 (perhaps_add_new_callers): Update.
12381 (decide_about_value): Update.
12382 (ipa_cp_c_finalize): Update.
12383 * ipa-devirt.c (struct odr_type_warn_count): Update.
12384 (struct decl_warn_count): Update.
12385 (struct final_warning_record): Update.
12386 (possible_polymorphic_call_targets): Update.
12387 (ipa_devirt): Update.
12388 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12389 * ipa-icf.c (sem_function::merge): Update.
12390 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12391 * ipa-inline.c (compute_uninlined_call_time): Update.
12392 (compute_inlined_call_time): Update.
12393 (want_inline_small_function_p): Update.
12394 (want_inline_self_recursive_call_p): Update.
12395 (edge_badness): Update.
12396 (lookup_recursive_calls): Update.
12397 (recursive_inlining): Update.
12398 (inline_small_functions): Update.
12399 (dump_overall_stats): Update.
12400 (dump_inline_stats): Update.
12401 * ipa-profile.c (ipa_profile_generate_summary): Update.
12402 (ipa_propagate_frequency): Update.
12403 (ipa_profile): Update.
12404 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12405 * ipa-utils.c (ipa_merge_profiles): Update.
12406 * loop-doloop.c (doloop_modify): Update.
12407 * loop-unroll.c (report_unroll): Update.
12408 (unroll_loop_runtime_iterations): Update.
12409 * lto-cgraph.c (lto_output_edge): Update.
12410 (lto_output_node): Update.
12411 (input_node): Update.
12412 (input_edge): Update.
12413 (merge_profile_summaries): Update.
12414 * lto-streamer-in.c (input_cfg): Update.
12415 * lto-streamer-out.c (output_cfg): Update.
12416 * mcf.c (create_fixup_graph): Update.
12417 (adjust_cfg_counts): Update.
12418 (sum_edge_counts): Update.
12419 * modulo-sched.c (sms_schedule): Update.
12420 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12421 * predict.c (maybe_hot_count_p): Update.
12422 (probably_never_executed): Update.
12423 (dump_prediction): Update.
12424 (combine_predictions_for_bb): Update.
12425 (propagate_freq): Update.
12426 (handle_missing_profiles): Update.
12427 (counts_to_freqs): Update.
12428 (rebuild_frequencies): Update.
12429 (force_edge_cold): Update.
12430 * predict.h: Include profile-count.h
12431 (maybe_hot_count_p, counts_to_freqs): UPdate.
12432 * print-rtl-function.c: Do not include cfg.h
12433 * print-rtl.c: Include basic-block.h
12434 * profile-count.c: New file.
12435 * profile-count.h: New file.
12436 * profile.c (is_edge_inconsistent): Update.
12437 (correct_negative_edge_counts): Update.
12438 (is_inconsistent): Update.
12439 (set_bb_counts): Update.
12440 (read_profile_edge_counts): Update.
12441 (compute_frequency_overlap): Update.
12442 (compute_branch_probabilities): Update; Initialize and deinitialize
12444 (branch_prob): Update.
12445 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12446 (edge_gcov_count): New.
12447 (bb_gcov_count): New.
12448 * shrink-wrap.c (try_shrink_wrapping): Update.
12449 * tracer.c (better_p): Update.
12450 * trans-mem.c (expand_transaction): Update.
12451 (ipa_tm_insert_irr_call): Update.
12452 (ipa_tm_insert_gettmclone_call): Update.
12453 * tree-call-cdce.c: Update.
12454 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12455 (gimple_duplicate_sese_tail): Update.
12456 (gimple_account_profile_record): Update.
12457 (execute_fixup_cfg): Update.
12458 * tree-inline.c (copy_bb): Update.
12459 (copy_edges_for_bb): Update.
12460 (initialize_cfun): Update.
12461 (freqs_to_counts): Update.
12462 (copy_cfg_body): Update.
12463 (expand_call_inline): Update.
12464 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12465 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12466 (try_unroll_loop_completely): Update.
12467 (try_peel_loop): Update.
12468 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12469 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12470 * tree-ssa-loop-split.c (connect_loops): Update.
12471 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12472 * tree-ssa-reassoc.c (branch_fixup): Update.
12473 * tree-ssa-tail-merge.c (replace_block_by): Update.
12474 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12475 (compute_path_counts): Update.
12476 (update_profile): Update.
12477 (recompute_probabilities): Update.
12478 (update_joiner_offpath_counts): Update.
12479 (estimated_freqs_path): Update.
12480 (freqs_to_counts_path): Update.
12481 (clear_counts_path): Update.
12482 (ssa_fix_duplicate_block_edges): Update.
12483 (duplicate_thread_path): Update.
12484 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12485 (struct switch_conv_info): Update.
12486 * tree-tailcall.c (decrease_profile): Update.
12487 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12488 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12489 * value-prof.c (check_counter): Update.
12490 (gimple_divmod_fixed_value): Update.
12491 (gimple_mod_pow2): Update.
12492 (gimple_mod_subtract): Update.
12493 (gimple_ic_transform): Update.
12494 (gimple_stringop_fixed_value): Update.
12495 * value-prof.h (gimple_ic): Update.
12497 2017-06-02 Carl Love <cel@us.ibm.com>
12499 * config/rs6000/rs6000-c: Add support for built-in functions
12500 vector double vec_doublee (vector signed int);
12501 vector double vec_doublee (vector unsigned int);
12502 vector double vec_doublee (vector float);
12503 vector double vec_doubleh (vector signed int);
12504 vector double vec_doubleh (vector unsigned int);
12505 vector double vec_doubleh (vector float);
12506 vector double vec_doublel (vector signed int);
12507 vector double vec_doublel (vector unsigned int);
12508 vector double vec_doublel (vector float);
12509 vector double vec_doubleo (vector signed int);
12510 vector double vec_doubleo (vector unsigned int);
12511 vector double vec_doubleo (vector float);.
12512 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12513 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12515 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12516 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12517 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12519 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12520 vec_doublel, vec_doubleh.
12521 * doc/extend.texi: Update the built-in documentation file for the
12522 new built-in functions.
12524 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12527 * ipa-inline-analysis.c (free_growth_caches): Set
12528 edge_removal_hook_holder to NULL after removing it.
12530 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12532 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12533 comparision with zero.
12535 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12536 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12537 for early expansion of vec_min and vec_max builtins.
12538 (builtin_function_type): Add min/max unsigned variants to those
12539 identified as having unsigned arguments.
12541 2017-06-02 Olivier Hainque <hainque@adacore.com>
12543 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12545 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12547 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12548 Use VALL_F16 iterator rather than VALL.
12550 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12552 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12553 Emit CBNZ inside loop when doing a strong exchange and comparing
12554 against zero. Generate the CC flags after the loop.
12556 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12558 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12559 (dl_section_ref): New.
12560 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
12561 On AIX, append an expression to subtract the size of the
12562 section length to dl_section_ref.
12564 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12566 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12567 for early expansion of vector absolute builtins.
12569 2017-06-02 Richard Biener <rguenther@suse.de>
12571 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
12572 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
12574 2017-06-02 Richard Biener <rguenther@suse.de>
12576 PR tree-optimization/80948
12577 * tree-tailcall.c (find_tail_calls): Track stmts to move in
12578 stmt order as well.
12580 2017-06-02 Richard Biener <rguenther@suse.de>
12582 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
12584 * tree-vect-stmts.c (process_use): Do not mark backedge defs
12585 for inductions as relevant.
12587 2017-06-02 Richard Biener <rguenther@suse.de>
12589 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
12590 (vectorizable_induction): ... this. Remove dead code.
12592 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
12594 * builtins. (expand_builtin_alloca): Remove second parameter and
12595 infer its value from the first parameter instead.
12596 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
12598 2017-06-02 Jakub Jelinek <jakub@redhat.com>
12600 PR rtl-optimization/80903
12601 * loop-doloop.c (add_test): Unshare sequence.
12603 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
12605 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
12607 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
12609 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
12611 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
12612 xlogue_layout::get_instance, logue_layout::xlogue_layout,
12613 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
12614 (xlogue_layout::get_stub_rtx): Make static.
12615 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
12616 (xlogue_layout::compute_stub_managed_regs): Rename to...
12617 (xlogue_layout::count_stub_managed_regs): ...this.
12618 (xlogue_layout::is_stub_managed_reg): New function.
12619 (xlogue_layout::m_stub_names): Rename to...
12620 (xlogue_layout::s_stub_names): ...this, make static.
12621 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
12622 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
12623 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
12624 xlogue_layout::s_stub_names): Instantiate statics.
12625 (stub_managed_regs): Remove.
12626 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
12627 (disable_call_ms2sysv_xlogues): Rename to...
12628 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
12629 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
12631 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
12632 change after reload_completed.
12633 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
12635 (ix86_expand_prologue): Likewise.
12636 (ix86_expand_epilogue): Likewise.
12637 (ix86_expand_split_stack_prologue): Likewise.
12638 (ix86_compute_frame_layout): Remove frame parameter ...
12639 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
12640 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
12642 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
12643 (ix86_frame): Move from here ...
12644 * config/i386/i386.h (ix86_frame): ... to here.
12645 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
12646 complete ix86_frame data structure instead. Remove some_ld_name.
12648 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12650 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
12651 symbols that hold a DECL_VALUE_EXPR.
12653 2017-06-01 Martin Jambor <mjambor@suse.cz>
12655 PR tree-optimization/80898
12656 * tree-sra.c (process_subtree_disqualification): Removed.
12657 (disqualify_candidate): Do not acll
12658 process_subtree_disqualification.
12659 (subtree_mark_written_and_enqueue): New function.
12660 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
12661 RHS has been disqualified and re-queue LHS if necessary. Apart
12662 from that, ignore disqualified RHS.
12664 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12666 * config/s390/s390.c (s390_emit_epilogue): Disable early return
12667 address fetch for z10 or later.
12669 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12671 * config/arc/arc.md (tst_movb): Add guard when splitting.
12673 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12675 * config/arc/arc.c (arc_can_eliminate): Test against
12676 arc_frame_pointer_needed.
12678 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12680 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
12681 to prevent store reordering.
12682 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
12683 (type): Add block type.
12684 (stack_tie): Define special instruction to be used in
12687 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12689 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
12690 constraint. It is not valid for the pattern.
12691 (noncommutative_binary_comparison): Likewise.
12693 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12695 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
12698 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12700 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
12701 be used by the reg-alloc.
12703 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12705 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
12706 reg-alloc when having mul64 or mul32x16 instructions.
12707 (mulsidi3): Likewise.
12708 (umulsidi3): Likewise.
12709 (mulsi32x16): New pattern.
12710 (mulsi64): Likewise.
12711 (mulsidi64): Likewise.
12712 (umulsidi64): Likewise.
12713 (MUL32x16_REG): Define.
12714 (mul64_600): Use MUL32x16_REG.
12715 (mac64_600): Likewise.
12716 (umul64_600): Likewise.
12717 (umac64_600): Likewise.
12719 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12721 * config/arc/arc.md (mulsi3_700): Make it commutative.
12723 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
12725 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
12727 (*sign_extendsidi2_insn): Likewise for movstosw.
12729 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12731 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12732 the type of the input discriminant value. Convert the
12733 discriminant value of signedness vary.
12735 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12737 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12738 Add to -Wall section.
12740 2017-06-01 Richard Biener <rguenther@suse.de>
12742 PR middle-end/66313
12743 * fold-const.c (fold_plusminus_mult_expr): If the factored
12744 factor may be zero use a wrapping type for the inner operation.
12745 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12746 and handle moved defs.
12747 (process_assignment): Properly guard the unary op case. Return a
12748 tri-state indicating that moving the stmt before the call may allow
12749 to continue. Pass through to_move.
12750 (find_tail_calls): Handle moving unrelated defs before
12753 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12756 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12757 splitter result in the canonical way.
12759 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12761 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
12762 also for 32bit target. Update insn attributes.
12763 (zero-extendsidi2 splitter): Allow all registers for operand 1.
12765 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
12767 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
12768 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
12769 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
12770 (_mm_maskz_min_ss): New intrinsics.
12772 2017-05-31 Martin Liska <mliska@suse.cz>
12774 * tree-vect-loop.c (vect_create_epilog_for_reduction):
12775 Change comment style to one we normally use.
12776 (vectorizable_reduction): Likewise.
12777 (vectorizable_induction): Likewise.
12778 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12779 (vectorizable_call): Likewise.
12780 (vectorizable_simd_clone_call): Likewise.
12781 (vectorizable_conversion): Likewise.
12782 (vectorizable_assignment): Likewise.
12783 (vectorizable_shift): Likewise.
12784 (vectorizable_operation): Likewise.
12785 (vectorizable_store): Likewise.
12786 (vectorizable_load): Likewise.
12787 * tree-vectorizer.h: Likewise.
12789 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
12791 * passes.c (emergency_dump_function): New.
12792 * tree-pass.h (emergency_dump_function): Declare.
12793 * plugin.c (plugins_internal_error_function): Remove.
12794 * plugin.h (plugins_internal_error_function): Remove declaration.
12795 * toplev.c (internal_error_function): New static function. Use it...
12796 (general_init): ...here.
12798 2017-05-31 Graham Markall <graham.markall@embecosm.com>
12800 * config/arc/arc.c (arc_print_operand): Handle constant operands.
12801 (arc_rtx_costs): Add costs for new patterns.
12802 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
12803 * config/arc/predicates.md: Add _1_2_3_operand predicate.
12805 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12807 * tree-ssa-strlen.c (get_next_strinfo): New function.
12808 (get_stridx_plus_constant): Use it.
12809 (zero_length_string): Likewise.
12810 (adjust_related_strinfos): Likewise.
12811 (adjust_last_stmt): Likewise.
12813 2017-05-31 Richard Biener <rguenther@suse.de>
12816 * config/i386/i386.c (ix86_expand_builtin): Remove assert
12817 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
12819 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12821 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
12822 loop_vinfo argument and use of dependence distance vectors.
12823 Check instead whether the two references differ only in their
12824 initial value and assume that they have the same alignment if the
12825 difference is a multiple of the vector alignment.
12826 (vect_analyze_data_refs_alignment): Update call accordingly.
12828 2017-05-31 Martin Liska <mliska@suse.cz>
12831 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
12833 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12835 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
12836 (create_intersect_range_checks): Move from ...
12837 * tree-data-ref.c (create_intersect_range_checks_index)
12838 (create_intersect_range_checks): ... to here.
12839 (create_runtime_alias_checks): New function factored from ...
12840 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
12841 here. Call above function.
12842 * tree-data-ref.h (create_runtime_alias_checks): New function.
12844 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12846 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
12847 segment length for dr_b and compute it in wide_int.
12849 2017-05-31 Richard Biener <rguenther@suse.de>
12851 PR tree-optimization/80906
12852 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
12853 and pass through iv_map.
12854 (copy_bb_and_scalar_dependences): Adjust.
12855 (translate_pending_phi_nodes): Likewise.
12856 (copy_loop_close_phi_args): Handle code-generating IVs instead
12859 2017-05-30 David Malcolm <dmalcolm@redhat.com>
12861 * diagnostic-color.c (color_dict): Add "type-diff".
12862 (parse_gcc_colors): Update comment.
12863 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
12864 -fdiagnostics-show-template-tree and -fno-elide-type.
12865 (GCC_COLORS): Add type-diff to example.
12867 (-fdiagnostics-show-template-tree): New.
12868 (-fno-elide-type): New.
12869 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
12870 the pp_format_decoder callback. Call any m_format_postprocessor's
12872 (pretty_printer::pretty_printer): Initialize
12873 m_format_postprocessor.
12874 (pretty_printer::~pretty_printer): Delete any
12875 m_format_postprocessor.
12876 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
12877 (class format_postprocessor): New class.
12878 (struct pretty_printer::format_decoder): Document the new parameters.
12879 (struct pretty_printer::m_format_postprocessor): New field.
12880 * tree-diagnostic.c (default_tree_printer): Update for new
12881 bool and const char ** params.
12882 * tree-diagnostic.h (default_tree_printer): Likewise.
12884 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
12886 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
12887 (lwa_operand): Delete rs6000_gen_cell_microcode test.
12888 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12889 rs6000_gen_cell_microcode code.
12890 (rs6000_final_prescan_insn): Delete.
12891 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
12892 "warn-cell-microcode" entries.
12893 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
12894 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
12895 throughout. Change cc_reg_not_micro_cr0_operand to
12896 cc_reg_not_cr0_operand throughout.
12897 (*extendhi<mode>2_noload): Delete.
12898 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
12899 (mwarn-cell-microcode): Delete.
12900 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
12901 -mgen-cell-microcode and -mwarn-cell-microcode.
12903 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
12906 * config/i386/constraints.md (Yd): New constraint.
12908 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
12909 and (?Yd, r) alternatives. Update insn attributes.
12910 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
12911 and (?*Yd, r) alternatives. Update insn attributes.
12912 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
12914 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
12916 * gimplify.c (gimplify_modify_expr): Don't create a
12917 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
12920 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
12922 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
12924 2017-05-30 Richard Biener <rguenther@suse.de>
12926 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
12927 and reduc_def fields.
12928 (STMT_VINFO_REDUC_TYPE): New define.
12929 (STMT_VINFO_REDUC_DEF): Likewise.
12930 (vect_force_simple_reduction): Adjust prototype.
12931 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
12932 (vect_is_simple_reduction): Remove check_reduction argument.
12933 (vect_force_simple_reduction): Adjust and set
12934 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12935 (vectorizable_reduction): Do not re-do reduction analysis
12936 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12937 * tree-parloops.c (gather_scalar_reductions): Adjust.
12939 2017-05-30 Richard Biener <rguenther@suse.de>
12941 PR middle-end/80901
12942 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
12945 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
12947 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
12948 Introduce unknown_misalignment parameter and remove vf.
12949 (vect_peeling_hash_get_lowest_cost):
12950 Pass unknown_misalignment parameter.
12951 (vect_enhance_data_refs_alignment):
12952 Fix unsupportable data ref treatment.
12954 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12956 * tree-vect-data-refs.c (vect_get_data_access_cost):
12957 Workaround for SLP handling.
12958 (vect_enhance_data_refs_alignment):
12959 Compute costs for doing no peeling at all, compare to the best
12960 peeling costs so far and avoid peeling if cheaper.
12962 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12964 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
12965 Return peeling info and set costs to zero for unlimited cost
12967 (vect_enhance_data_refs_alignment): Also inspect all datarefs
12968 with unknown misalignment. Compute and costs for unknown
12969 misalignment, compare them to the costs for known misalignment
12970 and choose the cheapest for peeling.
12972 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12974 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
12975 (vect_get_peeling_costs_all_drs): Create function.
12976 (vect_peeling_hash_get_lowest_cost):
12977 Use vect_get_peeling_costs_all_drs.
12978 (vect_peeling_supportable): Create function.
12979 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
12981 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12983 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
12984 DR_HAS_NEGATIVE_STEP.
12985 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
12986 (vect_enhance_data_refs_alignment): Use.
12987 (vect_duplicate_ssa_name_ptr_info): Use.
12988 * tree-vectorizer.h (dr_misalignment): Use.
12989 (known_alignment_for_access_p): Use.
12991 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12994 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
12996 (has_section_name): New function.
12998 2017-05-30 Martin Liska <mliska@suse.cz>
13001 * auto-profile.c (get_function_decl_from_block): Fix
13004 2017-05-30 Richard Biener <rguenther@suse.de>
13006 PR middle-end/80876
13007 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13009 2017-05-30 Martin Liska <mliska@suse.cz>
13011 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13012 * dumpfile.h (struct dump_file_info): Remove ctors.
13014 2017-05-30 Martin Liska <mliska@suse.cz>
13016 * predict.def: Fix GNU coding style.
13018 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13020 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13021 Mark 'to' argument with ATTRIBUTE_UNUSED.
13023 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13025 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13026 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13027 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13030 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13032 * doc/install.texi (Options specification): Restore entry of
13033 --enable-sjlj-exceptions.
13035 2017-05-27 Michael Eager <eager@eagercon.com>
13038 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13040 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13042 * config/microblaze/microblaze.h
13043 (FIXED_REGISTERS): Update in macro.
13044 (CALL_USED_REGISTERS): Update in macro.
13046 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13048 * doc/install.texi: Add links to macOS binary distributions.
13050 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13054 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13056 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13058 2017-05-26 Martin Liska <mliska@suse.cz>
13060 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13062 2017-05-26 Martin Liska <mliska@suse.cz>
13064 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13065 always leading ';; '.
13066 (dump_bb_info): Likewise.
13067 (brief_dump_cfg): Likewise.
13068 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13069 * dumpfile.c: Remove usage of TDF_VERBOSE.
13070 * dumpfile.h (enum dump_kind): Likewise.
13071 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13072 * print-tree.c (debug_verbose): Remove.
13073 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13074 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13075 * tree-diagnostic.c (default_tree_printer): Replace
13076 TDF_DIAGNOSTIC with TDF_SLIM.
13078 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13080 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13081 in parameter loop, rather than loop_vinfo.
13082 (create_intersect_range_checks): Ditto.
13083 (vect_create_cond_for_alias_checks): Update call to above functions.
13085 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13087 PR tree-optimization/80815
13088 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13089 for merging runtime alias checks. Handle negative DR_STEPs.
13091 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13093 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13095 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13096 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13097 out code pruning runtime alias checks.
13098 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13099 factored out from above.
13100 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13102 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13104 (prune_runtime_alias_test_list): New decalaration.
13106 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13108 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13109 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13110 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13111 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13112 (operator==, comp_dr_with_seg_len_pair): Ditto.
13113 (vect_prune_runtime_alias_test_list): Ditto.
13115 2017-05-26 Martin Liska <mliska@suse.cz>
13118 * params.def: Bound partial-inlining-entry-probability param.
13120 2017-05-26 Marek Polacek <polacek@redhat.com>
13123 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13126 2017-05-26 Richard Biener <rguenther@suse.de>
13128 PR tree-optimization/80842
13129 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13132 2017-05-26 Richard Biener <rguenther@suse.de>
13134 PR tree-optimization/80844
13135 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13137 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13139 * doc/md.texi (Machine Constraints): Update x86 family
13140 machine constraints section to match 'config/i386/constraints.md'.
13142 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13144 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13146 2017-05-25 Nathan Sidwell <nathan@acm.org>
13148 * doc/invoke.texi (--enable-languages): Update documentation.
13150 2017-05-25 Martin Liska <mliska@suse.cz>
13152 * dumpfile.c: Add TDF_FOLDING.
13153 * dumpfile.h (enum dump_kind): Likewise.
13154 * genmatch.c (dt_simplify::gen_1): Use it.
13156 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13158 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13160 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13162 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13163 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13165 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13167 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13168 * match.pd (X == C): Rewrite it here.
13169 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13170 with_certain_nonzero_bits2): New predicates.
13171 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13173 2017-05-24 Nathan Sidwell <nathan@acm.org>
13175 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13178 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13181 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13183 * config/powerpcspe: New port. Files are copied from the rs6000
13184 port, with "rs6000" in filenames replaced by "powerpcspe".
13186 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13188 PR rtl-optimization/80754
13189 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13191 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13193 * config/sparc/sparc.md (length): Return the correct value for -mflat
13194 sibcalls to match output_sibcall.
13196 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13200 * config/rs6000/rs6000.c (struct machine_function): Add new field
13202 (rs6000_get_separate_components): Init that field, use it.
13203 (rs6000_components_for_bb): Use the field.
13205 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13207 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13209 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13211 PR middle-end/80823
13212 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13214 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13217 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13218 address against address_operand predicate.
13219 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13221 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13223 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13224 parameters passed indirectly.
13226 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13228 * config/i386/i386.md (*movdi_internal): Remove SSE4
13229 alternative 18 (?r, *v). Update insn attributes.
13230 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13231 Update insn attributes.
13232 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13233 Update insn attributes.
13234 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13235 alternative 1 (r, v). Remove isa attribute.
13236 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13237 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13238 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13240 2017-05-23 Tom de Vries <tom@codesourcery.com>
13242 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13245 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13247 * cgraphunit.c (symbol_table::process_new_functions): Update.
13248 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13249 (inline_generate_summary): Rename to ...
13250 (ipa_fn_summary_generate): ... this one.
13251 (inline_read_summary): Rename to ...
13252 (ipa_fn_summary_read): ... this one.
13253 (inline_write_summary): Rename to ...
13254 (ipa_fn_summary_write): ... this one.
13255 (inline_free_summary): Rename to ...
13256 (ipa_free_fn_summary): ... this one.
13257 (pass_data_local_fn_summary, pass_local_fn_summary,
13258 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13259 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13260 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13261 make_pass_ipa_fn_summary): New.
13262 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13263 inline_write_summary, inline_free_summary): Remove.
13264 (ipa_free_fn_summary) : New.
13265 * ipa-inline.c (ipa_inline): Update.
13266 (pass_ipa_inline): Do not generate summaries.
13267 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13269 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13270 and add pass_ipa_fn_summary.
13271 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13273 (make_pass_inline_parameters): Remove.
13275 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13277 * omp-low.c (struct omp_context): Remove "default_kind" member.
13280 * omp-offload.c (execute_oacc_device_lower): Remove the
13281 parallelism dimensions function attributes for unparallelized
13282 OpenACC kernels constructs.
13284 2017-05-23 Martin Liska <mliska@suse.cz>
13286 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13288 (cgraph_edge::make_speculative): Likewise.
13289 (cgraph_edge::resolve_speculation): Likewise.
13290 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13291 (cgraph_node::dump): Likewise.
13292 * cgraph.h: Likewise.
13293 * cgraphunit.c (analyze_functions): Likewise.
13294 (symbol_table::compile): Likewise.
13295 * ipa-cp.c (print_all_lattices): Likewise.
13296 (determine_versionability): Likewise.
13297 (initialize_node_lattices): Likewise.
13298 (ipcp_verify_propagated_values): Likewise.
13299 (estimate_local_effects): Likewise.
13300 (update_profiling_info): Likewise.
13301 (create_specialized_node): Likewise.
13302 (perhaps_add_new_callers): Likewise.
13303 (decide_about_value): Likewise.
13304 (decide_whether_version_node): Likewise.
13305 (identify_dead_nodes): Likewise.
13306 (ipcp_store_bits_results): Likewise.
13307 * ipa-devirt.c (dump_targets): Likewise.
13308 (ipa_devirt): Likewise.
13309 * ipa-icf.c (sem_item::dump): Likewise.
13310 (sem_function::equals): Likewise.
13311 (sem_variable::equals): Likewise.
13312 (sem_item_optimizer::read_section): Likewise.
13313 (sem_item_optimizer::execute): Likewise.
13314 (congruence_class::dump): Likewise.
13315 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13316 (dump_inline_summary): Likewise.
13317 (estimate_node_size_and_time): Likewise.
13318 (inline_analyze_function): Likewise.
13319 * ipa-inline-transform.c (inline_call): Likewise.
13320 * ipa-inline.c (report_inline_failed_reason): Likewise.
13321 (want_early_inline_function_p): Likewise.
13322 (edge_badness): Likewise.
13323 (update_edge_key): Likewise.
13324 (inline_small_functions): Likewise.
13325 * ipa-profile.c (ipa_profile): Likewise.
13326 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13327 (ipa_make_edge_direct_to_target): Likewise.
13328 (remove_described_reference): Likewise.
13329 (ipa_impossible_devirt_target): Likewise.
13330 (propagate_controlled_uses): Likewise.
13331 (ipa_print_node_params): Likewise.
13332 (ipcp_transform_function): Likewise.
13333 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13334 (propagate_pure_const): Likewise.
13335 * ipa-reference.c (generate_summary): Likewise.
13336 (read_write_all_from_decl): Likewise.
13337 (propagate): Likewise.
13338 (ipa_reference_read_optimization_summary): Likewise.
13339 * ipa-utils.c (ipa_merge_profiles): Likewise.
13340 * ipa.c (walk_polymorphic_call_targets): Likewise.
13341 (symbol_table::remove_unreachable_nodes): Likewise.
13342 (ipa_single_use): Likewise.
13343 * passes.c (execute_todo): Likewise.
13344 * predict.c (drop_profile): Likewise.
13345 * symtab.c (symtab_node::get_dump_name): New function.
13346 (symtab_node::dump_name): Likewise.
13347 (symtab_node::dump_asm_name): Likewise.
13348 (symtab_node::dump_references): Likewise.
13349 (symtab_node::dump_referring): Likewise.
13350 (symtab_node::dump_base): Likewise.
13351 (symtab_node::debug_symtab): Likewise.
13352 * tree-sra.c (convert_callers_for_node): Likewise.
13353 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13354 * value-prof.c (init_node_map): Likewise.
13356 2017-05-23 Martin Liska <mliska@suse.cz>
13358 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13359 and symtab_node::debug_symtab to symbol_table::debug.
13360 * cgraphunit.c (analyze_functions): Use the renamed function.
13361 (symbol_table::compile): Likewise.
13362 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13363 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13364 * passes.c (execute_todo): Likewise.
13365 * symtab.c (symbol_table::dump): New function.
13366 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13368 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13370 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13371 that nonconst implies exec.
13373 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13375 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13376 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13377 (inline_edge_summary_vec): Turn into ...
13378 (ipa_call_summaries): ... this one.
13379 (redirect_to_unreachable, edge_set_predicate,
13380 evaluate_properties_for_edge, inline_summary_alloc,
13381 reset_ipa_call_summary, reset_inline_summary,
13382 inline_summary_t::duplicate): Update.
13383 (inline_edge_duplication_hook): Turn to ...
13384 (ipa_call_summary_t::duplicate): ... this one.
13385 (inline_edge_removal_hook): Turn to ...
13386 (ipa_call_summary_t::remove): ... this one.
13387 (dump_inline_edge_summary): Turn to ...
13388 (dump_ipa_call_summary): ... this one.
13389 (estimate_function_body_sizes): Update.
13390 (inline_update_callee_summaries): Update.
13391 (remap_edge_change_prob): Update.
13392 (remap_edge_summaries): Update.
13393 (inline_merge_summary): Update.
13394 (do_estimate_edge_time): Update.
13395 (inline_generate_summary): Update.
13396 (inline_read_section): Update.
13397 (inline_read_summary): Update.
13398 (inline_free_summary): Update.
13399 * ipa-inline.c (can_inline_edge_p): Update.
13400 (compute_inlined_call_time): Update.
13401 (want_inline_small_function_p): Update.
13402 (edge_badness): Update.
13403 (early_inliner): Update.
13404 * ipa-inline.h (inline_edge_summary): Turn to ...
13405 (ipa_call_summary): ... this one.
13406 (ipa_call_summary_t): New class.
13407 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13408 (ipa_call_summaries): New.
13409 (inline_edge_summary): Remove.
13410 (estimate_edge_growth): Update.
13411 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13412 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13413 * ipa-split.c (execute_split_functions): Update.
13414 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13416 2017-05-23 Tom de Vries <tom@codesourcery.com>
13418 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13419 attributes): Document rdrand effective target.
13421 2017-05-23 Tom de Vries <tom@codesourcery.com>
13423 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13424 attributes): Sort alphabetically.
13426 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13428 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13430 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13433 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13434 V2DF/V2DI splat into two separate patterns, one that handles
13435 registers, and the other that only handles memory. Drop support
13436 for splatting from a GPR on ISA 2.07 and then splitting the
13437 splat into direct move and splat.
13438 (vsx_splat_<mode>_reg): Likewise.
13439 (vsx_splat_<mode>_mem): Likewise.
13441 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13443 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13445 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13447 PR middle-end/80809
13448 * omp-low.c (finish_taskreg_remap): New function.
13449 (finish_taskreg_scan): If unit size of ctx->record_type
13450 is non-constant, unshare the size expression and replace
13451 decls in it with possible outer var refs.
13453 PR middle-end/80809
13454 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13455 GOVD_SHARED rather than GOVD_PRIVATE with it.
13456 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13457 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13459 PR middle-end/80853
13460 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13461 as last argument to build_outer_var_ref for pointer bases of array
13462 section reductions.
13464 2017-05-19 Martin Sebor <msebor@redhat.com>
13466 * print-tree.c (print_node): Print DECL_READ_P flag.
13468 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13470 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13471 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13472 * cgraph.c: Likewise.
13473 * cgraphunit.c: Likewise.
13474 * gengtype.c: Likewise.
13475 * ipa-cp.c: Likewise.
13476 * ipa-devirt.c: Likewise.
13477 * ipa-icf.c: Likewise.
13478 * ipa-predicate.c: Likewise.
13479 * ipa-profile.c: Likewise.
13480 * ipa-prop.c: Likewise.
13481 * ipa-split.c: Likewise.
13483 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13484 edge_predicate_pool, dump_inline_hints,
13485 inline_summary::account_size_time, redirect_to_unreachable,
13486 edge_set_predicate, set_hint_predicate,
13487 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13488 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13489 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13490 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13491 ipa_call_summary_t::remove, initialize_growth_caches,
13492 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13493 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13494 mark_modified, unmodified_parm_1, unmodified_parm,
13495 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13496 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13497 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13498 will_be_nonconstant_predicate, record_modified_bb_info,
13499 get_minimal_bb, record_modified, param_change_prob,
13500 phi_result_unknown_predicate, predicate_for_phi_result,
13501 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13502 estimate_function_body_sizes, compute_inline_parameters,
13503 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13504 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13505 inline_update_callee_summaries, remap_edge_change_prob,
13506 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13507 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13508 inline_analyze_function, inline_summary_t::insert,
13509 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13510 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13511 inline_free_summary): Move to ipa-fnsummary.h
13512 (predicate_t): Remove.
13513 * ipa-fnsummary.c: New file.
13514 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13515 (enum inline_hints_vals, inline_hints, agg_position_info,
13516 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13517 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13518 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13519 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13520 inline_read_summary, inline_write_summary, inline_free_summary,
13521 inline_analyze_function, initialize_inline_failed,
13522 inline_merge_summary, inline_update_overall_summary,
13523 compute_inline_parameters): Move to ipa-fnsummary.h
13524 * ipa-fnsummary.h: New file.
13525 * ipa-inline-transform.h: Include ipa-inline.h.
13526 * ipa-inline.c: LIkewise.
13528 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13530 * ipa-inline.c (edge_badness): Use inlined_time instead of
13531 inline_summaries->get.
13533 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13535 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13537 2017-05-22 Nathan Sidwell <nathan@acm.org>
13539 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13540 (fdump-lang): Document 'raw' option.
13541 * dumpfile.h (TDI_tu): Delete.
13542 * dumpfile.c (dump_files): Remove translation-unit.
13543 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13545 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13547 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13548 command option from $(AWK) call.
13549 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13550 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13551 [FORMAT]: Remove handling of variable.
13552 * config/avr/t-multilib: Regenerate.
13554 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13556 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13558 (dump_inline_summary): Do not print self_time.
13559 (estimate_function_body_sizes): Do not set self_time.
13560 (compute_inline_parameters): Likewise.
13561 (inline_read_section, inline_write_summary): Do not stream self_time.
13562 * ipa-inline.h (inline_summary): Drop self_time.
13564 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13566 * ipa-inline-analysis.c (account_size_time): Rename to ...
13567 (inline_summary::account_size_time): ... this one.
13568 (reset_ipa_call_summary): Turn to ...
13569 (ipa_call_summary::reset): ... this one.
13570 (reset_inline_summary): Turn to ...
13571 (inline_summary::reset): ... this one.
13572 (inline_summary_t::remove): Update.
13573 (inline_summary_t::duplicate): Update.
13574 (ipa_call_summary_t::remove): Update.
13575 (dump_inline_summary): Update.
13576 (estimate_function_body_sizes): Update.
13577 (compute_inline_parameters): Update.
13578 (estimate_node_size_and_time): Update.
13579 (inline_merge_summary): Update.
13580 (inline_update_overall_summary): Update.
13581 (inline_read_section): Update.
13582 (inline_write_summary): Update.
13583 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
13584 add account_size_time and reset member functions.
13585 (ipa_call_summary): Add reset function.
13586 * ipa-predicate.h (predicate::operator &): Constify.
13588 2017-05-22 Richard Biener <rguenther@suse.de>
13590 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
13592 2017-05-19 Jason Merrill <jason@redhat.com>
13594 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
13596 2017-05-19 Marek Polacek <polacek@redhat.com>
13599 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
13600 TYPE_OVERFLOW_WRAPS checks.
13602 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
13604 * tree-core.h (enum omp_clause_default_kind): Add
13605 "OMP_CLAUSE_DEFAULT_PRESENT".
13606 * tree-pretty-print.c (dump_omp_clause): Handle it.
13607 * gimplify.c (enum gimplify_omp_var_data): Add
13608 "GOVD_MAP_FORCE_PRESENT".
13609 (gimplify_adjust_omp_clauses_1): Map it to
13610 "GOMP_MAP_FORCE_PRESENT".
13611 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
13613 * gimplify.c (oacc_default_clause): Clarify.
13615 2017-05-19 Nathan Sidwell <nathan@acm.org>
13617 LANG_HOOK_REGISTER_DUMPS
13618 * toplev.c (general_init): Call register dump lang hook.
13619 * doc/invoke.texi: Document -fdump-lang option family.
13620 * dumpfile.c (dump_files): Remove class dump here.
13621 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
13622 * dumpfile.h (tree_dump_index): Remove TDI_class.
13623 * langhooks-def.h (lhd_register_dumps): Declare.
13624 (LANG_HOOKS_REGISTER_DUMPS): Define.
13625 (LANG_HOOKS_INITIALIZER): Add it.
13626 * langhooks.c (lhd_register_dumps): Define.
13627 * langhooks.h (struct lang_hooks): Add register_dumps.
13629 2017-05-19 Nathan Sidwell <nathan@acm.org>
13631 * context.h (context::set_passes): New.
13632 * context.c (context::context): Do not create pass manager.
13633 * toplev.c (general_init): Create pass manager here.
13635 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
13637 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
13638 use this splitter if two add or or instructions would also work for
13639 the constant we want to generate.
13641 2017-05-19 Richard Biener <rguenther@suse.de>
13644 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
13645 predicate evaluation.
13647 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13649 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
13651 * ipa-inline.c (want_inline_small_function_p): Do not cast to
13654 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13656 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13657 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13658 (inline_edge_summary_vec): Turn into ...
13659 (ipa_call_summaries): ... this one.
13660 (redirect_to_unreachable, edge_set_predicate,
13661 evaluate_properties_for_edge, inline_summary_alloc,
13662 reset_ipa_call_summary, reset_inline_summary,
13663 inline_summary_t::duplicate): Update.
13664 (inline_edge_duplication_hook): Turn to ...
13665 (ipa_call_summary_t::duplicate): ... this one.
13666 (inline_edge_removal_hook): Turn to ...
13667 (ipa_call_summary_t::remove): ... this one.
13668 (dump_inline_edge_summary): Turn to ...
13669 (dump_ipa_call_summary): ... this one.
13670 (estimate_function_body_sizes): Update.
13671 (inline_update_callee_summaries): Update.
13672 (remap_edge_change_prob): Update.
13673 (remap_edge_summaries): Update.
13674 (inline_merge_summary): Update.
13675 (do_estimate_edge_time): Update.
13676 (inline_generate_summary): Update.
13677 (inline_read_section): Update.
13678 (inline_read_summary): Update.
13679 (inline_free_summary): Update.
13680 * ipa-inline.c (can_inline_edge_p): Update.
13681 (compute_inlined_call_time): Update.
13682 (want_inline_small_function_p): Update.
13683 (edge_badness): Update.
13684 (early_inliner): Update.
13685 * ipa-inline.h (inline_edge_summary): Turn to ...
13686 (ipa_call_summary): ... this one.
13687 (ipa_call_summary_t): New class.
13688 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13689 (ipa_call_summaries): New.
13690 (inline_edge_summary): Remove.
13691 (estimate_edge_growth): Update.
13692 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13693 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13694 * ipa-split.c (execute_split_functions): Update.
13695 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13697 2017-05-19 Richard Biener <rguenther@suse.de>
13699 PR middle-end/80764
13700 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
13702 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
13704 * config/rs6000/rs6000.c (struct machine_function): Add field
13705 fpr_is_wrapped_separately.
13706 (rs6000_get_separate_components): Use 64 components. Handle the
13707 new FPR components.
13708 (rs6000_components_for_bb): Handle the FPR components.
13709 (rs6000_emit_prologue_components): Handle the FPR components.
13710 (rs6000_emit_epilogue_components): Handle the FPR components.
13711 (rs6000_set_handled_components): Handle the FPR components.
13712 (rs6000_emit_prologue): Don't output prologue code for those FPRs
13713 that are already separately shrink-wrapped.
13714 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
13715 that are already separately shrink-wrapped.
13717 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
13720 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
13723 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
13724 (define_peephole2 for Altivec d-form load): Add peepholes to catch
13725 cases where the register allocator uses a move and an offsettable
13726 memory operation to/from a FPR register on ISA 2.06/2.07.
13727 (define_peephole2 for Altivec d-form store): Likewise.
13729 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13732 * config/i386/mmx.md (*mov<mode>_internal): Enable
13733 alternatives 11, 12, 13 and 14 also for 32bit targets.
13734 Remove alternatives 15, 16, 17 and 18.
13735 * config/i386/sse.md (vec_concatv2di): Change
13736 alternative (!x, *y) to (x, ?!*Yn).
13738 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13740 * dumpfile.h (enum dump_kind): Remove stray comma.
13742 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13744 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13745 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13746 predicate::num_conditions
13747 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13748 (CHANGED): turn into predicate::changed.
13749 (agg_position_info): Move to ipa-predicate.h
13750 (add_condition, predicate::add_clause, predicate::operator &=,
13751 predicate::or_with, predicate::evaluate, predicate::probability,
13752 dump_condition, dump_clause, predicate::dump,
13753 predicate::remap_after_duplication, predicate::remap_after_inlining,
13754 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13755 (evaluate_conditions_for_known_args): Update.
13756 (set_cond_stmt_execution_predicate): Update.
13757 * ipa-inline.h: Include ipa-predicate.h
13758 (condition, inline_param_summary, conditions, agg_position_info,
13759 predicate): Move to ipa-predicate.h
13760 * ipa-predicate.c: New file.
13761 * ipa-predicate.h: New file.
13763 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
13765 * final.c (leaf_function_p): Check we are not in a sequence.
13767 2017-05-18 Martin Liska <mliska@suse.cz>
13769 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
13770 * dumpfile.c (dump_register): Use new enum dump_kind.
13771 (get_dump_file_name): Likewise.
13772 (dump_enable_all): Likewise.
13773 (dump_switch_p_1): Likewise.
13774 (enable_rtl_dump_file): Remove usage of TDF_RTL.
13775 * dumpfile.h (enum dump_kind): New enum type.
13776 (struct dump_file_info): Create constructor and
13777 format fields and comments.
13778 * passes.c (pass_manager::register_one_dump_file):
13780 * statistics.c (statistics_early_init): Likewise.
13781 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
13782 TDF_TREE with TDF_SLIM.
13783 (gather_memory_references_ref): Likewise.
13785 2017-05-18 Martin Liska <mliska@suse.cz>
13787 * vec.h (struct vnull): Use it.
13789 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13791 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
13792 (true_predicate, false_predicate, true_predicate_p,
13793 false_predicate_p): Remove.
13794 (single_cond_predicate, not_inlined_predicate): Turn to member function
13796 (add_condition): Update.
13797 (add_clause): Turn to...
13798 (predicate::add_clause): ... this one; update; allow passing NULL
13800 (and_predicates): Turn to ...
13801 (predicate::operator &=): ... this one.
13802 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
13803 (or_predicates): Turn to ...
13804 (predicate::or_with): ... this one.
13805 (evaluate_predicate): Turn to ...
13806 (predicate::evaluate): ... this one.
13807 (predicate_probability): Turn to ...
13808 (predicate::probability): ... this one.
13809 (dump_condition): Update.
13810 (dump_predicate): Turn to ...
13811 (predicate::dump): ... this one.
13812 (account_size_time): Update.
13813 (edge_set_predicate): Update.
13814 (set_hint_predicate): UPdate.
13815 (evaluate_conditions_for_known_args): Update.
13816 (evaluate_properties_for_edge): Update.
13817 (remap_predicate_after_duplication): Turn to...
13818 (predicate::remap_after_duplication): ... this one.
13819 (remap_hint_predicate_after_duplication): Update.
13820 (inline_summary_t::duplicate): UPdate.
13821 (dump_inline_edge_summary): Update.
13822 (dump_inline_summary): Update.
13823 (set_cond_stmt_execution_predicate): Update.
13824 (set_switch_stmt_execution_predicate): Update.
13825 (compute_bb_predicates): Update.
13826 (will_be_nonconstant_expr_predicate): Update.
13827 (will_be_nonconstant_predicate): Update.
13828 (phi_result_unknown_predicate): Update.
13829 (predicate_for_phi_result): Update.
13830 (array_index_predicate): Update.
13831 (estimate_function_body_sizes): Update.
13832 (estimate_node_size_and_time): Update.
13833 (estimate_ipcp_clone_size_and_time): Update.
13834 (remap_predicate): Rename to ...
13835 (predicate::remap_after_inlining): ... this one.
13836 (remap_hint_predicate): Update.
13837 (inline_merge_summary): Update.
13838 (inline_update_overall_summary): Update.
13839 (estimate_size_after_inlining): Update.
13840 (read_predicate): Rename to ...
13841 (predicate::stream_in): ... this one.
13842 (read_inline_edge_summary): Update.
13843 (write_predicate): Rename to ...
13844 (predicate::stream_out): ... this one.
13845 (write_inline_edge_summary): Update.
13846 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
13847 (clause_t): Turn to uint32_t
13848 (predicate): Turn to class; implement constructor and operators
13850 (size_time_entry): Update.
13851 (inline_summary): Update.
13852 (inline_edge_summary): Update.
13854 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
13856 * fold-const.c (fold_binary_loc): Move transformation...
13857 * match.pd (C - X CMP X): ... here.
13859 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
13861 * config/sparc/sparc.c (sparc_option_override): Set function
13862 alignment for -mcpu=niagara7 to 64 to match the I$ line.
13863 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
13865 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
13867 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
13869 2017-05-18 Marek Polacek <polacek@redhat.com>
13872 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
13873 (pass_ubsan::execute): Call gimple_assign_single_p instead of
13874 gimple_assign_load_p.
13876 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
13878 PR middle-end/80692
13879 * real.c (do_compare): Give decimal_do_compare preference over
13880 comparing just the signs.
13882 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
13884 * doc/md.texi (Canonicalization of Instructions): Describe the
13885 canonical form of instructions that inherently set a condition
13888 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
13890 PR middle-end/80775
13891 * tree-cfg.c: Move deletion of unreachable case statements to after
13892 the merging of consecutive case labels.
13894 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
13896 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
13897 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
13898 restoring of callee-saved registers.
13900 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
13902 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
13903 * config/visium/visium.c (single_set_and_flags): Likewise.
13904 * config/visium/visium.md (Substitutions): Likewise.
13906 2017-05-17 Martin Liska <mliska@suse.cz>
13908 * cfg.c: Introduce dump_flags_t type and
13909 use it instead of int type.
13911 * cfghooks.c: Likewise.
13912 * cfghooks.h (struct cfg_hooks): Likewise.
13913 * cfgrtl.c: Likewise.
13914 * cfgrtl.h: Likewise.
13915 * cgraph.c (cgraph_node::get_body): Likewise.
13916 * coretypes.h: Likewise.
13917 * domwalk.c: Likewise.
13918 * domwalk.h: Likewise.
13919 * dumpfile.c (struct dump_option_value_info): Likewise.
13920 (dump_enable_all): Likewise.
13921 (dump_switch_p_1): Likewise.
13922 (opt_info_switch_p): Likewise.
13923 * dumpfile.h (enum tree_dump_index): Likewise.
13924 (struct dump_file_info): Likewise.
13925 * genemit.c: Likewise.
13926 * generic-match-head.c: Likewise.
13927 * gengtype.c (open_base_files): Likewise.
13928 * gimple-pretty-print.c: Likewise.
13929 * gimple-pretty-print.h: Likewise.
13930 * graph.c (print_graph_cfg): Likewise.
13931 * graphite-scop-detection.c (dot_all_sese): Likewise.
13932 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
13933 * loop-unroll.c (report_unroll): Likewise.
13934 * passes.c (pass_manager::register_one_dump_file): Likewise.
13935 * print-tree.c: Likewise.
13936 * statistics.c: Likewise.
13937 * tree-cfg.c: Likewise.
13938 * tree-cfg.h: Likewise.
13939 * tree-dfa.c: Likewise.
13940 * tree-dfa.h: Likewise.
13941 * tree-dump.c (dump_function): Likewise.
13942 * tree-dump.h (struct dump_info): Likewise.
13943 * tree-pretty-print.c: Likewise.
13944 * tree-pretty-print.h: Likewise.
13945 * tree-ssa-live.c: Likewise.
13946 * tree-ssa-live.h: Likewise.
13947 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
13948 * tree-vect-loop.c: Likewise.
13949 * tree-vect-slp.c: Likewise.
13951 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
13952 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13954 PR tree-optimization/80457
13955 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
13956 of all arguments to a statement as scalar_to_vec operations.
13957 (vectorizable_call): Adjust call to vect_model_simple_cost for
13959 (vectorizable_conversion): Likewise.
13960 (vectorizable_assignment): Likewise.
13961 (vectorizable_shift): Likewise.
13962 (vectorizable_operation): Likewise.
13963 (vectorizable_comparison): Likewise.
13964 (vect_is_simple_cond): Record the def types for operands.
13965 (vectorizable_condition): Likewise, call vect_model_simple_cost.
13966 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
13967 for statement argument count.
13969 2017-05-16 Carl Love <cel@us.ibm.com>
13971 * config/rs6000/rs6000-c: Add support for built-in functions
13972 vector unsigned long long vec_bperm (vector unsigned long long,
13973 vector unsigned char)
13974 vector signed long long vec_mule (vector signed int,
13976 vector unsigned long long vec_mule (vector unsigned int,
13977 vector unsigned int)
13978 vector signed long long vec_mulo (vector signed int,
13980 vector unsigned long long vec_mulo (vector unsigned int,
13981 vector unsigned int)
13982 vector signed char vec_sldw (vector signed char,
13983 vector signed char,
13985 vector unsigned char vec_sldw (vector unsigned char,
13986 vector unsigned char,
13988 vector signed short vec_sldw (vector signed short,
13989 vector signed short,
13991 vector unsigned short vec_sldw (vector unsigned short,
13992 vector unsigned short,
13994 vector signed int vec_sldw (vector signed int,
13997 vector unsigned int vec_sldw (vector unsigned int,
13998 vector unsigned int,
14000 vector signed long long vec_sldw (vector signed long long,
14001 vector signed long long,
14003 vector unsigned long long vec_sldw (vector unsigned long long,
14004 vector unsigned long long,
14006 * config/rs6000/rs6000-c: Add support for built-in functions
14007 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14008 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14009 * doc/extend.texi: Update the built-in documentation for the
14010 new built-in functions.
14012 2017-05-16 Marek Polacek <polacek@redhat.com>
14016 * tree.c (save_expr): Don't fold the expression.
14018 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14020 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14021 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14022 and (?*y,m). Update insn attributes.
14024 2017-05-16 Martin Liska <mliska@suse.cz>
14026 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14027 flags argument of print_gimple_stmt, print_gimple_expr,
14028 print_generic_stmt and print_generic_expr.
14029 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14030 * coretypes.h: Likewise.
14031 * except.c (dump_eh_tree): Likewise.
14032 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14033 * gimple-pretty-print.h: Likewise.
14034 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14035 (backprop::push_to_worklist): Likewise.
14036 (backprop::pop_from_worklist): Likewise.
14037 (backprop::process_use): Likewise.
14038 (backprop::intersect_uses): Likewise.
14039 (note_replacement): Likewise.
14040 * gimple-ssa-store-merging.c
14041 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14042 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14043 (pass_store_merging::execute): Likewise.
14044 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14045 (ssa_base_cand_dump_callback): Likewise.
14046 (dump_incr_vec): Likewise.
14047 (replace_refs): Likewise.
14048 (replace_mult_candidate): Likewise.
14049 (create_add_on_incoming_edge): Likewise.
14050 (create_phi_basis): Likewise.
14051 (insert_initializers): Likewise.
14052 (all_phi_incrs_profitable): Likewise.
14053 (introduce_cast_before_cand): Likewise.
14054 (replace_one_candidate): Likewise.
14055 * gimplify.c (gimplify_expr): Likewise.
14056 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14057 (set_rename): Likewise.
14058 (rename_uses): Likewise.
14059 (copy_loop_phi_nodes): Likewise.
14060 (add_close_phis_to_merge_points): Likewise.
14061 (copy_loop_close_phi_args): Likewise.
14062 (copy_cond_phi_args): Likewise.
14063 (graphite_copy_stmts_from_block): Likewise.
14064 (translate_pending_phi_nodes): Likewise.
14065 * graphite-poly.c (print_pdr): Likewise.
14066 (dump_gbb_cases): Likewise.
14067 (dump_gbb_conditions): Likewise.
14068 (print_scop_params): Likewise.
14069 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14070 (build_cross_bb_scalars_use): Likewise.
14071 (gather_bbs::before_dom_children): Likewise.
14072 * hsa-dump.c (dump_hsa_immed): Likewise.
14073 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14074 (get_replacement_map): Likewise.
14075 * ipa-inline-analysis.c (dump_condition): Likewise.
14076 (estimate_function_body_sizes): Likewise.
14077 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14078 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14079 * ipa-prop.c (ipa_dump_param): Likewise.
14080 (ipa_print_node_jump_functions_for_edge): Likewise.
14081 (ipa_modify_call_arguments): Likewise.
14082 (ipa_modify_expr): Likewise.
14083 (ipa_dump_param_adjustments): Likewise.
14084 (ipa_dump_agg_replacement_values): Likewise.
14085 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14086 * ipa-pure-const.c (check_stmt): Likewise.
14087 (pass_nothrow::execute): Likewise.
14088 * ipa-split.c (execute_split_functions): Likewise.
14089 * omp-offload.c (dump_oacc_loop_part): Likewise.
14090 (dump_oacc_loop): Likewise.
14091 * trans-mem.c (tm_log_emit): Likewise.
14092 (tm_memopt_accumulate_memops): Likewise.
14093 (dump_tm_memopt_set): Likewise.
14094 (dump_tm_memopt_transform): Likewise.
14095 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14096 (print_loop): Likewise.
14097 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14098 (chkp_gather_checks_info): Likewise.
14099 (chkp_get_check_result): Likewise.
14100 (chkp_remove_check_if_pass): Likewise.
14101 (chkp_use_outer_bounds_if_possible): Likewise.
14102 (chkp_reduce_bounds_lifetime): Likewise.
14103 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14104 (chkp_mark_completed_bounds): Likewise.
14105 (chkp_register_incomplete_bounds): Likewise.
14106 (chkp_mark_invalid_bounds): Likewise.
14107 (chkp_maybe_copy_and_register_bounds): Likewise.
14108 (chkp_build_returned_bound): Likewise.
14109 (chkp_get_bound_for_parm): Likewise.
14110 (chkp_build_bndldx): Likewise.
14111 (chkp_get_bounds_by_definition): Likewise.
14112 (chkp_generate_extern_var_bounds): Likewise.
14113 (chkp_get_bounds_for_decl_addr): Likewise.
14114 * tree-chrec.c (chrec_apply): Likewise.
14115 * tree-data-ref.c (dump_data_reference): Likewise.
14116 (dump_subscript): Likewise.
14117 (dump_data_dependence_relation): Likewise.
14118 (analyze_overlapping_iterations): Likewise.
14119 * tree-inline.c (expand_call_inline): Likewise.
14120 (tree_function_versioning): Likewise.
14121 * tree-into-ssa.c (dump_defs_stack): Likewise.
14122 (dump_currdefs): Likewise.
14123 (dump_names_replaced_by): Likewise.
14124 (dump_update_ssa): Likewise.
14125 (update_ssa): Likewise.
14126 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14127 * tree-parloops.c (build_new_reduction): Likewise.
14128 (try_create_reduction_list): Likewise.
14129 (ref_conflicts_with_region): Likewise.
14130 (oacc_entry_exit_ok_1): Likewise.
14131 (oacc_entry_exit_single_gang): Likewise.
14132 * tree-pretty-print.h: Likewise.
14133 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14134 (get_scalar_evolution): Likewise.
14135 (add_to_evolution): Likewise.
14136 (get_loop_exit_condition): Likewise.
14137 (analyze_evolution_in_loop): Likewise.
14138 (analyze_initial_condition): Likewise.
14139 (analyze_scalar_evolution): Likewise.
14140 (instantiate_scev): Likewise.
14141 (number_of_latch_executions): Likewise.
14142 (gather_chrec_stats): Likewise.
14143 (final_value_replacement_loop): Likewise.
14144 (scev_const_prop): Likewise.
14145 * tree-sra.c (dump_access): Likewise.
14146 (disqualify_candidate): Likewise.
14147 (create_access): Likewise.
14148 (reject): Likewise.
14149 (maybe_add_sra_candidate): Likewise.
14150 (create_access_replacement): Likewise.
14151 (analyze_access_subtree): Likewise.
14152 (analyze_all_variable_accesses): Likewise.
14153 (sra_modify_assign): Likewise.
14154 (initialize_constant_pool_replacements): Likewise.
14155 (find_param_candidates): Likewise.
14156 (decide_one_param_reduction): Likewise.
14157 (replace_removed_params_ssa_names): Likewise.
14158 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14159 * tree-ssa-copy.c (dump_copy_of): Likewise.
14160 (copy_prop_visit_cond_stmt): Likewise.
14161 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14162 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14163 (record_equivalences_from_stmt): Likewise.
14164 * tree-ssa-dse.c (compute_trims): Likewise.
14165 (delete_dead_call): Likewise.
14166 (delete_dead_assignment): Likewise.
14167 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14168 (forward_propagate_into_cond): Likewise.
14169 (pass_forwprop::execute): Likewise.
14170 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14171 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14173 (move_computations_worker): Likewise.
14174 (execute_sm): Likewise.
14175 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14176 (remove_exits_and_undefined_stmts): Likewise.
14177 (remove_redundant_iv_tests): Likewise.
14178 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14179 (adjust_iv_update_pos): Likewise.
14180 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14181 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14182 (value_replacement): Likewise.
14183 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14184 * tree-ssa-pre.c (print_pre_expr): Likewise.
14185 (get_representative_for): Likewise.
14186 (create_expression_by_pieces): Likewise.
14187 (insert_into_preds_of_block): Likewise.
14188 (eliminate_insert): Likewise.
14189 (eliminate_dom_walker::before_dom_children): Likewise.
14190 (eliminate): Likewise.
14191 (remove_dead_inserted_code): Likewise.
14192 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14193 * tree-ssa-reassoc.c (get_rank): Likewise.
14194 (eliminate_duplicate_pair): Likewise.
14195 (eliminate_plus_minus_pair): Likewise.
14196 (eliminate_not_pairs): Likewise.
14197 (undistribute_ops_list): Likewise.
14198 (eliminate_redundant_comparison): Likewise.
14199 (update_range_test): Likewise.
14200 (optimize_range_tests_var_bound): Likewise.
14201 (optimize_vec_cond_expr): Likewise.
14202 (rewrite_expr_tree): Likewise.
14203 (rewrite_expr_tree_parallel): Likewise.
14204 (linearize_expr): Likewise.
14205 (break_up_subtract): Likewise.
14206 (linearize_expr_tree): Likewise.
14207 (attempt_builtin_powi): Likewise.
14208 (attempt_builtin_copysign): Likewise.
14209 (transform_stmt_to_copy): Likewise.
14210 (transform_stmt_to_multiply): Likewise.
14211 (dump_ops_vector): Likewise.
14212 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14213 (print_scc): Likewise.
14214 (set_ssa_val_to): Likewise.
14215 (visit_reference_op_store): Likewise.
14216 (visit_use): Likewise.
14217 (sccvn_dom_walker::before_dom_children): Likewise.
14218 (run_scc_vn): Likewise.
14219 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14221 (expr_hash_elt::print): Likewise.
14222 (const_and_copies::pop_to_marker): Likewise.
14223 (const_and_copies::record_const_or_copy_raw): Likewise.
14224 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14225 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14226 (dump_predicates): Likewise.
14227 (find_uninit_use): Likewise.
14228 (warn_uninitialized_phi): Likewise.
14229 (pass_late_warn_uninitialized::execute): Likewise.
14230 * tree-ssa.c (verify_vssa): Likewise.
14231 (verify_ssa): Likewise.
14232 (maybe_optimize_var): Likewise.
14233 * tree-vrp.c (dump_value_range): Likewise.
14234 (dump_all_value_ranges): Likewise.
14235 (dump_asserts_for): Likewise.
14236 (register_edge_assert_for_2): Likewise.
14237 (vrp_visit_cond_stmt): Likewise.
14238 (vrp_visit_switch_stmt): Likewise.
14239 (vrp_visit_stmt): Likewise.
14240 (vrp_visit_phi_node): Likewise.
14241 (simplify_cond_using_ranges_1): Likewise.
14242 (fold_predicate_in): Likewise.
14243 (evrp_dom_walker::before_dom_children): Likewise.
14244 (evrp_dom_walker::push_value_range): Likewise.
14245 (evrp_dom_walker::pop_value_range): Likewise.
14246 (execute_early_vrp): Likewise.
14248 2017-05-16 Richard Biener <rguenther@suse.de>
14250 * dwarf2out.c (loc_list_from_tree_1): Do not create
14251 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14253 2017-05-16 Richard Biener <rguenther@suse.de>
14255 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14257 (note_variable_value_in_expr): If we resolved the decl ref
14258 do not push to the stack.
14260 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14262 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14263 operations in fast-math mode.
14264 (vaddq_f16): Likewise.
14265 (vmul_f16): Likewise.
14266 (vmulq_f16): Likewise.
14267 (vsub_f16): Likewise.
14268 (vsubq_f16): Likewise.
14269 * config/arm/neon.md (add<mode>3): New.
14271 (fma:<VH:mode>3): New. Also remove outdated comment.
14274 2017-05-16 Martin Liska <mliska@suse.cz>
14278 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14279 (odr_types_equivalent_p): Likewise.
14281 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14283 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14285 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14288 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14289 non-interunit SSE move alternatives with '?'.
14290 (zero-extendsidi peephole2): New peephole to skip intermediate
14291 general register in SSE zero-extend sequence.
14293 2017-05-15 Jeff Law <law@redhat.com>
14295 * reorg.c (relax_delay_slots): Create a new variable to hold
14296 the temporary target rather than clobbering TARGET_LABEL.
14298 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14299 missing argument to extract_bit_field call.
14300 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14302 2017-05-15 Martin Liska <mliska@suse.cz>
14305 * gcc.c (process_command): Do not allow empty argument of -o option.
14307 2017-05-15 Renlin Li <renlin.li@arm.com>
14309 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14310 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14311 * config/aarch64/constraints.md (Usf): Add long call check.
14312 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14313 (call_value): Likewise.
14314 (sibcall): Likewise.
14315 (sibcall_value): Likewise.
14317 (call_value_insn): New.
14318 (sibcall_insn): Update rtx pattern.
14319 (sibcall_value_insn): Likewise.
14320 (call_internal): Remove.
14321 (call_value_internal): Likewise.
14322 (sibcall_internal): Likewise.
14323 (sibcall_value_internal): Likewise.
14324 (call_reg): Likewise.
14325 (call_symbol): Likewise.
14326 (call_value_reg): Likewise.
14327 (call_value_symbol): Likewise.
14329 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14332 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14334 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14336 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14337 compatible with CCGOCmode and with CCZmode.
14339 2017-05-14 Martin Sebor <msebor@redhat.com>
14341 PR middle-end/77671
14342 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14343 (gimple_fold_builtin_snprintf): Same.
14344 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14345 (gimple_fold_builtin_snprintf): Same.
14346 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14347 of character types.
14348 (is_call_safe): New function.
14349 (try_substitute_return_value): Call it.
14350 (try_simplify_call): New function.
14351 (pass_sprintf_length::handle_gimple_call): Call it.
14353 2017-05-14 Martin Sebor <msebor@redhat.com>
14355 PR middle-end/80669
14356 * builtins.c (expand_builtin_stpncpy): Simplify.
14358 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14360 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14361 * config/i386/i386.h
14362 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14363 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14364 (struct machine_function): Add new members call_ms2sysv,
14365 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14366 (struct machine_frame_state): New fields sp_realigned and
14367 sp_realigned_offset.
14368 * config/i386/i386.c
14369 (enum xlogue_stub): New enum.
14370 (enum xlogue_stub_sets): New enum.
14371 (class xlogue_layout): New class.
14372 (struct ix86_frame): New fields stack_realign_allocate_offset,
14373 stack_realign_offset and outlined_save_offset. Modify comments to
14374 detail stack layout when using out-of-line stubs.
14375 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14376 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14377 -mcall-ms2sysv-xlogues.
14378 (stub_managed_regs): New static variable.
14379 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14380 registers managed by out-of-line stub.
14381 (disable_call_ms2sysv_xlogues): New function.
14382 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14383 m->call_ms2sysv when appropriate and compute frame layout for
14385 (sp_valid_at, fp_valid_at): New inline functions.
14386 (choose_basereg): New function.
14387 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14389 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14390 Use align parameter of choose_baseaddr to generated aligned SSE movs
14392 (pro_epilogue_adjust_stack): Modify to track
14393 machine_frame_state::sp_realigned.
14394 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14395 (ix86_nsaved_sseregs): Likewise.
14396 (ix86_emit_save_regs): Likewise.
14397 (ix86_emit_save_regs_using_mov): Likewise.
14398 (ix86_emit_save_sse_regs_using_mov): Likewise.
14399 (get_scratch_register_on_entry): Likewise.
14400 (gen_frame_set): New function.
14401 (gen_frame_load): Likewise.
14402 (gen_frame_store): Likewise.
14403 (emit_outlined_ms2sysv_save): Likewise.
14404 (emit_outlined_ms2sysv_restore): Likewise.
14405 (ix86_expand_prologue): Modify stack re-alignment code and call
14406 emit_outlined_ms2sysv_save when appropriate.
14407 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14408 parameter rtx_insn *insn, which allows the function to be used to only
14409 generate the notes.
14410 (ix86_expand_epilogue): Modify validity checks of frame and stack
14411 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14412 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14413 * config/i386/predicates.md
14414 (save_multiple): New predicate.
14415 (restore_multiple): Likewise.
14416 * config/i386/sse.md
14417 (save_multiple<mode>): New pattern.
14418 (save_multiple_realign<mode>): Likewise.
14419 (restore_multiple<mode>): Likewise.
14420 (restore_multiple_and_return<mode>): Likewise.
14421 (restore_multiple_leave_return<mode>): Likewise.
14422 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14424 2017-05-14 Julia Koval <julia.koval@intel.com>
14426 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14427 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14428 (__builtin_ia32_xsetbv): New builtins.
14429 * config/i386/i386.c (ix86_expand_special_args_builtin):
14431 (ix86_expand_builtin): Special expand for new intrinsics.
14432 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14433 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14434 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14436 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14438 * cfganal.c (inverted_post_order_compute): Change argument type
14440 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14441 * df-core.c (rest_of_handle_df_initialize): Adjust.
14442 (rest_of_handle_df_finish): Likewise.
14443 (df_analyze_1): Likewise.
14444 (df_analyze): Likewise.
14445 (loop_inverted_post_order_compute): Change argument to be a vec *.
14446 (df_analyze_loop): Adjust.
14447 (df_get_n_blocks): Likewise.
14448 (df_get_postorder): Likewise.
14449 * df.h (struct df_d): Change field to be a vec.
14450 * lcm.c (compute_laterin): Adjust.
14451 (compute_available): Likewise.
14452 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14453 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14454 * tree-ssa-pre.c (compute_antic): Likewise.
14456 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14458 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14459 (depth_first_search::depth_first_search): Change structure init
14460 function to this constructor.
14461 (depth_first_search::add_bb): Rename function to this member.
14462 (depth_first_search::execute): Likewise.
14463 (flow_dfs_compute_reverse_finish): Adjust.
14465 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14467 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14468 (longest_simple_path): Likewise.
14469 * shrink-wrap.c (spread_components): Likewise.
14470 (disqualify_problematic_components): Likewise.
14471 (emit_common_heads_for_components): Likewise.
14472 (emit_common_tails_for_components): Likewise.
14473 (insert_prologue_epilogue_for_components): Likewise.
14475 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14477 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14480 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14482 * df-core.c (df_set_blocks): Start using auto_bitmap.
14483 (df_compact_blocks): Likewise.
14484 * df-problems.c (df_rd_confluence_n): Likewise.
14485 * df-scan.c (df_insn_rescan_all): Likewise.
14486 (df_process_deferred_rescans): Likewise.
14487 (df_update_entry_block_defs): Likewise.
14488 (df_update_exit_block_uses): Likewise.
14489 (df_entry_block_bitmap_verify): Likewise.
14490 (df_exit_block_bitmap_verify): Likewise.
14491 (df_scan_verify): Likewise.
14492 * lra-constraints.c (lra_constraints): Likewise.
14493 (undo_optional_reloads): Likewise.
14494 (lra_undo_inheritance): Likewise.
14495 * lra-remat.c (calculate_gen_cands): Likewise.
14496 (do_remat): Likewise.
14497 * lra-spills.c (assign_spill_hard_regs): Likewise.
14498 (spill_pseudos): Likewise.
14499 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14500 (bitmap_set_subtract_values): Likewise.
14502 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14504 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14505 management with auto_bitmap.
14506 (fix_inter_tick): Likewise.
14507 (fix_recovery_deps): Likewise.
14508 * ira.c (add_store_equivs): Likewise.
14509 (find_moveable_pseudos): Likewise.
14510 (split_live_ranges_for_shrink_wrap): Likewise.
14511 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14512 (rtx_reuse_manager::seen_def_p): Likewise.
14513 (rtx_reuse_manager::set_seen_def): Likewise.
14514 * print-rtl.h (class rtx_reuse_manager): Likewise.
14516 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14518 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14520 (migrate_btr_def): Likewise.
14521 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14522 * df-core.c (loop_post_order_compute): Likewise.
14523 (loop_inverted_post_order_compute): Likewise.
14524 * hsa-common.h: Likewise.
14525 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14526 * init-regs.c (initialize_uninitialized_regs): Likewise.
14527 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14528 (inline_small_functions): Likewise.
14529 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14530 * ira.c (combine_and_move_insns): Likewise.
14531 (build_insn_chain): Likewise.
14532 * loop-invariant.c (find_invariants): Likewise.
14533 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14534 * predict.c (tree_predict_by_opcode): Likewise.
14535 (predict_paths_leading_to): Likewise.
14536 (predict_paths_leading_to_edge): Likewise.
14537 (estimate_loops_at_level): Likewise.
14538 (estimate_loops): Likewise.
14539 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14540 (spread_components): Likewise.
14541 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14542 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14543 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14544 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14545 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14546 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14547 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14548 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14549 (mark_threaded_blocks): Likewise.
14550 (thread_through_all_blocks): Likewise.
14551 * tree-ssa.c (verify_ssa): Likewise.
14552 (execute_update_addresses_taken): Likewise.
14553 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14555 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14557 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14559 (post_order_compute): Likewise.
14560 (inverted_post_order_compute): Likewise.
14561 (pre_and_rev_post_order_compute_fn): Likewise.
14563 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14565 * genrecog.c (int_set::int_set): Explicitly construct our
14566 auto_vec base class.
14567 * vec.h (auto_vec::auto_vec): New constructor.
14569 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14571 * bitmap.h (class auto_bitmap): New constructor taking
14572 bitmap_obstack * argument.
14574 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14576 * bitmap.h (class auto_bitmap): Change type of m_bits to
14577 bitmap_head, and adjust ctor / dtor and member operators.
14579 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
14581 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
14582 when returned register mode doesn't match original mode.
14584 2017-05-12 Jeff Law <law@redhat.com>
14585 Jakub Jelinek <jakub@redhat.com>
14587 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
14588 we look for cc setter after the compare-elim changes.
14589 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
14590 within the vector to match what compare-elim now expects.
14591 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
14592 (xorsi3_flags, one_cmplsi2_flags): Likewise.
14594 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
14595 after the compare-elim changes.
14596 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
14597 the vector to match what compare-elim now expects.
14598 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
14599 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
14600 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
14601 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
14602 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
14604 * config/visium/visium.c (single_set_and_flags): Fix where
14605 we look for cc setter after the compare-elim changes.
14606 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
14607 with the vector to match what compare-elim now expects.
14608 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
14609 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
14610 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
14611 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
14612 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
14613 (neg<mode>2_insn_set_overflow): Likewise.
14615 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
14617 PR middle-end/79794
14618 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
14619 maybe_expand_insn call, set ops[0].target. If still set after call,
14620 set alt_rtl. Add extra arg to recursive calls.
14621 (extract_bit_field): Add alt_rtl argument. Pass to
14623 * expmed.h (extract_bit_field): Fix prototype.
14624 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
14625 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
14626 to extract_bit_field_calls.
14627 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
14628 Pass alt_rtl to extract_bit_field calls.
14629 * calls.c (store_unaligned_arguments_into_psuedos)
14630 load_register_parameters): Pass extra NULL to extract_bit_field calls.
14631 * optabs.c (maybe_legitimize_operand): Clear op->target when call
14633 * optabs.h (struct expand_operand): Add target bitfield.
14635 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14637 * compare-elim.c (try_eliminate_compare): Canonicalize
14638 operation with embedded compare to
14639 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
14640 (set (reg) (operation)].
14642 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
14644 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14647 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
14648 cost of adding a carry flag for ADC instruction.
14649 [case MINUS]: Ignore the cost of subtracting a carry flag
14650 for SBB instruction.
14652 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
14654 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
14656 * config/rs6000/bmiintrin.h: New file.
14657 * config/rs6000/bmi2intrin.h: New file.
14658 * config/rs6000/x86intrin.h: New file.
14660 2017-05-12 Jeff Law <law@redhat.com>
14662 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
14665 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
14667 PR middle-end/80707
14668 * tree-cfg.c: Remove cfg edges of unreachable case statements.
14670 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14672 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14673 early expansion of vector divide builtins.
14674 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
14675 builtins identified as having unsigned arguments.
14677 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14679 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
14680 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
14681 expansion of vector logical operations (and, andc, or, xor,
14684 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14686 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
14687 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
14689 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14691 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14692 early GIMPLE expansion of vector multiplies.
14694 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14696 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
14697 TARGET_HAVE_MOVT conditional.
14698 (movt splitter): Likewise.
14700 2017-05-12 Richard Biener <rguenther@suse.de>
14702 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
14703 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14704 Fold all stmts not inplace.
14706 2017-05-12 Richard Biener <rguenther@suse.de>
14708 PR tree-optimization/80713
14709 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
14710 inserted_exprs bit for not removed stmts.
14712 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
14714 PR middle-end/69921
14715 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
14716 parallelized" attribute for parallelized OpenACC kernels.
14717 * omp-offload.c (execute_oacc_device_lower): Use it.
14719 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
14720 Set "oacc kernels" attribute.
14721 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
14722 parameter. Adjust all users.
14723 (oacc_fn_attrib_kernels_p): Remove function.
14724 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
14725 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
14726 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
14727 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14728 assert "oacc kernels" attribute is set.
14730 2017-05-11 Carl Love <cel@us.ibm.com>
14732 * config/rs6000/rs6000-c: Add support for built-in functions
14733 vector unsigned char vec_popcnt (vector signed char)
14734 vector unsigned char vec_popcnt (vector unsigned char)
14735 vector unsigned short vec_popcnt (vector signed short)
14736 vector unsigned short vec_popcnt (vector unsigned short)
14737 vector unsigned int vec_popcnt (vector signed int)
14738 vector unsigned int vec_popcnt (vector unsigned int)
14739 vector unsigned long long vec_popcnt (vector signed long long)
14740 vector unsigned long long vec_popcnt (vector unsigned long long)
14741 vector signed long long vec_slo (vector signed long long,
14742 vector signed char)
14743 vector signed long long vec_slo (vector signed long long,
14744 vector unsigned char)
14745 vector unsigned long long vec_slo (vector unsigned long long,
14746 vector signed char)
14747 vector unsigned long long vec_slo (vector unsigned long long,
14748 vector unsigned char)
14749 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14750 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14751 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14752 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14753 * doc/extend.texi: Update the built-in documentation file for the
14754 new built-in functions.
14756 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14758 * attribs.h (sorted_attr_string): Move machine independent
14759 functions for target clone support from the i386 port to common
14760 code. Rename ix86_function_versions to common_function_versions.
14761 Rename make_name to make_unique_name.
14762 (common_function_versions): Likewise.
14763 (make_unique_name): Likewise.
14764 (make_dispatcher_decl): Likewise.
14765 (is_function_default_version): Likewise.
14766 * attribs.c (attr_strcmp): Likewise.
14767 (sorted_attr_string): Likewise.
14768 (common_function_versions): Likewise.
14769 (make_unique_name): Likewise.
14770 (make_dispatcher_decl): Likewise.
14771 (is_function_default_version): Likewise.
14772 * config/i386/i386.c (attr_strcmp): Likewise.
14773 (sorted_attr_string): Likewise.
14774 (ix86_function_versions): Likewise.
14775 (make_name): Likewise.
14776 (make_dispatcher_decl): Likewise.
14777 (is_function_default_version): Likewise.
14778 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
14780 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14783 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14784 Account for direct move costs for vec_construct of integer
14787 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
14790 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
14791 (UNSPEC_STX_ATOMIC): Ditto.
14792 (loaddi_via_sse): New insn.
14793 (storedi_via_sse): Ditto.
14794 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
14795 Update corresponding peephole2 patterns.
14796 (atomic_storedi_fpu): Ditto.
14798 2017-05-11 Julia Koval <julia.koval@intel.com>
14800 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
14801 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
14803 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
14804 (__builtin_ia32_rsqrt14ss_mask): New builtins.
14805 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
14807 2017-05-11 Nathan Sidwell <nathan@acm.org>
14809 * graphite-poly.c: Include dumpfile.h.
14811 * dumpfle.h (dump_function): Declare here ...
14812 * tree-dump.h (dump_function): ... not here.
14813 * dumpfile.c: #include tree-cfg.h.
14814 (dump_function): Move here from ...
14815 * tree-dump.c (dump_function): ... here.
14816 * gimplify.c: #include splay-tree.h, not tree-dump.h.
14817 * graphite-poly.c: Don't include tree-dump.h.
14818 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
14819 * print-tree.c: Likewise.
14820 * stor-layout.c: Likewise.
14821 * tree-nested.c: Likewise.
14823 * dumpfile.c (dump_start): Use TDF_FLAGS.
14824 (dump_enable_all): Fix TDF_KIND check thinko.
14826 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
14828 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14829 array entries to represent two legal parameterizations of the
14830 overloaded __builtin_cmpb function, as represented by the
14831 P6_OV_BUILTIN_CMPB constant.
14832 (altivec_resolve_overloaded_builtin): Add special case handling
14833 for the __builtin_cmpb function, as represented by the
14834 P6_OV_BUILTIN_CMPB constant.
14835 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
14836 (BU_P6_64BIT_2): New macro.
14837 (BU_P6_OVERLOAD_2): New macro
14838 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
14839 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
14840 (CMPB): Add overload support to represent both 32-bit and 64-bit
14841 compare-bytes function.
14842 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14843 support for TARGET_CMPB.
14844 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
14845 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14846 documentation of the __builtin_cmpb overloaded built-in function.
14848 2017-05-11 Richard Biener <rguenther@suse.de>
14850 PR tree-optimization/80705
14851 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
14852 bases are not vectorizable.
14854 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14856 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
14857 when counting register pressure.
14859 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14861 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
14862 (iv_ca_more_deps): Renamed to ...
14863 (iv_ca_compare_deps): ... this.
14864 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
14866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14868 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
14870 (determine_group_iv_costs): ... here.
14871 (find_inv_vars_cb): Record inv var if it's not recorded before.
14873 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14875 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
14876 (get_shiftadd_cost): Ditto.
14878 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14880 * tree-ssa-address.c: Include header file.
14881 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
14883 (add_to_parts): Refactor.
14884 (addr_to_parts): New parameter. Update use of move_hint_to_base.
14885 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
14888 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14890 PR tree-optimization/53090
14891 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
14893 (extract_cond_operands): Detect condition with IV on both sides
14894 and return COMP_IV_EXPR_2.
14895 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
14896 (rewrite_use_compare): Simplify by removing call to function
14897 extract_cond_operands.
14899 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14901 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
14902 (extract_cond_operands): Detect condition comparing against non-
14903 invariant bound and return appropriate enum value.
14904 (find_interesting_uses_cond): Update use of extract_cond_operands.
14905 Handle its return value accordingly.
14906 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
14908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14910 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
14911 nonlinear iv_use computation in loop invariant sensitive way.
14913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14915 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
14916 (find_iv_candidates): Call relate_compare_use_with_all_cands.
14918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14920 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
14921 (dump_cand): Support iv_cand.inv_exprs.
14922 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
14924 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
14927 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14929 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
14931 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
14932 as local function. Include necessary header files.
14933 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
14935 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14937 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
14939 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14941 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
14942 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
14943 RSHIFT_EXPR and BIT_NOT_EXPR.
14945 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14947 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
14948 (adjust_setup_cost): New parameter supporting round up adjustment.
14949 (struct address_cost_data): Delete.
14950 (force_expr_to_var_cost): Don't bound cost with spill_cost.
14951 (split_address_cost, ptr_difference_cost): Delete.
14952 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
14953 (struct ainc_cost_data): New struct.
14954 (get_address_cost_ainc): New function.
14955 (get_address_cost, get_computation_cost): Reimplement.
14956 (determine_group_iv_cost_address): Record inv_expr for all uses of
14958 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
14959 (iv_ca_has_deps): Reimplemented to ...
14960 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
14962 (iv_ca_extend): Call iv_ca_more_deps.
14964 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14966 * tree-ssa-address.c (struct mem_address): Move to header file.
14967 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
14968 * tree-ssa-address.h (struct mem_address): Move from C file.
14969 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
14971 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14973 * tree-affine.h (aff_combination_type): New interface.
14974 (aff_combination_zero_p): Remove static.
14975 (aff_combination_const_p): New interface.
14976 (aff_combination_singleton_var_p): New interfaces.
14978 2017-05-11 Richard Biener <rguenther@suse.de>
14980 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14981 Skip unreachable blocks and destinations.
14982 (eliminate): Move stmt removal and fixup ...
14983 (fini_eliminate): ... here. Skip inserted exprs.
14984 (pass_pre::execute): Move fini_pre after fini_eliminate.
14985 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
14986 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
14987 PRE to get rid of dead code that has invalid SSA form and
14988 split critical edges again.
14990 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14992 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14994 2017-05-11 Richard Biener <rguenther@suse.de>
14996 * passes.c (execute_function_todo): Verify loops if they are
14997 said to be up-to-date.
14998 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
14999 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15001 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15004 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15005 handle calling assemble_external ourself.
15008 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15009 modes with zero size. Enhance comment.
15011 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15013 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15014 built-ins for vec_xl and vec_xst with short and char pointer
15017 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15019 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15020 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15021 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15022 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15023 (_mm_maskz_min_round_ss): New intrinsics.
15024 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15025 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15026 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15027 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15028 (__builtin_ia32_minss_mask_round): New builtins.
15029 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15030 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15031 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15033 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15034 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15036 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15039 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15041 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15042 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15043 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15044 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15045 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15046 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15047 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15048 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15049 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15050 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15051 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15052 (__builtin_ia32_mulss_mask_round): New builtins.
15053 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15054 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15055 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15057 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15058 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15060 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15063 2017-05-10 Julia Koval <julia.koval@intel.com>
15065 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15066 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15067 (_mm256_setr_m128i): New intrinsics.
15069 2017-05-10 Julia Koval <julia.koval@intel.com>
15071 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15072 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15073 (_mm_maskz_rcp14_ss): New intrinsics.
15074 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15075 (__builtin_ia32_rcp14ss_mask): New builtins.
15076 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15078 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15080 PR tree-optimization/51513
15081 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15082 (assert_unreachable_fallthru_edge_p): Use it.
15083 (group_case_labels_stmt): Likewise.
15084 * tree-cfg.h: Prototype it.
15085 * stmt.c: Include cfghooks.h and tree-cfg.h.
15086 (emit_case_dispatch_table) <gap_label>: New local variable.
15087 Use it to fill dispatch table gaps.
15088 Test for default_label before updating probabilities.
15089 (expand_case) <default_label>: Remove unneeded initialization.
15090 Test for unreachable default case statement and remove its edge.
15091 Set default_label accordingly.
15092 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15094 2017-05-10 Carl Love <cel@us.ibm.com>
15096 * config/rs6000/rs6000-c: Add support for built-in functions
15097 vector signed char vec_neg (vector signed char)
15098 vector signed short int vec_neg (vector short int)
15099 vector signed int vec_neg (vector signed int)
15100 vector signed long long vec_neg (vector signed long long)
15101 vector float vec_neg (vector float)
15102 vector double vec_neg (vector double)
15103 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15105 * config/rs6000/altivec.h: Add define for vec_neg
15106 * doc/extend.texi: Update the built-in documentation for the
15107 new built-in functions.
15109 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15111 PR tree-optimization/77644
15112 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15114 2017-05-10 Nathan Sidwell <nathan@acm.org>
15116 * dumpfile.h (TDI_lang_all): New.
15117 (TDF_KIND): New. Renumber others
15118 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15120 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15122 (get_dump_file_name): Adjust suffix generation.
15123 (dump_enable_all): Use TDF_KIND.
15124 * doc/invoke.texi (-fdump-lang-all): Document.
15126 * dumpfile.h: Tabify.
15128 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15131 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15132 Move member access before delete.
15134 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15136 * tree-inline.c (expand_call_inline): Split block at stmt
15139 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15142 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15143 are now unused after splitting mov{sf,sd}_hardfloat.
15144 (f32_lr2): Likewise.
15145 (f32_lm): Likewise.
15146 (f32_lm2): Likewise.
15147 (f32_li): Likewise.
15148 (f32_li2): Likewise.
15149 (f32_lv): Likewise.
15150 (f32_sr): Likewise.
15151 (f32_sr2): Likewise.
15152 (f32_sm): Likewise.
15153 (f32_sm2): Likewise.
15154 (f32_si): Likewise.
15155 (f32_si2): Likewise.
15156 (f32_sv): Likewise.
15157 (f32_dm): Likewise.
15158 (f32_vsx): Likewise.
15159 (f32_av): Likewise.
15160 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15161 For movsf, order stores so the VSX stores occur before the GPR
15162 store which encourages the register allocator to use a traditional
15163 FPR instead of a GPR. For movsd, order the stores so that the GPR
15164 store comes before the VSX stores to allow the power6 to work.
15165 This is due to the power6 not having a 32-bit integer store
15166 instruction from a FPR.
15167 (movsf_hardfloat): Likewise.
15168 (movsd_hardfloat): Likewise.
15170 2017-05-09 Martin Sebor <msebor@redhat.com>
15172 PR translation/80280
15173 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15176 PR translation/80280
15177 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15178 data member added in r247778.
15179 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15181 2017-05-09 Nathan Sidwell <nathan@acm.org>
15183 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15185 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15188 2017-05-09 Marek Polacek <polacek@redhat.com>
15190 * doc/invoke.texi: Fix typo.
15192 2017-05-09 Richard Biener <rguenther@suse.de>
15194 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15195 (vrp_val_is_min): Likewise.
15196 (set_value_range_to_value): Likewise.
15197 (set_value_range_to_nonnegative): Likewise.
15198 (gimple_assign_nonzero_p): Likewise.
15199 (gimple_stmt_nonzero_p): Likewise.
15200 (vrp_int_const_binop): Likewise. Remove unreachable case.
15201 (adjust_range_with_scev): Adjust comments.
15202 (compare_range_with_value): Likewise.
15203 (extract_range_from_phi_node): Likewise.
15204 (test_for_singularity): Likewise.
15206 2017-05-09 Richard Biener <rguenther@suse.de>
15208 * tree-vrp.c (get_single_symbol): Add assert that we don't
15209 get overflowed constants as invariant part.
15210 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15211 checks. Use wi::cmp instead of recursing for integer constants.
15212 (compare_values): Just ignore whether we assumed undefined
15213 overflow instead of failing the compare.
15214 (extract_range_for_var_from_comparison_expr): Add comment before the
15215 TREE_NO_WARNING sets.
15216 (test_for_singularity): Likewise.
15217 (extract_range_from_comparison): Do not disable optimization
15218 when we assumed undefined overflow.
15219 (extract_range_basic): Remove init of unused var.
15221 2017-05-09 Richard Biener <rguenther@suse.de>
15223 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15224 (extract_range_from_multiplicative_op_1): Adjust.
15225 (extract_range_from_binary_expr_1): Use int_const_binop.
15227 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15230 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15231 rs6000_store_data_bypass_p in seven define_bypass directives and
15232 in several comments.
15233 * config/rs6000/rs6000-protos.h: Add prototype for
15234 rs6000_store_data_bypass_p function.
15235 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15236 function implements slightly different (rs6000-specific) semantics
15237 than store_data_bypass_p, returning false rather than aborting
15238 with assertion error when arguments do not satisfy the
15239 requirements of store data bypass.
15240 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15241 rs6000_store_data_bypass_p.
15243 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15245 * config/xtensa/xtensa-protos.h
15246 (xtensa_initial_elimination_offset): New declaration.
15247 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15248 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15249 macro definition, add case for FRAME_POINTER_REGNUM when
15250 FRAME_GROWS_DOWNWARD.
15251 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15252 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15253 xtensa_initial_elimination_offset.
15255 2017-05-08 Nathan Sidwell <nathan@acm.org>
15257 * doc/invoke.texi: Alphabetize -fdump options.
15259 2017-05-08 Martin Sebor <msebor@redhat.com>
15261 PR translation/80280
15262 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15264 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15266 * target.def (compute_frame_layout): New optional target hook.
15267 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15268 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15269 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15271 * reload1.c (verify_initial_elim_offsets): Likewise.
15272 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15273 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15274 (arm_get_frame_offsets): Split up into this ...
15275 (arm_compute_frame_layout): ... and this function.
15277 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15279 * config/aarch64/constraints.md (Usa): New constraint.
15280 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15282 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15284 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15285 with_multilib_list after it has been checked.
15287 2017-05-08 Richard Biener <rguenther@suse.de>
15289 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15290 (bitmap_set_subtract_values): Likewise.
15292 2017-05-08 Richard Biener <rguenther@suse.de>
15294 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15295 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15297 (gimple_stmt_nonzero_warnv_p): Rename to ...
15298 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15300 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15301 (extract_range_basic): Adjust, do not disable propagation on
15302 strict overflow sensitive simplification.
15303 (vrp_visit_cond_stmt): Likewise.
15305 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15307 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15308 body size unconditionally.
15310 2017-05-07 Jeff Law <law@redhat.com>
15313 2017-05-06 Jeff Law <law@redhat.com>
15314 PR tree-optimization/78496
15315 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15318 PR tree-optimization/78496
15319 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15320 (simplify_stmt_using_ranges): Call it.
15321 (vrp_dom_walker::before_dom_children): Extract equivalences
15322 from an ASSERT_EXPR with an equality comparison against a
15325 2017-05-06 Jeff Law <law@redhat.com>
15327 PR tree-optimization/78496
15328 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15331 PR tree-optimization/78496
15332 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15333 (simplify_stmt_using_ranges): Call it.
15334 (vrp_dom_walker::before_dom_children): Extract equivalences
15335 from an ASSERT_EXPR with an equality comparison against a
15338 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15340 * lra-constraints.c (lra_copy_reg_equiv): New function.
15341 (split_reg): Use it to copy equivalence information from the
15342 original register to the spill register.
15344 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15346 PR rtl-optimization/75964
15347 * simplify-rtx.c (simplify_const_relational_operation): Remove
15348 invalid handling of comparisons of integer ABS.
15350 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15352 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15353 initialize to zero.
15354 (init_regs): Remove declaration.
15355 (function_arg_advance_32): Initialize error_p as boolean variable.
15357 2017-05-05 Nathan Sidwell <nathan@acm.org>
15359 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15360 lines. Use for (;;).
15362 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15364 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15365 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15366 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15367 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15368 VF=2 that require versioning.
15370 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15372 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15375 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15377 * diagnostic.h (diagnostic_override_option_index): Convert from
15378 macro to inline function.
15380 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15382 * diagnostic.c (last_module_changed_p): New function.
15383 (set_last_module): New function.
15384 (diagnostic_report_current_module): Convert macro usage to
15385 the above functions.
15386 * diagnostic.h (diagnostic_context::last_module): Strengthen
15387 from const line_map * to const line_map_ordinary *.
15388 (diagnostic_last_module_changed): Delete macro.
15389 (diagnostic_set_last_module): Delete macro.
15391 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15393 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15394 with diagnostic_report_diagnostic.
15395 (diagnostic_n_impl_richloc): Likewise.
15396 * diagnostic.h (report_diagnostic): Delete macro.
15397 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15398 with diagnostic_report_diagnostic.
15399 * substring-locations.c (format_warning_va): Likewise.
15401 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15403 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15404 save/restor of format_spec. Move option-printing code to...
15405 (print_option_information): ...this new function, and
15406 reimplement by simply printing to the pretty_printer,
15407 rather than appending to the format string.
15409 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15411 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15412 handling logic into...
15413 (update_effective_level_from_pragmas): ...this new function.
15415 2017-05-04 Andrew Waterman <andrew@sifive.com>
15417 * config/riscv/riscv.opt (mstrict-align): New option.
15418 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15419 (SLOW_UNALIGNED_ACCESS): Define.
15420 (riscv_slow_unaligned_access): Declare.
15421 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15423 (riscv_slow_unaligned_access): New variable.
15424 (rocket_tune_info): Set slow_unaligned_access to true.
15425 (optimize_size_tune_info): Set slow_unaligned_access to false.
15426 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15427 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15428 (riscv_option_override): Set riscv_slow_unaligned_access.
15429 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15431 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15433 * config/riscv/riscv.md: Unify indentation.
15435 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15440 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15442 (extendsi<mode>2): Add support for doing sign extension via
15443 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15444 don't have ISA 3.0 instructions.
15445 (extendsi<mode>2 splitter): Likewise.
15446 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15447 generate the normal insns since SImode can now go in vector
15448 registers. Disallow the special UNSPECs needed for previous
15449 machines to hide SImode being used. Add new insns
15450 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15451 (fix_trunc<mode>si2_stfiwx): Likewise.
15452 (fix_trunc<mode>si2_internal): Likewise.
15453 (fixuns_trunc<mode>si2): Likewise.
15454 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15455 (fctiw<u>z_<mode>_smallint): Likewise.
15456 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15457 of floating point to 32-bit integer from doing a direct move to
15458 the GPR registers to do a store.
15459 (fctiwz_<mode>): Break long line.
15461 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15463 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15464 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15465 (addr_list, addr_offset_valid_p): New.
15466 (split_address_groups): Check offset validity with above function.
15467 (gt-tree-ssa-loop-ivopts.h): Include header file.
15469 2017-05-05 Nathan Sidwell <nathan@acm.org>
15471 * config.gcc (arm*-*-*): Add missing 'fi'.
15473 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15475 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15476 included in -fopt-info does not matter.
15477 * doc/optinfo.texi (-fopt-info): Fix description of default
15478 behavour. Explicitly say order of options included in -fopt-info
15481 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15483 * config.gcc: Allow combinations of aprofile and rmprofile values for
15484 --with-multilib-list.
15485 * config/arm/t-multilib: New file.
15486 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15487 variables. Remove setting of ISA and floating-point ABI in
15488 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15489 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15490 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15491 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15493 * config/arm/t-rmprofile: Likewise except for the matches changes.
15494 * doc/install.texi (--with-multilib-list): Document the combination of
15495 aprofile and rmprofile values and warn about pitfalls in doing that.
15497 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15499 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15500 (movdi_aarch64): Likewise.
15502 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15504 PR tree-optimization/80632
15505 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15507 (build_arrays): Initialize it for virtual phis.
15508 (fix_phi_nodes): Use it for virtual phis.
15510 PR tree-optimization/80558
15511 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15512 [x, y] op z into [x op, y op z] for op & or | if conditions
15515 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15516 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15519 * config/arm/arm.md (use_literal_pool): Remove.
15520 (64-bit immediate split): No longer takes cost into consideration
15521 if arm_disable_literal_pool is enabled.
15522 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15523 used when arm_disable_literal_pool is enabled.
15524 (arm_max_const_double_inline_cost): Remove use of
15525 arm_disable_literal_pool.
15526 (push_minipool_fix): Add assert.
15527 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15528 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15529 (no_literal_pool_sf_immediate): New.
15531 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15533 PR tree-optimization/80613
15534 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15535 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15537 2017-05-05 Richard Biener <rguenther@suse.de>
15539 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15541 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15543 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15544 of this flag from insn conditions due to removal from r247495.
15546 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15548 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15550 (arm_early_store_addr_dep_ptr): Likewise.
15551 * config/arm/aarch-common-protos.h
15552 (arm_early_load_addr_dep_ptr): Add prototype.
15553 (arm_early_store_addr_dep_ptr): Likewise.
15554 * config/arm/cortex-a53.md: Add new bypasses.
15556 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15558 * tree.c (next_type_uid): Change type to unsigned.
15559 (type_hash_canon): Decrement back next_type_uid if
15560 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
15561 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
15564 2017-05-04 Martin Sebor <msebor@redhat.com>
15566 * builtins.c: Fix a trivial typo in a comment.
15568 PR middle-end/79234
15569 * builtins.c (check_sizes): Adjust to handle reading past the end.
15570 Avoid printing excessive upper bound of ranges. Use %E to print
15571 tree nodes instead of converting them to %wu.
15572 (expand_builtin_memchr): New function.
15573 (compute_dest_size): Rename...
15574 (compute_objsize): ...to this.
15575 (expand_builtin_memcpy): Adjust.
15576 (expand_builtin_mempcpy): Adjust.
15577 (expand_builtin_strcat): Adjust.
15578 (expand_builtin_strcpy): Adjust.
15579 (check_strncat_sizes): Adjust.
15580 (expand_builtin_strncat): Adjust.
15581 (expand_builtin_strncpy): Adjust and simplify.
15582 (expand_builtin_memset): Adjust.
15583 (expand_builtin_bzero): Adjust.
15584 (expand_builtin_memcmp): Adjust.
15585 (expand_builtin): Handle memcmp.
15586 (maybe_emit_chk_warning): Check strncat just once.
15588 2017-05-04 Martin Sebor <msebor@redhat.com>
15590 PR preprocessor/79214
15591 PR middle-end/79222
15592 PR middle-end/79223
15593 * builtins.c (check_sizes): Add inlining context and issue
15594 warnings even when -Wno-system-headers is set.
15595 (check_strncat_sizes): Same.
15596 (expand_builtin_strncat): Same.
15597 (expand_builtin_memmove): New function.
15598 (expand_builtin_stpncpy): Same.
15599 (expand_builtin): Handle memmove and stpncpy.
15601 2017-05-04 Bin Cheng <bin.cheng@arm.com>
15603 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
15604 which is not used any more.
15606 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15608 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
15610 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15612 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
15613 (cortexa53_tunings): Likewise.
15614 (cortexa57_tunings): Likewise.
15615 (cortexa72_tunings): Likewise.
15616 (cortexa73_tunings): Likewise.
15618 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15620 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
15621 Set loop alignment to 8.
15623 2017-05-04 Martin Sebor <msebor@redhat.com>
15625 PR translation/80280
15626 * builtins.c (expand_builtin_object_size): Add missing quoting to
15627 %D and like directives.
15628 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
15629 (hsa_type_for_tree_type): Same.
15630 (verify_function_arguments): Same.
15631 * symtab.c (symbol_table::change_decl_assembler_name): Same.
15632 * varasm.c (get_section): Same.
15635 2017-05-04 Martin Sebor <msebor@redhat.com>
15637 PR translation/80280
15638 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
15640 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15642 * config/aarch64/aarch64.c (generic_addrcost_table):
15643 Change HI/TI mode setting.
15645 2017-05-04 Martin Jambor <mjambor@suse.cz>
15647 PR tree-optimization/80622
15648 * tree-sra.c (comes_initialized_p): New function.
15649 (build_accesses_from_assign): Only set write lazily when
15650 comes_initialized_p is false.
15651 (analyze_access_subtree): Use comes_initialized_p.
15652 (propagate_subaccesses_across_link): Assert !comes_initialized_p
15653 instead of testing for PARM_DECL.
15655 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15657 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
15658 constraint on operand 0 to allow more general addressing modes.
15659 Adjust output template.
15660 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
15662 * config/aarch64/aarch64-protos.h
15663 (aarch64_address_valid_for_prefetch_p): Declare prototype.
15664 * config/aarch64/constraints.md (Dp): New address constraint.
15665 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
15668 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
15670 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
15671 update use of estimate_ipcp_clone_size_and_time.
15672 (estimate_local_effects): Update use of
15673 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
15674 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
15675 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
15676 Return nonspecialized time.
15678 2017-05-04 Richard Biener <rguenther@suse.de>
15680 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
15681 for the last VUSE which def dominates the PHI. Directly call
15683 (get_continuation_for_phi_1): Remove.
15685 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
15687 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
15688 to explain the use of truncating division. Cap the number of
15689 iterations to the maximum given by nb_iterations_upper_bound,
15692 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15694 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
15695 * configure: Regenerate.
15696 * config.in: Regenerate.
15697 * config/i386/driver-mingw32.c: new file.
15698 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
15699 * config.host: Link driver-mingw32.o on MinGW host.
15700 * doc/install.texi: Document new --enable-mingw-wildcard configure
15703 2017-05-04 Marek Polacek <polacek@redhat.com>
15705 PR tree-optimization/80612
15706 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
15708 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15709 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
15711 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
15712 (movt splitter): Likewise.
15713 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
15714 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
15715 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
15716 block for Thumb-1 with MOVT.
15717 (thumb2_legitimate_address_p): Move code block ...
15718 (can_avoid_literal_pool_for_label_p): ... into this new function.
15719 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
15721 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
15722 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
15723 "M-profile targets with the MOVT instruction".
15725 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15727 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15728 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15729 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15731 2017-05-04 Martin Liska <mliska@suse.cz>
15733 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15734 variable cond_code.
15736 2017-05-04 Richard Biener <rguenther@suse.de>
15738 * tree.c (array_at_struct_end_p): Handle arrays at struct
15739 end with flexarrays more conservatively. Refactor and treat
15740 arrays of arrays or aggregates more strict. Fix
15741 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15742 * tree.c (array_at_struct_end_p): Adjust prototype.
15743 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15744 * gimple-fold.c (get_range_strlen): Likewise.
15745 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15747 2017-05-04 Richard Biener <rguenther@suse.de>
15749 PR tree-optimization/31130
15750 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15752 (supports_overflow_infinity): Likewise.
15753 (is_negative_overflow_infinity): Likewise.
15754 (is_positive_overflow_infinity): Likewise.
15755 (is_overflow_infinity): Likewise.
15756 (stmt_overflow_infinity): Likewise.
15757 (overflow_infinity_range_p): Likewise.
15758 (usable_range_p): Remove as always returning true.
15759 (make_overflow_infinity): Remove.
15760 (negative_overflow_infinity): Likewise.
15761 (positive_overflow_infinity): Likewise.
15762 (avoid_overflow_infinity): Likewise.
15763 (set_value_range): Adjust accordingly.
15764 (set_value_range_to_nonnegative): Likewise, remove now unused
15765 overflow_infinity arg.
15766 (vrp_operand_equal_p): Adjust.
15767 (update_value_range): Likewise.
15768 (range_int_cst_singleton_p): Likewise.
15769 (operand_less_p): Likewise.
15770 (compare_values_warnv): Likewise.
15771 (extract_range_for_var_from_comparison_expr): Likewise.
15772 (vrp_int_const_binop): Likewise.
15773 (zero_nonzero_bits_from_vr): Likewise.
15774 (extract_range_from_multiplicative_op_1): Likewise.
15775 (extract_range_from_binary_expr_1): Likewise.
15776 (extract_range_from_unary_expr): Likewise.
15777 (extract_range_from_comparison): Likewise.
15778 (extract_range_basic): Likewise.
15779 (adjust_range_with_scev): Likewise.
15780 (compare_ranges): Likewise.
15781 (compare_range_with_value): Likewise.
15782 (dump_value_range): Likewise.
15783 (test_for_singularity): Likewise, remove strict_overflow_p parameter
15785 (simplify_cond_using_ranges): Adjust.
15787 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15789 * brig-builtins.def: Added a builtin for class_f64.
15790 * builtin-types.def: Added a builtin type needed by class_f64.
15792 2017-05-03 Jason Merrill <jason@redhat.com>
15794 * timevar.def: Add TV_CONSTEXPR.
15796 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15798 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
15800 2017-05-03 Martin Jambor <mjambor@suse.cz>
15802 * ipa-prop.c (ipa_update_after_lto_read): Removed.
15803 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
15804 * ipa-cp.c (ipcp_propagate_stage): Do not call
15805 ipa_update_after_lto_read.
15806 * ipa-inline.c (ipa_inline): Likewise.
15808 2017-05-03 Martin Jambor <mjambor@suse.cz>
15810 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
15811 tag. Added a default constructor and a destructor.
15812 (ipa_edge_args_sum_t): New class;
15813 (ipa_edge_args_sum): Declare.
15814 (ipa_edge_args_vector): Remove declaration.
15815 (IPA_EDGE_REF): Use ipa_edge_args_sum.
15816 (ipa_free_edge_args_substructures): Remove declaration.
15817 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
15818 (ipa_edge_args_info_available_for_edge_p): Likewise.
15819 * ipa-prop.c (ipa_edge_args_vector): Removed.
15820 (edge_removal_hook_holder): Likewise.
15821 (edge_duplication_hook_holder): Likewise.
15822 (ipa_edge_args_sum): New variable.
15823 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
15824 ipa_edge_args_vector.
15825 (ipa_free_edge_args_substructures): Likewise.
15826 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
15827 ipa_edge_args_vector.
15828 (ipa_edge_removal_hook): Turned into method
15829 ipa_edge_args_sum_t::remove.
15830 (ipa_edge_duplication_hook): Turned into method
15831 ipa_edge_args_sum_t::duplicate.
15832 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
15833 registering edge hooks.
15834 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
15835 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
15836 ipa_edge_args_sum instead of ipa_edge_args_vector.
15837 * ipa-profile.c (ipa_profile): Likewise.
15839 2017-05-03 Martin Jambor <mjambor@suse.cz>
15841 * symbol-summary.h (function_summary): New method exists.
15842 (function_summary::symtab_removal): Deallocate through release.
15843 (call_summary): New class.
15844 (gt_ggc_mx): New overload.
15845 (gt_pch_nx): Likewise.
15846 (gt_pch_nx): Likewise.
15848 2017-05-03 Jeff Law <law@redhat.com>
15850 PR tree-optimization/78496
15851 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
15852 from simplify_cond_using_ranges. Split off code to walk
15853 backwards through casts into ...
15854 (simplify_cond_using_ranges_2): New function.
15855 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
15856 (execute_vrp): After identifying jump threads, call
15857 simplify_cond_using_ranges_2.
15859 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
15862 * ipa-inline.h (inline_summary): Add ctor.
15863 (create_ggc): Do not use ggc_cleared_alloc.
15865 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
15866 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15868 * gcc.c (handle_braces): Support escaping in switch matching
15870 * doc/invoke.texi (Spec Files): Document it.
15871 Remove superfluous @code markup in items.
15873 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15875 * diagnostic-show-locus.c (struct column_range): New struct.
15876 (get_affected_columns): New function.
15877 (get_printed_columns): New function.
15878 (struct correction): New struct.
15879 (correction::ensure_capacity): New function.
15880 (correction::ensure_terminated): New function.
15881 (struct line_corrections): New struct.
15882 (line_corrections::~line_corrections): New dtor.
15883 (line_corrections::add_hint): New function.
15884 (layout::print_trailing_fixits): Reimplement in terms of the new
15886 (selftest::test_overlapped_fixit_printing): New function.
15887 (selftest::diagnostic_show_locus_c_tests): Call it.
15889 2017-05-03 Nathan Sidwell <nathan@acm.org>
15891 Canonicalize canonical type hashing
15892 * tree.h (type_hash_canon_hash): Declare.
15893 * tree.c (type_hash_list, attribute_hash_list): Move into
15894 type_hash_canon_hash.
15895 (build_type_attribute_qual_variant): Break out hash code calc into
15896 type_hash_canon_hash.
15897 (type_hash_canon_hash): New. Generic type hash computation.
15898 (build_range_type_1, build_array_type_1, build_function_type,
15899 build_method_type_directly, build_offset_type, build_complex_type,
15900 make_vector_type): Call it.
15902 2017-05-03 Richard Biener <rguenther@suse.de>
15904 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15905 When all DRs have unknown misaligned do not always peel
15906 when there is a store but apply the same costing model as if
15907 there were only loads.
15909 2017-05-03 Richard Biener <rguenther@suse.de>
15912 PR tree-optimization/80492
15913 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15914 compare_base_decls returning dont-know properly.
15916 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
15918 * config/arm/iterators.md (CCSI): New mode iterator.
15919 (arch): New mode attribute.
15920 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
15921 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
15922 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
15923 code iterator for success result mode.
15924 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
15925 the corresponding new insn generators.
15927 2017-05-03 Bin Cheng <bin.cheng@arm.com>
15930 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15931 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15933 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
15935 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
15936 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
15937 (DDR_A): Wrap DDR argument in brackets.
15938 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
15939 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
15940 (DDR_REVERSED_P): Likewise.
15942 2017-05-03 Jakub Jelinek <jakub@redhat.com>
15944 PR tree-optimization/79472
15945 * tree-switch-conversion.c (struct switch_conv_info): Add
15946 contiguous_range and default_case_nonstandard fields.
15947 (collect_switch_conv_info): Compute contiguous_range and
15948 default_case_nonstandard fields, don't clear final_bb if
15949 contiguous_range and only the default case doesn't have the required
15951 (check_all_empty_except_final): Set default_case_nonstandard instead
15952 of failing if contiguous_range and the default case doesn't have empty
15954 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
15955 and only the default case doesn't have the required constants. Skip
15957 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
15958 if default_case_nonstandard.
15959 (build_constructors): Build constant 1 just once. Assert that default
15960 values aren't inserted in between cases if contiguous_range. Skip
15962 (build_arrays): Skip virtual phis.
15963 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
15964 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
15965 Handle virtual phis.
15966 (gen_inbound_check): Handle default_case_nonstandard case.
15967 (process_switch): Adjust check_final_bb caller. Call
15968 gather_default_values with the first non-default case instead of
15969 default case if default_case_nonstandard.
15971 2017-05-02 Nathan Sidwell <nathan@acm.org>
15973 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
15974 check. Fix formatting.
15976 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
15978 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
15979 errors when comparing specialized and unspecialized times.
15981 2017-05-02 David Malcolm <dmalcolm@redhat.com>
15983 * diagnostic-show-locus.c
15984 (layout::should_print_annotation_line_p): Make private.
15985 (layout::print_annotation_line): Make private.
15986 (layout::annotation_line_showed_range_p): Make private.
15987 (layout::show_ruler): Make private.
15988 (layout::print_source_line): Make private. Pass in line and
15989 line_width, rather than calling location_get_source_line. Drop
15991 (layout::print_leading_fixits): New method.
15992 (layout::print_any_fixits): Rename to...
15993 (layout::print_trailing_fixits): ...this, and make private.
15994 Don't print newline fixits.
15995 (diagnostic_show_locus): Move logic for printing one row into...
15996 (layout::print_line): ...this new function. Move the
15997 location_get_source_line call and error-handling from
15998 print_source_line to here. Call print_leading_fixits, and rename
15999 print_any_fixits to print_trailing_fixits.
16000 (selftest::test_fixit_insert_containing_newline): Update now that
16001 newlines are partially supported.
16002 (selftest::test_fixit_insert_containing_newline_2): New test.
16003 (selftest::test_fixit_replace_containing_newline): Update comments.
16004 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16005 * edit-context.c (class added_line): New class.
16006 (class edited_line): Describe newline handling in comment.
16007 (edited_line::actually_edited_p): New method.
16008 (edited_line::print_content): Delete redundant decl.
16009 (edited_line::m_predecessors): New field.
16010 (edited_file::print_content): Call edited_line::print_content.
16011 (edited_file::print_diff): Update to support newlines.
16012 (edited_file::print_diff_hunk): Likewise.
16013 (edited_file::print_run_of_changed_lines): New function.
16014 (edited_file::print_diff_line): Convert to...
16015 (print_diff_line): ...this.
16016 (edited_file::get_effective_line_count): New function.
16017 (edited_line::edited_line): Initialize new field m_predecessors.
16018 (edited_line::~edited_line): Clean up m_predecessors.
16019 (edited_line::apply_fixit): Handle newlines.
16020 (edited_line::get_effective_line_count): New function.
16021 (edited_line::print_content): New function.
16022 (edited_line::print_diff_lines): New function.
16023 (selftest::test_applying_fixits_insert_containing_newline): New
16025 (selftest::test_applying_fixits_replace_containing_newline): New
16027 (selftest::insert_line): New function.
16028 (selftest::test_applying_fixits_multiple_lines): Add example of
16030 (selftest::edit_context_c_tests): Call the new tests.
16032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16034 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16035 parameter cand. Update dump information.
16036 (get_computation_cost): Update uses.
16038 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16040 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16041 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16042 (get_computation_at, rewrite_use_address): Update use of
16043 get_computation_aff.
16045 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16047 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16048 (get_computation): Delete.
16049 (get_computation_cost): Implement like get_computation_cost_at.
16050 Use get_computation_at.
16051 (get_computation_cost_at): Delete.
16052 (rewrite_use_nonlinear_expr): Use get_computation_at.
16053 (rewrite_use_compare, remove_unused_ivs): Ditto.
16055 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16057 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16059 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16061 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16062 (ivopts_global_cost_for_size): Rename parameter and update uses.
16063 (iv_ca_recount_cost): Update uses.
16064 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16065 candidates seperately in n_invs and n_cands.
16066 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16068 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16070 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16071 (find_inv_vars_cb): New.
16072 (find_depends): Renamed to ...
16073 (find_inv_vars): ... this.
16074 (add_candidate_1, force_var_cost): Call find_inv_vars.
16075 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16077 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16079 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16080 inv_vars. Add inv_exprs.
16081 (struct iv_cand): Rename depends_on to inv_vars.
16082 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16083 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16084 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16085 (dump_cand): Dump inv_vars.
16086 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16087 (record_invariant, find_depends, add_candidate_1): Ditto.
16088 (set_group_iv_cost, force_var_cost): Ditto.
16089 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16090 (get_computation_cost_at, get_computation_cost): Ditto.
16091 (determine_group_iv_cost_generic): Ditto.
16092 (determine_group_iv_cost_address): Ditto.
16093 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16094 (determine_group_iv_costs): Ditto.
16095 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16096 (iv_ca_set_remove_invariants): Renamed to ...
16097 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16098 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16099 (iv_ca_set_add_invariants): Renamed to ...
16100 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16101 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16102 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16103 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16104 (create_new_ivs): Remove useless dump.
16106 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16108 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16110 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16111 (iv_ca_set_no_cp, create_new_iv): Ditto.
16113 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16115 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16117 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16119 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16120 function tree_check2.
16122 2017-05-02 Martin Liska <mliska@suse.cz>
16124 * doc/gcov.texi: Add missing preposition.
16125 * gcov.c (function_info::function_info): Properly fill up
16126 all member variables.
16128 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16130 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16132 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16134 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16136 2017-05-02 Martin Liska <mliska@suse.cz>
16139 * lto-streamer-in.c (lto_read_tree_1): Remove
16140 LTO_STREAMER_DEBUG.
16141 * lto-streamer.c (struct tree_hash_entry): Likewise.
16142 (struct tree_entry_hasher): Likewise.
16143 (tree_entry_hasher::hash): Likewise.
16144 (tree_entry_hasher::equal): Likewise.
16145 (lto_streamer_init): Likewise.
16146 (lto_orig_address_map): Likewise.
16147 (lto_orig_address_get): Likewise.
16148 (lto_orig_address_remove): Likewise.
16149 * lto-streamer.h: Likewise.
16150 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16151 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16153 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16155 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16156 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16157 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16158 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16159 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16160 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16161 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16162 (mm_maskz_sub_ss): New intrinsics.
16163 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16164 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16165 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16166 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16167 (__builtin_ia32_subss_mask_round): New builtins.
16168 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16169 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16170 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16172 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16173 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16175 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16178 2017-05-02 Martin Jambor <mjambor@suse.cz>
16180 PR tree-optimization/78687
16181 * tree-sra.c (access): New field parent.
16182 (process_subtree_disqualification): New function.
16183 (disqualify_candidate): Call it.
16184 (build_accesses_from_assign): Reset write flag if creating an
16186 (build_access_subtree): Fill in parent field and also prpagate
16187 down grp_write flag.
16188 (create_artificial_child_access): New parameter set_grp_write, set
16189 grp_write to its value.
16190 (propagate_subaccesses_across_link): Also propagate grp_write flag
16192 (propagate_all_subaccesses): Push the closest parent back to work
16193 queue if add_access_to_work_queue returned true.
16195 2017-05-02 Richard Biener <rguenther@suse.de>
16197 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16198 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16199 -fstrict-overflow documentation.
16200 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16201 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16202 flag_strict_overflow.
16203 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16204 * lto-opts.c (lto_write_options): Do not stream it.
16205 * lto-wrapper.c (merge_and_complain): Do not handle it.
16206 * opts.c (default_options_table): Do not set -fstrict-overflow.
16207 (finish_options): Likewise do not clear it when sanitizing.
16208 * simplify-rtx.c (simplify_const_relational_operation): Do not
16209 test flag_strict_overflow.
16211 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16213 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16214 using enabled attribute.
16215 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16216 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16217 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16218 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16219 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16220 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16221 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16222 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16223 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16224 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16226 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16228 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16230 2017-05-02 Richard Biener <rguenther@suse.de>
16232 PR tree-optimization/80591
16234 2017-04-10 Richard Biener <rguenther@suse.de>
16236 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16239 2017-05-02 Richard Biener <rguenther@suse.de>
16241 PR tree-optimization/80549
16242 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16243 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16244 headers if they do not have a preheader.
16246 2017-05-02 Martin Liska <mliska@suse.cz>
16249 * common.opt: Fix typo.
16250 * doc/invoke.texi: Likewise.
16252 2017-05-01 Jan Beulich <jbeulich@suse.com>
16254 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16255 swapping, add (x,x,m,x,n) alternative.
16257 2017-05-01 Nathan Sidwell <nathan@acm.org>
16259 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16260 unnecessary unadjusted_alignment check.
16262 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16265 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16267 * gimplify.c (gimplify_cilk_detach): New function.
16268 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16269 * tree-core.h: Document EXPR_CILK_SPAWN.
16270 * tree.h (EXPR_CILK_SPAWN): Define.
16272 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16274 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16275 to use new fixit_hint representation, using the "replace" logic.
16276 (get_line_span_for_fixit_hint): Likewise.
16277 (layout::print_any_fixits): Likewise.
16278 (selftest::test_one_liner_many_fixits): Rename to...
16279 (selftest::test_one_liner_many_fixits_1): ...this, and update
16280 comment and expected output to reflect that the multiple fix-it
16281 hints are now consolidated into one insertion.
16282 (selftest::test_one_liner_many_fixits_2): New test.
16283 (selftest::test_diagnostic_show_locus_one_liner): Update for
16285 (selftest::test_fixit_consolidation): Update for fix-it API
16287 * diagnostic.c (print_parseable_fixits): Likewise.
16288 * edit-context.c (edited_line::m_line_events): Convert from
16289 auto_vec <line_event *> to auto_vec <line_event>.
16290 (class line_event): Convert from abstract base class to a concrete
16291 class, taking over the role of replace_event.
16292 (class insert_event): Delete.
16293 (class replace_event): Rename to class line_event. Convert to
16295 (edit_context::add_fixits): Reimplement.
16296 (edit_context::apply_insert): Delete.
16297 (edit_context::apply_replace): Rename to...
16298 (edit_context::apply_fixit): ...this. Convert to half-open range.
16299 (edited_file::apply_insert): Delete.
16300 (edited_file::apply_replace): Rename to...
16301 (edited_file::apply_fixit): ...this.
16302 (edited_line::~edited_line): Drop deletion of events.
16303 (edited_line::apply_insert): Delete.
16304 (edited_line::apply_replace): Rename to...
16305 (edited_line::apply_fixit): ...this. Convert to half-open range.
16306 Update for change to type of m_line_events.
16307 * edit-context.h (edit_context::apply_insert): Delete.
16308 (edit_context::apply_replace): Rename to...
16309 (edit_context::apply_fixit): ...this.
16311 2017-05-01 Martin Sebor <msebor@redhat.com>
16313 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16316 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16319 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16320 __get_cpuid_max returns 0.
16321 (__get_cpuid_count): Ditto.
16323 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16325 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16326 replacement expression is another instance of one of its arguments.
16328 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16331 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16332 check for stack push/pop autoinc.
16333 * config/i386/i386.c (ix86_agi_dependent): Return false
16334 if the only reason why modified_in_p returned true is that
16335 addr is SP based and set_insn is a push or pop.
16337 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16339 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16342 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16345 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16346 (account_size_time): Use two predicates - exec_pred and
16348 (evaluate_conditions_for_known_args): Compute both clause and
16350 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16351 (inline_summary_t::duplicate): Update.
16352 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16354 (compute_inline_parameters): Likewise.
16355 (estimate_edge_size_and_time): Update caluclation of time.
16356 (estimate_node_size_and_time): Compute both time and nonspecialized
16358 (estimate_ipcp_clone_size_and_time): Update.
16359 (inline_merge_summary): Update.
16360 (do_estimate_edge_time): Update.
16361 (do_estimate_edge_size): Update.
16362 (do_estimate_edge_hints): Update.
16363 (inline_read_section, inline_write_summary): Stream both new predicates.
16364 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16366 (compute_inlined_call_time): Cleanup.
16367 (big_speedup_p): Update.
16368 (edge_badness): Update.
16369 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16370 (size_time_entry): Replace predicate by exec_predicate and
16371 nonconst_predicate.
16372 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16373 (estimate_edge_time): Return also nonspec_time.
16374 (reset_edge_growth_cache): Update.
16376 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16378 PR rtl-optimization/80491
16379 * ifcvt.c (noce_process_if_block): When looking for x setter
16380 with missing else_bb, don't check only the insn right before
16381 cond_earliest, but look for the last insn that x is modified in
16382 within the same bb.
16384 PR rtl-optimization/80491
16385 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16387 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16389 PR tree-optimization/80487
16390 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16392 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16394 PR tree-optimization/79697
16395 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16396 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16397 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16399 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16400 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16402 2017-04-28 Martin Sebor <msebor@redhat.com>
16404 PR tree-optimization/80523
16405 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16406 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16408 (maybe_warn, format_directive, parse_directive): Use new functions.
16409 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16411 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16413 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16415 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16417 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16418 target_header_dir): Set correctly.
16419 * configure: Regenerated.
16420 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16421 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16422 instead of SYSTEM_HEADER_DIR.
16424 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16426 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16427 (estimate_local_effects): Likewise.
16428 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16429 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16430 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16431 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16432 do_estimate_edge_time, estimate_edge_time): Likewise.
16433 * ipa-inline-analysis.c (estimate_node_size_and_time,
16434 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16435 (estimate_time_after_inlining): Remove.
16437 2017-04-28 Martin Liska <mliska@suse.cz>
16439 * doc/gcov.texi: Enhance documentation of gcov.
16441 2017-04-28 Martin Liska <mliska@suse.cz>
16443 * doc/gcov.texi: Sort options in alphabetic order.
16444 * doc/gcov-dump.texi: Likewise.
16445 * doc/gcov-tool.texi: Likewise.
16446 * gcov.c (print_usage): Likewise.
16447 * gcov-dump.c (print_usage): Likewise.
16448 * gcov-tool.c (print_merge_usage_message): Likewise.
16449 (print_rewrite_usage_message): Likewise.
16450 (print_overlap_usage_message): Likewise.
16452 2017-04-28 Martin Liska <mliska@suse.cz>
16454 PR gcov-profile/53915
16455 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16457 2017-04-28 Martin Liska <mliska@suse.cz>
16459 PR gcov-profile/79891
16460 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16461 is marked by compiler as living on a line.
16462 (get_cycles_count): Remove usage of the union.
16463 (output_intermediate_file): Likewise.
16464 (find_source): Fix GNU coding style.
16465 (accumulate_line_counts): Remove old non-all block mode.
16466 (output_lines): Remove usage of the union.
16467 * profile.c (output_location): Include all BBs, even if
16468 belonging to a same line (and file) as a previous BB.
16470 2017-04-28 Martin Liska <mliska@suse.cz>
16472 * gcov.c (process_args): Handle new argument 'w'.
16473 (read_graph_file): Assign ID to BBs.
16474 (output_branch_count): Display BB # if verbose flag is set.
16475 (output_lines): Likewise for arcs.
16476 (print_usage): Add '--verbose' option help.
16477 * doc/gcov.texi: Document --verbose (-w) option.
16479 2017-04-28 Martin Liska <mliska@suse.cz>
16481 * gcov.c (struct block_location_info): New struct.
16482 (process_file): Fill up the new structure.
16483 (read_graph_file): Replace usage of encoding by the newly added
16485 (add_line_counts): Likewise.
16486 (accumulate_line_counts): Remove usage of the union.
16487 (function_info::function_info): New function.
16488 (function_info::~function_info): Likewise.
16489 (process_file): Call delete instead of release_function.
16490 (release_function): Release the function.
16491 (release_structures): Call delete instead of release_function.
16492 (solve_flow_graph): Replace usage of num_blocks.
16493 (find_exception_blocks): Likewise.
16494 (output_lines): Fix GNU coding style.
16496 2017-04-28 Martin Liska <mliska@suse.cz>
16499 * coverage.c (coverage_remove_note_file): New function.
16500 * coverage.h: Declare the function.
16501 * toplev.c (finalize): Clean if an error has been seen.
16503 2017-04-28 Martin Liska <mliska@suse.cz>
16505 PR gcov-profile/80031
16506 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16507 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16508 * gcov.c (read_graph_file): Read just number of blocks.
16509 * profile.c (branch_prob): Do not stream 0 flags per a basic
16512 2017-04-28 Martin Liska <mliska@suse.cz>
16514 * gcov-dump.c (tag_*): Add new argument to declarations.
16515 (dump_gcov_file): Likewise.
16516 (tag_blocks): Add and use new argument depth.
16517 (tag_arcs): Likewise.
16518 (tag_lines): Likewise.
16519 (tag_counters): Likewise.
16520 (tag_summary): Likewise.
16521 (dump_working_sets): Use depth to do a proper indentation.
16523 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16526 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16527 * symtab.c (symtab_node::debug_symtab): Move definition here.
16529 2017-04-28 Richard Biener <rguenther@suse.de>
16531 * lto-streamer.h (LTO_major_version): Bump to 7.
16533 2017-04-28 Richard Biener <rguenther@suse.de>
16535 * tree-vrp.c (assert_info): New struct.
16536 (add_assert_info): New helper.
16537 (register_edge_assert_for_2): Refactor to add asserts to a vector
16539 (register_edge_assert_for_1): Likewise.
16540 (register_edge_assert_for): Likewise.
16541 (finish_register_edge_assert_for): New helper actually registering
16542 asserts where live on edge.
16543 (find_conditional_asserts): Adjust.
16544 (find_switch_asserts): Likewise.
16545 (evrp_dom_walker::try_find_new_range): Generalize.
16546 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16548 2017-04-27 Marek Polacek <polacek@redhat.com>
16551 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16552 arg10 and arg11 to itype.
16554 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16556 * doc/extend.texi (Object Size Checking): Improve grammar.
16558 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
16561 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
16562 that the logic for permitting reciprocal estimates matches that
16565 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16568 * tree.c (type_cache_hasher::equal): Only compare
16569 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
16570 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
16571 non-aggregate element types.
16572 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
16573 about the flag on ARRAY_TYPEs in the comment, formatting fix.
16575 2017-04-27 Richard Biener <rguenther@suse.de>
16577 PR middle-end/80533
16578 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
16579 stripping ARRAY_REFs from MEM_EXPR make sure we're not
16580 keeping a reference to a trailing array.
16582 2017-04-27 Richard Biener <rguenther@suse.de>
16584 PR middle-end/80539
16585 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
16586 being in loop-closed SSA form conservatively.
16587 (chrec_fold_multiply_poly_poly): Likewise.
16589 2017-04-27 Tamar Christina <tamar.christina@arm.com>
16591 PR middle-end/79665
16592 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
16593 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
16595 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16598 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
16599 (aarch64_function_arg_alignment): Return unsigned int again, but still
16600 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
16601 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16602 Don't emit -Wpsabi note.
16603 (aarch64_function_arg_boundary): Likewise.
16604 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16607 2017-04-26 Nathan Sidwell <nathan@acm.org>
16609 * tree.h (crc32_unsigned_n): Declare.
16610 (crc32_unsigned, crc32_unsigned): Make inline.
16611 * tree.c (crc32_unsigned_bits): Replace with ...
16612 (crc32_unsigned_n): ... this.
16613 (crc32_unsigned, crc32_byte): Remove.
16614 (crc32_string): Remove unnecessary braces.
16616 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16618 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
16619 * ipa-inline-analysis.c (MAX_TIME): Remove.
16620 (account_size_time): Use sreal for time.
16621 (dump_inline_summary): Update.
16622 (estimate_function_body_sizes): Update.
16623 (estimate_edge_size_and_time): Update.
16624 (estimate_calls_size_and_time): Update.
16625 (estimate_node_size_and_time): Update.
16626 (inline_merge_summary): Update.
16627 (inline_update_overall_summary): Update.
16628 (estimate_time_after_inlining): Update.
16629 (inline_read_section): Update.
16630 (inline_write_summary): Update.
16631 * ipa-inline.c (compute_uninlined_call_time): Update.
16632 (compute_inlined_call_time): Update.
16633 (recursive_inlining): Update.
16634 (inline_small_functions): Update.
16635 (dump_overall_stats): Update.
16636 * ipa-inline.h: Include sreal.h.
16637 (size_time_entry): Turn time to sreal.
16638 (inline_summary): Turn self_time nad time to sreal.
16640 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16642 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
16644 (sreal::stream_out, sreal::stream_in): New.
16645 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
16647 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16649 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
16652 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
16655 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
16657 (dimode_scalar_chain::compute_convert_gain): Ditto.
16658 (dimode_scalar_chain::make_vector_copies): Ditto.
16659 (dimode_scalar_chain::convert_reg): Ditto.
16660 (dimode_scalar_chain::convert_insn): Ditto.
16661 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
16662 (VI248_AVX512BW_1): New mode iterator.
16663 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
16664 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
16667 2017-04-25 Martin Sebor <msebor@redhat.com>
16669 PR tree-optimization/80497
16670 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
16671 constants are representable in HOST_WIDE_INT.
16672 (parse_directive): Ditto.
16674 2017-04-25 Martin Sebor <msebor@redhat.com>
16677 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
16678 (new_zero_array): Adjust signature.
16679 (dom_info::dom_init): Used unsigned rather that size_t.
16680 (dom_info::dom_info): Same.
16682 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16683 Jakub Jelinek <jakub@redhat.com>
16686 * config/arm/arm.c: Include gimple.h.
16687 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16688 returns negative, increment ncrn only if it returned positive.
16689 (arm_needs_doubleword_align): Return int instead of bool,
16690 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
16691 members, but if there is any such non-FIELD_DECL
16692 > PARM_BOUNDARY aligned decl, return -1 instead of false.
16693 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16694 returns negative, increment nregs only if it returned positive.
16695 (arm_setup_incoming_varargs): Likewise.
16696 (arm_function_arg_boundary): Emit -Wpsabi note if
16697 arm_needs_doubleword_align returns negative, return
16698 DOUBLEWORD_ALIGNMENT only if it returned positive.
16700 2017-04-25 Marek Polacek <polacek@redhat.com>
16703 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16704 first argument to type.
16706 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
16709 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
16710 type checks to test for compatibility instead of equality.
16712 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16713 Jakub Jelinek <jakub@redhat.com>
16716 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
16718 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
16719 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
16720 the alignment computation, but return their maximum in warn_alignment.
16721 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16722 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
16724 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
16725 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16730 * config/arc/simdext.md (dmpyh): Fix typo.
16732 2017-04-25 Richard Biener <rguenther@suse.de>
16734 PR tree-optimization/80492
16735 * alias.c (compare_base_decls): Handle registers with asm
16736 specification conservatively.
16737 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16738 compare_base_decls returning dont-know properly.
16740 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16742 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16743 (legitimate_offset_address_p): New function.
16744 (arc_legitimate_address_p): Use above function.
16746 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16748 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16750 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16752 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16753 ACCH registers whenever they are available.
16755 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16757 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16758 double regs fix when not used.
16760 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16762 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
16764 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
16765 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
16767 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16769 * config/arc/arc.c (arc_output_addsi): Check for h-register class
16770 when emitting short ADD instructions.
16772 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16774 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
16776 (cmpsi_cc_c_insn): Likewise.
16777 (cbranchsi4_scratch): Compute proper instruction length using
16778 compact_hreg_operand.
16779 * config/arc/predicates.md (compact_hreg_operand): New predicate.
16781 2017-04-25 Richard Biener <rguenther@suse.de>
16783 PR middle-end/80509
16784 * passes.c (pass_manager::pass_manager): Initialize
16785 m_name_to_pass_map.
16787 2017-04-25 Richard Biener <rguenther@suse.de>
16789 PR tree-optimization/79201
16790 * tree-ssa-sink.c (statement_sink_location): Handle calls.
16792 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16795 * config/s390/vector.md: Split MEM->GPR vector moves for
16796 non-s_operand addresses.
16798 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16801 * config/s390/predicates.md (reload_const_wide_int_operand): New
16803 * config/s390/s390.md ("movti"): Remove d/P alternative.
16804 ("movti_bigconst"): New pattern definition.
16806 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16809 * s390-protos.h (s390_expand_cs_hqi): Removed.
16810 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
16811 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
16812 modes as well as CCZ1mode and CCZmode.
16813 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
16814 signature of s390_emit_compare_and_swap.
16815 (s390_expand_cs_hqi): Likewise, make static.
16816 (s390_expand_cs_tdsi): Generate an explicit compare before trying
16817 compare-and-swap, in some cases.
16818 (s390_expand_cs): Wrapper function.
16819 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
16821 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
16822 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
16823 patterns for small and large integers. Forbid symref memory operands.
16824 Move expander to s390.c. Require cc register.
16825 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
16826 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
16827 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
16828 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
16829 symref memory operands. Remove CC mode and call s390_match_ccmode
16831 ("atomic_exchange<mode>"): Allow and implement all integer modes.
16833 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16835 * config/s390/s390.md (define_peephole2): New peephole to help
16836 combining the load-and-test pattern with volatile memory.
16838 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16840 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
16841 with CCZmode for TARGET_Z196.
16843 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16845 PR rtl-optimization/80501
16846 * combine.c (make_compound_operation_int): Set subreg_code to SET
16847 even for AND with mask of the sign bit of mode.
16849 PR rtl-optimization/80500
16850 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
16851 sum's initial value.
16853 2017-04-25 Julian Brown <julian@codesourcery.com>
16854 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16856 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
16858 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
16860 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
16862 2017-04-25 Julian Brown <julian@codesourcery.com>
16863 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16865 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
16866 (thunderx2t99_sha): New Reservation.
16868 2017-04-25 Julian Brown <julian@codesourcery.com>
16869 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16871 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
16872 type for 1-element load.
16874 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
16876 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
16878 2017-04-24 Martin Jambor <mjambor@suse.cz>
16880 PR tree-optimization/80293
16881 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
16882 char arrays not totally scalarizable if it is false.
16883 (analyze_all_variable_accesses): Pass correct value in the new
16884 parameter. Add a statistics counter.
16886 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
16888 PR middle-end/79931
16889 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
16891 2017-04-24 Richard Biener <rguenther@suse.de>
16893 PR tree-optimization/80494
16894 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
16895 out for complex types.
16897 2017-04-24 Richard Biener <rguenther@suse.de>
16899 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
16900 * tree-ssa-sccvn.c (print_scc): Print SCC size.
16901 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
16902 (DFS): Adjust and never fail.
16903 (sccvn_dom_walker::fail): Remove.
16904 (sccvn_dom_walker::before_dom_children): Adjust.
16905 (run_scc_vn): Likewise and never fail.
16906 * tree-ssa-pre.c (pass_pre::execute): Adjust.
16907 (pass_fre::execute): Likewise.
16909 2017-04-24 Richard Biener <rguenther@suse.de>
16911 PR tree-optimization/79725
16912 * tree-ssa-sink.c (statement_sink_location): Return whether
16913 failure reason was zero uses. Move that check later.
16914 (sink_code_in_bb): Deal with zero uses by removing the stmt
16917 2017-04-24 Richard Biener <rguenther@suse.de>
16920 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
16921 pointer-based references.
16923 2017-04-24 Richard Biener <rguenther@suse.de>
16926 * pass_manager.h (pass_manager::operator new): Remove.
16927 (pass_manager::operator delete): Likewise.
16928 * passes.c (pass_manager::operator new): Remove.
16929 (pass_manager::operator delete): Likewise.
16930 (pass_manager::pass_manager): Zero individual pass members.
16932 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
16935 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
16936 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
16937 Check "XEXP (src, 1)" operand here.
16938 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
16939 Check "XEXP (src, 1)" operand here.
16940 (dimode_scalar_chain::make_vector_copies): Detect count register
16941 of a shift instruction. Zero extend count register from QImode
16942 to DImode to satisfy vector shift pattern count operand predicate.
16943 Substitute vector shift count operand with a DImode copy.
16944 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
16947 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
16949 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
16950 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16951 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
16952 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16953 (UNSPEC_NOREX_MEM): Remove definition.
16955 2017-04-21 Richard Biener <rguenther@suse.de>
16957 PR tree-optimization/79547
16958 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
16959 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
16960 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
16961 without any constraints.
16963 2017-04-21 Richard Biener <rguenther@suse.de>
16965 PR tree-optimization/78847
16966 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
16968 2017-04-21 Richard Biener <rguenther@suse.de>
16970 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
16971 (build_distinct_type_copy): Likewise.
16972 (build_variant_type_copy): Likewise.
16973 * tree.c (build_qualified_type): Pass down mem-stat info.
16974 (build_distinct_type_copy): Likewise.
16975 (build_variant_type_copy): Likewise.
16977 2017-04-21 Richard Biener <rguenther@suse.de>
16979 PR tree-optimization/80237
16980 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
16982 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
16983 for a simplified result.
16985 2016-04-21 Richard Biener <rguenther@suse.de>
16987 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
16988 sth as strict as a simple_iv but a chrec without symbols and an
16989 operand defined in the loop we are peeling (and not some subloop).
16990 (propagate_constants_for_unrolling): Propagate all constants.
16992 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16995 * config/i386/i386.c (print_reg): Remove assert for disalowed
16996 regno values, call output_operand_lossage instead.
16998 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17001 * config/i386/constraints.md (Yc): New register constraint.
17002 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17003 Use Yc constraint for alternative 2 of operand 0. Remove
17004 preferred_for_speed attribute.
17006 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17008 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17009 lastprivate clauses in SIMT case.
17011 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17013 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17015 2017-04-20 Richard Biener <rguenther@suse.de>
17017 PR tree-optimization/57796
17018 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17019 as N scalar stores.
17020 (vect_model_load_cost): Cost gathers as N scalar loads.
17022 2017-04-20 Richard Biener <rguenther@suse.de>
17024 * ggc-page.c (ggc_allocated_p): Rename to ...
17025 (safe_lookup_page_table_entry): ... this and return the lookup
17027 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17029 2017-04-20 Richard Biener <rguenther@suse.de>
17031 PR tree-optimization/80453
17032 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17033 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17034 from the conditions.
17035 (vn_phi_eq): Pass them down.
17036 (vn_phi_lookup): Record them.
17037 (vn_phi_insert): Likewise.
17039 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17041 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17042 uninitialized variable warning to avoid buffer overrun.
17044 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17047 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17048 is suppressed for '{ 0 }' in C.
17050 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17052 * BASE-VER: Set to 8.0.0.
17054 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17056 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17057 priority .init_array and .fini_array section with SECTION_NOTYPE
17060 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17062 PR middle-end/80423
17063 * tree.h (build_array_type): Add typeless_storage default argument.
17064 * tree.c (type_cache_hasher::equal): Also compare
17065 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17066 (build_array_type): Add typeless_storage argument, set
17067 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17069 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17070 (build_array_type): Likewise. Add typeless_storage argument.
17072 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17073 Jakub Jelinek <jakub@redhat.com>
17075 PR tree-optimization/80426
17076 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17077 operation on symbolic operands, also compute the overflow for the
17078 invariant part when the operation degenerates into a negation.
17080 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17083 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17084 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17087 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17089 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17092 * config/avr/avr.c (tree.h): Include it.
17093 (cgraph.h): Include it.
17094 (avr_encode_section_info): Don't warn for uninitialized progmem
17095 variable if it's just an alias.
17097 2017-04-19 Richard Biener <rguenther@suse.de>
17100 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17101 when needed by AutoPGO.
17103 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17106 * doc/lto.texi: Remove an extra 'that'.
17108 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17110 PR rtl-optimization/80429
17111 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17112 are only used in debug insns.
17114 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17115 Vladimir Makarov <vmakarov@redhat.com>
17117 * config/sparc/predicates.md (input_operand): Add comment. Return
17118 true for any memory operand when LRA is in progress.
17119 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17121 2017-04-18 Jeff Law <law@redhat.com>
17124 * mips.md ({return,simple_return}_internal): Do not overwrite
17127 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17129 PR tree-optimization/80443
17130 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17131 instead of adding 1, subtract -1 and similarly instead of subtracting
17134 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17136 PR rtl-optimization/80357
17137 * haifa-sched.c (tmp_bitmap): New variable.
17138 (model_recompute): Handle duplicate use records.
17139 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17140 (free_global_sched_pressure_data): Free it.
17142 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17145 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17146 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17147 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17148 instead of SYSTEM_HEADER_DIR.
17150 2017-04-18 Jeff Law <law@redhat.com>
17152 PR middle-end/80422
17153 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17154 predecessors after walking up the insn chain.
17156 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17159 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17160 sizetype type into debug info.
17162 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17165 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17166 unneeded test for TARGET_UPPER_REGS_SF.
17167 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17169 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17172 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17173 instead of gsi_after_labels.
17175 2017-04-18 Jeff Law <law@redhat.com>
17177 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17181 2017-04-13 Jeff Law <law@redhat.com>
17182 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17183 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17185 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17188 * config/avr/avr.c (intl.h): Include it.
17189 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17191 2017-04-18 Martin Liska <mliska@suse.cz>
17193 PR gcov-profile/78783
17194 * gcov-tool.c (gcov_output_files): Validate that destination
17195 file is either removed by the tool or by a user.
17197 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17198 Guy Benyei <guybe@mellanox.com>
17200 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17201 block, and do not negate it, the stored id is already negative.
17203 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17205 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17207 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17210 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17211 masks of options that should be turned off if the VSX vector
17212 options are turned off.
17213 (OTHER_P8_VECTOR_MASKS): Likewise.
17214 (OTHER_VSX_VECTOR_MASKS): Likewise.
17215 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17216 rs6000_disable_incompatible_switches to validate no type switches
17218 (rs6000_incompatible_switch): New function to disallow turning on
17219 other vector options if -mno-vsx, -mno-power8-vector, or
17220 -mno-power9-vector are specified.
17222 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17224 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17226 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17228 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17229 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17230 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17231 (ARG_POINTER_CFA_OFFSET): Likewise.
17233 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17235 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17236 conditions to take advantage of various optimizations.
17238 2017-04-13 Jeff Law <law@redhat.com>
17240 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17241 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17242 (zero_extendsidi2_dext): Likewise.
17244 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17247 * fold-const.c (fold_ternary_loc): Revert
17248 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17251 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17253 PR rtl-optimization/80343
17254 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17255 new scratch pseudo.
17257 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17260 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17261 to ubsan_encode_value.
17263 2017-04-13 Jeff Law <law@redhat.com>
17265 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17266 appearing in DEBUG_INSNs.
17268 2017-04-13 Martin Liska <mliska@suse.cz>
17270 PR gcov-profile/80413
17271 * gcov-io.c (gcov_write_string): Copy to buffer just when
17272 allocated size is greater than zero.
17274 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17277 * dwarf2out.c (decls_for_scope): Ignore declarations of
17278 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17280 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17283 * ipa-visibility.c (non_local_p): Fix typos.
17284 (localize_node): When localizing symbol in same comdat group,
17285 dissolve the group only when we know external symbols are going
17287 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17289 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17291 PR tree-optimization/79390
17292 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17293 order does not result in usable sequence, retry with reversed operand
17299 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17300 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17301 op0 instead of fold_convert_loc (loc, type, arg0).
17303 2017-04-12 Jeff Law <law@redhat.com>
17305 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17306 has a delay slot in the generated code.
17308 * config/cris/cris.md (cris_preferred_reload_class): Return
17309 GENNONACR_REGS rather than GENERAL_REGS.
17311 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17314 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17315 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17316 signedness of the result type.
17318 2017-04-12 Richard Biener <rguenther@suse.de>
17319 Jeff Law <law@redhat.com>
17321 PR tree-optimization/80359
17322 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17323 trim stores to TARGET_MEM_REFs.
17325 2017-04-12 Richard Biener <rguenther@suse.de>
17327 PR tree-optimization/79390
17328 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17329 threading case even more.
17331 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17334 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17335 for quad_address_p for TImode, instead of just not indexed_address.
17337 2017-04-12 Richard Biener <rguenther@suse.de>
17338 Bernd Edlinger <bernd.edlinger@hotmail.de>
17340 PR middle-end/79671
17341 * alias.c (component_uses_parent_alias_set_from): Handle
17342 TYPE_TYPELESS_STORAGE.
17343 (get_alias_set): Likewise.
17344 * tree-core.h (tree_type_common): Add typeless_storage flag.
17345 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17346 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17347 for types containing members with TYPE_TYPELESS_STORAGE.
17348 (place_field): Likewise.
17349 (layout_type): Likewise for ARRAY_TYPE.
17350 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17351 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17352 TYPE_TYPELESS_STORAGE.
17353 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17355 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17358 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17359 first argument to type.
17361 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17365 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17366 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17367 (rs6000_expand_binop_builtin): Likewise.
17368 (rs6000_expand_ternop_builtin): Likewise; also add missing
17369 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17370 vshasigma built-ins.
17371 * doc/extend.texi: Document that vec_xxpermdi's third argument
17372 must be a constant.
17374 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17376 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17377 Use shift_const cost parameter when calculating gain of STV shifts.
17379 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17381 PR rtl-optimization/70478
17382 * lra-constraints.c (process_alt_operands): Check memory for
17383 disfavoring memory insn operand.
17385 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17387 PR middle-end/80100
17388 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17389 left shift in unsigned HOST_WIDE_INT type.
17391 PR rtl-optimization/80385
17392 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17393 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17396 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17397 if they have any depend clauses.
17399 2017-04-11 Martin Liska <mliska@suse.cz>
17402 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17403 * ipa-split.c (split_function): Create a local comdat symbol
17404 if caller is in a comdat group.
17406 2017-04-11 Martin Liska <mliska@suse.cz>
17409 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17412 2017-04-11 Martin Sebor <msebor@redhat.com>
17414 PR middle-end/80364
17415 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17416 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17418 (directive::set_width, directive::set_precision, format_character):
17420 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17423 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17426 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17427 conflict, set target->arch_name instead of target->cpu_name.
17429 2017-04-11 Richard Biener <rguenther@suse.de>
17431 PR tree-optimization/80374
17432 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17433 build_zero_cst, remove fold_convertible_p check again.
17435 2017-04-11 Martin Liska <mliska@suse.cz>
17438 * ubsan.c (instrument_object_size): Do not instrument register
17441 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17444 * config/i386/i386-builtin-types.def
17445 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17446 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17447 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17448 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17449 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17450 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17451 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17452 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17453 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17454 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17455 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17456 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17457 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17458 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17459 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17460 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17461 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17462 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17463 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17464 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17465 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17466 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17467 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17468 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17469 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17470 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17471 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17472 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17473 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17474 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17475 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17476 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17477 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17478 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17479 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17480 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17481 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17482 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17483 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17484 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17485 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17486 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17487 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17488 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17489 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17490 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17492 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17493 flag to second_arg_count, handle 4 argument function type _COUNT
17494 aliases, handle second_arg_count on second argument rather than last.
17496 2017-04-10 Jeff Law <law@redhat.com>
17498 PR tree-optimization/80374
17499 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17500 record anything if we can not convert integer_zero_node to the
17503 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17506 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17507 Enhance special handling given to the TARGET_P9_MINMAX option in
17508 relation to certain other options.
17510 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17512 PR tree-optimization/80153
17513 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17514 remove POINTER_PLUS_EXPR's base part directly, rather than through
17517 2017-04-10 Richard Biener <rguenther@suse.de>
17518 Bin Cheng <bin.cheng@arm.com>
17520 PR tree-optimization/80153
17521 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17522 the first element of pointer type aff_tree. Build result expr in
17524 (add_elt_to_tree): Convert to type unconditionally. Remove other
17525 fold_convert calls.
17526 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17527 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17529 2017-04-10 Richard Biener <rguenther@suse.de>
17531 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17534 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17536 PR rtl-optimization/70478
17537 * lra-constraints.c (curr_small_class_check): New.
17538 (update_and_check_small_class_inputs): New.
17539 (process_alt_operands): Update curr_small_class_check. Disfavor
17540 alternative insn memory operands. Check available regs for small
17543 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17546 * config/mips/mips.opt (-mvirt): Update description.
17547 * doc/invoke.texi (-mvirt): Likewise.
17549 2017-04-10 Richard Biener <rguenther@suse.de>
17551 PR middle-end/80362
17552 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17553 looking for NEGATE_EXPR in -A / -B to A / B folding.
17555 2017-04-10 Martin Liska <mliska@suse.cz>
17557 PR gcov-profile/80224
17558 * gcov.c (print_usage): Fix usage string.
17559 (get_gcov_intermediate_filename): Remove.
17560 (output_gcov_file): Use both for normal and intermediate format.
17561 (generate_results): Do not initialize special file for
17562 intermediate format.
17564 2017-04-10 Richard Biener <rguenther@suse.de>
17566 PR tree-optimization/80304
17567 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
17570 2017-04-10 Nathan Sidwell <nathan@acm.org>
17573 * config/rs6000/rs6000.c (rs6000_vector_type): New.
17574 (rs6000_init_builtins): Use it.
17576 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17578 * config/arm/arm.md (<mrc>): Add mode to SET source.
17579 (<mrrc>): Likewise.
17581 2017-04-10 Richard Biener <rguenther@suse.de>
17583 PR middle-end/80344
17584 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
17586 2017-04-10 Jakub Jelinek <jakub@redhat.com>
17589 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
17590 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
17591 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
17592 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
17593 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
17594 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
17595 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
17596 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
17597 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
17598 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
17599 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
17600 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
17601 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
17602 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
17603 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
17604 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
17605 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
17606 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
17607 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
17608 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
17609 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
17610 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
17611 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
17613 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
17615 PR rtl-optimization/70478
17616 * lra-constraints.c: Reverse the last patch.
17618 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
17620 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
17621 Add comment for WCHAR_T.
17623 2017-04-08 Martin Liska <mliska@suse.cz>
17626 2017-04-07 Martin Liska <mliska@suse.cz>
17629 * ipa-split.c (split_function): Add function part to a same comdat
17632 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17635 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
17637 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
17639 * rs6000/rs6000.c (vec_load_pendulum): Rename...
17640 (vec_pairing): ...to this.
17641 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
17642 (rs6000_sched_init): Adjust for name change.
17643 (struct rs6000_sched_context): Likewise.
17644 (rs6000_init_sched_context): Likewise.
17645 (rs6000_set_sched_context): Likewise.
17647 2017-04-07 Jakub Jelinek <jakub@redhat.com>
17653 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
17655 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
17656 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
17657 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
17659 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
17661 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
17663 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17665 PR rtl-optimization/70703
17666 * ira-color.c (update_conflict_hard_regno_costs): Use
17667 int64_t instead of HOST_WIDE_INT.
17669 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17671 PR rtl-optimization/70478
17672 * lra-constraints.c (process_alt_operands): Disfavor alternative
17673 insn memory operands.
17675 2017-04-07 Jeff Law <law@redhat.com>
17677 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
17678 CALL and NOTE_INSN_CALL_ARG_LOCATION.
17680 2017-04-07 Martin Liska <mliska@suse.cz>
17683 * config/aarch64/aarch64.c (aarch64_process_target_attr):
17684 Show error message instead of an ICE.
17686 2017-04-07 Martin Liska <mliska@suse.cz>
17689 * ipa-split.c (split_function): Add function part to a same comdat
17692 2017-04-07 Richard Biener <rguenther@suse.de>
17694 PR middle-end/80341
17695 * tree.c (get_unwidened): Also handle ! for_type case for
17697 * convert.c (do_narrow): Split out from ...
17698 (convert_to_integer_1): ... here. Do not pass final truncation
17699 type to get_unwidened for TRUNC_DIV_EXPR.
17701 2017-04-07 Richard Biener <rguenther@suse.de>
17703 * tree-affine.c (wide_int_ext_for_comb): Take type rather
17705 (aff_combination_const): Adjust.
17706 (aff_combination_scale): Likewise.
17707 (aff_combination_add_elt): Likewise.
17708 (aff_combination_add_cst): Likewise.
17709 (aff_combination_convert): Likewise.
17710 (add_elt_to_tree): Likewise. Remove unused argument.
17711 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
17713 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
17715 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
17717 * config/arm/arm.c (arm_default_short_enums): Use
17718 ARM_DEFAULT_SHORT_ENUMS.
17719 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
17721 2017-04-06 Jakub Jelinek <jakub@redhat.com>
17724 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
17725 members with redundant out-of-class redeclaration.
17727 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17730 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17731 * config/i386/i386.md (*zero_extendsidi2):
17732 Add (?*x,*x) and (?*v,*v) alternatives.
17734 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17737 * config/i386/i386.c (ix86_expand_builtin)
17738 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17739 mode from insn data. Convert operands to insn operand mode.
17740 Copy operands that don't satisfy insn predicate to a register.
17742 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17744 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17747 2017-04-06 Richard Biener <rguenther@suse.de>
17749 PR tree-optimization/80334
17750 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17751 preserve alignment of accesses.
17753 2017-04-06 Richard Biener <rguenther@suse.de>
17755 PR tree-optimization/80262
17756 * tree-sra.c (build_ref_for_offset): Preserve address-space
17758 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17759 Drop useless address-space information on MEM_REF offsets.
17761 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
17763 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
17765 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17767 PR rtl-optimization/70703
17768 * ira-color.c (update_conflict_hard_regno_costs): Use
17769 HOST_WIDE_INT instead of long.
17771 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
17774 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
17775 not defined for x86_64 target. Add -mmmx target option when __SSE2__
17777 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
17778 for x86_64 target. Handle -m3dnowa option.
17780 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17782 PR rtl-optimization/70703
17783 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
17784 (update_conflict_hard_regno_costs): Use long instead of unsigned
17785 arithmetic for cost calculation.
17787 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17788 Bernd Edlinger <bernd.edlinger@hotmail.de>
17791 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
17794 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
17797 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
17798 ptr_mode with Pmode throughout.
17799 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
17800 into probe_stack_range and use DImode.
17802 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
17805 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
17806 call_eh_return is true.
17808 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17810 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
17811 Initialize last_match_fntype_index.
17813 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17816 * tree-nvr.c: Include internal-fn.h.
17817 (pass_return_slot::execute): Ignore internal calls without
17820 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17821 Richard Biener <rguenther@suse.de>
17824 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
17825 captures used multiple times, except for the last use.
17826 * generic-match-head.c: Include gimplify.h.
17828 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17830 PR tree-optimization/79390
17831 * target.h (struct noce_if_info): Declare.
17832 * targhooks.h (default_noce_conversion_profitable_p): Declare.
17833 * target.def (noce_conversion_profitable_p): New target hook.
17834 * ifcvt.h (struct noce_if_info): New type, moved from ...
17835 * ifcvt.c (struct noce_if_info): ... here.
17836 (noce_conversion_profitable_p): Renamed to ...
17837 (default_noce_conversion_profitable_p): ... this. No longer
17839 (noce_try_store_flag_constants, noce_try_addcc,
17840 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
17841 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
17842 instead of noce_conversion_profitable_p.
17843 * config/i386/i386.c: Include ifcvt.h.
17844 (ix86_option_override_internal): Don't override
17845 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
17846 (ix86_noce_conversion_profitable_p): New function.
17847 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
17848 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
17849 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
17850 * doc/tm.texi: Regenerated.
17852 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17854 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
17857 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17860 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
17861 instructions for small multiply cores.
17863 2017-04-04 Jeff Law <law@redhat.com>
17865 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
17867 (mips_expand_vec_perm_const): Initialize elements in orig_perm
17868 that are not set by the loop over the elements.
17870 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17873 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
17874 int mode, convert_modes it to mode as unsigned, otherwise use
17875 lowpart_subreg to mode rather than SImode.
17876 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
17877 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
17878 Use DImode instead of SImode for the shift count operand.
17879 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
17882 2017-04-04 Richard Biener <rguenther@suse.de>
17884 PR middle-end/80281
17885 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
17886 arithmetic done for the negate or the plus. Simplify.
17887 (A - (-B) -> A + B): Likewise.
17888 * fold-const.c (split_tree): Make sure to not negate pointers.
17890 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
17892 PR rtl-optimization/60818
17893 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
17894 a compare of comparisons with the thing compared if this results
17895 in a different machine mode.
17897 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
17899 * alias.c (base_alias_check): Fix typo in comment.
17900 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
17901 * cgraphunit.c (symbol_table::compile): Likewise.
17902 * collect2.c (maybe_run_lto_and_relink): Likewise.
17903 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
17904 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
17905 * config/avr/avr.c (avr_map_op_t): Likewise.
17906 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
17907 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
17908 * config/epiphany/epiphany.md (movcc): Likewise.
17909 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
17910 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
17912 * config/mips/mips.c (mips_save_restore_reg): Likewise.
17913 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
17914 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
17915 * config/sh/sh.c (sh_rtx_costs): Likewise.
17916 * fold-const.c (fold_truth_andor): Likewise.
17917 * genautomata.c (collapse_flag): Likewise.
17918 * gengtype.h (struct type::u::s): Likewise.
17919 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
17920 * input.c (FORMAT_AMOUNT): Likewise.
17921 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
17922 (known_aggs_to_agg_replacement_list): Likewise.
17923 * ipa-inline-analysis.c: Likewise.
17924 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
17925 * ipa-polymorphic-call.c
17926 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
17927 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
17928 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
17930 * modulo-sched.c (apply_reg_moves): Likewise.
17931 * omp-expand.c (build_omp_regions_1): Likewise.
17932 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
17933 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
17934 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
17935 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
17936 * value-prof.c: Likewise.
17937 * var-tracking.c (val_reset): Likewise.
17939 2017-04-03 Richard Biener <rguenther@suse.de>
17941 PR tree-optimization/80275
17942 * fold-const.c (split_address_to_core_and_offset): Handle
17945 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
17947 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
17948 descriptors is at least equal to that of functions.
17950 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17952 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
17954 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17957 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
17958 (mov<IMOD4:mode>): New expander.
17959 (*mov<IMOD4:mode>_internal): New insn and split pattern.
17961 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
17963 PR rtl-optimization/79405
17964 * fwprop.c (propagations_left): New variable.
17965 (forward_propagate_into): Decrement it.
17966 (fwprop_init): Initialize it.
17967 (fw_prop): If the variable has reached zero, stop propagating.
17968 (fwprop_addr): Ditto.
17970 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17973 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
17974 a FUNCTION_DECL, pass it as decl instead of origin to
17977 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
17979 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
17982 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
17985 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
17986 TARGET_VSX_SMALL_INTEGER.
17988 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17990 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17991 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
17993 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17995 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
17996 extraction from odd-numbered MSA register.
17998 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18000 PR middle-end/80173
18001 * expmed.c (store_bit_field_1): Don't attempt to create
18002 a word subreg out of hard registers wider than word if they
18003 have HARD_REGNO_NREGS of 1 for their mode.
18005 PR middle-end/80163
18006 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18007 conversions to integer types wider than word and pointer.
18010 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18011 (rtx_equal_for_cselib_p): Pass 0 to it.
18012 * cselib.c (cselib_hasher::equal): Likewise.
18013 (rtx_equal_for_cselib_1): Add depth argument. If depth
18014 is 128, don't look up VALUE locs and punt. Increment
18015 depth in recursive calls when walking VALUE locs.
18017 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18019 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18020 (make_gcov_file_name): Use the canonical path name for generating
18022 (read_line): Fix handling of files with ascii null bytes.
18024 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18026 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18027 to initialise a vector register instead
18028 of using a const_int.
18030 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18032 PR translation/80189
18033 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18034 diagnostic messages.
18036 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18039 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18040 (dfp_diex_<mode>): Update mode of operand 1.
18041 * doc/extend.texi (dxex, dxexq): Document change to return type.
18042 (diex, diexq): Document change to argument type.
18044 2017-03-30 Martin Jambor <mjambor@suse.cz>
18047 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18048 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18049 it reflects the signature changes performed at the callee side.
18050 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18051 to cgraph_build_function_type_skip_args.
18052 (build_function_decl_skip_args): Adjust call to the above function.
18054 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18057 * config/i386/sse.md
18058 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18059 register as dest whenever it is a MEM not rtx_equal_p to the
18060 corresponding dup operand, and when forcing into reg move the
18061 reg into the memory afterwards.
18062 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18063 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18064 for the force_reg mode.
18065 (avx512vl_vextractf128<mode>): Use register as dest either
18066 always when a MEM, or when it is a MEM not rtx_equal_p to the
18067 corresponding dup operand, or even not when it is a CONST_VECTOR
18068 depending on the mode and lo vs. hi.
18069 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18071 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18072 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18073 Likewise. Require that operands[2] is even.
18074 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18075 Remove extraneous parens. Require that operands[2] is a multiple
18077 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18078 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18079 disallow memory then.
18081 2017-03-30 Richard Biener <rguenther@suse.de>
18083 PR tree-optimization/77498
18084 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18085 to non-constants over backedges.
18087 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18089 PR rtl-optimization/80233
18090 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18091 as last_combined_insn. Do not test for BARRIER_P separately.
18093 2017-03-29 Andreas Schwab <schwab@suse.de>
18096 * calls.c (prepare_call_address): Convert funexp to Pmode before
18097 copying to temp reg.
18099 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18101 PR tree-optimization/80158
18102 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18103 Handle possible future case of more than one alternate
18105 (replace_rhs_if_not_dup): Likewise.
18106 (replace_one_candidate): Likewise.
18108 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18110 PR rtl-optimization/80193
18111 * ira.c (ira): Do not check allocation for LRA.
18113 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18115 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18116 (nvptx_output_simt_exit): Declare.
18117 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18118 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18119 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18120 (nvptx_declare_function_name): ...here. Emit declaration of local
18121 memory space buffer for omp_simt_enter insn.
18122 (nvptx_output_unisimt_switch): New.
18123 (nvptx_output_softstack_switch): New.
18124 (nvptx_output_simt_enter): New.
18125 (nvptx_output_simt_exit): New.
18126 * config/nvptx/nvptx.h (struct machine_function): New fields
18127 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18128 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18129 (UNSPECV_SIMT_EXIT): Ditto.
18130 (omp_simt_enter_insn): New insn.
18131 (omp_simt_enter): New expansion.
18132 (omp_simt_exit): New insn.
18133 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18135 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18136 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18137 (expand_GOMP_SIMT_EXIT): New.
18138 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18139 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18140 (GOMP_SIMT_EXIT): Ditto.
18141 * target-insns.def (omp_simt_enter): New insn.
18142 (omp_simt_exit): Ditto.
18143 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18145 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18146 (lower_rec_input_clauses): Likewise.
18147 (lower_lastprivate_clauses): Handle SIMT privatization.
18149 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18150 (ompdevlow_adjust_simt_enter): New.
18151 (find_simtpriv_var_op): New.
18152 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18153 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18155 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18156 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18157 (copy_decl_for_dup_finish): Ditto.
18159 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18161 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18164 * config/i386/i386.c (ix86_option_override_internal): Always
18165 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18167 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18169 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18171 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18173 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18174 mark new edge's irreducible flag accordign to it.
18175 (vect_do_peeling): Check loop preheader edge's irreducible flag
18176 and pass it to function slpeel_add_loop_guard.
18178 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18180 PR tree-optimization/80218
18181 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18182 Update block frequencies and counts.
18184 2017-03-28 Richard Biener <rguenther@suse.de>
18186 PR tree-optimization/78644
18187 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18188 of a simplification result we may not use it at all.
18190 2017-03-28 Richard Biener <rguenther@suse.de>
18193 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18194 without arguments, generate default definition of a SSA name.
18196 2017-03-28 Richard Biener <rguenther@suse.de>
18198 PR middle-end/80222
18199 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18200 TYPE_REF_CAN_ALIAS_ALL references.
18201 * fold-const.c (fold_indirect_ref_1): Likewise.
18203 2017-03-28 Martin Liska <mliska@suse.cz>
18206 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18207 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18209 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18210 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18212 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18213 (EXTRA_SPECS): Define.
18214 (SUBTARGET_EXTRA_SPECS): Likewise.
18215 (SUBTARGET_CPP_SPEC): Likewise.
18216 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18217 SUBTARGET_EXTRA_SPECS.
18218 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18220 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18222 * config/arc/simdext.md (vst64_insn): Update pattern.
18223 (vld32wh_insn): Likewise.
18224 (vld32wl_insn): Likewise.
18225 (vld64_insn): Likewise.
18226 (vld32_insn): Likewise.
18228 2017-03-28 Marek Polacek <polacek@redhat.com>
18231 * fold-const.c (fold_comparison): Use protected_set_expr_location
18232 instead of SET_EXPR_LOCATION.
18234 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18236 * tree.c (add_expr): Avoid name lookup warning.
18238 2017-03-27 Jeff Law <law@redhat.com>
18240 PR tree-optimization/80216
18241 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18242 function name. Limit recursion depth.
18243 (record_temporary_equivalences): Corresponding changes.
18245 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18247 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18250 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18253 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18255 * cfgcleanup.c (reg_note_cfa_p): New array.
18256 (insns_have_identical_cfa_notes): New function.
18257 (old_insns_match_p): Don't cross-jump in between /f
18258 and non-/f instructions. If both i1 and i2 are frame related,
18259 verify all CFA notes, their order and content.
18261 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18264 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18265 HImode and SImode with zero extend to DImode to one insn.
18266 (bswap<mode>2_extenddi): Likewise.
18267 (bswapsi2_extenddi): Likewise.
18268 (bswaphi2_extendsi): Likewise.
18269 (bswaphi2): Combine bswap HImode and SImode into one insn.
18270 Separate memory insns from swapping register.
18271 (bswapsi2): Likewise.
18272 (bswap<mode>2): Likewise.
18273 (bswaphi2_internal): Delete, no longer used.
18274 (bswapsi2_internal): Likewise.
18275 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18276 store, and gpr<-gpr swap insns.
18277 (bswap<mode>2_store): Likewise.
18278 (bswaphi2_reg): Register only splitter, combine with the splitter.
18279 (bswaphi2 splitter): Likewise.
18280 (bswapsi2_reg): Likewise.
18281 (bswapsi2 splitter): Likewise.
18282 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18283 the insns into load, store, and register/register insns.
18284 (bswapdi2_ldbrx): Likewise.
18285 (bswapdi2_load): Likewise.
18286 (bswapdi2_store): Likewise.
18287 (bswapdi2_reg): Likewise.
18289 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18291 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18292 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18294 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18297 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18299 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18300 special handling for target option conflicts between dform
18301 options (-mpower9-dform, -mpower9-dform-vector,
18302 -mpower9-dform-scalar) and -mno-direct-move.
18304 2017-03-27 Richard Biener <rguenther@suse.de>
18306 PR tree-optimization/80181
18307 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18309 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18311 * config/arc/predicates.md (move_double_src_operand): Replace the
18312 call to move_double_src_operand with a call to address_operand.
18314 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18316 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18317 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18318 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18320 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18322 * config/arc/predicates.md (long_immediate_loadstore_operand):
18323 Consider scaled addresses cases.
18325 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18327 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18328 restored when in interrupt.
18329 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18330 doesn't have delay slot.
18332 2017-03-27 Richard Biener <rguenther@suse.de>
18335 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18336 inlined thunk clones.
18338 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18341 * asan.c (instrument_derefs): Copy over last operand from
18342 original COMPONENT_REF to the new COMPONENT_REF with
18343 DECL_BIT_FIELD_REPRESENTATIVE.
18344 * ubsan.c (instrument_object_size): Likewise.
18346 2017-03-27 Richard Biener <rguenther@suse.de>
18348 PR tree-optimization/80170
18349 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18350 sure DR/SCEV didnt fold in constants we do not see when looking
18351 at the reference base alignment.
18353 2017-03-27 Richard Biener <rguenther@suse.de>
18355 PR middle-end/80171
18356 * gimple-fold.c (fold_ctor_reference): Properly guard against
18357 NULL return value from canonicalize_constructor_val.
18359 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18362 * config/i386/i386.c (ix86_expand_builtin)
18363 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18364 flags reg setting and flags reg using instructions.
18365 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18366 clobbering instructions to zero extend op2.
18368 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18370 * doc/install.texi (Configuration) <--with-aix-soname>:
18371 Update link to AIX ld.
18373 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18375 PR rtl-optimization/80160
18376 PR rtl-optimization/80159
18377 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18378 reg_alternate_class into account.
18380 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18383 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18384 to consider in curr_insn_transform.
18386 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18388 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18389 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18390 and emit_mode_inner.
18392 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18394 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18395 argument to the overloaded builtin variants. Use the new flag to
18396 deprecate certain builtin variants.
18397 * config/s390/s390-builtin-types.def: Add new builtin types.
18398 * config/s390/s390-builtins.h: Support new flags field for
18399 overloaded builtins.
18400 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18401 (s390_macro_to_expand): Enable vector float data type.
18402 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18403 builtins by incrementing the __VEC__ version number.
18404 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18406 (s390_resolve_overloaded_builtin): Emit error messages depending
18407 on the builtin flags.
18408 * config/s390/s390.c (s390_expand_builtin): Support additional
18409 flags argument. Change error message to match the messages
18410 emitted in s390-c.c.
18411 * config/s390/s390.md: New UNSPEC_* constants.
18412 (op_type): Add new instruction types.
18413 * config/s390/vecintrin.h: Add new builtins and test data class
18415 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18416 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18417 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18418 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18419 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18420 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18422 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18423 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18424 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18425 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18427 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18428 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18429 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18430 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18432 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18433 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18434 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18435 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18436 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18437 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18438 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18440 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18441 ("vec_scatter_element<V_HW_4:mode>_DI")
18442 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18443 ("vec_fpint<mode>", "vflls")
18444 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18445 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18446 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18447 ("*vec_cmphe<mode>_cc"): ... these.
18449 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18450 mode constant instead of magic value.
18452 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18454 * config/s390/s390.c (s390_expand_vec_compare): Support other
18455 vector floating point modes than just V2DF.
18456 (s390_expand_vcond): Likewise.
18457 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18458 (s390_cannot_change_mode_class): Prevent mode changes between TF
18459 and V1TF in vector registers.
18460 * config/s390/s390.md (DF, SF): New mode attributes.
18461 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18462 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18463 SFmode support for VRs.
18464 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18466 (VFT, VF_HW): New mode iterators.
18467 (vw, sdx): New mode attributes.
18468 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18469 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18470 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18471 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18472 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18473 also the new vector floating point modes. Renaming to ...
18475 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18476 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18477 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18478 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18479 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18480 ("vec_unordered<mode>"): ... these.
18482 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18483 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18484 ("*vec_extendv2df"): New insn definitions.
18486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18488 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18489 ("mulditi3_2", "*muldi3_sign"): New patterns.
18490 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18491 rename the pattern definition.
18493 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18495 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18497 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18499 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18501 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18502 instruction if possible.
18503 * config/s390/vector.md (vec_halfnumelts): New mode
18505 ("*vec_vllezlf<mode>"): New pattern.
18507 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18509 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18510 ("popcountv4si2", "popcountv2di2"): Rename to ...
18511 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18512 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18514 ("popcount<mode>2_vxe"): New pattern.
18516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18518 * common/config/s390/s390-common.c (processor_flags_table): Add
18520 * config.gcc: Add arch12.
18521 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18522 Default to arch12 for unknown CPU model numbers.
18523 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18524 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18525 PROCESSOR_max sanity check.
18526 * config/s390/s390-opts.h (enum processor_type): Add
18528 * config/s390/s390.c (processor_table): Add arch12.
18529 (s390_expand_builtin): Add check for B_VXE flag.
18530 (s390_issue_rate): Add PROCESSOR_ARCH12.
18531 (s390_get_sched_attrmask): Likewise.
18532 (s390_get_unit_mask): Likewise.
18533 (s390_sched_score): Enable z13 scheduling for arch12.
18534 (s390_sched_reorder): Likewise.
18535 (s390_sched_variable_issue): Likewise.
18536 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18538 (s390_tune_attr): Use z13 scheduling also for arch12.
18539 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18540 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18541 (TARGET_VXE_P): New macros.
18542 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18543 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18544 * config/s390/s390.opt: Add arch12 as processor_type.
18546 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18548 * config/s390/s390.md
18549 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18550 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18551 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18553 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18554 Rename expanders to ...
18556 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18557 ("fixuns_truncdddi2_emu"): ... these.
18559 ("fixuns_trunc<mode>si2_emu"): New expander.
18561 ("*fixuns_truncdfdi2_z13"): Rename to ...
18562 ("*fixuns_truncdfdi2_vx"): ... this.
18564 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18566 * config/s390/2964.md: Remove the single element vector compare
18567 instructions which are no longer used.
18568 * config/s390/s390.c (s390_select_ccmode): Remove handling of
18570 (s390_canonicalize_comparison): Remove handling of DFmode
18572 (s390_expand_vec_compare_scalar): Remove function.
18573 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
18574 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
18576 ("*cmp<mode>_ccs"): Add wfcdb instruction.
18578 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18580 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
18582 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
18583 will anyway by matched by mov<mode>_64dfp.
18585 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18587 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
18588 vlef/vstef. Add missing operand to vleif.
18590 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18592 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
18593 pair for all vector types with 64 bit elements.
18594 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
18595 * config/s390/vector.md (V_HW_64): ... here.
18596 (V_128_NOSINGLE): New mode iterator.
18597 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
18598 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
18599 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
18600 ("*vec_load_pairv2di"): Change to ...
18601 ("*vec_load_pair<mode>"): ... this one.
18603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18605 * config/s390/constraints.md: Add comments.
18606 (jKK): Reject element sizes > 8 bytes.
18607 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
18609 * config/s390/s390.md: Add the s_operand checks formerly in
18610 s390_split_ok_p to various splitters where they are still
18612 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
18613 for 128 bit vectors. Plus two splitters.
18615 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18617 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
18620 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18623 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
18624 error if the boundary argument is not constant.
18626 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18628 PR rtl-optimization/80112
18629 * loop-doloop.c (doloop_condition_get): Don't check condition
18630 if cmp isn't SET with IF_THEN_ELSE src.
18632 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18634 PR tree-optimization/80158
18635 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
18636 replacing a candidate statement, also replace it for the
18637 candidate's alternate interpretation.
18638 (replace_rhs_if_not_dup): Likewise.
18639 (replace_one_candidate): Likewise.
18641 2017-03-24 Richard Biener <rguenther@suse.de>
18643 PR tree-optimization/80167
18644 * graphite-isl-ast-to-gimple.c
18645 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
18647 (translate_isl_ast_to_gimple::get_rename): Likewise.
18649 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
18651 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
18652 handling of certain combinations of target options, including the
18653 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
18654 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
18656 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18659 * config/arm/arm.md (*load_multiple): Add reload_completed to
18660 matching condition.
18662 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18663 Richard Biener <rguenth@suse.de>
18665 PR tree-optimization/79908
18666 PR tree-optimization/80136
18667 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18668 been cast away, gimplify_and_add suffices.
18670 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
18672 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
18674 2017-03-23 Richard Biener <rguenther@suse.de>
18676 PR tree-optimization/80032
18677 * gimplify.c (gimple_push_cleanup): Forced unconditional
18678 cleanups still have to go to the conditional_cleanups
18681 2017-03-22 Jakub Jelinek <jakub@redhat.com>
18683 PR tree-optimization/80072
18684 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
18686 (next_operand_entry_id): Change type to unsigned int.
18687 (sort_by_operand_rank): Make sure to return the right return value
18688 even if unsigned fields are bigger than INT_MAX.
18689 (struct oecount): Change cnt and id type to unsigned int.
18690 (oecount_hasher::equal): Formatting fix.
18691 (oecount_cmp): Make sure to return the right return value
18692 even if unsigned fields are bigger than INT_MAX.
18693 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
18696 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
18697 TREE_READONLY on result if writing it more than once.
18700 * doc/invoke.texi (-fsanitize=thread): Document that with
18701 -fnon-call-exceptions atomics are not able to throw
18705 * tsan.c: Include tree-eh.h.
18706 (instrument_builtin_call): Call maybe_clean_eh_stmt or
18707 maybe_clean_or_replace_eh_stmt where needed.
18708 (instrument_memory_accesses): Add cfg_changed argument.
18709 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
18710 if it returned true.
18711 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
18713 PR rtl-optimization/63191
18714 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
18715 wrapper function, moved the whole old content into ...
18716 (ix86_delegitimize_address_1): ... this. New inline function.
18717 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
18718 true as last argument instead of ix86_delegitimize_address.
18720 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18722 * config/aarch64/aarch64.c (generic_branch_cost): Copy
18723 cortexa57_branch_cost.
18725 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18727 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18729 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18732 * doc/md.texi (Constraints): Document wA constraint.
18733 * config/rs6000/constraints.md (wA): New.
18734 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18735 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18736 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18737 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18739 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18742 * gimplify.c (is_oacc_declared): New function.
18743 (oacc_default_clause): Use it to set default flags for acc declared
18744 variables inside parallel regions.
18745 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18746 declared variables.
18747 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18748 declare attribute to any decl as necessary.
18750 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18753 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18754 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18755 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18756 (arm_arch_lpae): This.
18757 * config/arm/arm.c (arm_arch7ve): Rename into ...
18758 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18759 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
18762 2017-03-22 Martin Liska <mliska@suse.cz>
18765 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
18766 error message instead of an ICE.
18768 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18770 * doc/extend.texi (6.11 Additional Floating Types): Revise.
18772 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
18774 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18776 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18779 2017-03-21 Martin Sebor <msebor@redhat.com>
18781 * doc/extend.texi: Use "cannot" instead of "can't."
18782 * doc/hostconfig.texi: Same.
18783 * doc/install.texi: Same.
18784 * doc/invoke.texi: Same.
18785 * doc/loop.texi: Same.
18786 * doc/md.texi: Same.
18787 * doc/objc.texi: Same.
18788 * doc/rtl.texi: Same.
18789 * doc/tm.texi: Same.
18790 * doc/tm.texi.in: Same.
18791 * doc/trouble.texi: Same.
18793 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
18796 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
18797 (collect_checksum_attributes): Set it.
18798 (die_checksum_ordered): Use it.
18800 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18802 PR tree-optimization/79908
18803 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
18804 change: For a VA_ARG whose LHS has been cast away, use
18805 force_gimple_operand to construct the side effects.
18807 2017-03-21 David Malcolm <dmalcolm@redhat.com>
18809 PR translation/80001
18810 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
18811 more amenable to translation.
18812 (oacc_loop_auto_partitions): Likewise.
18814 2017-03-21 Marek Polacek <polacek@redhat.com>
18815 Martin Sebor <msebor@redhat.com>
18817 PR tree-optimization/80109
18818 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
18819 on INTEGRAL_TYPE_P.
18821 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18822 Segher Boessenkool <segher@kernel.crashing.org>
18825 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
18826 check reg_used_between_p between insn and one of succ or succ2
18827 depending on if succ is artificial insn not inserted into insn
18830 2017-03-21 Martin Liska <mliska@suse.cz>
18832 PR gcov-profile/80081
18833 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
18834 * doc/gcc.texi: Include gcov-dump stuff.
18835 * doc/gcov-dump.texi: New file.
18837 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
18839 PR rtl-optimization/79150
18840 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
18841 conditional jump, if the jump is the last insn of the loop.
18843 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18844 Richard Biener <rguenth@suse.de>
18846 PR tree-optimization/79908
18847 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18848 been cast away, use force_gimple_operand to construct the side
18851 2017-03-21 Martin Liska <mliska@suse.cz>
18853 PR libfortran/79956
18854 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
18857 2017-03-21 Brad Spengler <spender@grsecurity.net>
18860 * plugin.c (htab_hash_plugin): New function.
18861 (add_new_plugin): Use it and adjust.
18862 (parse_plugin_arg_opt): Adjust.
18863 (init_one_plugin): Likewise.
18865 2017-03-21 Richard Biener <rguenther@suse.de>
18867 PR tree-optimization/80032
18868 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
18869 if set force the cleanup to happen unconditionally.
18870 (gimplify_target_expr): Push inserted clobbers with force_uncond
18871 to avoid them being removed by control-dependent DCE.
18873 2017-03-21 Richard Biener <rguenther@suse.de>
18875 PR tree-optimization/80122
18876 * tree-inline.c (copy_bb): Do not expans va-arg packs or
18877 va_arg_pack_len when the inlined call stmt requires pack
18879 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
18881 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18884 * tsan.c (instrument_builtin_call): If the memory model argument
18885 is not a constant, assume it is valid.
18888 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
18891 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18893 PR rtl-optimization/79910
18894 * combine.c (can_combine_p): Do not allow combining an I0 or I1
18895 if its dest is used by an insn before I2 (other than the combined
18896 insns themselves, which are properly handled already).
18898 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18901 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18903 * combine.c (record_used_regs): New static function.
18904 (try_combine): Handle situations where there is an additional
18905 instruction between I2 and I3 which needs to have a LOG_LINK
18909 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18911 * combine.c (try_combine): Delete redundant i1 test. Call
18912 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18914 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18917 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
18918 alternatives 13/14.
18920 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18922 PR tree-optimization/80054
18923 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
18924 the optimization if a PHI or any of its arguments is not dominated
18925 by the candidate's basis. Use gphi* rather than gimple* as
18927 (replace_profitable_candidates): Clean up a gimple* variable that
18928 should be a gphi* variable.
18930 2017-03-20 Martin Sebor <msebor@redhat.com>
18933 * doc/extend.texi (attribute constructor): Document present limitation.
18935 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
18938 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
18939 __POWER9_VECTOR__ #ifdef control, change template definition to
18940 use Power9-specific built-in function.
18941 (vec_any_eq): Likewise.
18942 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
18943 to control outcomes from this test.
18944 (vector_ae_<mode>p): For VEC_F modes, likewise.
18946 2017-03-20 Ian Lance Taylor <iant@google.com>
18948 * config/i386/i386.c (ix86_function_regparm): Save an extra
18949 register for -fsplit-stack with DECL_STATIC_CHAIN.
18951 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18954 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
18955 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
18957 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18959 * config/riscv/riscv.c (riscv_print_operand): Use "fence
18961 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
18964 2017-03-20 Marek Polacek <polacek@redhat.com>
18967 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
18969 2017-03-20 Richard Biener <rguenther@suse.de>
18971 PR tree-optimization/80113
18972 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
18973 allocate extra SSA name for PHI def.
18974 (add_close_phis_to_outer_loops): Likewise.
18975 (add_close_phis_to_merge_points): Likewise.
18976 (copy_loop_close_phi_args): Likewise.
18977 (copy_cond_phi_nodes): Likewise.
18979 2017-03-20 Martin Liska <mliska@suse.cz>
18981 PR middle-end/79753
18982 * tree-chkp.c (chkp_build_returned_bound): Do not build
18983 returned bounds for a LHS that's not a BOUNDED_P type.
18985 2017-03-20 Martin Liska <mliska@suse.cz>
18989 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
18990 COMPLEX_CST and VECTOR_CST.
18992 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18995 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
18996 target operand. A new splitter adds the clobber statement in case
18997 the target operand is dead anyway.
18999 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19001 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19002 to age-old versions of binutils and glibc.
19004 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19006 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19008 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19010 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19012 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19014 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19015 requirement for binutils 2.13.
19017 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19019 * combine.c (try_combine): Delete redundant i1 test. Call
19020 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19022 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19024 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19025 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19027 <riscv64-*-elf>: Re-arrange section
19028 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19029 <riscv32-*-linux>: Likewise.
19030 <riscv64-*-elf>: Likewise
19031 <riscv64-*-linux>: Likewise.
19033 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19036 * aarch64.opt(verbose-cost-dump): Fix typo.
19038 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19041 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19042 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19044 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19046 * reload.c (find_reloads): When reloading a nonoffsettable address,
19047 use RELOAD_OTHER for it and its address reloads.
19049 PR rtl-optimization/79910
19050 * combine.c (record_used_regs): New static function.
19051 (try_combine): Handle situations where there is an additional
19052 instruction between I2 and I3 which needs to have a LOG_LINK
19055 2017-03-17 Jeff Law <law@redhat.com>
19057 PR tree-optimization/71437
19058 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19059 conditional in the hash table first.
19060 (vrp_dom_walker::before_dom_children): Extract condition from
19061 ASSERT_EXPR. Record condition, its inverion and any implied
19062 conditions as well.
19064 2017-03-17 Marek Polacek <polacek@redhat.com>
19065 Markus Trippelsdorf <markus@trippelsdorf.de>
19067 PR tree-optimization/80079
19068 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19071 2017-03-17 Richard Biener <rguenther@suse.de>
19073 PR middle-end/80075
19074 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19075 Properly verify the LHS before the RHS possibly claims to be
19077 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19080 2017-03-17 Martin Jambor <mjambor@suse.cz>
19082 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19083 (List of -O2 options): Likewise.
19084 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19087 2017-03-17 Tom de Vries <tom@codesourcery.com>
19089 * gcov-dump.c (print_usage): Print bug_report_url.
19091 2017-03-17 Richard Biener <rguenther@suse.de>
19093 PR middle-end/80050
19094 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19095 (parser::peek): Likewise.
19097 2017-03-17 Richard Biener <rguenther@suse.de>
19099 PR tree-optimization/80048
19100 * sese.c (free_sese_info): Properly release rename_map and
19101 copied_bb_map elements.
19103 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19105 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19106 Add linked-list forward and backlinks. Insert on
19107 construction, remove on destruction.
19108 (class pass_store_merging): Add m_stores_head field.
19109 (pass_store_merging::terminate_and_process_all_chains):
19110 Iterate over m_stores_head list.
19111 (pass_store_merging::terminate_all_aliasing_chains):
19113 (pass_store_merging::execute): Check for debug stmts first.
19114 Push new chains onto the m_stores_head stack.
19116 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19119 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19120 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19121 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19123 2017-03-16 Jeff Law <law@redhat.com>
19125 PR tree-optimization/71437
19126 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19127 member function. Implementation moved into after_dom_children
19128 member function and into the threader's thread_outgoing_edges
19130 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19131 some code into new thread_outgoing_edges.
19132 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19133 definition. Simplify marker handling (do it here). Assume we always
19134 have the available expression and the const/copies tables.
19135 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19137 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19138 * tree-vrp.c (equiv_stack): No longer file scoped.
19139 (vrp_dom_walker): New class.
19140 (vrp_dom_walker::before_dom_children): New member function.
19141 (vrp_dom_walker::after_dom_children): Likewise.
19142 (identify_jump_threads): Setup domwalker. Use it rather than
19143 walking edges in a random order by hand. Simplify setup/finalization.
19144 (finalize_jump_threads): Remove.
19145 (vrp_finalize): Do not call identify_jump_threads here.
19146 (execute_vrp): Do it here instead and call thread_through_all_blocks
19149 PR tree-optimization/71437
19150 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19152 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19154 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19155 (dom_opt_dom_walker::thread_across_edge): Remove
19156 handle_dominating_asserts argument. All callers changed.
19157 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19158 changes. Remove calls to lhs_of_dominating_assert. Other
19159 uses of handle_dominating_asserts turn into unconditional code
19160 (simplify_control_stmt_condition_1): Likewise.
19161 (simplify_control_stmt_condition): Likewise.
19162 (thread_through_normal_block, thread_across_edge): Likewise.
19163 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19164 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19165 object if it is not an SSA_NAME.
19166 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19167 before calling into the VRP specific simplifiers.
19168 (identify_jump_threads): Remove handle_dominating_asserts
19171 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19174 * tree-diagnostic.c (default_tree_printer): No longer static.
19175 * tree-diagnostic.h (default_tree_printer): New prototype.
19177 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19179 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19180 Change ins into fmov.
19182 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19184 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19185 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19186 Use h_con constraint for operand 1.
19187 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19188 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19190 2017-03-15 Jeff Law <law@redhat.com>
19192 PR tree-optimization/71437
19193 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19194 (record_temporary_equivalences): Use it.
19196 PR tree-optimization/71437
19197 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19198 tree-ssa-scopedtables.
19199 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19200 (record_conditions, record_cond, vuse_eq): Likewise.
19201 (record_edge_info): Adjust to API tweak of record_conditions.
19202 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19203 (record_temporary_equivalences, optimize_stmt): Likewise.
19204 (eliminate_redundant_computations): Likewise.
19205 (record_equivalences_from_stmt): Likewise.
19206 * tree-ssa-scopedtables.c: Include options.h and params.h.
19207 (vuse_eq): New function, moved from tree-ssa-dom.c
19208 (build_and_record_new_cond): Likewise.
19209 (record_conditions): Likewise. Accept vector of conditions rather
19210 than edge_equivalence structure for first argument.
19211 for the first argument.
19212 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19213 from tree-ssa-dom.c.
19214 (avail_exprs_stack::record_cond): Likewise.
19215 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19216 from tree-ssa-dom.c.
19217 (avail_exprs_stack): Add new member functions lookup_avail_expr
19219 (record_conditions): Declare.
19221 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19224 * lra-constraints.c (process_alt_operands): Increase reject for
19225 reloading an input/output operand.
19227 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19230 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19231 insns to convert from signed/unsigned char/short to IEEE 128-bit
19233 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19235 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19238 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19239 subreg of inner mode for values already in registers.
19241 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19243 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19244 iteration reg is used after the loop.
19246 2017-03-14 Martin Sebor <msebor@redhat.com>
19248 PR tree-optimization/79800
19249 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19250 precision in negative-positive range.
19251 (format_floating): Call non-const overload with adjusted precision.
19253 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19256 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19259 2017-03-14 Martin Sebor <msebor@redhat.com>
19261 PR middle-end/80020
19262 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19263 * builtins.def (aligned_alloc): Use it.
19266 * Makefile.in (GTFILES): Add calls.c.
19267 * calls.c: Include "gt-calls.h".
19269 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19271 PR rtl-optimization/79728
19272 * regs.h (struct target_regs): New field
19273 x_contains_allocatable_regs_of_mode.
19274 (contains_allocatable_regs_of_mode): New macro.
19275 * reginfo.c (init_reg_sets_1): Initialize it, and change
19276 contains_reg_of_mode so it includes global regs as well.
19277 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19278 rather than contains_regs_of_mode.
19280 2017-03-14 Martin Liska <mliska@suse.cz>
19282 * doc/invoke.texi: Document options that can't be combined with
19283 -fcheck-pointer-bounds.
19285 2017-03-14 Martin Liska <mliska@suse.cz>
19287 PR middle-end/79831
19288 * doc/invoke.texi (-Wchkp): Document the option.
19290 2017-03-14 Martin Liska <mliska@suse.cz>
19292 * Makefile.in: Install gcov-dump.
19294 2017-03-14 Martin Liska <mliska@suse.cz>
19296 * multiple_target.c (expand_target_clones): Bail out for
19297 an invalid attribute.
19299 2017-03-14 Richard Biener <rguenther@suse.de>
19301 * alias.c (struct alias_set_entry): Pack properly.
19302 * cfgloop.h (struct loop): Likewise.
19303 * cse.c (struct set): Likewise.
19304 * ipa-utils.c (struct searchc_env): Likewise.
19305 * loop-invariant.c (struct invariant): Likewise.
19306 * lra-remat.c (struct cand): Likewise.
19307 * recog.c (struct change_t): Likewise.
19308 * rtl.h (struct address_info): Likewise.
19309 * symbol-summary.h (function_summary): Likewise.
19310 * tree-loop-distribution.c (struct partition): Likewise.
19311 * tree-object-size.c (struct object_size_info): Likewise.
19312 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19313 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19314 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19315 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19316 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19317 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19318 (struct _stmt_vec_info): Likewise.
19320 2017-03-14 Martin Liska <mliska@suse.cz>
19323 * multiple_target.c (create_dispatcher_calls): Check that
19324 a target can create a function dispatcher.
19326 2017-03-14 Martin Liska <mliska@suse.cz>
19329 * multiple_target.c (expand_target_clones): Drop local.local
19330 flag for default implementation.
19332 2017-03-14 Richard Biener <rguenther@suse.de>
19334 PR tree-optimization/80030
19335 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19337 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19339 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19340 gcc_fallthrough() instead of __attribute__((fallthrough));
19342 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19344 * doc/gcc.texi: Remove "up" link to (DIR).
19345 * doc/gccint.texi: Ditto.
19347 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19349 * doc/install.texi (Specific) <avr>: Remove reference to
19352 2017-03-13 Jeff Law <law@redhat.com>
19354 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19355 attribute rather than comments.
19357 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19358 match_scratch operand is highest.
19360 2017-03-13 Martin Liska <mliska@suse.cz>
19362 PR middle-end/78339
19363 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19364 is a CHKP clone, use original declaration.
19366 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19368 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19369 (arc_conditional_register_usage): Use a different allocation order
19370 when optimizing for size.
19371 * common/config/arc/arc-common.c (arc_option_optimization_table):
19372 Section anchors default on when optimizing for size.
19374 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19376 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19378 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19380 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19381 * config/arc/arc.md (cpu_facility): Add cd variant.
19382 (*movqi_insn): Add code density variant.
19383 (*movhi_insn): Likewise.
19384 (*movqi_insn): Likewise.
19385 (*addsi3_mixed): Likewise.
19386 (subsi3_insn): Likewise.
19388 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19390 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19392 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19394 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19395 expressions with MINUS and UNARY ops.
19397 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19400 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19402 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19403 between vec_select and vector argument.
19404 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19405 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19406 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19407 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19408 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19409 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19411 2017-03-13 Richard Biener <rguenther@suse.de>
19414 * params.def (vect-max-peeling-for-alignment): Fix typo.
19416 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19418 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19419 issue that only occurred with binutils below 2.18.
19421 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19423 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19424 refer to binutils 2.11/2.12 minimum.
19426 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19428 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19429 ftp.kernel.org and simplify binutils requirement.
19431 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19433 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19435 (Optimize Options): Ditto. Also remove redundancy.
19437 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19439 PR translation/79848
19440 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19442 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19443 to G_ to avoid double translation.
19445 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19447 PR translation/79923
19448 * auto-profile.c (get_combined_location): Convert leading
19449 character of diagnostics to lower case and remove trailing period.
19450 (read_profile): Likewise for various diagnostics.
19451 * config/arm/arm.c (arm_option_override): Remove trailing period
19452 from various diagnostics.
19453 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19454 (msp430_expand_delay_cycles): Likewise.
19456 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19459 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19460 full command-line argument, rather than just "str".
19461 (aarch64_validate_march): Likewise.
19462 (aarch64_validate_mtune): Likewise.
19464 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19466 PR rtl-optimization/78911
19467 * lra-assigns.c (must_not_spill_p): New function.
19468 (spill_for): Use it.
19470 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19472 PR tree-optimization/79981
19473 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19474 ATOMIC_COMPARE_EXCHANGE ifn result.
19475 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19476 IFN_ATOMIC_COMPARE_EXCHANGE.
19478 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19481 * opts.c (parse_sanitizer_options): Add missing question mark to
19482 "did you mean" message.
19484 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19486 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19488 (VMULEUH_UNS): Likewise.
19489 (VMULOUB_UNS): Likewise.
19490 (VMULOUH_UNS): Likewise.
19491 * config/rs6000/rs6000.c (builtin_function_type): Remove
19492 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19494 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19497 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19498 x with result of extra_parsing_for_operand_code_0.
19499 (function_reader::extra_parsing_for_operand_code_0): Convert
19500 return type from void to rtx, returning x. When reading
19501 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19502 larger size containing struct block_symbol.
19504 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19506 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19507 -mfloat128-hardware without -m64.
19509 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19512 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19513 entries to the case statement that marks unsigned arguments to
19514 overloaded functions.
19516 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19518 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19519 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19521 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19524 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19525 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19527 2017-03-10 Martin Liska <mliska@suse.cz>
19531 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19532 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19535 2017-03-10 Olivier Hainque <hainque@adacore.com>
19537 * tree-switch-conversion (array_value_type): Start by resetting
19538 candidate type to it's main variant.
19540 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19542 PR rtl-optimization/79909
19543 * combine.c (try_combine): Use simplify_replace_rtx on individual
19544 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19545 of the whole CALL_INSN_FUNCTION_USAGE.
19547 PR tree-optimization/79972
19548 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19549 get_range_info on SSA_NAMEs. Formatting fixes.
19551 2017-03-10 Richard Biener <rguenther@suse.de>
19552 Jakub Jelinek <jakub@redhat.com>
19554 PR tree-optimization/77975
19555 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19556 edge to be constant.
19557 (get_val_for): For constant x return it. Formatting fix.
19558 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19559 would use the same bases as the current one.
19561 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19563 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
19564 instead of vec_select for V1TImode.
19565 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
19567 (VSX_LE_128): Add V1TI to this mode iterator.
19568 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
19569 (*vsx_le_perm_store_<mode>): Likewise.
19570 (pre-reload splitter for VSX stores): Likewise.
19571 (post-reload splitter for VSX stores): Likewise.
19572 (*vsx_xxpermdi2_le_<mode>): Likewise.
19573 (*vsx_lxvd2x2_le_<mode>): Likewise.
19574 (*vsx_stxvd2x2_le_<mode>): Likewise.
19576 2017-03-09 Michael Eager <eager@eagercon.com>
19578 Correct failures with --enable-checking=yes,rtl.
19580 * config/microblaze/microblaze.c (microblaze_expand_shift):
19581 Replace GET_CODE test with CONST_INT_P and INTVAL test with
19582 test for const0_rtx.
19583 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
19584 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
19586 2017-03-09 Richard Biener <rguenther@suse.de>
19588 PR tree-optimization/79977
19589 * graphite-scop-detection.c (scop_detection::merge_sese):
19590 Handle the case of extra exits to blocks dominating the entry.
19592 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
19594 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19597 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
19599 PR rtl-optimization/79949
19600 * lra-constraints.c (process_alt_operands): Check memory when
19601 trying to predict a cycle. Print about the overall increase.
19603 2017-03-09 Richard Biener <rguenther@suse.de>
19605 PR middle-end/79971
19606 * gimple-expr.c (useless_type_conversion_p): Preserve
19607 TYPE_SATURATING for fixed-point types.
19609 2017-03-09 Richard Biener <rguenther@suse.de>
19612 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
19613 alignment of BLKmode params.
19615 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19618 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
19619 (VALL_NO_V2Q): Likewise.
19621 * config/aarch64/aarch64-simd.md
19622 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
19624 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
19625 VALL_NO_V2Q mode iterator.
19626 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
19628 2017-03-09 Martin Liska <mliska@suse.cz>
19630 PR tree-optimization/79631
19631 * tree-chkp-opt.c (chkp_is_constant_addr): Call
19632 tree_int_cst_sign_bit just for INTEGER constants.
19634 2017-03-09 Martin Liska <mliska@suse.cz>
19638 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
19641 2017-03-09 Marek Polacek <polacek@redhat.com>
19644 * tree.c (inchash::add_expr): Handle TREE_VEC.
19646 2017-03-09 Martin Liska <mliska@suse.cz>
19649 (chkp_narrow_size_and_offset): New function.
19650 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
19651 (void chkp_parse_bit_field_ref): New function.
19652 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
19653 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
19655 2017-03-09 Martin Liska <mliska@suse.cz>
19658 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
19659 (chkp_find_bounds_1): Remove gcc_unreachable.
19661 2017-03-09 Jakub Jelinek <jakub@redhat.com>
19664 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
19665 BUILT_IN_SYNC*, determine the access type from the size suffix and
19666 always build a MEM_REF with that type. Handle forgotten
19667 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
19670 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
19671 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
19672 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
19673 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
19674 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
19675 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
19676 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
19677 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
19678 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
19679 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
19680 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
19681 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
19682 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
19683 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
19684 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
19685 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
19686 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
19687 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
19688 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
19689 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
19690 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
19691 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
19692 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
19693 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
19694 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
19695 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
19696 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
19697 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
19698 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
19699 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
19700 definitions outside of __OPTIMIZE__ guarded section.
19703 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
19704 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
19705 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
19706 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
19709 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19711 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
19712 ("vfenez<mode>"): Add missing constraints.
19714 2017-03-08 Martin Sebor <msebor@redhat.com>
19717 * config/nds32/nds32.c (nds32_option_override):
19718 Fix misspelled diagnostic.
19720 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19723 * gimplify.c (gimplify_omp_for): Replace index var in outer
19724 taskloop statement with an artificial variable and add
19725 OMP_CLAUSE_PRIVATE clause for it.
19727 2017-03-08 Richard Biener <rguenther@suse.de>
19729 PR tree-optimization/79955
19730 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19731 for accesses that are completely outside of the variable.
19733 2017-03-08 Andrew Haley <aph@redhat.com>
19735 PR tree-optimization/79943
19736 * tree-ssa-loop-split.c (compute_new_first_bound): When
19737 calculating the new upper bound, (END-BEG) should be added, not
19740 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19742 * config/avr/avr.md (setmemhi): Make sure match_dup
19743 operand number comes before match_scratch.
19745 2017-03-08 Richard Biener <rguenther@suse.de>
19747 PR tree-optimization/79920
19748 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19749 with ncopies == 1 to ...
19750 (vect_transform_slp_perm_load): ... here. Properly compute
19751 all element loads by iterating VF times over the group. Do
19752 not handle ncopies (computed in a broken way) in
19753 vect_create_mask_and_perm.
19755 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19758 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19759 is a uniform vector, use uniform_vector_p return value instead of
19760 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
19762 2017-03-07 Marek Polacek <polacek@redhat.com>
19764 PR middle-end/79809
19765 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
19766 (alloca_call_type): Likewise.
19768 2017-03-07 Martin Liska <mliska@suse.cz>
19770 * gcov.c (process_args): Put comment to correct location.
19772 2017-03-07 Martin Liska <mliska@suse.cz>
19774 PR middle-end/68270
19775 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
19776 Use array_at_struct_end_p instead of DECL_CHAIN (field).
19777 (chkp_narrow_bounds_for_field): Likewise.
19778 (chkp_parse_array_and_component_ref): Pass one more argument to
19781 2017-03-07 Richard Biener <rguenther@suse.de>
19783 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
19786 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19788 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
19789 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
19791 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19794 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
19795 to end of description.
19796 (PARAM_MAX_STORES_TO_MERGE): Likewise.
19798 2017-03-07 Jakub Jelinek <jakub@redhat.com>
19800 PR rtl-optimization/79901
19801 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
19803 (*avx512f_<code><mode>3<mask_name>): ... this.
19804 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
19805 iterator instead of VI8_AVX2_AVX512BW.
19807 PR rtl-optimization/79901
19808 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
19809 min/max expander, expand it using expand_vec_cond_expr.
19812 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
19815 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19818 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
19819 to void * for PCH reasons.
19820 * dwarf2out.c (output_loc_operands, output_die): Cast
19821 v.val_vec.array to unsigned char *.
19823 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
19826 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
19829 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
19831 PR rtl-optimization/79571
19832 * lra-constraints.c (process_alt_operands): Calculate static
19833 reject and subtract it from overall when only addresses will be
19836 2017-03-06 Julia Koval <julia.koval@intel.com>
19839 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
19840 incoming stack boundary to 128 for 64-bit targets.
19842 2017-03-06 Richard Biener <rguenther@suse.de>
19844 PR tree-optimization/79894
19845 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
19846 to NULL after folding it.
19848 2017-03-06 Richard Biener <rguenther@suse.de>
19850 PR tree-optimization/79824
19851 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
19852 check disabling peeling for gaps.
19854 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
19856 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19857 attributes): Document gettimeofday.
19859 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19861 * config/s390/s390.c (s390_option_override_internal): Set
19862 PARAM_MIN_VECT_LOOP_BOUND
19864 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19866 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
19867 * config/s390/s390.md: Likewise.
19869 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19872 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
19873 (<avx2_avx512>_perm<mode>): Rename to ...
19874 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
19876 (<avx512>_perm<mode>_mask): Rename to ...
19877 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
19879 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
19880 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
19881 instead of VI8F_256_512.
19882 (avx512f_perm<mode>): New define_expand.
19883 (avx512f_perm<mode>_mask): Likewise.
19884 (avx512f_perm<mode>_1<mask_name>): New define_insn.
19885 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
19887 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19889 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
19890 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
19892 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
19894 2017-03-06 Martin Liska <mliska@suse.cz>
19897 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
19898 when having a SSA NAME w/o VAR_DECL assigned to it.
19900 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19902 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
19903 msa_dpsub_<su>_d): Fix MODE for vec_select.
19905 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19907 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
19909 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
19911 2017-03-06 Richard Biener <rguenther@suse.de>
19913 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
19914 * plugin.c (register_plugin_info): Likewise.
19915 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
19917 2017-03-05 Jakub Jelinek <jakub@redhat.com>
19919 * config/i386/sse.md (sse_storehps, sse_storelps,
19920 avx_<castmode><avxsizesuffix>_<castmode>,
19921 avx512f_<castmode><avxsizesuffix>_<castmode>,
19922 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
19923 in condition that at least one operand is not a MEM.
19925 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19927 PR middle-end/79805
19928 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
19929 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
19931 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
19932 gimple_call_nothrow_p flag based on whether original builtin can throw.
19933 If it can, emit following stmts on the fallthrough edge.
19934 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
19935 don't create new bb if inserting just debug stmts on the edge, try to
19936 insert them on the fallthru bb or just reset debug stmts.
19938 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
19941 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
19942 restore recog_data (including the operand rtxes inside it) around
19943 the call to get_insn_template.
19945 2017-03-03 Martin Sebor <msebor@redhat.com>
19947 PR tree-optimization/79699
19948 * context.c (context::~context): Free MPFR caches to avoid
19949 a memory leak on program exit.
19951 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19953 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
19954 Use wide_int::ulow () instead of .elt (0).
19956 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19958 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
19959 (*pushxf): Limit oF constraint to 32bit targets and add oC
19960 constraint for 64bit targets.
19961 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
19962 (*pushdf): Change rmF constraint to rmC.
19964 2017-03-03 Martin Liska <mliska@suse.cz>
19966 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
19967 Remove unused variable.
19969 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19972 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
19973 is a memory operand, increase num_memory.
19974 (ix86_expand_args_builtin): Likewise.
19976 2017-03-03 Jan Hubicka <jh@suse.cz>
19979 * ipa-devirt.c (maybe_record_node): Properly handle
19980 __cxa_pure_virtual visibility.
19982 2017-03-03 Martin Liska <mliska@suse.cz>
19984 PR tree-optimization/79803
19985 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
19987 (pass_loop_prefetch::execute): Disabled optimization if an
19988 assumption about L1 cache size is not met.
19990 2017-03-03 Martin Liska <mliska@suse.cz>
19992 PR rtl-optimization/79574
19993 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
19994 (hash_scan_set): Likewise.
19995 (dump_hash_table): Likewise.
19996 (hoist_code): Likewise.
19998 2017-03-03 Richard Biener <rguenther@suse.de>
20000 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20001 in place of uhigh ().
20002 (fixed_convert_from_real): Likewise.
20004 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20007 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20009 2017-03-03 Richard Biener <rguenther@suse.de>
20011 PR middle-end/79818
20012 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20013 TYPE_OVERFLOW_UNDEFINED check.
20015 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20017 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20019 (vector_ae_<mode>_p): Likewise.
20020 (vector_nez_<mode>_p): Likewise.
20021 (vector_ne_v2di_p): Likewise.
20022 (vector_ae_v2di_p): Likewise.
20023 (vector_ne_<mode>_p): Likewise.
20024 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20026 (vsx_tsqrt<mode>2_fe): Likewise.
20028 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20031 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20033 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20035 PR rtl-optimization/79780
20036 * cprop.c (one_cprop_pass): When second and further conditional trap
20037 in a single basic block is turned into an unconditional trap, turn it
20038 into a deleted note to avoid RTL verification failures.
20040 2017-03-02 Richard Biener <rguenther@suse.de>
20042 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20044 2017-03-02 Richard Biener <rguenther@suse.de>
20046 PR tree-optimization/79345
20048 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20049 param and abort the walk, returning -1 if it is hit.
20050 (walk_aliased_vdefs): Take a limit param and pass it on.
20051 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20052 defaulting to 0 and return a signed int.
20053 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20054 (check_defs): New helper.
20055 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20056 about uninitialized memory.
20057 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20058 bogus uninitialized warning.
20059 (fixed_convert_from_real): Likewise.
20061 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20063 PR tree-optimization/66768
20064 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20065 iv_use if base object can't be determined.
20067 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20069 PR tree-optimization/79345
20070 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20071 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20072 (get_pattern_stats): Initialize it.
20073 * genemit.c (gen_expand): Verify match_scratch numbers come after
20074 match_operand/match_dup numbers.
20075 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20076 match_scratch numbers.
20077 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20079 * config/s390/s390.md (trunctdsd2): Likewise.
20081 2017-03-02 Richard Biener <rguenther@suse.de>
20083 * wide-int.h (wide_int_storage::operator=): Implement in terms
20086 2017-03-02 Richard Biener <rguenther@suse.de>
20088 PR tree-optimization/79777
20089 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20090 the to insert expression to sth existing.
20092 2017-03-01 Martin Sebor <msebor@redhat.com>
20094 PR middle-end/79692
20095 * gimple-ssa-sprintf.c
20096 (directive::known_width_and_precision): New function.
20097 (format_integer): Use it.
20098 (get_mpfr_format_length): Consider the full range of precision
20099 when computing %g output with the # flag. Set the likely byte
20100 count to 3 rather than 1 when precision is indeterminate.
20101 (format_floating): Correct the lower bound of precision.
20103 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20105 * doc/invoke.texi: Document default code model for 64-bit Linux.
20107 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20110 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20111 udiv rather than div since input pattern is unsigned.
20113 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20115 * config/i386/i386.c (print_reg): Warn for values of
20116 unsupported size in integer register.
20118 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20121 * config/rs6000/predicates.md (current_file_function_operand): Do
20122 not allow self calls to be local if the function is replaceable.
20124 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20127 * config/rs6000/altivec.h (vec_ctz and others): Change the
20128 preprocessor macro that controls conditional compilation from
20129 _ARCH_PWR9 to __POWER9_VECTOR__.
20130 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20131 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20132 control (instead of _ARCH_PWR9 control) so that template
20133 definition uses power9-specific function.
20134 (vec_any_eq): Likewise.
20135 (vec_all_ne): Change macro definition to use a power9-specific
20136 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20137 _ARCH_PWR9 control).
20138 (vec_any_eq) Likewise.
20139 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20140 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20141 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20142 support for xvcmpnedp instruction.
20143 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20144 macro expansion so that Power9 implementation of vec_all_ne does
20145 not use the AltiVec predicate framework.
20146 (VCMPNEH_P): Likewise.
20147 (VCMPNEW_P): Likewise.
20148 (VCMPNED_P): Likewise.
20149 (VCMPNEFP_P): Likewise.
20150 (VCMPNEDP_P): Likewise.
20151 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20152 implementation of vec_any_eq to not use AltiVec predicate
20154 (VCMPAEH_P): Likewise.
20155 (VCMPAEW_P): Likewise.
20156 (VCMPAED_P): Likewise.
20157 (VCMPAEFP_P): Likewise.
20158 (VCMPAEDP_P): Likewise.
20159 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20160 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20161 not use the AltiVec predicate framework.
20162 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20163 of vec_any_eq to not use AltiVec predicate framework.
20164 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20165 support for predefined __POWER9_VECTOR__ macro to indicate that
20166 Power9 instruction selection is enabled.
20167 (altivec_overloaded_builtins): Remove extraneous
20168 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20169 function argument types RS6000_BTI_bool_V16QI and
20170 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20171 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20172 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20173 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20174 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20175 Power9 for implementations of vec_cmpne. Change the signature for
20176 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20177 (representing vec_all_ne) to remove the previously described first
20178 argument of type RS6000_BTI_INTSI, as this was an artifact of
20179 reliance on the AltiVec predicate framework, which is no longer
20180 used in the implementation of these functions. Add
20181 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20182 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20183 since, unlike the AltiVec predicate framework implementation, we
20184 do not share function descriptors between vec_alle and vec_anyeq.
20185 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20186 set of modes that receive special treatment even when
20187 TARGET_P9_VECTOR is true. The special treatment emits code that
20188 does not depend on Power9 instructions.
20189 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20190 define_expand to not rely on AltiVec predicate framework.
20191 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20193 (vector_ne_v2di_p): Change this define_expand to not rely on
20194 AltiVec predicate framework.
20195 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20197 (vector_ne_<mode>_p): Change this define_expand to not rely on
20198 AltiVec predicate framework.
20199 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20201 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20202 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20203 define_insn pattern.
20204 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20205 define_insn pattern because the xvcmpne<VSs>. instruction is not
20207 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20208 instruction is not supported.
20210 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20212 * config/nvptx/nvptx.c: Include intl.h.
20214 2017-03-01 Martin Jambor <mjambor@suse.cz>
20217 * ipa-prop.h (ipa_bits): Removed field known.
20218 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20219 to pointers. Adjusted their comments to warn about their sharing.
20220 (ipcp_transformation_summary): Change bits to a vector of pointers.
20221 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20222 (ipa_get_ipa_bits_for_value): Declare.
20223 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20224 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20225 (ipa_bits_hash_table): Likewise.
20226 (ipa_vr_ggc_hash_traits): Likewise.
20227 (ipa_vr_hash_table): Likewise.
20228 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20229 being pointers and vr_known being removed.
20230 (ipa_set_jf_unknown): Likewise.
20231 (ipa_get_ipa_bits_for_value): New function.
20232 (ipa_set_jfunc_bits): Likewise.
20233 (ipa_get_value_range): New overloaded functions.
20234 (ipa_set_jfunc_vr): Likewise.
20235 (ipa_compute_jump_functions_for_edge): Use the above functions to
20236 construct bits and vr parts of jump functions.
20237 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20238 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20240 (ipcp_grow_transformations_if_necessary): Also allocate
20241 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20243 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20244 them. Fix too long lines.
20245 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20246 vr_known being removed.
20247 (ipa_read_jump_function): Use new setter functions to construct bits
20248 and vr parts of jump functions or set them to NULL.
20249 (write_ipcp_transformation_info): Adjust for bits being pointers.
20250 (read_ipcp_transformation_info): Likewise.
20251 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20253 Include gt-ipa-prop.h.
20254 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20256 (ipcp_store_bits_results): Likewise.
20257 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20258 Do not write to existing jump functions but use a temporary instead.
20260 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20263 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20264 attempt to use its first operand as BIT_FIELD_REF base.
20266 2017-03-01 Richard Biener <rguenther@suse.de>
20268 PR middle-end/79721
20269 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20270 interpolating formula in wrapping arithmetic.
20271 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20273 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20275 PR tree-optimization/79734
20276 * tree-vect-generic.c (expand_vector_condition): Optimize
20277 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20278 Handle VEC_COND_EXPR where comparison has different inner width from
20279 type's inner width.
20281 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20283 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20284 markup, and similar issues. Remove @opindex entries for things
20285 that aren't options. Add missing -mmpy-option entries.
20287 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20289 PR tree-optimization/79737
20290 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20291 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20292 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20293 instead of byte_size. Formatting fix.
20294 (shift_bytes_in_array_right): Formatting fix.
20296 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20299 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20300 condition on optimize for the leaf function test.
20302 2017-02-28 Martin Liska <mliska@suse.cz>
20305 * read-rtl-function.c (function_reader::handle_unknown_directive):
20306 Bail out when one uses -flto.
20308 2017-02-28 Martin Liska <mliska@suse.cz>
20310 * common.opt: Replace space with tabular for options of <number>
20312 * config/i386/i386.opt: Show <number> value for
20313 -mlarge-data-threshold.
20314 * opts.c (print_filtered_help): Do not display number in hexadecimal
20317 2017-02-28 Martin Liska <mliska@suse.cz>
20319 * common.opt: Fix --help=option -Q for options which are of
20322 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20324 * config/i386/i386.c (print_reg): Error out for values
20325 of 8-bit size in invalid integer register.
20327 2017-02-28 Martin Sebor <msebor@redhat.com>
20329 PR tree-optimization/79691
20330 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20332 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20335 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20336 gcc_unreachable with output_operand_lossage.
20338 2017-02-28 Richard Biener <rguenther@suse.de>
20340 PR tree-optimization/79740
20341 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20343 (visit_nary_op): Insert the nary into the hashtable if we
20344 pattern-matched sth.
20345 * tree-ssa-pre.c (eliminate_insert): Robustify.
20347 2017-02-28 Richard Biener <rguenther@suse.de>
20349 PR middle-end/79731
20350 * fold-const.c (decode_field_reference): Reject out-of-bound
20353 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20355 * config/i386/i386.c: Include intl.h.
20356 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20357 instead of just cond ? "..." : "...".
20358 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20359 * coverage.c (read_counts_file): Likewise.
20360 * omp-offload.c: Include intl.h.
20361 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20362 of just cond ? "..." : "...".
20363 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20364 of just cond ? "..." : "...".
20366 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20369 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20371 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20372 'tune for' CPU name.
20373 * config/arm/arm-cpu-data.h: Regenerated.
20375 2017-02-28 Richard Biener <rguenther@suse.de>
20377 PR tree-optimization/79732
20378 * tree-inline.c (expand_call_inline): Do not shadow var.
20380 2017-02-28 Richard Biener <rguenther@suse.de>
20382 PR tree-optimization/79723
20383 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20384 address-space properly.
20386 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20388 * doc/optinfo.texi (Optimization groups): Fix option used for
20390 * doc/invoke.texi (-fopt-info): Document "omp".
20391 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20392 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20393 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20394 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20395 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20397 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20399 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20400 OPTGROUP_OMP with "omp".
20402 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20405 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20406 for arithmetic shift of unsigned V2DI.
20408 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20410 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20411 arc/linux.h headers.
20412 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20413 (LINK_SPEC): Likewise.
20414 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20415 (EXTRA_SPECS): Likewise.
20416 (STARTFILE_SPEC): Likewise.
20417 (ENDFILE_SPEC): Likewise.
20418 (LIB_SPEC): Likewise.
20419 (TARGET_SDATA_DEFAULT): Likewise.
20420 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20421 (MULTILIB_DEFAULTS): Likewise.
20422 (DWARF2_UNWIND_INFO): Likewise.
20423 * config/arc/big.h: New file.
20424 * config/arc/elf.h: Likewise.
20425 * config/arc/linux.h: Likewise.
20426 * config/arc/t-uClibc: Remove.
20428 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20430 PR tree-optimization/77536
20431 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20432 (tree_transform_and_unroll_loop): Use above function to compute the
20433 estimated niter of unrolled loop and use it when scaling profile.
20434 Also use count info rather than frequency if it's non-zero.
20435 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20436 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20437 (vect_transform_loop): Call above function.
20439 2017-02-27 Richard Biener <rguenther@suse.de>
20441 PR tree-optimization/45397
20442 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20443 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20444 (visit_nary_op): Add pattern matching for CSEing sign-changed
20445 or truncated operations with wider ones.
20447 2017-02-27 Richard Biener <rguenther@suse.de>
20449 PR tree-optimization/79690
20450 * tree-vect-stmts.c (vectorizable_store): Use vector type
20451 built from the DR with address-space.
20453 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20455 * doc/invoke.texi (Optimize Options): Refine the description
20456 of asan-use-after-return.
20458 2017-02-25 Alan Modra <amodra@gmail.com>
20460 PR rtl-optimization/79584
20461 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20462 base, not ad->base_term, the reg within base. Remove assertion
20463 that ad->base == ad->base_term. Replace gen_int_mode using
20464 bogus mode with const0_rtx.
20466 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20468 PR middle-end/79396
20469 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20470 FMA_EXPR like tcc_binary or tcc_unary.
20472 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20475 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20477 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20478 (output_loc_operands): Handle DW_OP_call_ref and
20479 DW_OP_GNU_variable_value.
20480 (struct variable_value_struct): New type.
20481 (struct variable_value_hasher): Likewise.
20482 (variable_value_hash): New variable.
20483 (string_types): Remove.
20484 (copy_loc_descr): New function.
20485 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20486 (prepend_loc_descr_to_each): New function.
20487 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20488 instead of add_loc_descr_to_each if the first argument is single
20489 location list and the second has multiple.
20490 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20491 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20492 when looking for variable value which doesn't have other location info.
20493 (loc_list_from_tree): Formatting fix.
20494 (gen_array_type_die): Simplify DW_AT_string_length handling.
20495 (adjust_string_types): Remove.
20496 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20497 string_types. Call resolve_variable_values.
20498 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20499 (resolve_addr_in_expr): Likewise. Add A argument.
20500 (copy_deref_exprloc): Remove deref argument. Adjust for the
20501 original expression being DW_OP_GNU_variable_value with optionally
20502 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20503 optionally after it.
20504 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20505 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20506 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20507 (variable_value_hasher::hash, variable_value_hasher::equal): New
20509 (resolve_variable_value_in_expr, resolve_variable_value,
20510 resolve_variable_values, note_variable_value_in_expr,
20511 note_variable_value): New functions.
20512 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20515 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20518 * opts.h (handle_generated_option): Add GENERATED_P argument.
20519 * opts-common.c (handle_option): Adjust function comment.
20520 (handle_generated_option): Add GENERATED_P argument, pass it to
20522 (control_warning_option): Pass false to handle_generated_option
20524 * opts.c (maybe_default_option): Pass true to handle_generated_option
20526 * optc-gen.awk: Likewise.
20528 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20530 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20531 a REG, look at the REG it is a SUBREG of.
20532 (splitter for cmpeqsi_t): Ditto.
20534 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20536 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20537 the special USEs with the pattern of the insn, not the insn itself.
20539 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20542 * doc/invoke.texi: Document -mload-store-pairs.
20544 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20545 Sandra Loosemore <sandra@codesourcery.com>
20547 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20548 argument isn't a CONST_INT.
20549 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20550 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20551 (nios2_validate_compare): Bypass alternate compare logic if *op2
20552 is not a CONST_INT.
20553 (ldstwm_operation_p): Return false if first_base is not a REG or
20554 if first_offset is not a CONST_INT.
20556 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20558 * config/cris/cris.md: Use correct operand in a define_peephole2.
20560 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20562 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
20564 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20566 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
20567 this_insn if it is an INSN or JUMP_INSN.
20568 (force_offsettable): Look at base, not at addr.
20569 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
20570 on things that aren't necessarily CONST_INTs.
20572 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
20574 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
20575 -mfpmath=sse is the default also for x86-32 targets with SSE2
20576 instruction set when @option{-ffast-math} is enabled
20578 2017-02-24 Jeff Law <law@redhat.com>
20580 PR rtl-optimizatoin/79286
20581 * ira.c (update_equiv_regs): Drop may_trap_p exception to
20584 2017-02-24 Richard Biener <rguenther@suse.de>
20586 PR tree-optimization/79389
20587 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
20590 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
20592 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
20593 function comment to reflect reality.
20594 (loop_exits_before_overflow): Fix typo in function description.
20596 2017-02-24 Richard Biener <rguenther@suse.de>
20598 PR tree-optimization/79389
20599 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
20600 properly that a threading opportunity exists. Detect conditional
20601 copy/constant propagation opportunities.
20603 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
20605 * config/visium/visium.md (type): Add trap.
20606 (b): New mode attribute.
20607 (*btst): Rename into...
20608 (*btst<mode>): ...this and adjust.
20609 (*cbranchsi4_btst_insn): Rename into...
20610 (*cbranch<mode>4_btst_insn): ...this and adjust.
20611 (trap): New define_insn.
20613 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20615 PR tree-optimization/79389
20616 * ifcvt.c (struct noce_if_info): Add rev_cond field.
20617 (noce_reversed_cond_code): New function.
20618 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
20619 reversed_comparison_code. Formatting fix.
20620 (noce_try_store_flag): Test rev_cond != NULL in addition to
20621 reversed_comparison_code.
20622 (noce_try_store_flag_constants): Likewise.
20623 (noce_try_store_flag_mask): Likewise.
20624 (noce_try_addcc): Use rev_cond if non-NULL instead of
20625 reversed_comparison_code.
20626 (noce_try_cmove_arith): Likewise. Formatting fixes.
20627 (noce_try_minmax, noce_try_abs): Clear rev_cond.
20628 (noce_find_if_block): Initialize rev_cond.
20629 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
20630 instead of false as last argument never attempt to reverse it
20633 2017-02-23 Bin Cheng <bin.cheng@arm.com>
20635 PR tree-optimization/79663
20636 * tree-predcom.c (combine_chains): Process refs in reverse order
20637 only for ZERO length chains, and add explaining comment.
20639 2017-02-23 Jeff Law <law@redhat.com>
20641 PR tree-optimization/79578
20642 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
20643 in call to operand_equal_p.
20645 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
20648 * config/i386/cpuid.h: Fix another undefined behavior.
20650 2017-02-23 Richard Biener <rguenther@suse.de>
20652 PR tree-optimization/79683
20653 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
20654 vector types for data-refs.
20656 2017-02-23 Martin Liska <mliska@suse.cz>
20658 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
20660 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20662 PR middle-end/79665
20663 * internal-fn.c (get_range_pos_neg): Moved to ...
20664 * tree.c (get_range_pos_neg): ... here. No longer static.
20665 * tree.h (get_range_pos_neg): New prototype.
20666 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
20667 are known to be in between 0 and signed maximum inclusive, try to
20668 expand both unsigned and signed divmod and use the cheaper one from
20671 2017-02-22 Jeff Law <law@redhat.com>
20673 PR tree-optimization/79578
20674 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
20675 to compare base operands.
20677 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
20680 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
20681 gpc_reg_operand instead of fpr_reg_operand.
20683 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
20685 * config/mips/mips.c (mips_return_in_memory): Force FP
20686 vector types to be returned in memory for o32 ABI.
20688 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20690 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
20691 instead of DW_TAG_member for static data member declarations and don't
20692 set no_linkage_name for static inline data members.
20693 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
20696 2017-02-22 Martin Liska <mliska@suse.cz>
20698 * doc/invoke.texi: Replace inequality signs with square brackets
20701 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20703 PR tree-optimization/68644
20704 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
20706 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
20709 * lra-constraints.c (simplify_operand_subreg): Handle
20710 WORD_REGISTER_OPERATIONS targets.
20712 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20715 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
20716 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
20717 elimination by swapping fld*.
20719 2017-02-22 Richard Biener <rguenther@suse.de>
20721 PR tree-optimization/79673
20722 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
20723 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
20724 irrelevant address-space qualifiers and avoiding a
20725 ADDR_SPACE_CONVERT_EXPR from fold_convert.
20727 2017-02-22 Richard Biener <rguenther@suse.de>
20729 PR tree-optimization/79666
20730 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20731 to not symbolically negate if that may introduce undefined
20734 2017-02-22 Martin Liska <mliska@suse.cz>
20737 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20738 * data-streamer-out.c (streamer_write_gcov_count_stream):
20740 * value-prof.c (stream_out_histogram_value): Make assert more
20741 precise based on type of counter.
20743 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20746 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20747 Use nonimmediate_operand instead of memory_operand for operand 1.
20748 (float-extend standard_x87sse_constant_load splitter): Ditto.
20750 2017-02-21 Jeff Law <law@redhat.com>
20752 PR tree-optimization/79621
20753 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20754 blocks with edges to themselves.
20756 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20759 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20760 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20761 Use gimple_call_builtin_p.
20764 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
20765 on temporarily removed DEBUG_INSNs.
20767 PR tree-optimization/79649
20768 * tree-loop-distribution.c (classify_partition): Give up on
20769 non-generic address space loads/stores.
20771 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
20773 * doc/loop.texi (Loop manipulation): Remove nonexistent
20774 tree_ssa_loop_version from the documentation.
20775 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
20777 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20780 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
20781 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
20782 * config/rs6000/rs6000.c: Include except.h.
20783 (rs6000_expand_split_stack_prologue): Call
20784 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
20786 2017-02-21 Martin Jambor <mjambor@suse.cz>
20789 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
20790 have been analyzed.
20792 2017-02-21 Martin Jambor <mjambor@suse.cz>
20794 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
20795 for backward compatibility only.
20796 * doc/invoke.texi (Option Summary): Remove all references to
20797 -fipa-cp-alignment.
20799 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
20803 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20805 * lra-constraints.c (curr_insn_transform): Handle
20806 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20808 2017-02-21 Martin Liska <mliska@suse.cz>
20810 * config/i386/i386.opt: Replace -masm-dialect with -masm.
20812 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
20814 PR translation/79638
20815 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
20817 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
20820 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
20821 (arm_function_ok_for_sibcall): Return false for an indirect call by
20822 descriptor if all the argument registers are used.
20823 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
20824 alignment of the function.
20826 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20828 PR tree-optimization/61441
20829 * simplify-rtx.c (simplify_const_unary_operation): For
20830 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
20831 the sNaN unmodified.
20833 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20835 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20836 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20837 instead of SYSTEM_HEADER_DIR.
20839 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
20840 Martin Liška <mliska@suse.cz>
20842 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
20843 Fix typos and grammar, use active voice, and clarify.
20845 2017-02-20 Marek Polacek <polacek@redhat.com>
20847 PR middle-end/79537
20848 * gimplify.c (gimplify_expr): Handle unused *&&L;.
20851 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
20853 2017-02-20 Jakub Jelinek <jakub@redhat.com>
20856 * config/i386/i386.c (ix86_expand_builtin): Handle
20857 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
20858 ix86_builtins_isa[fcode].isa as a requirement of those
20859 flags and any other flag in the bitmask.
20860 (ix86_init_mmx_sse_builtins): Use 0 instead of
20861 ~OPTION_MASK_ISA_64BIT as mask.
20862 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
20863 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
20864 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
20865 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
20867 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20870 * lra-constraints.c (split_reg): Check requested split mode
20871 is supported by the register.
20873 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20875 * lra-constraints.c (simplify_operand_subreg): Remove early
20878 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20881 * lra-constraints.c (curr_insn_transform): Tighten condition
20882 for converting SUBREG reloads from OP_OUT to OP_INOUT.
20884 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20887 * lra-constraints.c (curr_insn_transform): Handle
20888 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20890 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
20893 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
20895 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
20897 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
20900 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
20903 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20905 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
20908 2017-02-18 Jakub Jelinek <jakub@redhat.com>
20910 * final.c (last_columnnum, override_columnnum): New variables.
20911 (final_start_function): Set last_columnnum, pass it to begin_prologue
20912 hook and pass 0 to dwarf2out_begin_prologue.
20913 (final_scan_insn): Update override_columnnum. Pass last_columnnum
20914 to source_line debug hook.
20915 (notice_source_line): Compute last_columnnum and for debug_column_info
20916 return true on column changes.
20917 * debug.h (struct gcc_debug_hooks): Add column argument to
20918 source_line and begin_prologue hooks.
20919 (debug_nothing_int_charstar_int_bool): Remove prototype.
20920 (debug_nothing_int_int_charstar,
20921 debug_nothing_int_int_charstar_int_bool): New prototypes.
20922 (dwarf2out_begin_prologue): Add column argument.
20923 * debug.c (do_nothing_debug_hooks): Adjust source_line and
20924 begin_prologue hooks.
20925 (debug_nothing_int_charstar_int_bool): Remove.
20926 (debug_nothing_int_int_charstar,
20927 debug_nothing_int_int_charstar_int_bool): New functions.
20928 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
20929 through to dwarf2out_source_line.
20930 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
20931 (dwarf2out_source_line): Add column argument, emit it if requested.
20932 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
20934 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20935 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20936 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
20937 through to dwarf2out_begin_prologue.
20938 (vmsdbgout_source_line): Add column argument, pass it through to
20939 dwarf2out_source_line.
20940 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
20941 dbxout_source_line caller.
20942 (dbxout_source_line): Add column argument.
20944 * common.opt (gno-column-info, gcolumn-info): New options.
20945 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
20946 (check_die): Also test for multiple DW_AT_decl_column attributes.
20947 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
20948 DW_AT_decl_column if requested.
20949 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
20951 (gen_variable_die): Likewise.
20952 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
20953 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
20956 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
20957 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
20958 (ix86_handle_option): Handle OPT_m3dnowa.
20959 * doc/invoke.texi (-m3dnowa): Document.
20960 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
20961 -m3dnowa instead of -m3dnow -march=athlon.
20964 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
20965 instead of gcc_assert for K, r and R code checks. Formatting fixes.
20967 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20970 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
20971 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
20972 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
20973 generator for vsx_xxpermdi_<mode>_be.
20974 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
20975 force big-endian semantics.
20976 (vsx_xxpermdi_<mode>_be): New define_expand with same
20977 implementation as previous version of vsx_xxpermdi_<mode>.
20979 2017-02-17 Jakub Jelinek <jakub@redhat.com>
20981 PR tree-optimization/79327
20982 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
20983 variable, its initialization and use.
20985 2017-02-17 Julia Koval <julia.koval@intel.com>
20987 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
20988 (OPTION_MASK_ISA_PKU_UNSET): New.
20989 (ix86_handle_option): Handle -mrdpid.
20990 * config/i386/cpuid.h (bit_RDPID): New.
20991 * config/i386/driver-i386.c (host_detect_local_cpu):
20992 Detect RDPID feature.
20993 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
20994 * config/i386/i386-c.c (ix86_target_macros_internal):
20996 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
20997 (ix86_valid_target_attribute_inner_p): Add "rdpid".
20998 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20999 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21000 * config/i386/i386.md (define_insn "rdpid"): New.
21001 * config/i386/i386.opt Add -mrdpid.
21002 * config/i386/immintrin.h (_rdpid_u32): New.
21004 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21006 PR rtl-optimization/79541
21007 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21008 instead of transforming it into USE.
21010 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21012 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21013 If HONOR_SNANS (SFmode) force the input to a register.
21014 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21015 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21016 an frsp or similar insn.
21018 2017-02-17 Martin Liska <mliska@suse.cz>
21020 PR rtl-optimization/79577
21021 * params.def (selsched-max-sched-times): Increase minimum to 1.
21023 2017-02-17 Martin Liska <mliska@suse.cz>
21025 PR rtl-optimization/79574
21026 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21028 2017-02-17 Martin Liska <mliska@suse.cz>
21030 PR tree-optimization/79529
21031 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21032 ssa_defined_default_def_p to handle cases which are implicitly
21034 * tree-ssa.c (ssa_defined_default_def_p): New function.
21035 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21036 which are implicitly defined.
21037 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21039 2017-02-17 Richard Biener <rguenther@suse.de>
21041 PR middle-end/79576
21042 * params.def (max-ssa-name-query-depth): Limit to 10.
21044 2017-02-17 Richard Biener <rguenther@suse.de>
21046 PR tree-optimization/79552
21047 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21050 2017-02-17 Richard Biener <rguenther@suse.de>
21053 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21055 2017-02-17 Marek Polacek <polacek@redhat.com>
21057 PR middle-end/79536
21058 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21059 (fold_negate_expr): New wrapper.
21061 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21063 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21064 Correct terminology and de-emphasize pre-standard behavior.
21066 2017-02-16 Alan Modra <amodra@gmail.com>
21068 PR rtl-optimization/79286
21069 * ira.c (def_dominates_uses): New function.
21070 (update_equiv_regs): Don't create an equivalence for insns that
21071 may trap where the register def does not dominate the use.
21073 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21075 PR rtl-optimization/78127
21076 * lra.c (lra): Call lra_eliminate before finish the loop after
21079 2017-02-16 Richard Biener <rguenther@suse.de>
21081 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21083 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21084 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21085 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21086 (isl_val_int_from_wi): New function.
21087 (extract_affine_gmp): Rename to ...
21088 (extract_affine_wi): ... this, take a widest_int.
21089 (extract_affine_int): Just wrap extract_affine_wi.
21090 (add_param_constraints): Use isl_val_int_from_wi.
21091 (add_loop_constraints): Likewise, and extract_affine_wi.
21093 2017-02-15 Jeff Law <law@redhat.com>
21095 PR middle-end/79521
21096 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21097 ira_init_register_move_cost_if_necessary.
21099 2017-02-15 Martin Sebor <msebor@redhat.com>
21101 PR middle-end/32003
21102 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21103 removed in a prior commit.
21105 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21107 PR tree-optimization/79347
21108 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21109 counters during peeling.
21111 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21113 * Makefile.in (site.exp): Remove "set ISLVER".
21115 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21118 * real.c (real_from_integer): Call real_convert even for decimal.
21120 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21123 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21125 2017-02-14 Andrew Pinski <apinski@cavium.com>
21127 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21128 cores and change the partno/implementer to be correct.
21129 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21130 the 'B" as the implementer.
21131 * config/aarch64/aarch64-tune.md: Regenerate.
21133 2017-02-14 Carl Love <cel@us.ibm.com>
21135 * config/rs6000/rs6000.c: Add case statement entry to make the
21136 xvcvuxdsp built-in argument unsigned.
21137 * config/rs6000/vsx.md: Fix the source and return operand types so they
21138 match the instruction definitions from the ISA document. Fix typo
21139 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21142 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21145 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21146 member early_clobber_alts.
21147 * lra-lives.c (reg_early_clobber_p): New.
21148 (process_bb_lives): Use it.
21149 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21150 (debug_operand_data): Initialize early_clobber_alts.
21151 (setup_operand_alternative): Set up early_clobber_alts.
21152 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21153 alternatives to new_insn_reg.
21154 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21156 (lra_update_insn_regno_info): Pass the new arg.
21158 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21160 PR middle-end/79505
21161 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21162 (new_oacc_loop_raw): Don't clear already cleared fields.
21165 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21166 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21167 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21169 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21172 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21174 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21177 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21178 the extra instruction to the right place to store 128-bit constant
21181 2017-02-14 Martin Sebor <msebor@redhat.com>
21183 PR middle-end/79448
21184 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21185 warning for strings of unknown length.
21187 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21189 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21191 2017-02-14 Jeff Law <law@redhat.com>
21194 * ira-costs.c (scan_one_insn): Initialize register move costs
21195 for pseudos seen in USE/CLOBBER insns.
21197 PR tree-optimization/79095
21198 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21199 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21200 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21201 if the operands are known to be not equal, then the resulting range
21203 (intersect_ranges): If the new range is ~[0,0] and the old range is
21204 wide, then prefer ~[0,0].
21205 * tree-vrp.c (overflow_comparison_p_1): New function.
21206 (overflow_comparison_p): New function.
21207 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21208 if NAME is used in an overflow test.
21209 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21210 overflow check that can be expressed as an equality test, then adjust
21211 ops to be that equality test.
21213 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21215 * config/s390/s390-builtin-types.def: Remove flags argument.
21216 * config/s390/s390.c (s390_init_builtins): Likewise.
21218 2017-02-14 Martin Liska <mliska@suse.cz>
21220 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21221 vector. Fix trailing white spaces.
21223 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21225 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21228 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21230 PR rtl-optimization/68664
21231 * config/arm/arm.c (arm_sched_can_speculate_insn):
21232 New function. Declare prototype.
21233 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21235 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21237 PR rtl-optimization/68664
21238 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21240 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21242 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21244 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21245 max skip bytes for function, loop and jump.
21247 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21249 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21250 ABS_EXPR for gimple dump.
21252 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21255 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21257 PR tree-optimization/79408
21258 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21259 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21260 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21261 also if rhs1 is INTEGER_CST.
21263 2017-02-14 Richard Biener <rguenther@suse.de>
21265 PR middle-end/79432
21266 * tree-into-ssa.c (insert_phi_nodes): When the function can
21267 have abnormal edges rewrite SSA names with broken use-def
21268 dominance out of SSA and register them for PHI insertion.
21270 2017-02-13 Martin Sebor <msebor@redhat.com>
21272 PR middle-end/79496
21273 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21274 clearing info.nowrite flag when snprintf size argument is a range.
21276 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21278 * cprop.c (cprop_jump): Add missing space in string literal.
21279 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21280 (get_constraint_for_component_ref): Likewise.
21281 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21282 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21283 * lra-constraints.c (process_alt_operands): Likewise.
21284 * ipa-inline.c (inline_small_functions): Likewise.
21285 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21286 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21287 * trans-mem.c (diagnose_tm_1_op): Likewise.
21288 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21289 (grid_parallel_clauses_gridifiable): Likewise.
21291 * config/nvptx/mkoffload.c (process): Add space in between
21294 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21295 "MOD4_SSE_REGS" and "ALL_REGS".
21297 * spellcheck.c (test_data): Add , in between "foo" and "food".
21299 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21302 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21303 boundary crossing check and subsequent code generation agree.
21305 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21307 * config/aarch64/aarch64.c (has_memory_op): Delete.
21308 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21311 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21313 PR rtl-optimization/79388
21314 PR rtl-optimization/79450
21315 * combine.c (distribute_notes): When removing TEM_INSN for which
21316 corresponding dest has last value recorded, invalidate that last
21319 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21321 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21322 of explicit '@'. Add missing assembly comment marker on branch costs
21325 2017-02-13 Nathan Sidwell <nathan@acm.org>
21327 * gengtype-lex.l (<in_struct>): Add '/'.
21329 2017-02-13 Martin Liska <mliska@suse.cz>
21332 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21334 2017-02-13 Richard Biener <rguenther@suse.de>
21336 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21338 * configure: Re-generate.
21339 * config.in: Likewise.
21340 * graphite-dependences.c: Simplify as if
21341 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21342 * graphite-isl-ast-to-gimple.c: Likewise.
21343 * graphite-optimize-isl.c: Likewise.
21344 * graphite-poly.c: Likewise.
21345 * graphite-sese-to-poly.c: Likewise.
21346 * graphite.h: Likewise.
21347 * toplev.c: Include isl/version.h and use isl_version () for
21348 printing the ISL version.
21349 * doc/install.texi: Update ISL requirement.
21351 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21353 * doc/standards.texi (Standards): Update reference to
21356 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21358 * doc/extend.texi (Named Address Spaces): sourceware.org now
21360 * doc/install.texi (Binaries): Ditto.
21363 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21365 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21366 standard terminology "stringize"/"stringizing" throughout.
21367 * doc/cppinternals.texi: Likewise.
21369 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21371 * doc/extend.texi: Fix some spelling mistakes and typos.
21372 * doc/invoke.texi: Likewise.
21374 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21377 * params.def (inline-min-speedup) Change from 10 to 8.
21379 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21381 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21384 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21387 * ipa-inline-analysis.c (get_minimal_bb): New function.
21388 (record_modified): Use it.
21389 (remap_edge_change_prob): Handle also ancestor functions.
21391 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21393 * doc/contrib.texi (Contributors): Remove broken link into
21394 the Mauve CVS repository.
21396 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21398 PR middle-end/79454
21399 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21400 result computation whenever lhs doesn't have vector mode, not
21401 just when it has BLKmode.
21403 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21405 * doc/makefile.texi (profiledbootstrap): Refer to the
21406 installation instructions only in textual form.
21408 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21411 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21413 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21415 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21416 (Specific): Update mingw-w64 reference.
21418 (Specific): Remove broken link to Renesas RX processor.
21420 2017-02-10 Richard Biener <rguenther@suse.de>
21422 * toplev.c (process_options): Do not mention obsolete graphite
21423 options when printing sorry message about missing graphite support.
21424 Mention -floop-nest-optimize.
21426 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21428 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21429 (vtst_p16): Likewise.
21430 (vtstq_p8): Likewise.
21431 (vtstq_p16): Likewise.
21433 (vtstq_p64): Likewise.
21434 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21435 (vset_lane_p64): New.
21436 (vsetq_lane_p64): New.
21438 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21440 PR tree-optimization/79411
21441 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21442 stmt operands are SSA_NAMEs used in abnormal phis.
21443 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21446 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21449 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21452 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21454 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21456 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21458 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21462 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21463 not arbitrary TREE_CONSTANT.
21466 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21467 "omp declare target link" attribute unless is_global_var.
21468 * omp-offload.c (find_link_var_op): Likewise.
21470 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21471 Chung-Lin Tang <cltang@codesourcery.com>
21473 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21475 (gimplify_adjust_omp_clauses): Don't delete TILE.
21476 (gimplify_omp_for): Deal with TILE.
21477 * internal-fn.c (expand_GOACC_TILE): New function.
21478 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21480 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21481 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21483 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21484 avoid DIV for outermost collapse var.
21485 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21486 Remove out of date comments, fix whitespace.
21487 * omp-general.c (omp_extract_for_data): Deal with tiling.
21488 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21489 adjust OLF_DIM_BASE value.
21490 (struct omp_for_data): Add tiling field.
21491 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21492 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21493 for auto loops. Remove default auto determining, moved to
21494 oacc_loop_fixed_partitions.
21495 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21496 stmts, add e_mask field.
21497 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21498 (oacc_thread_numbers): Use oacc_dim_call.
21499 (oacc_xform_tile): New.
21500 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21501 (finish_oacc_loop): Adjust for ifns vector.
21502 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21503 add case for GOACC_TILE fns.
21504 (oacc_loop_xform_loop): Delete.
21505 (oacc_loop_process): Iterate over call list directly, and add
21506 handling for GOACC_TILE fns.
21507 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21509 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21510 vector partitioning to outer loops. Assign 2 partitions to loops
21511 when available. Add TILE handling.
21512 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21513 (execite_oacc_device_lower): Process GOACC_TILE fns,
21514 ignore unknown specs.
21515 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21516 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21517 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21519 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21521 * configure.ac (ACX_BUGURL): Update.
21522 * configure: Regenerate.
21524 2017-02-09 Richard Biener <rguenther@suse.de>
21526 PR tree-optimization/69823
21527 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21528 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21530 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21532 * config/arc/arc-c.def: Add __NPS400__ definition.
21533 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21534 (TARGET_NPS400): Define.
21536 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21538 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21540 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21541 pointer, arch_info.
21542 (arc_cpu_types): Fill the arch_info field with a pointer into the
21543 arc_arch_types table.
21544 (arc_selected_cpu): Declare.
21545 * config/arc/arc.c (arc_selected_cpu): Make global.
21546 (arc_selected_arch): Delete.
21547 (arc_base_cpu): Delete.
21548 (arc_override_options): Remove references to deleted variables,
21549 update access to arch information.
21550 (ARC_OPT): Update access to arch information.
21551 (ARC_OPTX): Likewise.
21552 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21553 (TARGET_ARC600): Update access to arch information.
21554 (TARGET_ARC601): Likewise.
21555 (TARGET_ARC700): Likewise.
21556 (TARGET_EM): Likewise.
21557 (TARGET_HS): Likewise.
21558 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
21561 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
21564 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
21565 condition/operands for integer GE/LE/GEU/LEU operations.
21567 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
21569 PR translation/79397
21570 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
21573 2017-02-08 Martin Jambor <mjambor@suse.cz>
21576 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
21577 whether allocation happened.
21578 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
21579 nothing was allocated.
21581 2017-02-08 Jakub Jelinek <jakub@redhat.com>
21583 PR tree-optimization/79408
21584 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
21585 constant, but SSA_NAME with a known integer range, use the minimum
21586 of that range instead of op1 to determine if modulo can be replaced
21587 with its first operand.
21589 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21591 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
21593 2017-02-08 Richard Biener <rguenther@suse.de>
21595 PR tree-optimization/71824
21596 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21597 Check all loops contained in the merged region.
21599 2017-02-07 Andrew Pinski <apinski@cavium.com>
21601 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
21603 2017-02-07 Andrew Pinski <apinski@cavium.com>
21605 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
21606 (thunderxt88): Likewise.
21607 (thunderxt81): Disable LSE and change v8.1 to v8.
21608 (thunderxt83): Likewise.
21610 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21611 Richard Biener <rguenther@suse.de>
21613 PR middle-end/79399
21614 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
21615 type from int to size_t.
21616 * ira-costs.c (struct_costs_size): Change type from int to size_t.
21618 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21620 PR rtl-optimization/79386
21621 * cprop.c (bypass_conditional_jumps): Initialize
21622 bypass_last_basic_block already before splitting bbs after
21623 unconditional traps...
21624 (bypass_conditional_jumps): ... rather than here.
21627 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
21628 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
21629 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
21630 fix -masm=intel patterns.
21632 2017-02-07 Richard Biener <rguenther@suse.de>
21634 PR tree-optimization/79256
21635 PR middle-end/79278
21636 * builtins.c (get_object_alignment_2): Use min_align_of_type
21637 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
21638 and ADJUST_FIELD_ALIGN.
21640 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
21642 * doc/tm.texi: Regenerate.
21643 * stor-layout.c (layout_decl): Adjust.
21644 (update_alignment_for_field): Likewise.
21645 (place_field): Likewise.
21646 (min_align_of_type): Likewise.
21647 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
21648 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
21649 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
21650 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
21651 * config/frv/frv.c (frv_adjust_field_align): Likewise.
21652 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
21653 * config/i386/i386.c (x86_field_alignment): Likewise.
21654 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
21655 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
21656 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
21657 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
21658 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
21659 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
21663 2017-01-30 Richard Biener <rguenther@suse.de>
21665 PR tree-optimization/79256
21666 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21667 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21670 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
21672 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
21673 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
21674 builtins to SImode and emit a zero-extend, if necessary.
21676 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21678 * docs/invoke.texi (RISC-V Options): Alphabetize.
21680 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21682 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
21685 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21687 * config/riscv/riscv.c: New file.
21688 * common/config/riscv/riscv-common.c: Likewise.
21689 * config.gcc: Likewise.
21690 * config/riscv/constraints.md: Likewise.
21691 * config/riscv/elf.h: Likewise.
21692 * config/riscv/generic.md: Likewise.
21693 * config/riscv/linux.h: Likewise.
21694 * config/riscv/multilib-generator: Likewise.
21695 * config/riscv/peephole.md: Likewise.
21696 * config/riscv/pic.md: Likewise.
21697 * config/riscv/predicates.md: Likewise.
21698 * config/riscv/riscv-builtins.c: Likewise.
21699 * config/riscv/riscv-c.c: Likewise.
21700 * config/riscv/riscv-ftypes.def: Likewise.
21701 * config/riscv/riscv-modes.def: Likewise.
21702 * config/riscv/riscv-opts.h: Likewise.
21703 * config/riscv/riscv-protos.h: Likewise.
21704 * config/riscv/riscv.h: Likewise.
21705 * config/riscv/riscv.md: Likewise.
21706 * config/riscv/riscv.opt: Likewise.
21707 * config/riscv/sync.md: Likewise.
21708 * config/riscv/t-elf-multilib: Likewise.
21709 * config/riscv/t-linux: Likewise.
21710 * config/riscv/t-linux-multilib: Likewise.
21711 * config/riscv/t-riscv: Likewise.
21712 * configure.ac: Likewise.
21713 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
21714 Waterman as RISC-V maintainers.
21715 * doc/install.texi: Add RISC-V entries.
21716 * doc/invoke.texi: Add RISC-V options section.
21717 * doc/md.texi: Add RISC-V constraints section.
21718 * configure: Regenerated.
21720 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21723 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
21724 false values to be constant vectors with all 0 or all 1 bits set.
21725 (vcondu<mode><mode>): Likewise.
21726 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21728 (fpmask_comparison_operator): Update comment.
21729 (vecint_comparison_operator): New predicate.
21730 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21731 vector conditionals when the true and false values are constant
21732 vectors with all 0 bits or all 1 bits set.
21734 2017-02-06 Martin Sebor <msebor@redhat.com>
21736 PR tree-optimization/79376
21737 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21739 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21741 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21742 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21743 to simplify split condition.
21745 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21747 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21748 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21751 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21753 PR rtl-optimization/68664
21754 * target.def (can_speculate_insn): New hook.
21755 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21756 * doc/tm.texi: Regenerate.
21757 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21758 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21759 (rs6000_sched_can_speculate_insn): New function.
21761 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21763 PR tree-optimization/79284
21764 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
21765 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
21766 vectorizable_mask_load_store, vectorizable_operation,
21767 vect_is_simple_cond, get_same_sized_vectype): Use it instead
21768 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
21769 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
21770 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
21771 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
21772 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
21773 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
21774 is_gimple_assign (stmt). Replace another such test with
21775 is_gimple_assign (stmt).
21777 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21780 * config/avr/avr.c (rtl-iter.h): Include it.
21781 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
21782 (avr_legitimate_combined_insn): ...and implementation.
21784 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
21786 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
21787 * config/s390/s390.c (s390_const_operand_ok)
21788 (s390_canonicalize_comparison, s390_extract_part)
21789 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
21790 (s390_contiguous_bitmask_p, s390_rtx_costs)
21791 (legitimize_pic_address): Likewise.
21792 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
21793 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
21794 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
21795 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
21796 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
21798 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21800 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
21801 REGNO($0) == REGNO($1).
21803 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21805 * config/s390/linux.h(SIZE_TYPE): Add comment.
21807 2017-02-06 Julian Brown <julian@codesourcery.com>
21808 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21809 Virendra Pathak <virendra.pathak@broadcom.com>
21811 * config/aarch64/aarch64-cores.def: Change the scheduler
21813 * config/aarch64/aarch64.md: Include thunderx2t99.md.
21814 * config/aarch64/thunderx2t99.md: New file.
21816 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21818 * doc/standards.texi (Go Language): Update link to language
21821 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21823 * tree-eh.c (lower_resx): Sanitize profile.
21824 (cleanup_empty_eh_move_lp): Likewise.
21826 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21829 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
21831 * cfgloopmanip.h (loop_version): Update prototype.
21832 * modulo-sched.c (sms_schedule): Update call of loop_version.
21833 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
21834 * tree-parloops.c (gen_parallel_loop): Likewise.
21835 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
21836 * tree-ssa-loop-split.c (split_loop): Likewise.
21837 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
21838 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21840 2017-02-05 Martin Liska <mliska@suse.cz>
21843 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
21845 (print_operand_address): Initialize a struct to zero.
21847 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21849 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
21850 garbage collector only in textual form.
21852 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21854 * doc/extend.texi (x86 specific memory model extensions for
21855 transactional memory): Simplify a phrase.
21857 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
21860 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
21861 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
21862 (atomic_storedi_1): Likewise.
21864 2017-02-04 Jakub Jelinek <jakub@redhat.com>
21866 PR tree-optimization/79338
21867 * tree-parloops.c (gather_scalar_reductions): Don't call
21868 vect_analyze_loop_form for loop->inner before destroying loop's
21871 2017-02-03 Martin Sebor <msebor@redhat.com>
21873 PR tree-optimization/79327
21874 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
21875 when precision has resulted in leading zeros.
21876 (format_integer): Adjust the likely counter to assume an unknown
21877 argument that may be zero is non-zero.
21879 2017-02-03 Jason Merrill <jason@redhat.com>
21882 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
21883 avoid copying non-taken branch.
21885 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21887 PR tree-optimization/79340
21888 * tree-vect-loop.c (vectorizable_reduction): Release
21889 vec_defs elements after safe_splicing them into other vectors.
21892 PR tree-optimization/79327
21893 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
21894 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
21896 (format_integer): Use wide_int_to_tree instead of build_int_cst
21897 + to_?hwi. If argmin is NULL, just set argmin and argmax to
21898 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
21899 of shortest and longest sequence.
21901 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
21903 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
21904 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
21906 2017-02-03 Walter Lee <walt@tilera.com>
21909 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
21910 after initial stackframe link reg save.
21911 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
21913 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21916 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
21917 wu for stxssp alternative.
21919 2017-02-03 Martin Sebor <msebor@redhat.com>
21921 PR tree-optimization/79352
21922 * gimple-fold.c (get_range_strlen): Add argument.
21923 (get_range_strlen): Change return type to bool.
21924 (get_maxval_strlen): Pass in a dummy argument.
21925 * gimple-fold.h (get_range_strlen): Change return type to bool.
21926 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
21927 * tree.h (array_at_struct_end_p): Add argument.
21928 * tree.c (array_at_struct_end_p): Handle it.
21930 2017-02-03 Martin Liska <mliska@suse.cz>
21933 * multiple_target.c (create_dispatcher_calls): Redirect edge
21934 from a caller of a dispatcher.
21935 (expand_target_clones): Make the clones local.
21936 (ipa_target_clone): Do both target clones and resolvers.
21937 (ipa_dispatcher_calls): Remove the pass.
21938 (pass_dispatcher_calls::gate): Likewise.
21939 (make_pass_dispatcher_calls): Likewise.
21940 * passes.def (pass_target_clone): Put as very first IPA early
21943 2017-02-03 Martin Liska <mliska@suse.cz>
21945 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
21946 in case of a function with ifunc attribute.
21948 2017-02-03 Martin Liska <mliska@suse.cz>
21950 * cgraph.c (cgraph_node::dump): Dump function version info.
21951 * symtab.c (symtab_node::dump_base): Add missing new line.
21953 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21955 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
21956 (ifcombine_ifandif): Use it.
21958 2017-02-03 Martin Liska <mliska@suse.cz>
21960 * doc/invoke.texi: Document default value for
21961 use-after-scope-direct-emission-threshold.
21963 2017-02-03 Martin Liska <mliska@suse.cz>
21965 PR tree-optimization/79339
21966 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
21967 (format_floating): Likewise.
21969 2017-02-03 Martin Liska <mliska@suse.cz>
21972 * ipa-prop.c (ipa_node_params_t::insert): Remove current
21974 (ipa_node_params_t::remove): Likewise.
21975 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
21976 initialization from removed ipa_node_params_t::insert.
21977 (ipa_node_params::~ipa_node_params): Move from removed
21978 ipa_node_params_t::release.
21979 * symbol-summary.h (symbol_summary::m_released): New member.
21980 Do not release a summary twice. Do not allow to call finalizer
21981 for types of a summary that live in GGC memory.
21983 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21985 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
21988 2017-02-02 Martin Sebor <msebor@redhat.com>
21990 PR middle-end/79275
21991 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
21992 (format_string): Tighten up the range of output for non-constant
21993 strings and correct the expected range for wide non-constant strings.
21995 2017-02-02 Martin Sebor <msebor@redhat.com>
21997 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
21999 PR middle-end/32003
22000 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22002 (-fdump-tree-@var): Add to index and document how to come up
22003 with pass-specific option and dump file names.
22004 (-fdump-passes): Clarify where to look for output.
22006 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22008 PR middle-end/77445
22009 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22010 statistics of the analyzed path; allow threading for speed when
22011 any of BBs along the path are optimized for speed.
22013 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22015 PR middle-end/78468
22016 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22017 settings of the virtual registers.
22020 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22022 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22023 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22026 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22028 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22029 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22031 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22033 * config/s390/s390.md: Add missing comments with the expanded
22035 * config/s390/vector.md: Likewise.
22036 * config/s390/vx-builtins.md: Likewise.
22038 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22041 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22042 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22043 conditions on a single line.
22045 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22047 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22048 __S390_VX__ to __VX__.
22050 2017-02-01 Andrew Pinski <apinski@cavium.com>
22052 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22053 stmt_info to record_stmt_cost.
22054 (vect_get_known_peeling_cost): Pass stmt_info if known to
22056 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22057 cpu_vector_cost field into
22058 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22059 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22060 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22061 splitting of scalar_stmt_cost and vec_stmt_cost.
22062 (thunderx_vector_cost): Likewise.
22063 (cortexa57_vector_cost): LIkewise.
22064 (exynosm1_vector_cost): Likewise.
22065 (xgene1_vector_cost): Likewise.
22066 (thunderx2t99_vector_cost): Improve after the splitting of the two
22068 (aarch64_builtin_vectorization_cost): Update for the splitting of
22069 scalar_stmt_cost and vec_stmt_cost.
22071 2017-02-01 Torvald Riegel <triegel@redhat.com>
22072 Richard Henderson <rth@redhat.com>
22074 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22075 conditional on existance of a fast atomic load.
22076 * optabs-query.c (can_atomic_load_p): New function.
22077 * optabs-query.h (can_atomic_load_p): Declare it.
22078 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22079 no fast atomic load is available for the particular size of access.
22080 (expand_atomic_compare_and_swap): Likewise.
22081 (expand_atomic_load): Likewise.
22082 (expand_atomic_store): Likewise.
22083 (expand_atomic_fetch_op): Likewise.
22084 * testsuite/lib/target-supports.exp
22085 (check_effective_target_sync_int_128): Remove x86 because it provides
22086 no fast atomic load.
22087 (check_effective_target_sync_int_128_runtime): Likewise.
22089 2017-02-01 Richard Biener <rguenther@suse.de>
22091 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22092 (graphite_transform_loops): Provide opt-info for optimized nests.
22093 * tree-parloop.c (parallelize_loops): Provide opt-info for
22094 parallelized loops.
22096 2017-02-01 Richard Biener <rguenther@suse.de>
22098 PR middle-end/79315
22099 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22100 was not set before.
22102 2017-02-01 Richard Biener <rguenther@suse.de>
22104 PR tree-optimization/71824
22105 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22106 Verify the loops are valid in the merged SESE region.
22107 (scop_detection::can_represent_loop_1): Check analyzing the
22108 evolution of the number of iterations in the region succeeds.
22110 2017-01-31 Ian Lance Taylor <iant@golang.org>
22112 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22113 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22115 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22117 PR preprocessor/79210
22118 * input.c (get_substring_ranges_for_loc): Replace line_width
22119 assertion with error-handling.
22121 2017-01-31 Richard Biener <rguenther@suse.de>
22123 PR tree-optimization/77318
22124 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22125 (create_pw_aff_from_tree): Take loop parameter.
22126 (add_condition_to_pbb): Pass loop of the condition to
22127 create_pw_aff_from_tree.
22129 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22131 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22132 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22134 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22138 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22140 (convert_int_to_float128): Likewise.
22141 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22142 (convert_int_to_float128): Likewise.
22143 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22144 (UNSPEC_IEEE128_CONVERT): Likewise.
22145 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22146 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22147 Use local variables for IBM extended format.
22148 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22149 (fix_trunc<mode>si2_fprs): Likewise.
22150 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22151 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22152 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22153 to know that we can now have integers of all sizes in vector
22155 (fix<uns>_<mode>di2_hw): Likewise.
22156 (float<uns>_<mode>si2_hw): Likewise.
22157 (fix_<mode>si2_hw): Likewise.
22158 (fixuns_<mode>si2_hw): Likewise.
22159 (float<uns>_<mode>di2_hw): Likewise.
22160 (float_<mode>di2_hw): Likewise.
22161 (float_<mode>si2_hw): Likewise.
22162 (floatuns_<mode>di2_hw): Likewise.
22163 (floatuns_<mode>si2_hw): Likewise.
22164 (xscvqp<su>wz_<mode>): Delete, no longer used.
22165 (xscvqp<su>dz_<mode>): Likewise.
22166 (xscv<su>dqp_<mode>): Likewise.
22167 (ieee128_mfvsrd_64bit): Likewise.
22168 (ieee128_mfvsrd_32bit): Likewise.
22169 (ieee128_mfvsrwz): Likewise.
22170 (ieee128_mtvsrw): Likewise.
22171 (ieee128_mtvsrd_64bit): Likewise.
22172 (ieee128_mtvsrd_32bit): Likewise.
22174 2017-01-31 Martin Liska <mliska@suse.cz>
22177 * ipa-prop.c (ipa_free_all_node_params): Call release method
22178 instead of ~sumbol_summary to not to trigger double times
22181 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22183 PR tree-optimization/71691
22184 * bitmap.h (class auto_bitmap): New.
22185 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22186 is_maybe_undefined instead of ssa_undefined_value_p.
22188 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22190 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22191 __S390_ARCH_LEVEL__ to __ARCH__.
22193 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22195 PR tree-optimization/79267
22196 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22197 if should_remove_lhs_p is true.
22199 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22202 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22203 (add_alignment_attribute): New.
22204 (base_type_die): Add alignment attribute.
22205 (subrange_type_die): Likewise.
22206 (modified_type_die): Likewise.
22207 (gen_array_type_die): Likewise.
22208 (gen_descr_array_type_die: Likewise.
22209 (gen_enumeration_type_die): Likewise.
22210 (gen_subprogram_die): Likewise.
22211 (gen_variable_die): Likewise.
22212 (gen_field_die): Likewise.
22213 (gen_ptr_to_mbr_type_die): Likewise.
22214 (gen_struct_or_union_type_die): Likewise.
22215 (gen_subroutine_type_die): Likewise.
22216 (gen_typedef_die): Likewise.
22217 (base_type_cmp): Compare alignment attribute.
22219 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22222 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22223 (setb_unsigned) New pattern for setb with CCUNS.
22224 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22225 subfc./subfe sequence to avoid overflow problems. Generate a
22226 shorter sequence with cmpld/setb for power9.
22227 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22228 for generating subfc. instruction.
22229 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22230 now uses this instruction.
22232 2017-01-30 Ian Lance Taylor <iant@google.com>
22235 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22236 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22238 2017-01-30 Martin Sebor <msebor@redhat.com>
22240 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22241 Move constant to the right of a relational operator.
22242 (get_mpfr_format_length, format_character, format_string): Ditto.
22243 (should_warn_p, maybe_warn): Same.
22245 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22247 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22250 * asan.c (get_translation_unit_decl): Remove function.
22251 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22253 2017-01-30 Martin Liska <mliska@suse.cz>
22255 PR gcov-profile/79259
22256 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22257 -fprofile-generate.
22259 2017-01-30 Martin Liska <mliska@suse.cz>
22262 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22263 Initialize variables with NULL value.
22265 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22268 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22270 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22272 2017-01-30 Richard Biener <rguenther@suse.de>
22274 PR tree-optimization/79276
22275 * tree-vrp.c (process_assert_insertions): Properly adjust common
22276 when removing a duplicate.
22278 * gcc.dg/torture/pr79276.c: New testcase.
22280 2017-01-30 Richard Biener <rguenther@suse.de>
22282 PR tree-optimization/79256
22283 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22284 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22286 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22288 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22291 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22292 ("*r<noxa>sbg_<mode>_sll_bitmask")
22293 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22294 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22295 Use contiguous_bitmask_nowrap_operand.
22297 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22300 * config/rs6000/altivec.h (vec_xl): Revise #define.
22301 (vec_xst): Likewise.
22303 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22305 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22307 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22309 PR rtl-optimization/79194
22310 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22311 traps before call to bypass_conditional_jumps.
22313 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22315 PR tree-optimization/71374
22316 * lra-constraints.c (check_conflict_input_operands): New.
22317 (match_reload): Use it.
22319 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22322 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22323 account to calculate conflict_set.
22325 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22327 PR rtl-optimization/78559
22328 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22329 other_insn in combine.
22331 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22333 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22334 uint16_type_node for BT_UINT16.
22336 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22338 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22339 "RTL Tests" to menu.
22340 (GIMPLE Tests): New node.
22341 (RTL Tests): New node.
22343 2017-01-27 Richard Biener <rguenther@suse.de>
22345 PR tree-optimization/79245
22346 * tree-loop-distribution.c (distribute_loop): Apply cost
22347 modeling also to detected patterns.
22349 2017-01-27 Richard Biener <rguenther@suse.de>
22351 PR tree-optimization/71433
22352 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22353 (compare_assert_loc): New function.
22354 (process_assert_insertions): Sort and optimize assert locations
22355 to remove duplicates and push down identical assertions on
22356 edges to their destination block.
22358 2017-01-27 Richard Biener <rguenther@suse.de>
22360 PR tree-optimization/79244
22361 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22362 out SSA names even if abnormal.
22364 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22366 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22367 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22368 instead of MPFR_RNDN.
22370 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22373 * arm.c (arm_option_override): Don't call build_target_option_node
22374 until after doing all option overrides.
22375 (arm_valid_target_attribute_tree): Likewise.
22377 2017-01-27 Martin Liska <mliska@suse.cz>
22379 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22380 for {cd}tors and C++ {cd}tors.
22382 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22384 * config/s390/s390.md ("*setmem_long_and")
22385 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22387 2017-01-26 Martin Sebor <msebor@redhat.com>
22389 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22392 2017-01-26 Martin Sebor <msebor@redhat.com>
22394 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22395 HAVE_DFmode before using XFmode or DFmode.
22396 (parse_directive): Avoid using the z length modifier to avoid
22397 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22399 PR middle-end/78703
22400 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22401 to accept adjustment as an array.
22402 (get_int_range): New function.
22403 (struct directive): Make width and prec arrays.
22404 (directive::set_width, directive::set_precision): Call get_int_range.
22405 (format_integer, format_floating): Handle width and precision ranges.
22406 (format_string, parse_directive): Same.
22408 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22411 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22412 comdat_type_p set, just clone them, but keep the children in the
22416 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22417 which have direct callers with -fvar-tracking-assignments enabled
22419 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22420 inside of type units.
22422 2017-01-26 Martin Sebor <msebor@redhat.com>
22424 PR middle-end/78703
22425 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22427 (struct format_result): Replace number_chars, number_chars_min,
22428 and number_chars_max with a single member of struct result_range.
22430 (format_result::operator+=): Adjust.
22431 (struct fmtresult): Remove bounded. Handle likely and unlikely
22433 (fmtresult::adjust_for_width_or_precision): New function.
22434 (fmtresult:type_max_digits): New function.
22435 (bytes_remaining): Handle likely and unlikely counters.
22436 (min_bytes_remaining): Remove.
22437 (format_percent): Simplify.
22438 (format_integer, format_floating): Set likely and unlikely counters.
22439 (get_string_length, format_character, format_string): Same.
22440 (format_plain, should_warn_p): New function.
22441 (maybe_warn): Call should_warn_p. Update diagnostic messages
22442 and handle those for all directives, including plain strings.
22443 (format_directive): Handle likely and unlikely counters.
22444 Remove unnecessary quoting from diagnostics. Add an informational
22446 (add_bytes): Remove.
22447 (pass_sprintf_length::compute_format_length): Simplify.
22448 (try_substitute_return_value): Handle likely and unlikely counters.
22450 2017-01-26 Carl Love <cel@us.ibm.com>
22452 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22453 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22455 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22458 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22459 endianess for subregs into account.
22460 * lra-constraints.c (lra_constraints): Do risky transformations
22461 always on the first iteration.
22462 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22463 last_call_used_reg_set.
22464 (process_bb_lives): Define and use last_call_used_reg_set.
22465 * lra.c (lra): Always continue after lra_constraints on the first
22468 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22470 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22472 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22474 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22476 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22477 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22478 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22479 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22480 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22481 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22482 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22483 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22484 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22486 2017-01-26 Marek Polacek <polacek@redhat.com>
22489 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22490 for the third operand.
22492 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22494 PR middle-end/79236
22495 * omp-low.c (struct omp_context): Add simt_stmt field.
22496 (scan_omp_for): Return omp_context *.
22497 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22498 context to the _simt_ SIMD stmt.
22499 (lower_omp_for): For combined SIMD with sibling _simt_
22500 SIMD, make sure to use the same decls in _looptemp_
22501 clauses as in the sibling.
22503 2017-01-26 David Sherwood <david.sherwood@arm.com>
22505 PR middle-end/79212
22506 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22509 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22512 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22513 emit fld b; fld a; if possible.
22515 * brig-builtins.def: Update copyright years.
22516 * config/arm/arm_acle_builtins.def: Update copyright years.
22518 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22521 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22522 constraint instead of o for the stxsd instruction.
22524 2017-01-25 Carl Love <cel@us.ibm.com>
22526 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22527 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22529 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22531 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22533 2017-01-25 Richard Biener <rguenther@suse.de>
22535 PR tree-optimization/69264
22536 * target.def (vector_alignment_reachable): Improve documentation.
22537 * doc/tm.texi: Regenerate.
22538 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22540 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22541 earlier changes with respect to TYPE_USER_ALIGN.
22542 (vector_alignment_reachable_p): Likewise. Improve dumping.
22544 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22547 * config/arm/arm.md (xordi3): Force constant operand into a register
22550 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22552 * doc/invoke.texi (-fstore-merging): Correct default optimization
22553 levels at which it is enabled.
22554 (-O): Move -fstore-merging from list to...
22557 2017-01-25 Richard Biener <rguenther@suse.de>
22560 * omp-expand.c: Include debug.h.
22561 (expand_omp_taskreg): Make sure to generate early debug before
22562 outlining anything from a function.
22563 (expand_omp_target): Likewise.
22564 (grid_expand_target_grid_body): Likewise.
22566 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
22569 * asan.c (get_translation_unit_decl): New function.
22570 (asan_add_global): Extract modules file name from globals
22571 TRANSLATION_UNIT_DECL name.
22573 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
22576 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
22577 for long calls with APCS frame and VFP.
22579 2017-01-24 David Malcolm <dmalcolm@redhat.com>
22581 * cfg.c (original_copy_tables_initialized_p): New function.
22582 * cfg.h (original_copy_tables_initialized_p): New decl.
22583 * cfgrtl.c (relink_block_chain): Guard the call to
22584 free_original_copy_tables with a call to
22585 original_copy_tables_initialized_p.
22586 * cgraph.h (symtab_node::native_rtl_p): New decl.
22587 * cgraphunit.c (symtab_node::native_rtl_p): New function.
22588 (symtab_node::needed_p): Don't assert for early assembly output
22589 for __RTL functions.
22590 (cgraph_node::finalize_function): Set "force_output" for __RTL
22592 (cgraph_node::analyze): Bail out early for __RTL functions.
22593 (analyze_functions): Update assertion to support __RTL functions.
22594 (cgraph_node::expand): Bail out early for __RTL functions.
22595 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
22597 * function.h (struct function): Update comment for field
22599 * gimple-expr.c: Include "tree-pass.h".
22600 (gimple_has_body_p): Return false for __RTL functions.
22601 * Makefile.in (OBJS): Add run-rtl-passes.o.
22602 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
22604 (gcc::pass_manager::get_clean_slate): New accessor.
22605 * passes.c: Include "insn-addr.h".
22606 (should_skip_pass_p): Add logging. Update logic for running
22607 "expand" to be compatible with both __GIMPLE and __RTL. Guard
22608 property-provider override so it is only done for gimple passes.
22610 (skip_pass): New function.
22611 (execute_one_pass): Call skip_pass when skipping passes.
22612 * read-md.c (md_reader::read_char): Support filtering
22613 the input to a subset of line numbers.
22614 (md_reader::md_reader): Initialize fields
22615 m_first_line and m_last_line.
22616 (md_reader::read_file_fragment): New function.
22617 * read-md.h (md_reader::read_file_fragment): New decl.
22618 (md_reader::m_first_line): New field.
22619 (md_reader::m_last_line): New field.
22620 * read-rtl-function.c (function_reader::create_function): Only
22621 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
22622 curr_properties. Set DECL_INITIAL to a dummy block.
22623 (read_rtl_function_body_from_file_range): New function.
22624 * read-rtl-function.h (read_rtl_function_body_from_file_range):
22626 * run-rtl-passes.c: New file.
22627 * run-rtl-passes.h: New file.
22629 2017-01-24 Jeff Law <law@redhat.com>
22631 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
22634 2017-01-24 Bin Cheng <bin.cheng@arm.com>
22636 PR tree-optimization/79159
22637 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
22638 (record_nonwrapping_iv): Improve boundary using above function if no
22639 value range information.
22641 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
22642 Martin Jambor <mjambor@suse.cz>
22644 * brig-builtins.def: New file.
22645 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
22646 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
22647 (DEF_HSAIL_SAT_BUILTIN): Likewise.
22648 (DEF_HSAIL_INTR_BUILTIN): Likewise.
22649 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
22650 * builtin-types.def (BT_INT8): New.
22651 (BT_INT16): Likewise.
22652 (BT_UINT8): Likewise.
22653 (BT_UINT16): Likewise.
22654 (BT_FN_ULONG): Likewise.
22655 (BT_FN_UINT_INT): Likewise.
22656 (BT_FN_UINT_ULONG): Likewise.
22657 (BT_FN_UINT_LONG): Likewise.
22658 (BT_FN_UINT_PTR): Likewise.
22659 (BT_FN_ULONG_PTR): Likewise.
22660 (BT_FN_INT8_FLOAT): Likewise.
22661 (BT_FN_INT16_FLOAT): Likewise.
22662 (BT_FN_UINT32_FLOAT): Likewise.
22663 (BT_FN_UINT16_FLOAT): Likewise.
22664 (BT_FN_UINT8_FLOAT): Likewise.
22665 (BT_FN_UINT64_FLOAT): Likewise.
22666 (BT_FN_UINT16_UINT32): Likewise.
22667 (BT_FN_UINT32_UINT16): Likewise.
22668 (BT_FN_UINT16_UINT16_UINT16): Likewise.
22669 (BT_FN_INT_PTR_INT): Likewise.
22670 (BT_FN_UINT_PTR_UINT): Likewise.
22671 (BT_FN_LONG_PTR_LONG): Likewise.
22672 (BT_FN_ULONG_PTR_ULONG): Likewise.
22673 (BT_FN_VOID_UINT64_UINT64): Likewise.
22674 (BT_FN_UINT8_UINT8_UINT8): Likewise.
22675 (BT_FN_INT8_INT8_INT8): Likewise.
22676 (BT_FN_INT16_INT16_INT16): Likewise.
22677 (BT_FN_INT_INT_INT): Likewise.
22678 (BT_FN_UINT_FLOAT_UINT): Likewise.
22679 (BT_FN_FLOAT_UINT_UINT): Likewise.
22680 (BT_FN_ULONG_UINT_UINT): Likewise.
22681 (BT_FN_ULONG_UINT_PTR): Likewise.
22682 (BT_FN_ULONG_ULONG_ULONG): Likewise.
22683 (BT_FN_UINT_UINT_UINT): Likewise.
22684 (BT_FN_VOID_UINT_PTR): Likewise.
22685 (BT_FN_UINT_UINT_PTR: Likewise.
22686 (BT_FN_UINT32_UINT64_PTR): Likewise.
22687 (BT_FN_INT_INT_UINT_UINT): Likewise.
22688 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
22689 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
22690 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
22691 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
22692 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
22693 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
22694 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
22695 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
22696 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
22697 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
22698 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
22699 * doc/frontends.texi: List BRIG FE.
22700 * doc/install.texi (Testing): Add BRIG tesring requirements.
22701 * doc/invoke.texi (Overall Options): Mention BRIG.
22702 * doc/standards.texi (Standards): Doucment BRIG HSA version.
22704 2017-01-24 Richard Biener <rguenther@suse.de>
22706 PR translation/79208
22707 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
22709 2017-01-24 Martin Jambor <mjambor@suse.cz>
22712 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
22713 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
22714 and known_contexts.
22716 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
22718 PR middle-end/79123
22719 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
22720 casts from signed to unsigned really don't have a range.
22722 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
22724 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
22725 GMP_RNDx for compatiblity.
22727 2017-01-24 Martin Liska <mliska@suse.cz>
22730 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22731 that would prevent us to call alloca with -1 as argument.
22733 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22735 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22736 Avoid -Wformat-security warning.
22738 2017-01-23 Andrew Pinski <apinski@cavium.com>
22740 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22743 2017-01-23 Martin Sebor <msebor@redhat.com>
22745 PR middle-end/78703
22746 * gimple-ssa-sprintf.c (warn_level): New global.
22747 (format_integer): Use it here and throughout the rest of the file.
22748 Use the same switch to compute sign as base.
22749 (maybe_warn): New function.
22750 (format_directive): Factor out warnings into maybe_warn.
22751 Add debugging output. Use warn_level.
22752 (add_bytes): Use warn_level.
22753 (pass_sprintf_length::compute_format_length): Add debugging output.
22754 (try_substitute_return_value): Same.
22755 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22757 PR middle-end/78703
22758 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22759 (struct fmtresult, format_integer, format_floating): Adjust.
22760 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22761 (get_string_length, format_string,format_directive): Same.
22762 (pass_sprintf_length::compute_format_length): Same.
22763 (try_substitute_return_value): Simplify slightly.
22765 PR middle-end/78703
22766 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
22767 (fmtresult::operator+=): Outlined.
22768 (struct fmtresult): Add ctors.
22769 (struct conversion_spec): Rename...
22770 (struct directive): ...to this. Add and remove data members.
22771 (directive::set_width, directive::set_precision): New functions.
22772 (format_percent): Use fmtresult ctor.
22773 (get_width_and_precision): Remove.
22774 (format_integer): Make naming changes. Avoid computing width and
22776 (format_floating): Same. Adjust indentation.
22777 (format_character, format_none): New functions.
22778 (format_string): Moved character handling to format_character.
22779 (format_directive): Remove arguments, change return type.
22780 (parse_directive): New function.
22781 (pass_sprintf_length::compute_format_length): Move directive
22782 parsing to parse_directive.
22784 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22786 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
22787 (assign_assembler_name_if_needed): ... this.
22788 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
22789 (assign_assembler_name_if_needed): ... this.
22790 (free_lang_data_in_cgraph): Adjust callers.
22791 * cgraphunit.c (cgraph_node::analyze): Likewise.
22792 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
22795 2017-01-23 Richard Biener <rguenther@suse.de>
22797 PR tree-optimization/79088
22798 PR tree-optimization/79188
22799 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
22800 resetting loop bounds after last path deletion. Reset loop
22801 bounds of the target loop, make code match the comments.
22802 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
22803 Make sure loops need no fixups.
22805 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22807 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
22808 exponent support with double type for first argument.
22809 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
22810 type returned by __builtin_vec_extract_sig,
22811 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
22812 functions from "vector int" to "vector unsigned int" or from
22813 "vector long long int" to "vector unsigned long long int".
22814 Changed type returned by __builtin_vec_extract_exp,
22815 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
22816 functions from "vector int" to "vector unsigned int" or from
22817 "vector long long int" to "vector unsigned long long int".
22818 Changed return type of __builtin_vec_test_data_class,
22819 __builtin_vec_test_data_class_sp, and
22820 __builtin_vec_test_data_class_dp from "vector int" to
22821 "vector bool int" or from "vector long long int" to "vector bool
22822 long long int" and changed second argument type from "unsigned
22823 int" to "int". Added new overloaded function forms "vector float
22824 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
22825 "vector float __builtin_vec_insert_exp_sp (vector float, vector
22826 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
22827 double, vector unsigned long long int)" and "vector double
22828 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
22829 long int)". Changed return type of
22830 __builtin_scalar_test_data_class and
22831 __builtin_scalar_test_data_class_sp and
22832 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
22833 int" and changed second argument from "unsigned int" to "int".
22834 Changed type returned by __builtin_scalar_test_neg,
22835 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
22836 from "int" to "bool int". Added new overloaded function form
22837 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
22838 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
22839 exponent double-precision with floating point first argument.
22840 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
22841 documentation of scalar_test_data_class, scalar_test_neg,
22842 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
22843 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
22844 vec_test_data_class built-in functions to reflect refinements in
22845 their type signatures.
22847 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
22849 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
22851 (aarch64_elf_asm_destructor): Likewise.
22853 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
22855 PR rtl-optimization/78634
22856 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
22857 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
22858 * ifcvt.c (noce_try_cmove): Add missing cost check.
22860 PR rtl-optimization/71724
22861 * combine.c (if_then_else_cond): Look for situations where it is
22862 beneficial to undo the work of one of the recursive calls.
22864 2017-01-23 Bin Cheng <bin.cheng@arm.com>
22866 PR tree-optimization/70754
22867 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
22868 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
22869 combined stmt before it if not NULL.
22870 (combine_chains): Process refs reversely and compute dominance point
22873 2017-01-23 Martin Liska <mliska@suse.cz>
22875 PR tree-optimization/79196
22876 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
22877 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
22879 (strlen_optimize_stmt): Call the renamed function.
22881 2017-01-23 Michael Matz <matz@suse.de>
22883 PR tree-optimization/78384
22884 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
22886 2017-01-23 Richard Biener <rguenther@suse.de>
22888 PR tree-optimization/79186
22889 * tree-vrp.c (register_new_assert_for): Make sure we've seen
22890 both incoming edges before moving an assert.
22892 2017-01-23 Martin Jambor <mjambor@suse.cz>
22894 * ipa-prop.c (load_from_param_1): Removed.
22895 (load_from_unmodified_param): Bits from load_from_param_1 put back
22897 (load_from_param): Removed.
22898 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
22899 with stmt. Reverted back to use of load_from_unmodified_param.
22901 2017-01-23 Martin Jambor <mjambor@suse.cz>
22904 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
22905 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
22906 field a pointer to garbage collected vector, mark lattices and
22907 ipcp_orig_node with GTY((skip)).
22908 (ipa_get_param_count): Adjust to descriptors being a pointer.
22909 (ipa_get_param): Likewise.
22910 (ipa_get_type): Likewise.
22911 (ipa_get_param_move_cost): Likewise.
22912 (ipa_set_param_used): Likewise.
22913 (ipa_get_controlled_uses): Likewise.
22914 (ipa_set_controlled_uses): Likewise.
22915 (ipa_is_param_used): Likewise.
22916 (ipa_node_params_t): Move into garbage collector. New methods insert
22918 (ipa_node_params_sum): Annotate wth GTY(()).
22919 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
22921 (ipa_load_from_parm_agg): Adjust declaration.
22922 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
22923 * ipa-profile.c (ipa_profile): Likewise.
22924 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
22925 (ipa_populate_param_decls): Make descriptors parameter garbage
22927 (ipa_dump_param): Adjust to descriptors being a pointer.
22928 (ipa_alloc_node_params): Likewise.
22929 (ipa_initialize_node_params): Likewise.
22930 (load_from_param_1): Make descriptors parameter garbage collected.
22931 (load_from_unmodified_param): Likewise.
22932 (load_from_param): Likewise.
22933 (ipa_load_from_parm_agg): Likewise.
22934 (ipa_node_params::~ipa_node_params): Removed.
22935 (ipa_free_all_node_params): Remove call to delete operator.
22936 (ipa_node_params_t::insert): New.
22937 (ipa_node_params_t::remove): Likewise.
22938 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
22939 copy known_csts and known_contexts vectors.
22940 (ipa_read_node_info): Adjust to descriptors being a pointer.
22941 (ipcp_modif_dom_walker): Make m_descriptors field garbage
22943 (ipcp_transform_function): Make descriptors variable garbage
22946 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
22948 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
22949 * config/i386/avx512dqintrin.h: Ditto.
22950 * config/i386/avx512fintrin.h: Ditto.
22951 * config/i386/i386.c: Handle new builtins.
22952 * config/i386/i386-builtin.def: Add new builtins.
22953 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
22954 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
22956 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22957 Martin Liska <mliska@suse.cz>
22959 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
22960 * asan.c (asan_expand_poison_ifn): Support stores and use
22961 appropriate ASAN report function.
22962 * internal-fn.c (expand_ASAN_POISON_USE): New function.
22963 * internal-fn.def (ASAN_POISON_USE): Declare.
22964 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
22965 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
22966 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
22967 ASAN_POISON calls w/o LHS.
22968 * tree-ssa.c (execute_update_addresses_taken): Create clobber
22969 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
22970 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
22971 * gimplify.c (asan_poison_variables): Add attribute
22972 use_after_scope_memory to variables that really needs to live
22974 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
22975 having the attribute.
22977 2017-01-23 Martin Liska <mliska@suse.cz>
22979 * asan.c (create_asan_shadow_var): New function.
22980 (asan_expand_poison_ifn): Likewise.
22981 * asan.h (asan_expand_poison_ifn): New declaration.
22982 * internal-fn.c (expand_ASAN_POISON): Likewise.
22983 * internal-fn.def (ASAN_POISON): New builtin.
22984 * sanopt.c (pass_sanopt::execute): Expand
22985 asan_expand_poison_ifn.
22986 * tree-inline.c (copy_decl_for_dup_finish): Make function
22988 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
22989 * tree-ssa.c (is_asan_mark_p): New function.
22990 (execute_update_addresses_taken): Rewrite local variables
22991 (identified just by use-after-scope as addressable) into SSA.
22993 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22995 * doc/install.texi (Specific): opensource.apple.com uses https
22996 now. Remove trailing slash.
22998 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23000 * README.Portability: Remove note on an Irix compatibility issue.
23002 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23004 * gcov.c (INCLUDE_ALGORITHM): Define.
23005 (INCLUDE_VECTOR): Define.
23006 No longer include <vector> and <algorithm> directly.
23008 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23010 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23012 * doc/invoke.texi (Code Gen Options): Ditto.
23014 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23017 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23019 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23021 rtl-optimization/79125
23022 * cprop.c (local_cprop_pass): Handle cases where we make an
23023 unconditional trap.
23025 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23029 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23030 read from, for big endian.
23032 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23034 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23035 register pauth builtins for LP64 only.
23037 2017-01-20 Marek Polacek <polacek@redhat.com>
23040 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23043 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23045 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23047 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23048 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23049 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23051 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23054 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23055 in big-endian mode when they are not a single duplicated value.
23057 2017-01-20 Richard Biener <rguenther@suse.de>
23059 * BASE-VER: Bump to 7.0.1.
23061 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23063 * omp-low.c (omplow_simd_context): New struct. Use it...
23064 (lower_rec_simd_input_clauses): ...here and...
23065 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23066 references to idx, lane, max_vf, is_simt.
23068 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23070 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23073 2017-01-20 Martin Jambor <mjambor@suse.cz>
23075 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23076 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23078 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23079 (GTFILES): Rename hsa.c to hsa-common.c.
23080 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23081 * hsa-dump.c: Likewise.
23082 * hsa-gen.c: Likewise.
23083 * hsa-regalloc.c: Likewise.
23084 * ipa-hsa.c: Likewise.
23085 * omp-expand.c: Likewise.
23086 * omp-low.c: Likewise.
23087 * toplev.c: Likewise.
23089 2017-01-20 Marek Polacek <polacek@redhat.com>
23092 * doc/invoke.texi: Document -Wduplicated-branches.
23093 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23094 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23095 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23096 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23097 return 0 only when not OEP_LEXICOGRAPHIC.
23098 (fold_build_cleanup_point_expr): Use the expression
23099 location when building CLEANUP_POINT_EXPR.
23100 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23101 * tree.c (add_expr): Handle error_mark_node.
23103 2017-01-20 Martin Liska <mliska@suse.cz>
23106 * tree-profile.c (init_ic_make_global_vars): Do not call
23108 (gimple_init_gcov_profiler): Likewise.
23110 2017-01-20 Martin Liska <mliska@suse.cz>
23113 * cgraph.h (maybe_create_reference): Remove argument and
23115 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23117 * ipa-cp.c (create_specialized_node): Likewise.
23118 * symtab.c (symtab_node::maybe_create_reference): Handle
23119 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23121 2017-01-20 Martin Liska <mliska@suse.cz>
23123 * read-rtl-function.c (function_reader::create_function): Use
23124 build_decl instread of build_decl_stat.
23126 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23128 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23129 * config/i386/avx512dqintrin.h: Ditto.
23130 * config/i386/avx512fintrin.h: Ditto.
23131 * config/i386/i386-builtin-types.def: Add new types.
23132 * config/i386/i386.c: Handle new types.
23133 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23134 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23135 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23136 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23137 (__builtin_ia32_kshiftridi): New.
23138 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23140 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23144 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23145 define to rs6000_init_stack_protect_guard.
23146 (rs6000_init_stack_protect_guard): New function.
23148 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23149 Yunqiang Su <yunqiang.su@imgtec.com>
23151 * config.gcc (supported_defaults): Add madd4.
23152 (with_madd4): Add validation.
23153 (all_defaults): Add madd4.
23154 * config/mips/mips.opt (mmadd4): New option.
23155 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23157 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23159 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23160 (ISA_HAS_FUSED_MADD4): Likewise.
23161 * doc/invoke.texi (-mmadd4): Document the new option.
23162 * doc/install.texi (--with-madd4): Document the new option.
23164 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23166 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23167 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23168 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23169 (aarch64_init_pauth_hint_builtins): New.
23170 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23171 (aarch64_expand_builtin): Expand new builtins.
23173 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23175 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23176 * combine-stack-adj.c (no_unhandled_cfa): Handle
23177 REG_CFA_TOGGLE_RA_MANGLE.
23178 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23179 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23180 info for return address signing.
23181 (aarch64_expand_epilogue): Likewise.
23183 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23185 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23186 * config/aarch64/aarch64-protos.h
23187 (aarch64_return_address_signing_enabled): New declaration.
23188 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23190 (aarch64_expand_prologue): Sign return address before it's pushed onto
23192 (aarch64_expand_epilogue): Authenticate return address fetched from
23194 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23195 (aarch64_attributes): New function attributes for "sign-return-address".
23196 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23197 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23198 ("*do_return"): Generate combined instructions according to key index.
23199 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23200 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23202 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23203 * config/aarch64/aarch64.opt (msign-return-address=): New.
23204 * doc/extend.texi (AArch64 Function Attributes): Documents
23205 "sign-return-address=".
23206 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23208 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23210 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23211 overall option summary.
23213 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23215 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23216 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23217 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23218 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23220 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23222 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23223 -mpower9-minmax by default for -mcpu=power9.
23224 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23225 128-bit floating point.
23227 2017-01-20 Alan Modra <amodra@gmail.com>
23229 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23230 optimizing for size.
23232 2017-01-20 Alan Modra <amodra@gmail.com>
23235 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23236 for strcmp and strncmp from corresponding builtin decl.
23238 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23240 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23241 instead of i386/rtems-64.h.
23242 * config/i386/rtems-64.h: Remove.
23244 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23248 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23250 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23252 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23254 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23255 Change int to HOST_WIDE_INT.
23256 * config/aarch64/aarch64-protos.h
23257 (aarch64_simd_gen_const_vector_dup): Likewise.
23258 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23260 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23262 * langhooks-def.h (lhd_type_for_size): New decl.
23263 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23264 * langhooks.c (lhd_type_for_size): New function, taken from
23267 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23269 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23270 define_bypass for CR latency.
23271 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23272 (power9-alu2): Add define_bypass for CR latency.
23274 (power9-mul): Update insn latency.
23275 (power9-mul-compare): Update insn latency, bypass latency and remove
23278 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23280 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23282 * config/aarch64/aarch64.md
23283 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23284 aarch64_nopcrelative_literal_loads.
23285 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23287 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23289 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23290 TARGET_LOONGSON_3A.
23291 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23293 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23296 * config.gcc (supported_defaults): Add lxc1-sxc1.
23297 (with_lxc1_sxc1): Add validation.
23298 (all_defaults): Add lxc1-sxc1.
23299 * config/mips/mips.opt (mlxc1-sxc1): New option.
23300 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23302 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23303 __mips_no_lxc1_sxc1.
23304 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23305 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23306 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23308 2017-01-19 Richard Biener <rguenther@suse.de>
23310 PR tree-optimization/72488
23311 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23312 sure to restore SSA info.
23313 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23315 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23317 PR rtl-optimization/79121
23318 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23319 of the inner type when shifting an extended value.
23321 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23324 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23325 interposable aliases.
23327 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23330 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23331 Use the evmergelohi instruction.
23332 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23333 (mov_sitf_e500_subreg8_2_be): Likewise.
23334 (mov_sitf_e500_subreg12_2_le): Likewise.
23335 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23336 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23337 (mov_sitf_e500_subreg8_2_le): Likewise.
23338 (mov_sitf_e500_subreg12_2_be): Likewise.
23340 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23342 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23343 attribute from vecsimple to vecperm.
23344 (altivec_vbpermq2): Likewise.
23346 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23349 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23351 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23352 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23353 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23354 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23355 case where N arg is SIZE_MAX.
23356 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23357 (cmpstrsi): Add pattern.
23359 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23361 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23362 __builtin_vec_revb builtins.
23363 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23364 built-in functions to support generation of the ISA 3.0 XXBR<x>
23365 vector byte reverse instructions.
23366 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23367 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23368 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23369 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23370 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23371 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23372 (P9V_BUILTIN_VEC_REVB): Likewise.
23373 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23374 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23375 (p9_xxbrq_v16qi): Likewise.
23376 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23377 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23378 (p9_xxbrh_v8hi): Likewise.
23379 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23380 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23381 vec_revb built-in functions.
23383 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23385 PR rtl-optimization/78952
23386 * config/i386/i386.md (any_extract): New code iterator.
23387 (*insvqi_2): Use any_extract for source operand.
23388 (*insvqi_3): Use any_shiftrt for source operand.
23390 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23392 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23394 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23396 2017-01-18 Matthias Klose <doko@ubuntu.com>
23398 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23400 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23402 * config/rs6000/altivec.h (vec_bperm): Change #define.
23403 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23404 (altivec_vbpermq2): New define_insn.
23405 (altivec_vbpermd): Likewise.
23406 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23407 function interface.
23408 (VBPERMD): Likewise.
23409 (VBPERM): New polymorphic function interface.
23410 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23411 Add entries for P9V_BUILTIN_VEC_VBPERM.
23412 * doc/extend.texi: Add interfaces for vec_bperm.
23414 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23416 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23417 first letter of error messages.
23418 (s390_resolve_overloaded_builtin): Likewise.
23419 * config/s390/s390.c (s390_expand_builtin): Likewise.
23420 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23421 (s390_valid_target_attribute_inner_p): Likewise.
23422 * config/s390/s390.md ("tabort"): Likewise.
23424 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23426 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23427 (ISA_AVOID_DIV_HILO): New macro.
23428 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23429 (ISA_HAS_DDIV): Likewise.
23431 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23433 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23435 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23437 * doc/invoke.texi (fabi-version): Spelling fix.
23439 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23442 * doc/invoke.texi (fabi-version): Mention mangling fix for
23445 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23448 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23450 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23453 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23454 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23456 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23458 (stack_protect_test): Ditto.
23459 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23460 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23462 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23463 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23464 -mstack-protector-guard-offset=.
23465 (RS/6000 and PowerPC Options): Ditto.
23467 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23469 * config/i386/i386.h (MASK_CLASS_P): New define.
23470 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23471 there are no registers from different register sets also when
23472 mask registers are used. Update function comment.
23473 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23474 to (*k/*r) and (*k/*km) alternatives.
23476 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23478 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23479 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23480 (EH_RETURN_HANDLER_RTX): New define.
23481 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23482 Force frame pointer in EH return functions.
23483 (aarch64_expand_epilogue): Add barrier for eh_return.
23484 (aarch64_final_eh_return_addr): Remove.
23485 (aarch64_eh_return_handler_rtx): New function.
23486 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23488 (aarch64_eh_return_handler_rtx): New prototype.
23490 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23492 * config/rs6000/altivec.h (vec_rlmi): New #define.
23493 (vec_vrlnm): Likewise.
23494 (vec_rlnm): Likewise.
23495 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23496 (UNSPEC_VRLNM): Likewise.
23497 (VIlong): New mode iterator.
23498 (altivec_vrl<VI_char>mi): New define_insn.
23499 (altivec_vrl<VI_char>nm): Likewise.
23500 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23502 (VRLDNM): Likewise.
23503 (RLNM): New polymorphic function entry.
23504 (VRLWMI): New monomorphic function entry.
23505 (VRLDMI): Likewise.
23506 (RLMI): New polymorphic function entry.
23507 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23508 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23509 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23512 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23515 * dwarf2out.c (field_byte_offset): Restore the
23516 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23517 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23518 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23521 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23524 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23526 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23529 * dwarf2out.c (add_data_member_location_attribute): For constant
23530 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23531 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23532 DW_AT_byte_size attributes.
23534 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23536 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23537 after forcing to constant memory when the code model is medium.
23539 2017-01-17 Julia Koval <julia.koval@intel.com>
23542 * config/i386/avx512fintrin.h
23543 (_mm512_i32gather_ps): Change __addr type to void const*.
23544 (_mm512_mask_i32gather_ps): Ditto.
23545 (_mm512_i32gather_pd): Ditto.
23546 (_mm512_mask_i32gather_pd): Ditto.
23547 (_mm512_i64gather_ps): Ditto.
23548 (_mm512_mask_i64gather_ps): Ditto.
23549 (_mm512_i64gather_pd): Ditto.
23550 (_mm512_mask_i64gather_pd): Ditto.
23551 (_mm512_i32gather_epi32): Ditto.
23552 (_mm512_mask_i32gather_epi32): Ditto.
23553 (_mm512_i32gather_epi64): Ditto.
23554 (_mm512_mask_i32gather_epi64): Ditto.
23555 (_mm512_i64gather_epi32): Ditto.
23556 (_mm512_mask_i64gather_epi32): Ditto.
23557 (_mm512_i64gather_epi64): Ditto.
23558 (_mm512_mask_i64gather_epi64): Ditto.
23559 (_mm512_i32scatter_ps): Change __addr type to void*.
23560 (_mm512_mask_i32scatter_ps): Ditto.
23561 (_mm512_i32scatter_pd): Ditto.
23562 (_mm512_mask_i32scatter_pd): Ditto.
23563 (_mm512_i64scatter_ps): Ditto.
23564 (_mm512_mask_i64scatter_ps): Ditto.
23565 (_mm512_i64scatter_pd): Ditto.
23566 (_mm512_mask_i64scatter_pd): Ditto.
23567 (_mm512_i32scatter_epi32): Ditto.
23568 (_mm512_mask_i32scatter_epi32): Ditto.
23569 (_mm512_i32scatter_epi64): Ditto.
23570 (_mm512_mask_i32scatter_epi64): Ditto.
23571 (_mm512_i64scatter_epi32): Ditto.
23572 (_mm512_mask_i64scatter_epi32): Ditto.
23573 (_mm512_i64scatter_epi64): Ditto.
23574 (_mm512_mask_i64scatter_epi64): Ditto.
23575 * config/i386/avx512pfintrin.h
23576 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
23577 (_mm512_mask_prefetch_i32gather_ps): Ditto.
23578 (_mm512_mask_prefetch_i64gather_pd): Ditto.
23579 (_mm512_mask_prefetch_i64gather_ps): Ditto.
23580 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
23581 (_mm512_prefetch_i32scatter_ps): Ditto.
23582 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
23583 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
23584 (_mm512_prefetch_i64scatter_pd): Ditto.
23585 (_mm512_prefetch_i64scatter_ps): Ditto.
23586 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
23587 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
23588 * config/i386/avx512vlintrin.h
23589 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
23590 (_mm_mmask_i32gather_ps): Ditto.
23591 (_mm256_mmask_i32gather_pd): Ditto.
23592 (_mm_mmask_i32gather_pd): Ditto.
23593 (_mm256_mmask_i64gather_ps): Ditto.
23594 (_mm_mmask_i64gather_ps): Ditto.
23595 (_mm256_mmask_i64gather_pd): Ditto.
23596 (_mm_mmask_i64gather_pd): Ditto.
23597 (_mm256_mmask_i32gather_epi32): Ditto.
23598 (_mm_mmask_i32gather_epi32): Ditto.
23599 (_mm256_mmask_i32gather_epi64): Ditto.
23600 (_mm_mmask_i32gather_epi64): Ditto.
23601 (_mm256_mmask_i64gather_epi32): Ditto.
23602 (_mm_mmask_i64gather_epi32): Ditto.
23603 (_mm256_mmask_i64gather_epi64): Ditto.
23604 (_mm_mmask_i64gather_epi64): Ditto.
23605 (_mm256_i32scatter_ps): Change __addr type to void*.
23606 (_mm256_mask_i32scatter_ps): Ditto.
23607 (_mm_i32scatter_ps): Ditto.
23608 (_mm_mask_i32scatter_ps): Ditto.
23609 (_mm256_i32scatter_pd): Ditto.
23610 (_mm256_mask_i32scatter_pd): Ditto.
23611 (_mm_i32scatter_pd): Ditto.
23612 (_mm_mask_i32scatter_pd): Ditto.
23613 (_mm256_i64scatter_ps): Ditto.
23614 (_mm256_mask_i64scatter_ps): Ditto.
23615 (_mm_i64scatter_ps): Ditto.
23616 (_mm_mask_i64scatter_ps): Ditto.
23617 (_mm256_i64scatter_pd): Ditto.
23618 (_mm256_mask_i64scatter_pd): Ditto.
23619 (_mm_i64scatter_pd): Ditto.
23620 (_mm_mask_i64scatter_pd): Ditto.
23621 (_mm256_i32scatter_epi32): Ditto.
23622 (_mm256_mask_i32scatter_epi32): Ditto.
23623 (_mm_i32scatter_epi32): Ditto.
23624 (_mm_mask_i32scatter_epi32): Ditto.
23625 (_mm256_i32scatter_epi64): Ditto.
23626 (_mm256_mask_i32scatter_epi64): Ditto.
23627 (_mm_i32scatter_epi64): Ditto.
23628 (_mm_mask_i32scatter_epi64): Ditto.
23629 (_mm256_i64scatter_epi32): Ditto.
23630 (_mm256_mask_i64scatter_epi32): Ditto.
23631 (_mm_i64scatter_epi32): Ditto.
23632 (_mm_mask_i64scatter_epi32): Ditto.
23633 (_mm256_i64scatter_epi64): Ditto.
23634 (_mm256_mask_i64scatter_epi64): Ditto.
23635 (_mm_i64scatter_epi64): Ditto.
23636 (_mm_mask_i64scatter_epi64): Ditto.
23637 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
23638 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
23639 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
23640 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
23641 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
23642 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
23643 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
23644 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
23645 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
23646 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
23647 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
23648 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
23649 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
23650 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
23651 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
23652 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
23653 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
23654 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
23655 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
23656 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
23657 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
23658 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
23659 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
23660 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
23661 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
23662 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
23663 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
23664 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
23665 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
23666 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
23667 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
23668 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
23669 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
23670 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
23671 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
23672 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
23673 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
23674 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
23675 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
23676 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
23677 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
23678 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
23679 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
23680 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
23681 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
23682 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
23683 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
23684 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
23685 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
23686 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
23687 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
23688 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
23689 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
23690 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
23691 definitions accordingly.
23693 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
23694 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23697 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
23700 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
23703 * ira-conflicts.c (ira_build_conflicts): Update total conflict
23704 hard regs for inner regno.
23706 2017-01-17 Martin Liska <mliska@suse.cz>
23709 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
23710 assumption and add comment.
23712 2017-01-17 Nathan Sidwell <nathan@acm.org>
23714 * ipa-visibility.c (localize_node): New function, broken out of ...
23715 (function_and_variable_visibility): ... here. Call it.
23717 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23719 PR middle-end/77445
23720 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
23721 correctly set frequency of oudgoing edge.
23722 (duplicate_thread_path): Fix profile updating.
23724 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23727 * configure.ac: Add GCC_BASE_VER.
23728 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23729 version from BASE-VER file.
23730 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23731 (gcc.o): Depend on $(BASEVER).
23732 * common.opt (dumpfullversion): New option.
23733 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23734 * doc/invoke.texi: Document -dumpfullversion.
23735 * doc/install.texi: Document --with-gcc-major-version-only.
23736 * configure: Regenerated.
23738 2017-01-17 Richard Biener <rguenther@suse.de>
23740 PR tree-optimization/71433
23741 * tree-vrp.c (register_new_assert_for): Merge same asserts
23742 on all incoming edges.
23743 (process_assert_insertions_for): Handle insertions at the
23746 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23748 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23749 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23751 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23754 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23757 2017-01-17 Alan Modra <amodra@gmail.com>
23760 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23761 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
23762 symbolic stack limit when pic.
23764 2017-01-16 Martin Sebor <msebor@redhat.com>
23766 PR tree-optimization/78608
23767 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
23769 2017-01-16 Jeff Law <law@redhat.com>
23772 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
23773 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
23774 for several include directories that may be relative to sysroot.
23775 * config/i386/x-mingw32 (gplus_includedir): Define.
23776 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
23777 (native_system_includedir): Likewise.
23778 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
23779 override if TARGET_SYSTEM_ROOT is defined.
23780 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
23782 PR tree-optimization/79090
23783 PR tree-optimization/33562
23784 PR tree-optimization/61912
23785 PR tree-optimization/77485
23786 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
23787 and computed trims into the dump file.
23789 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23791 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
23793 2017-01-16 Jakub Jelinek <jakub@redhat.com>
23796 * gimplify.c (gimplify_init_constructor): If want_value and
23797 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
23801 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
23802 sequence. Formatting fixes.
23803 (doloop_optimize): Formatting fixes.
23806 * gcc.c (debug_level_greater_than_spec_func): New function.
23807 (static_spec_functions): Add debug-level-gt spec function.
23808 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
23810 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23811 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23812 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
23813 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
23814 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
23815 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
23817 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
23819 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
23820 QImode fixups to general and mask registers only.
23822 2017-01-16 Carl Love <cel@us.ibm.com>
23824 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
23825 for built-in functions
23826 vector signed char vec_nabs (vector signed char)
23827 vector signed short vec_nabs (vector signed short)
23828 vector signed int vec_nabs (vector signed int)
23829 vector signed long long vec_nabs (vector signed long long)
23830 vector float vec_nabs (vector float)
23831 vector double vec_nabs (vector double)
23832 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
23834 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
23835 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
23836 * doc/extend.texi: Update the documentation file for the new built-in
23839 2017-01-16 Martin Sebor <msebor@redhat.com>
23841 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
23844 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23846 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
23847 UNSPEC_VSX__XXSPLTD to require special splat handling.
23849 2017-01-16 David Malcolm <dmalcolm@redhat.com>
23852 * system.h: Poison strndup.
23854 2017-01-16 Alan Modra <amodra@gmail.com>
23857 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
23860 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
23862 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
23864 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
23866 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
23867 call recog here. Assert that INSN_CODE (insn) is non-negative.
23869 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23872 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
23874 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
23875 in the currently scheduled RTL fragment.
23877 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23879 PR rtl-optimization/78751
23880 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
23883 2017-01-14 Jeff Law <law@redhat.com>
23885 PR tree-optimization/79090
23886 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
23887 variable length stores.
23888 (compute_trims): Delete dead assignment to *trim_tail.
23889 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
23892 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
23894 PR rtl-optimization/78626
23895 PR rtl-optimization/78727
23896 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
23897 of a block, and split such blocks after everything else is finished.
23899 2017-01-14 Alan Modra <amodra@gmail.com>
23902 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
23903 target legitimate_combined_insn.
23904 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
23905 (rs6000_legitimate_combined_insn): New function.
23906 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
23908 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
23909 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
23910 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
23912 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
23914 * doc/frontends.texi (G++ and GCC): Remove references to Java.
23916 2017-01-13 Jeff Law <law@redhat.com>
23918 PR tree-optimization/33562
23919 PR tree-optimization/61912
23920 PR tree-optimization/77485
23921 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
23923 (delete_dead_assignment): Likewise.
23924 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
23925 statement to delete_dead_call and delete_dead_assignment.
23927 2017-01-13 David Malcolm <dmalcolm@redhat.com>
23930 * substring-locations.c (format_warning_va): Strengthen case 1 so
23931 that both endpoints of the substring must be within the format
23932 range for just the substring to be printed.
23934 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
23936 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
23937 * config/i386/i386.c (ix86_target_string): Add missing options
23938 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
23939 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
23940 flags_other and ix86_target_other to flags2_other. Display unknown
23942 (ix86_valid_target_attribute_inner_p): Add missing options and
23943 reorder options by implied ISAs, as in ix86_target_string.
23945 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23947 * hash-table.h (hash_table::too_empty_p): New function.
23948 (hash_table::expand): Use it.
23949 (hash_table::traverse): Likewise.
23950 (hash_table::empty_slot): Use sizeof (value_type) instead of
23951 sizeof (PTR) to convert bytes to elements. Shrink the table
23952 if the current size is excessive for the current number of
23955 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23957 * ira-costs.c (record_reg_classes): Break from the inner loop
23958 early once alt_fail is known to be true. Update outer loop
23959 handling accordingly.
23961 2017-01-13 Jeff Law <law@redhat.com>
23963 * tree-ssa-dse.c (decrement_count): New function.
23964 (increment_start_addr, maybe_trim_memstar_call): Likewise.
23965 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
23966 when we know the partially dead statement is a mem* function.
23968 PR tree-optimization/61912
23969 PR tree-optimization/77485
23970 * tree-ssa-dse.c: Include expr.h.
23971 (maybe_trim_constructor_store): New function.
23972 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
23974 PR tree-optimization/33562
23975 PR tree-optimization/61912
23976 PR tree-optimization/77485
23977 * doc/invoke.texi: Document new dse-max-object-size param.
23978 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
23979 * tree-ssa-dse.c: Include params.h.
23980 (dse_store_status): New enum.
23981 (initialize_ao_ref_for_dse): New, partially extracted from
23983 (valid_ao_ref_for_dse, normalize_ref): New.
23984 (setup_live_bytes_from_ref, compute_trims): Likewise.
23985 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
23986 (maybe_trim_partially_dead_store): Likewise.
23987 (maybe_trim_complex_store): Likewise.
23988 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
23989 Track what bytes live from the original store. Return tri-state
23990 for dead, partially dead or live.
23991 (dse_dom_walker): Add constructor, destructor and new private members.
23992 (delete_dead_call, delete_dead_assignment): New extracted from
23994 (dse_optimize_stmt): Make a member of dse_dom_walker.
23995 Use initialize_ao_ref_for_dse.
23997 PR tree-optimization/33562
23998 PR tree-optimization/61912
23999 PR tree-optimization/77485
24000 * sbitmap.h (bitmap_count_bits): Prototype.
24001 (bitmap_clear_range, bitmap_set_range): Likewise.
24002 * sbitmap.c (bitmap_clear_range): New function.
24003 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24005 2017-01-13 Martin Liska <mliska@suse.cz>
24008 * function.c (set_cfun): Add new argument force.
24009 * function.h (set_cfun): Likewise.
24010 * ipa-inline-transform.c (inline_call): Use the function when
24011 strict alising from is dropped for function we inline to.
24013 2017-01-13 Richard Biener <rguenther@suse.de>
24015 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24016 for dumping GIMPLE INTEGER_CSTs.
24018 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24020 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24021 to 201112L since C++17.
24023 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24026 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24027 if -fsanitize=kernel-address is present.
24029 2017-01-13 Richard Biener <rguenther@suse.de>
24031 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24032 as _Literal ( type ) number in case usual suffixes do not
24033 preserve all information.
24035 2017-01-13 Richard Biener <rguenther@suse.de>
24037 PR tree-optimization/77283
24038 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24039 and ssa-iterators.h.
24040 (is_feasible_trace): Implement a cost model based on joiner
24043 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24046 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24047 char or short to __float128/_Float128 directly.
24049 2017-01-12 Martin Sebor <msebor@redhat.com>
24051 to -Wformat-overflow.
24052 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24053 (min_bytes_remaining): Same.
24054 (get_string_length): Same.
24055 (format_string): Same.
24056 (format_directive): Same.
24058 (pass_sprintf_length::handle_gimple_call): Same.
24060 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24062 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24063 info.nowrite calls with no lhs that can't throw. Return bool
24064 whether gsi_remove has been called or not.
24065 (pass_sprintf_length::handle_gimple_call): Return bool whether
24066 try_substitute_return_value called gsi_remove. Formatting fix.
24067 (pass_sprintf_length::execute): Don't use gsi_remove if
24068 handle_gimple_call returned true.
24071 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24072 be removed due to side-effects, don't remove following barrier nor
24073 turn the successor edge into fallthru edge.
24075 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24078 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24079 element-reversing loads and stores as not swappable.
24081 2017-01-12 Nathan Sidwell <nathan@acm.org>
24082 Nicolai Stange <nicstange@gmail.com>
24084 * combine.c (try_combine): Don't ignore result of overlap checking
24085 loop. Combine overlap & asm check into single loop.
24087 2017-01-12 Richard Biener <rguenther@suse.de>
24089 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24090 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24092 2017-01-12 Richard Biener <rguenther@suse.de>
24094 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24095 and TS_TARGET_OPTION directly derive from TS_BASE.
24096 * tree-core.h (tree_optimization_option): Derive from tree_base.
24097 (tree_target_option): Likewise.
24099 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24101 * config/i386/i386.c (memory_address_length): Increase len
24102 only when rip_relative_addr_p returns false.
24104 2017-01-11 Julia Koval <julia.koval@intel.com>
24106 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24107 (OPTION_MASK_ISA_SGX_SET): New.
24108 (ix86_handle_option): Handle OPT_msgx.
24109 * config.gcc: Added sgxintrin.h.
24110 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24111 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24112 * config/i386/i386.c (ix86_target_string): Add -msgx.
24114 (ix86_option_override_internal): Handle new options.
24115 (ix86_valid_target_attribute_inner_p): Add sgx.
24116 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24117 * config/i386/i386.opt: Add msgx.
24118 * config/i386/sgxintrin.h: New file.
24119 * config/i386/x86intrin.h: Add sgxintrin.h.
24121 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24124 * fold-const.c (maybe_nonzero_address): Return 1 for function
24126 (tree_single_nonzero_warnv_p): Don't handle function local objects
24130 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24133 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24136 * opts.c: Include "spellcheck.h"
24137 (struct string_fragment): New struct.
24138 (struct edit_distance_traits<const string_fragment &>): New
24140 (get_closest_sanitizer_option): New function.
24141 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24143 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24145 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24147 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24148 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24149 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24150 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24151 for initial die_offset if dwarf_split_debug_info.
24152 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24153 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24155 (output_skeleton_debug_sections): Formatting fix. Use
24156 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24157 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24159 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24161 * config/arm/cortex-a53.md: Add bypasses for
24162 cortex_a53_r2f_cvt.
24163 (cortex_a53_r2f): Only use for transfers.
24164 (cortex_a53_f2r): Likewise.
24165 (cortex_a53_r2f_cvt): Add reservation for conversions.
24166 (cortex_a53_f2r_cvt): Likewise.
24168 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24170 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24171 to all inlined functions, change static to extern.
24173 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24176 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24178 (arm_assemble_integer): Likewise.
24180 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24182 * config.gcc: Use new awk script to check CPU, FPU and architecture
24183 parameters for --with-... options.
24184 * config/arm/parsecpu.awk: New file
24185 * config/arm/arm-cpus.in: New file.
24186 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24188 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24190 * config/arm/t-arm: Update dependency rules.
24191 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24192 of processing .def files.
24193 * config/arm/genopt.sh: Deleted.
24194 * config/arm/gentune.sh: Deleted.
24195 * config/arm/arm-cores.def: Deleted.
24196 * config/arm/arm-arches.def: Deleted.
24197 * config/arm/arm-fpus.def: Deleted.
24198 * config/arm/arm-tune.md: Regenerated.
24199 * config/arm/arm-tables.opt: Regenerated.
24200 * config/arm/arm-cpu.h: New generated file.
24201 * config/arm/arm-cpu-data.h: New generated file.
24202 * config/arm/arm-cpu-cdata.h: New generated file.
24204 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24207 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24209 (input_varpool_node): Unpack dynamically_initialized bit.
24211 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24213 PR rtl-optimization/79032
24214 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24215 the alignment of the adjusted memory reference against that of MODE,
24216 instead of the alignment of the original memory reference.
24218 2017-01-11 Martin Jambor <mjambor@suse.cz>
24220 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24222 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24223 decorated functions.
24225 2017-01-11 Richard Biener <rguenther@suse.de>
24227 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24228 set range/nonnull info for PHI results. Do not set it on
24229 stmts marked for removal.
24231 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24233 * expr.c (store_field): In the bitfield case, fetch the return value
24234 from the registers before applying a single big-endian adjustment.
24235 Always do a final load for a BLKmode value not larger than a word.
24237 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24240 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24241 that we correctly handle column numbers greater than
24242 LINE_MAP_MAX_COLUMN_NUMBER.
24244 2017-01-10 Martin Sebor <msebor@redhat.com>
24246 PR middle-end/78245
24247 * gimple-ssa-sprintf.c (get_destination_size): Call
24248 {init,fini}object_sizes.
24249 * tree-object-size.c (addr_object_size): Adjust.
24250 (pass_through_call): Adjust.
24251 (pass_object_sizes::execute): Adjust.
24252 * tree-object-size.h (fini_object_sizes): Declare.
24254 2017-01-10 Martin Sebor <msebor@redhat.com>
24256 PR tree-optimization/78775
24257 * builtins.c (get_size_range): Move...
24258 * calls.c: ...to here.
24259 (alloc_max_size): Accept zero argument.
24260 (operand_signed_p): Remove.
24261 (maybe_warn_alloc_args_overflow): Call get_size_range.
24262 * calls.h (get_size_range): Declare.
24264 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24266 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24267 from TI's devices.csv file as of September 2016.
24268 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24270 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24272 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24273 * doc/invoke.texi: Likewise.
24274 * doc/md.texi: Likewise.
24275 * doc/objc.texi: Likewise.
24277 2017-01-10 Joshua Conner <joshconner@google.com>
24279 * config/arm/fuchsia-elf.h: New file.
24280 * config/fuchsia.h: New file.
24281 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24282 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24284 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24286 2016-01-10 Richard Biener <rguenther@suse.de>
24288 PR tree-optimization/79034
24289 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24290 Propagate out degenerate PHIs in the joiner.
24292 2017-01-10 Martin Liska <mliska@suse.cz>
24294 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24295 (sort_congruence_classes_by_decl_uid): Likewise.
24296 (sort_congruence_class_groups_by_decl_uid): Likewise.
24297 (sem_item_optimizer::merge_classes): Sort class, groups in these
24298 classes and members in the groups by DECL_UID of declarations.
24299 This would make merge operations stable.
24301 2017-01-10 Martin Liska <mliska@suse.cz>
24303 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24304 usage of m_classes_vec.
24305 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24306 (sem_item_optimizer::get_group_by_hash): Likewise.
24307 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24308 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24309 (sem_item_optimizer::verify_classes): Likewise.
24310 (sem_item_optimizer::process_cong_reduction): Likewise.
24311 (sem_item_optimizer::dump_cong_classes): Likewise.
24312 (sem_item_optimizer::merge_classes): Likewise.
24313 * ipa-icf.h (congruence_class_hash): Rename from
24314 congruence_class_group_hash. Remove declaration of m_classes_vec.
24316 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24318 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24319 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24320 * config.gcc: Add avx512vpopcntdqintrin.h.
24321 * config/i386/avx512vpopcntdqintrin.h: New.
24322 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24323 * config/i386/i386-builtin-types.def: Add new types.
24324 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24325 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24326 __builtin_ia32_vpopcountq_v8di_mask): New.
24327 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24328 __AVX512VPOPCNTDQ__.
24329 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24330 (PTA_AVX512VPOPCNTDQ): Define.
24331 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24332 TARGET_AVX512VPOPCNTDQ_P): Define.
24333 * config/i386/i386.opt: Add mavx512vpopcntdq.
24334 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24335 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24337 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24339 PR middle-end/77484
24340 * predict.def (PRED_CALL): Set to 67.
24342 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24344 * expr.c (store_field): In the bitfield case, if the value comes from
24345 a function call and is of an aggregate type returned in registers, do
24346 not modify the field mode; extract the value in all cases if the mode
24347 is BLKmode and the size is not larger than a word.
24349 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24352 * config/i386/cpuid.h: Fix undefined behavior.
24354 2017-01-04 Jeff Law <law@redhat.com>
24356 PR tree-optimization/79007
24357 PR tree-optimization/67955
24358 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24359 conservative for pt.null when flag_non_call_exceptions is on.
24361 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24363 PR translation/79019
24364 PR translation/79020
24365 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24366 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24367 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24369 * config/avr/avr.opt (maccumulate-args): Likewise.
24370 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24371 * common.opt (freport-bug): Likewise.
24372 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24373 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24374 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24375 translatable string.
24376 * config/i386/i386.c (function_value_32): Likewise.
24377 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24378 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24380 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24381 * common/config/msp430/msp430-common.c (msp430_handle_option):
24383 * symtab.c (symtab_node::verify_base): Likewise.
24384 * opts.c (set_debug_level): Likewise.
24385 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24386 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24387 missing whitespace to translatable strings.
24388 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24389 * config/sh/superh.h: Likewise.
24390 * config/i386/xopintrin.h: Likewise.
24391 * config/i386/znver1.md: Likewise.
24392 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24393 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24394 * double-int.h (struct double_int): Likewise.
24395 * double-int.c (div_and_round_double): Likewise.
24396 * wide-int.cc: Likewise.
24397 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24398 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24399 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24400 (crossjumps_occurred): ... this.
24401 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24404 PR tree-optimization/78899
24405 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24406 returning bool return struct loop *, NULL for failure and the new
24408 (versionable_outer_loop_p): Don't version outer loop if it has
24409 dont_vectorized bit set.
24410 (tree_if_conversion): When versioning outer loop, ensure
24411 tree_if_conversion is performed also on the inner loop of the
24412 non-vectorizable outer loop copy.
24413 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24414 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24415 prevent vectorization of it.
24416 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24417 the outer loop vectorization of the non-scalar version is attempted
24418 before vectorization of the inner loop in scalar version. If
24419 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24420 vectorization of its inner loop.
24421 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24422 has 2 inner loops, rename also on edges from bb whose single pred
24423 is outer_loop->header. Fix typo in function comment.
24425 2017-01-09 Martin Sebor <msebor@redhat.com>
24428 * asan.c (asan_emit_stack_protection): Increase local buffer size
24429 to avoid snprintf truncation warning.
24431 2017-01-09 Andrew Pinski <apinski@cavium.com>
24433 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24434 to reference thunderx2t99 for the tuning structure
24435 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24437 (thunderx2t99_extra_costs): This.
24438 * config/aarch64/aarch64-tune.md: Regenerate.
24439 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24440 (vulcan_addrcost_table): This.
24441 (vulcan_regmove_cost): Rename to ...
24442 (thunderx2t99_regmove_cost): This.
24443 (vulcan_vector_cost): Rename to ...
24444 (thunderx2t99_vector_cost): this.
24445 (vulcan_branch_cost): Rename to ...
24446 (thunderx2t99_branch_cost): This.
24447 (vulcan_tunings): Rename to ...
24448 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24449 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24451 2017-01-09 Martin Jambor <mjambor@suse.cz>
24455 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24456 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24457 (propagate_vr_accross_jump_function): Use the above function for all
24458 value range computations for pass-through jump functions and type
24459 converasion from explicit value range values.
24460 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24461 parameters from TYPE_ARG_TYPES.
24462 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24463 (ipa_write_node_info): Stream type of the actual argument.
24464 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24466 2017-01-09 Martin Liska <mliska@suse.cz>
24469 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24470 (lookup_compiler): Do not show error message with have_E.
24472 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24474 PR tree-optimization/78938
24475 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24476 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24477 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24478 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24481 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24483 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24486 2017-01-09 Richard Biener <rguenther@suse.de>
24488 PR tree-optimization/78997
24489 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24490 name condition properly.
24492 2017-01-09 Richard Biener <rguenther@suse.de>
24495 * dwarf2out.c (is_cxx): New overload with context.
24496 (is_naming_typedef_decl): Use it.
24498 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24500 * invoke.texi (Option Summary): Correct spacing in option lists
24501 and add line breaks to fix over-long lines.
24503 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24505 PR middle-end/17660
24507 * extend.texi (Common Variable Attributes): Add xref to GCC
24508 Internals manual to explain mode attribute keywords.
24510 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24513 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24514 and Preprocessor Options.
24515 (Options for Linking): Document -pthread here....
24516 (RS/6000 and PowerPC Options): ...not here.
24517 (Solaris 2 Options): ...or here.
24518 * doc/cppopts.texi: Document -pthread.
24520 2017-01-08 Martin Sebor <msebor@redhat.com>
24522 PR middle-end/77708
24523 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24524 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24525 New member functions.
24526 (format_directive): Used them.
24528 (pass_sprintf_length::handle_gimple_call): Same.
24529 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24530 to avoid truncation for any argument.
24531 (extract_affine_mul): Same.
24532 * tree.c (get_file_function_name): Same.
24534 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24536 PR middle-end/77484
24537 * predict.def (PRED_INDIR_CALL): Set to 86.
24539 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24541 PR preprocessor/54124
24542 * doc/cppopts.texi: Reformat -d subtable to list the full name
24543 of the options. Add cross-reference to the docs for the general
24544 compiler -d options.
24545 * doc/invoke.texi (Developer Options): Add cross-reference to the
24546 preprocessor-specific -d option documentation.
24548 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24550 PR preprocessor/13498
24551 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24552 redudant material, and reflect new command-line options.
24553 (System Headers): Likewise.
24555 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24557 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24558 -isystem, and -idirafter. Copy-edit.
24559 * doc/cppopts.texi: Copy-edit. Remove contradiction about
24560 default for -ftrack-macro-expansion. Delete obsolete and
24561 badly-formatted implementation details about -fdebug-cpp output.
24562 * doc/cppwarnopts.texi: Copy-edit.
24564 2017-01-07 David Malcolm <dmalcolm@redhat.com>
24567 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24568 that the transition from a max line width >= 1<<10 to narrower
24569 lines works correctly.
24571 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
24573 * doc/options.texi (PerFunction): New.
24574 * opt-functions.awk (switch_flags): Map both Optimization and
24575 PerFunction to CL_OPTIMIZATION.
24576 * opth-gen.awk: Test for PerFunction flag along with
24578 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
24579 it only when the latter is present. Skip those that don't in
24580 the hash function generator.
24581 * common.opt (fvar-tracking): Mark as PerFunction instead of
24583 (fvar-tracking-assignments): Likewise.
24584 (fvar-tracking-assignments-toggle): Likewise.
24585 (fvar-tracking-uninit): Likewise.
24587 2017-01-07 Jakub Jelinek <jakub@redhat.com>
24589 PR translation/79018
24590 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
24593 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
24596 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
24597 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
24598 TARGET_LONG_JUMP_TABLE_OFFSETS.
24599 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
24600 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
24601 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
24602 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
24603 * config/m68k/m68k.md (tablejump expander): Likewise.
24604 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
24605 TARGET_LONG_JUMP_TABLE_OFFSETS.
24606 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
24607 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
24609 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
24610 David Holsgrove <david.holsgrove@xilinx.com>
24612 * common/config/microblaze/microblaze-common.c
24613 (TARGET_EXCEPT_UNWIND_INFO): Remove.
24614 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
24616 * config/microblaze/microblaze.c (microblaze_must_save_register)
24617 (microblaze_expand_epilogue, microblaze_return_addr): Handle
24619 (microblaze_eh_return): New function.
24620 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
24621 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
24622 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
24623 * config/microblaze/microblaze.md (eh_return): New pattern.
24625 2017-01-06 Jakub Jelinek <jakub@redhat.com>
24627 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
24628 GCC_DIAGNOSTIC_STRINGIFY): Define.
24630 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
24632 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24634 * config/arm/arm.md (<mcrr>): New.
24636 * config/arm/arm.c (arm_arch5te): New.
24637 (arm_option_override): Set arm_arch5te.
24638 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
24640 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
24641 (arm_mcrr_qualifiers): ... this. New.
24642 (MRRC_QUALIFIERS): Define to...
24643 (arm_mrrc_qualifiers): ... this. New.
24644 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
24646 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
24647 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
24648 (MRRCI, mrrc, MRRC): New.
24649 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
24650 VUNSPEC_MRRC2): New.
24652 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24654 * config/arm/arm.md (<mcr>): New.
24656 * config/arm/arm.c (arm_coproc_builtin_available): Add
24657 support for mcr, mrc, mcr2 and mrc2.
24658 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
24659 (arm_mcr_qualifiers): ... this. New.
24660 (MRC_QUALIFIERS): Define to ...
24661 (arm_mrc_qualifiers): ... this. New.
24662 (MCR_QUALIFIERS): Define to ...
24663 (arm_mcr_qualifiers): ... this. New.
24664 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
24666 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
24667 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
24668 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
24669 VUNSPEC_MRC2): New.
24671 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24673 * config/arm/arm.md (*ldc): New.
24677 * config/arm/arm.c (arm_coproc_builtin_available): Add
24678 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
24679 (arm_coproc_ldc_stc_legitimate_address): New.
24680 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
24681 'qualifier_const_pointer'.
24682 (LDC_QUALIFIERS): Define to...
24683 (arm_ldc_qualifiers): ... this. New.
24684 (STC_QUALIFIERS): Define to...
24685 (arm_stc_qualifiers): ... this. New.
24686 * config/arm/arm-protos.h
24687 (arm_coproc_ldc_stc_legitimate_address): New.
24688 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
24689 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
24690 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
24691 stc2, stcl, stc2l): New.
24692 * config/arm/constraints.md (Uz): New.
24693 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
24694 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
24695 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
24696 VUNSPEC_STC2L): New.
24698 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24700 * config/arm/arm.md (<cdp>): New.
24701 * config/arm/arm.c (neon_const_bounds): Rename this ...
24702 (arm_const_bounds): ... this.
24703 (arm_coproc_builtin_available): New.
24704 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
24705 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
24706 (CDP_QUALIFIERS): Define to...
24707 (arm_cdp_qualifiers): ... this. New.
24709 (arm_expand_builtin_args): Add case for 6 arguments.
24710 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
24711 (arm_const_bounds): ... this.
24712 (arm_coproc_builtin_available): New.
24713 * config/arm/arm_acle.h (__arm_cdp): New.
24715 * config/arm/arm_acle_builtins.def (cdp): New.
24717 * config/arm/iterators.md (CDPI,CDP,cdp): New.
24718 * config/arm/neon.md: Rename all 'neon_const_bounds' to
24719 'arm_const_bounds'.
24720 * config/arm/types.md (coproc): New.
24721 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
24722 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
24723 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
24724 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
24726 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24728 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24729 (UBINOP_QUALIFIERS): New.
24731 (acle_builtin_data): New. Change comment.
24732 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24733 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24734 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24735 arm_acle_builtins.def.
24736 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24737 (arm_init_acle_builtins): New.
24738 (CRC32_BUILTIN): Remove.
24739 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24740 crc32cb, crc32ch and crc32cw.
24741 (arm_init_crc32_builtins): Remove.
24742 (arm_init_builtins): Use arm_init_acle_builtins rather
24743 than arm_init_crc32_builtins.
24744 (arm_expand_acle_builtin): New.
24745 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24746 * config/arm/arm_acle_builtins.def: New.
24748 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24750 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24751 (arm_builtin_datum): ... this.
24752 (arm_init_neon_builtin): Rename to ...
24753 (arm_init_builtin): ... this. Add a new parameters PREFIX
24754 and USE_SIG_IN_NAME.
24755 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24756 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24757 'arm_builtin_datum'.
24758 (arm_init_vfp_builtins): Likewise.
24759 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24760 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24761 (arm_expand_neon_args): Rename to ...
24762 (arm_expand_builtin_args): ... this. Rename builtin_arg
24763 enum values and differentiate between ARG_BUILTIN_MEMORY
24764 and ARG_BUILTIN_NEON_MEMORY.
24765 (arm_expand_neon_builtin_1): Rename to ...
24766 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
24767 values, arm_expand_builtin_args and add bool parameter NEON.
24768 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
24769 (arm_expand_vfp_builtin): Likewise.
24770 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
24772 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24774 PR middle-end/77484
24775 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
24776 * predict.c (tree_estimate_probability_bb): Reverse direction of
24777 polymorphic call predictor.
24779 2017-01-06 David Malcolm <dmalcolm@redhat.com>
24781 * passes.c (execute_one_pass): Split out pass-skipping logic into...
24782 (determine_pass_name_match): ...this new function and...
24783 (should_skip_pass_p): ...this new function.
24785 2017-01-06 Nathan Sidwell <nathan@acm.org>
24787 * ipa-visibility.c (function_and_variable_visibility): Reformat
24788 comments and long lines. Remove extrneous if.
24789 * symtab.c (symtab_node::make_decl_local): Fix code format.
24790 (symtab_node::set_section_for_node): Fix comment typo.
24792 2017-01-06 Martin Liska <mliska@suse.cz>
24795 * lra-constraints.c: Rename invariant to lra_invariant.
24796 * predict.c (set_even_probabilities): Initialize e to NULL.
24798 2017-01-05 Martin Sebor <msebor@redhat.com>
24800 PR tree-optimization/78910
24801 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
24802 (format_integer): Correct off-by-one error in the handling
24803 of precision with negative numbers in signed conversions..
24805 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
24807 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
24809 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24811 PR tree-optimization/71016
24812 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
24813 factor_out_conditional_conversion. Formatting fix.
24814 (factor_out_conditional_conversion): Add cond_stmt argument.
24815 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
24816 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
24819 2017-01-05 David Malcolm <dmalcolm@redhat.com>
24821 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
24822 read-rtl-function.o, and selftest-rtl.o.
24823 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
24824 (selftest::aarch64_test_loading_full_dump): New function.
24825 (selftest::aarch64_run_selftests): New function.
24826 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
24827 selftest::aarch64_run_selftests.
24828 * config/i386/i386.c
24829 (selftest::ix86_test_loading_dump_fragment_1): New function.
24830 (selftest::ix86_test_loading_call_insn): New function.
24831 (selftest::ix86_test_loading_full_dump): New function.
24832 (selftest::ix86_test_loading_unspec): New function.
24833 (selftest::ix86_run_selftests): Call the new functions.
24834 * emit-rtl.c (maybe_set_max_label_num): New function.
24835 * emit-rtl.h (maybe_set_max_label_num): New decl.
24836 * function.c (instantiate_decls): Guard call to
24837 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
24838 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
24840 * gensupport.c (gen_reader::gen_reader): Pass "false"
24841 for new "compact" param of rtx_reader.
24842 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
24843 rather than an empty string for NULL strings.
24844 * read-md.c: Potentially include config.h rather than bconfig.h.
24845 Wrap include of errors.h with #ifdef GENERATOR_FILE.
24846 (have_error): New global, copied from errors.c.
24847 (md_reader::read_name): Rename to...
24848 (md_reader::read_name_1): ...this, adding "out_loc" param,
24849 and converting "missing name or number" to returning false, rather
24851 (md_reader::read_name): Reimplement in terms of read_name_1.
24852 (md_reader::read_name_or_nil): New function.
24853 (md_reader::read_string): Handle "(nil)" by returning NULL.
24854 (md_reader::md_reader): Add new param "compact".
24855 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
24856 (md_reader::read_file): New method.
24857 * read-md.h (md_reader::md_reader): Add new param "compact".
24858 (md_reader::read_file): New method.
24859 (md_reader::is_compact): New accessor.
24860 (md_reader::read_name): Convert return type from void to file_location.
24861 (md_reader::read_name_or_nil): New decl.
24862 (md_reader::read_name_1): New decl.
24863 (md_reader::m_compact): New field.
24864 (noop_reader::noop_reader): Pass "false" for new "compact" param
24866 (rtx_reader::rtx_reader): Add new "compact" param.
24867 (rtx_reader::read_rtx_operand): Make virtual and convert return
24868 type from void to rtx.
24869 (rtx_reader::read_until): New decl.
24870 (rtx_reader::handle_any_trailing_information): New virtual function.
24871 (rtx_reader::postprocess): New virtual function.
24872 (rtx_reader::finalize_string): New virtual function.
24873 (rtx_reader::m_in_call_function_usage): New field.
24874 (rtx_reader::m_reuse_rtx_by_id): New field.
24875 * read-rtl-function.c: New file.
24876 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
24877 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
24878 (selftest::verify_three_block_rtl_cfg): New decl.
24879 * read-rtl-function.h: New file.
24880 * read-rtl.c: Potentially include config.h rather than bconfig.h.
24881 For host, include function.h, memmodel.h, and emit-rtl.h.
24882 (one_time_initialization): New function.
24883 (struct compact_insn_name): New struct.
24884 (compact_insn_names): New array.
24885 (find_code): Handle insn codes in compact dumps.
24886 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
24887 (bind_subst_iter_and_attr): Likewise.
24888 (add_condition_to_string): Likewise.
24889 (add_condition_to_rtx): Likewise.
24890 (apply_attribute_uses): Likewise.
24891 (add_current_iterators): Likewise.
24892 (apply_iterators): Likewise.
24893 (initialize_iterators): Guard usage of apply_subst_iterator with
24894 #ifdef GENERATOR_FILE.
24895 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
24896 (md_reader::read_mapping): Likewise.
24897 (add_define_attr_for_define_subst): Likewise.
24898 (add_define_subst_attr): Likewise.
24899 (read_subst_mapping): Likewise.
24900 (check_code_iterator): Likewise.
24901 (rtx_reader::read_rtx): Likewise. Move one-time initialization
24903 (one_time_initialization): New function.
24904 (rtx_reader::read_until): New method.
24905 (read_flags): New function.
24906 (parse_reg_note_name): New function.
24907 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
24908 Handle reuse_rtx ids.
24909 Wrap iterator lookup within #ifdef GENERATOR_FILE.
24910 Add parsing support for RTL dumps, mirroring the special-cases in
24911 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
24912 values, and calling handle_any_trailing_information.
24913 (rtx_reader::read_rtx_operand): Convert return type from void
24914 to rtx, returning return_rtx. Handle case 'e'. Call
24915 finalize_string on XSTR and XTMPL fields.
24916 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
24917 "(nil)" values were omitted. Call the postprocess vfunc on the
24919 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
24920 class ctor. Initialize m_in_call_function_usage. Call
24921 one_time_initialization.
24922 * rtl-tests.c (selftest::test_uncond_jump): Call
24923 set_new_first_and_last_insn.
24924 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
24925 * selftest-rtl.c: New file.
24926 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
24927 (selftest::get_insn_by_uid): New decl.
24928 * selftest-run-tests.c (selftest::run_tests): Call
24929 read_rtl_function_c_tests.
24930 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
24931 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
24934 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
24936 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
24937 operands in a special way. Assert that pos+len <= mode precision.
24939 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24941 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
24942 3 argument Alias with unlimited for the negative form.
24943 (fno-vect-cost-model): Removed.
24945 2017-01-05 Martin Liska <mliska@suse.cz>
24947 * hsa-gen.c (gen_hsa_divmod): New function.
24948 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
24950 2017-01-05 Martin Liska <mliska@suse.cz>
24953 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
24956 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24958 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
24959 small constant length operands.
24961 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24963 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
24964 between loop iterations.
24966 2017-01-05 Martin Liska <mliska@suse.cz>
24969 * gimplify.c (gimplify_decl_expr): Compare to
24970 asan_poisoned_variables instread of checking flags.
24971 (gimplify_target_expr): Likewise.
24972 (gimplify_expr): Likewise.
24973 (gimplify_function_tree): Conditionally initialize
24974 asan_poisoned_variables.
24976 2017-01-04 Jeff Law <law@redhat.com>
24978 PR tree-optimizatin/78812
24979 * rtl.h (contains_mem_rtx_p): Prototype.
24980 * ifcvt.c (containts_mem_rtx_p): Move from here to...
24981 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
24982 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
24983 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
24984 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
24986 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24988 * input.c (assert_char_at_range): Default-initialize actual_range.
24990 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24992 * df-scan.c (df_ref_create_structure): Make regno unsigned,
24993 to match the caller.
24995 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24997 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
24998 insns after final jump in test to emit dummy move.
25000 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25002 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25003 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25005 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25007 * multiple_target.c (create_dispatcher_calls): Init e_next.
25008 * tree-ssa-loop-split.c (split_loop): Init border.
25009 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25012 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25017 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25018 (altivec_register_operand): Do not return true if the operand
25019 contains a SUBREG mixing SImode and SFmode.
25020 (vsx_register_operand): Likewise.
25021 (vsx_reg_sfsubreg_ok): New predicate.
25022 (vfloat_operand): Do not return true if the operand contains a
25023 SUBREG mixing SImode and SFmode.
25024 (vint_operand): Likewise.
25025 (vlogical_operand): Likewise.
25026 (gpc_reg_operand): Likewise.
25027 (int_reg_operand): Likewise.
25028 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25029 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25030 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25032 (rs6000_emit_move_si_sf_subreg): New helper function.
25033 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25034 fixup SUBREGs involving SImode and SFmode.
25035 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25036 numbers for the new peephole2 optimization.
25037 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25038 the GLIBC math library that do AND/IOR/XOR operations on single
25039 precision floating point.
25040 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25041 target macros to say whether we need to avoid SUBREGs mixing
25043 (TARGET_ALLOW_SF_SUBREG): Likewise.
25044 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25045 (UNSPEC_SI_FROM_SF): Likewise.
25046 (iorxor): Change spacing.
25047 (and_ior_xor): New iterator for AND, IOR, and XOR.
25048 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25049 (movdi_from_sf_zero_ext): Likewise.
25050 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25051 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25052 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25053 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25054 (fms<mode>4): Likewise.
25055 (fnma<mode>4): Likewise.
25056 (fnms<mode>4): Likewise.
25057 (nfma<mode>4): Likewise.
25058 (nfms<mode>4): Likewise.
25060 2017-01-04 Marek Polacek <polacek@redhat.com>
25063 * doc/invoke.texi: Document -Wpointer-compare.
25065 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25067 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25070 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25071 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25074 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25077 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25078 documentation of the powerpc_popcntb_ok attribute.
25079 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25080 code to issue warning messages if a requested CPU configuration is
25081 not supported by the binary (assembler and loader) toolchain.
25082 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25083 made to define a built-in function that has been disabled.
25084 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25085 made to define a built-in function that has been disabled.
25086 (altivec_init_builtins): Add comment explaining why definition
25087 of the DST built-in functions is not preceded by an assertion
25088 check. Add assertions to prevent ICE if attempts are made to
25089 define an altivec predicate or an abs* built-in function that has
25091 (htm_init_builtins): Add comment explaining why definition of the
25092 htm built-in functions is not preceded by an assertion check.
25094 2017-01-04 Jeff Law <law@redhat.com>
25096 PR tree-optimizatin/67955
25097 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25098 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25099 the points-to solution does not include pt_null. Use DECL_PT_UID
25102 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25104 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25105 Use gen_int_mode instead of gen_lopwart for const_int operands.
25107 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25109 PR tree-optimization/71563
25110 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25111 out of range value - has low bits known to be zero.
25113 2017-01-04 Alan Modra <amodra@gmail.com>
25115 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25116 * configure: Regenerate.
25117 * config.in: Regenerate.
25119 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25122 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25123 a substring of the message, but strcmp with the whole message. Ifdef
25124 ENABLE_NLS, translate the message first using dgettext.
25126 2017-01-03 Jeff Law <law@redhat.com>
25128 PR tree-optimizatin/78856
25129 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25130 (mark_threaded_blocks): Remove code to truncate thread paths that
25131 cross multiple loop headers. Instead invalidate the cached loop
25132 iteration information and handle case of a thread path walking
25133 into an irreducible region.
25135 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25138 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25139 assertions. Add support for doing the signbit if the IEEE 128-bit
25140 floating point value is in a GPR.
25141 * config/rs6000/rs6000.md (Fsignbit): Delete.
25142 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25143 Update the length attribute if the value is in a GPR.
25144 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25145 the sign or zero extension instruction, since the value is always 0/1.
25146 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25149 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25150 extracting SImode to a GPR register so that we can generate a
25151 store, limit the vector to be in a traditional Altivec register
25152 for the vextuwrx instruction.
25154 2017-01-03 Ian Lance Taylor <iant@google.com>
25156 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25158 2017-01-03 Martin Sebor <msebor@redhat.com>
25160 PR tree-optimization/78696
25161 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25162 precision. Use MPFR for %f for greater fidelity. Correct handling
25164 (pass_sprintf_length::compute_format_length): Set width and precision
25165 specified by asrerisk to void_node for vararg functions.
25166 (try_substitute_return_value): Adjust dump output.
25168 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25170 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25172 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25174 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25175 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25176 -mlra/-mno-lra was passed to the compiler.
25178 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25180 PR rtl-optimization/65618
25181 * emit-rtl.c (try_split): Move initialization of "before" and
25182 "after" to just before the call to emit_insn_after_setloc.
25184 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25186 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25188 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25190 * dwarf2out.c (gen_enumeration_type_die): When
25191 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25193 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25195 PR tree-optimization/78965
25196 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25197 Change first argument from const call_info & to call_info &. For %n
25198 set info.nowrite to false.
25200 PR middle-end/78901
25201 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25202 possibly throwing calls.
25204 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25205 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25206 and fns handling, rather than in a separate case SSA_NAME.
25208 2017-01-02 Jeff Law <law@redhat.com>
25210 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25211 fixes for first_period and second_period variables.
25213 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25216 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25217 (*insvqi_1): New insn pattern.
25218 (*insvqi_1_mem_rex64): Ditto.
25219 (*insvqi_2): Ditto.
25220 (*insvqi_3): Rename from *insvqi.
25222 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25224 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25226 * doc/cfg.texi (Edges): Remove reference to Java.
25227 (Maintaining the CFG): Ditto.
25229 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25231 PR middle-end/77674
25232 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25233 transparent aliases.
25235 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25237 PR middle-end/77484
25238 * predict.def (PRED_CALL): Update hitrate.
25239 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25240 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25241 into direct/indirect/polymorphic variants.
25243 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25245 Update copyright years.
25247 * gcc.c (process_command): Update copyright notice dates.
25248 * gcov-dump.c (print_version): Ditto.
25249 * gcov.c (print_version): Ditto.
25250 * gcov-tool.c (print_version): Ditto.
25251 * gengtype.c (create_file): Ditto.
25252 * doc/cpp.texi: Bump @copying's copyright year.
25253 * doc/cppinternals.texi: Ditto.
25254 * doc/gcc.texi: Ditto.
25255 * doc/gccint.texi: Ditto.
25256 * doc/gcov.texi: Ditto.
25257 * doc/install.texi: Ditto.
25258 * doc/invoke.texi: Ditto.
25260 Copyright (C) 2017 Free Software Foundation, Inc.
25262 Copying and distribution of this file, with or without modification,
25263 are permitted in any medium without royalty provided the copyright
25264 notice and this notice are preserved.